diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,50936 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999784343325426,
+  "eval_steps": 25,
+  "global_step": 6954,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00043131334914815614,
+      "grad_norm": 0.4778429865837097,
+      "learning_rate": 0.0,
+      "loss": 2.4396,
+      "step": 1
+    },
+    {
+      "epoch": 0.0008626266982963123,
+      "grad_norm": 0.6504954695701599,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 2.5284,
+      "step": 2
+    },
+    {
+      "epoch": 0.0012939400474444684,
+      "grad_norm": 0.4514806568622589,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 2.3895,
+      "step": 3
+    },
+    {
+      "epoch": 0.0017252533965926246,
+      "grad_norm": 0.4726278483867645,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 2.5381,
+      "step": 4
+    },
+    {
+      "epoch": 0.0021565667457407807,
+      "grad_norm": 0.46150511503219604,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 2.401,
+      "step": 5
+    },
+    {
+      "epoch": 0.002587880094888937,
+      "grad_norm": 0.47438162565231323,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 2.5229,
+      "step": 6
+    },
+    {
+      "epoch": 0.003019193444037093,
+      "grad_norm": 0.5883695483207703,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 2.4581,
+      "step": 7
+    },
+    {
+      "epoch": 0.003450506793185249,
+      "grad_norm": 0.6027829647064209,
+      "learning_rate": 2.1e-05,
+      "loss": 2.4734,
+      "step": 8
+    },
+    {
+      "epoch": 0.0038818201423334053,
+      "grad_norm": 289971.125,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 2.4135,
+      "step": 9
+    },
+    {
+      "epoch": 0.004313133491481561,
+      "grad_norm": 0.46024876832962036,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 2.5216,
+      "step": 10
+    },
+    {
+      "epoch": 0.004744446840629717,
+      "grad_norm": 0.3895394206047058,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 2.4507,
+      "step": 11
+    },
+    {
+      "epoch": 0.005175760189777874,
+      "grad_norm": 0.47679391503334045,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 2.631,
+      "step": 12
+    },
+    {
+      "epoch": 0.005607073538926029,
+      "grad_norm": 0.37820500135421753,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 2.2789,
+      "step": 13
+    },
+    {
+      "epoch": 0.006038386888074186,
+      "grad_norm": 0.30506932735443115,
+      "learning_rate": 3.9e-05,
+      "loss": 2.3152,
+      "step": 14
+    },
+    {
+      "epoch": 0.006469700237222342,
+      "grad_norm": 0.23102450370788574,
+      "learning_rate": 4.2e-05,
+      "loss": 2.485,
+      "step": 15
+    },
+    {
+      "epoch": 0.006901013586370498,
+      "grad_norm": 19.553178787231445,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 2.4329,
+      "step": 16
+    },
+    {
+      "epoch": 0.007332326935518654,
+      "grad_norm": 0.17712759971618652,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 2.3715,
+      "step": 17
+    },
+    {
+      "epoch": 0.0077636402846668106,
+      "grad_norm": 0.14594106376171112,
+      "learning_rate": 5.1e-05,
+      "loss": 2.1465,
+      "step": 18
+    },
+    {
+      "epoch": 0.008194953633814967,
+      "grad_norm": 0.1296325922012329,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 2.3163,
+      "step": 19
+    },
+    {
+      "epoch": 0.008626266982963123,
+      "grad_norm": 0.13177795708179474,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.4841,
+      "step": 20
+    },
+    {
+      "epoch": 0.009057580332111279,
+      "grad_norm": 0.11691755056381226,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 2.2407,
+      "step": 21
+    },
+    {
+      "epoch": 0.009488893681259434,
+      "grad_norm": 0.10889497399330139,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 2.3919,
+      "step": 22
+    },
+    {
+      "epoch": 0.009920207030407592,
+      "grad_norm": 0.1357669234275818,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 2.6173,
+      "step": 23
+    },
+    {
+      "epoch": 0.010351520379555747,
+      "grad_norm": 216.71868896484375,
+      "learning_rate": 6.9e-05,
+      "loss": 2.3811,
+      "step": 24
+    },
+    {
+      "epoch": 0.010782833728703903,
+      "grad_norm": 0.12138977646827698,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 2.2775,
+      "step": 25
+    },
+    {
+      "epoch": 0.010782833728703903,
+      "eval_loss": 2.243220806121826,
+      "eval_runtime": 199.0847,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 25
+    },
+    {
+      "epoch": 0.011214147077852059,
+      "grad_norm": 0.13480618596076965,
+      "learning_rate": 7.5e-05,
+      "loss": 2.422,
+      "step": 26
+    },
+    {
+      "epoch": 0.011645460427000216,
+      "grad_norm": 0.11292761564254761,
+      "learning_rate": 7.8e-05,
+      "loss": 2.3631,
+      "step": 27
+    },
+    {
+      "epoch": 0.012076773776148372,
+      "grad_norm": 0.10692216455936432,
+      "learning_rate": 8.1e-05,
+      "loss": 2.4068,
+      "step": 28
+    },
+    {
+      "epoch": 0.012508087125296528,
+      "grad_norm": 0.10275225341320038,
+      "learning_rate": 8.4e-05,
+      "loss": 2.1603,
+      "step": 29
+    },
+    {
+      "epoch": 0.012939400474444683,
+      "grad_norm": 0.10391239821910858,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 2.2908,
+      "step": 30
+    },
+    {
+      "epoch": 0.01337071382359284,
+      "grad_norm": 0.10926996171474457,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 2.4315,
+      "step": 31
+    },
+    {
+      "epoch": 0.013802027172740997,
+      "grad_norm": 0.11844368278980255,
+      "learning_rate": 9.3e-05,
+      "loss": 2.4053,
+      "step": 32
+    },
+    {
+      "epoch": 0.014233340521889152,
+      "grad_norm": 0.11695211380720139,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 2.3072,
+      "step": 33
+    },
+    {
+      "epoch": 0.014664653871037308,
+      "grad_norm": 0.13541661202907562,
+      "learning_rate": 9.9e-05,
+      "loss": 2.106,
+      "step": 34
+    },
+    {
+      "epoch": 0.015095967220185465,
+      "grad_norm": 107.78008270263672,
+      "learning_rate": 0.000102,
+      "loss": 2.1436,
+      "step": 35
+    },
+    {
+      "epoch": 0.015527280569333621,
+      "grad_norm": 0.11143426597118378,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 2.1563,
+      "step": 36
+    },
+    {
+      "epoch": 0.015958593918481777,
+      "grad_norm": 0.12790513038635254,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 2.3599,
+      "step": 37
+    },
+    {
+      "epoch": 0.016389907267629934,
+      "grad_norm": 0.24229316413402557,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 2.2941,
+      "step": 38
+    },
+    {
+      "epoch": 0.016821220616778088,
+      "grad_norm": 0.12360761314630508,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 2.3284,
+      "step": 39
+    },
+    {
+      "epoch": 0.017252533965926246,
+      "grad_norm": 0.10814772546291351,
+      "learning_rate": 0.000117,
+      "loss": 2.3827,
+      "step": 40
+    },
+    {
+      "epoch": 0.017683847315074403,
+      "grad_norm": 0.12808896601200104,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 2.2024,
+      "step": 41
+    },
+    {
+      "epoch": 0.018115160664222557,
+      "grad_norm": 0.11168445646762848,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 2.3587,
+      "step": 42
+    },
+    {
+      "epoch": 0.018546474013370715,
+      "grad_norm": 0.1182243749499321,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 2.3013,
+      "step": 43
+    },
+    {
+      "epoch": 0.01897778736251887,
+      "grad_norm": 0.1579306572675705,
+      "learning_rate": 0.000129,
+      "loss": 2.4032,
+      "step": 44
+    },
+    {
+      "epoch": 0.019409100711667026,
+      "grad_norm": 0.10591604560613632,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 2.2099,
+      "step": 45
+    },
+    {
+      "epoch": 0.019840414060815183,
+      "grad_norm": 0.10893277078866959,
+      "learning_rate": 0.000135,
+      "loss": 2.2341,
+      "step": 46
+    },
+    {
+      "epoch": 0.020271727409963337,
+      "grad_norm": 0.10329752415418625,
+      "learning_rate": 0.000138,
+      "loss": 2.2163,
+      "step": 47
+    },
+    {
+      "epoch": 0.020703040759111495,
+      "grad_norm": 0.09824061393737793,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 2.3557,
+      "step": 48
+    },
+    {
+      "epoch": 0.021134354108259652,
+      "grad_norm": 0.12618935108184814,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 2.3195,
+      "step": 49
+    },
+    {
+      "epoch": 0.021565667457407806,
+      "grad_norm": 0.11831948906183243,
+      "learning_rate": 0.000147,
+      "loss": 2.624,
+      "step": 50
+    },
+    {
+      "epoch": 0.021565667457407806,
+      "eval_loss": 2.210568428039551,
+      "eval_runtime": 197.8736,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 50
+    },
+    {
+      "epoch": 0.021996980806555964,
+      "grad_norm": 0.33164137601852417,
+      "learning_rate": 0.00015,
+      "loss": 2.3776,
+      "step": 51
+    },
+    {
+      "epoch": 0.022428294155704118,
+      "grad_norm": 0.13195934891700745,
+      "learning_rate": 0.00014999999223521492,
+      "loss": 2.3575,
+      "step": 52
+    },
+    {
+      "epoch": 0.022859607504852275,
+      "grad_norm": 0.11495663225650787,
+      "learning_rate": 0.00014999996894086122,
+      "loss": 2.2736,
+      "step": 53
+    },
+    {
+      "epoch": 0.023290920854000433,
+      "grad_norm": 0.13961608707904816,
+      "learning_rate": 0.0001499999301169438,
+      "loss": 2.2999,
+      "step": 54
+    },
+    {
+      "epoch": 0.023722234203148587,
+      "grad_norm": 0.14567121863365173,
+      "learning_rate": 0.00014999987576347067,
+      "loss": 1.983,
+      "step": 55
+    },
+    {
+      "epoch": 0.024153547552296744,
+      "grad_norm": 0.14206400513648987,
+      "learning_rate": 0.00014999980588045305,
+      "loss": 2.5026,
+      "step": 56
+    },
+    {
+      "epoch": 0.0245848609014449,
+      "grad_norm": 0.17789866030216217,
+      "learning_rate": 0.00014999972046790544,
+      "loss": 2.2247,
+      "step": 57
+    },
+    {
+      "epoch": 0.025016174250593055,
+      "grad_norm": 0.13413818180561066,
+      "learning_rate": 0.00014999961952584555,
+      "loss": 2.3253,
+      "step": 58
+    },
+    {
+      "epoch": 0.025447487599741213,
+      "grad_norm": 0.13895240426063538,
+      "learning_rate": 0.00014999950305429424,
+      "loss": 2.2533,
+      "step": 59
+    },
+    {
+      "epoch": 0.025878800948889367,
+      "grad_norm": 0.11061283200979233,
+      "learning_rate": 0.00014999937105327564,
+      "loss": 2.1234,
+      "step": 60
+    },
+    {
+      "epoch": 0.026310114298037524,
+      "grad_norm": 0.14688056707382202,
+      "learning_rate": 0.00014999922352281712,
+      "loss": 2.4367,
+      "step": 61
+    },
+    {
+      "epoch": 0.02674142764718568,
+      "grad_norm": 0.16806663572788239,
+      "learning_rate": 0.00014999906046294915,
+      "loss": 2.0712,
+      "step": 62
+    },
+    {
+      "epoch": 0.027172740996333836,
+      "grad_norm": 0.14308182895183563,
+      "learning_rate": 0.00014999888187370553,
+      "loss": 2.4362,
+      "step": 63
+    },
+    {
+      "epoch": 0.027604054345481993,
+      "grad_norm": 0.1595005989074707,
+      "learning_rate": 0.00014999868775512326,
+      "loss": 2.3226,
+      "step": 64
+    },
+    {
+      "epoch": 0.028035367694630147,
+      "grad_norm": 0.12989497184753418,
+      "learning_rate": 0.00014999847810724253,
+      "loss": 2.2128,
+      "step": 65
+    },
+    {
+      "epoch": 0.028466681043778305,
+      "grad_norm": 0.13269419968128204,
+      "learning_rate": 0.00014999825293010675,
+      "loss": 2.1651,
+      "step": 66
+    },
+    {
+      "epoch": 0.028897994392926462,
+      "grad_norm": 0.12201163172721863,
+      "learning_rate": 0.00014999801222376248,
+      "loss": 2.0443,
+      "step": 67
+    },
+    {
+      "epoch": 0.029329307742074616,
+      "grad_norm": 0.14177387952804565,
+      "learning_rate": 0.00014999775598825967,
+      "loss": 2.2544,
+      "step": 68
+    },
+    {
+      "epoch": 0.029760621091222773,
+      "grad_norm": 0.1548364758491516,
+      "learning_rate": 0.00014999748422365127,
+      "loss": 2.2396,
+      "step": 69
+    },
+    {
+      "epoch": 0.03019193444037093,
+      "grad_norm": 10.743372917175293,
+      "learning_rate": 0.00014999719692999365,
+      "loss": 2.3391,
+      "step": 70
+    },
+    {
+      "epoch": 0.030623247789519085,
+      "grad_norm": 0.13508769869804382,
+      "learning_rate": 0.00014999689410734622,
+      "loss": 2.3851,
+      "step": 71
+    },
+    {
+      "epoch": 0.031054561138667242,
+      "grad_norm": 1.441625952720642,
+      "learning_rate": 0.00014999657575577173,
+      "loss": 2.2796,
+      "step": 72
+    },
+    {
+      "epoch": 0.031485874487815396,
+      "grad_norm": 0.17903247475624084,
+      "learning_rate": 0.00014999624187533606,
+      "loss": 2.4809,
+      "step": 73
+    },
+    {
+      "epoch": 0.031917187836963554,
+      "grad_norm": 0.23883047699928284,
+      "learning_rate": 0.0001499958924661084,
+      "loss": 2.3531,
+      "step": 74
+    },
+    {
+      "epoch": 0.03234850118611171,
+      "grad_norm": 0.14161138236522675,
+      "learning_rate": 0.00014999552752816104,
+      "loss": 2.2301,
+      "step": 75
+    },
+    {
+      "epoch": 0.03234850118611171,
+      "eval_loss": 2.1953680515289307,
+      "eval_runtime": 197.8389,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 75
+    },
+    {
+      "epoch": 0.03277981453525987,
+      "grad_norm": 0.18088607490062714,
+      "learning_rate": 0.00014999514706156956,
+      "loss": 2.4097,
+      "step": 76
+    },
+    {
+      "epoch": 0.03321112788440802,
+      "grad_norm": 0.16779634356498718,
+      "learning_rate": 0.00014999475106641276,
+      "loss": 2.4324,
+      "step": 77
+    },
+    {
+      "epoch": 0.033642441233556176,
+      "grad_norm": 0.16245946288108826,
+      "learning_rate": 0.00014999433954277264,
+      "loss": 2.3828,
+      "step": 78
+    },
+    {
+      "epoch": 0.034073754582704334,
+      "grad_norm": 0.16727153956890106,
+      "learning_rate": 0.00014999391249073435,
+      "loss": 2.3144,
+      "step": 79
+    },
+    {
+      "epoch": 0.03450506793185249,
+      "grad_norm": 0.15430496633052826,
+      "learning_rate": 0.00014999346991038639,
+      "loss": 2.2893,
+      "step": 80
+    },
+    {
+      "epoch": 0.03493638128100065,
+      "grad_norm": 0.13965576887130737,
+      "learning_rate": 0.00014999301180182036,
+      "loss": 2.3668,
+      "step": 81
+    },
+    {
+      "epoch": 0.035367694630148806,
+      "grad_norm": 0.14725452661514282,
+      "learning_rate": 0.00014999253816513113,
+      "loss": 2.2724,
+      "step": 82
+    },
+    {
+      "epoch": 0.03579900797929696,
+      "grad_norm": 0.1368856281042099,
+      "learning_rate": 0.00014999204900041678,
+      "loss": 2.2663,
+      "step": 83
+    },
+    {
+      "epoch": 0.036230321328445114,
+      "grad_norm": 0.134723961353302,
+      "learning_rate": 0.00014999154430777857,
+      "loss": 2.4259,
+      "step": 84
+    },
+    {
+      "epoch": 0.03666163467759327,
+      "grad_norm": 0.1524089127779007,
+      "learning_rate": 0.00014999102408732104,
+      "loss": 2.4348,
+      "step": 85
+    },
+    {
+      "epoch": 0.03709294802674143,
+      "grad_norm": 0.1420392543077469,
+      "learning_rate": 0.00014999048833915185,
+      "loss": 2.3566,
+      "step": 86
+    },
+    {
+      "epoch": 0.03752426137588959,
+      "grad_norm": 0.13361221551895142,
+      "learning_rate": 0.000149989937063382,
+      "loss": 2.3743,
+      "step": 87
+    },
+    {
+      "epoch": 0.03795557472503774,
+      "grad_norm": 0.14212311804294586,
+      "learning_rate": 0.00014998937026012557,
+      "loss": 2.2588,
+      "step": 88
+    },
+    {
+      "epoch": 0.038386888074185894,
+      "grad_norm": 0.12589286267757416,
+      "learning_rate": 0.00014998878792949998,
+      "loss": 2.2884,
+      "step": 89
+    },
+    {
+      "epoch": 0.03881820142333405,
+      "grad_norm": 1.3771302700042725,
+      "learning_rate": 0.00014998819007162577,
+      "loss": 2.2237,
+      "step": 90
+    },
+    {
+      "epoch": 0.03924951477248221,
+      "grad_norm": 0.14674894511699677,
+      "learning_rate": 0.00014998757668662678,
+      "loss": 2.4529,
+      "step": 91
+    },
+    {
+      "epoch": 0.03968082812163037,
+      "grad_norm": 0.15564079582691193,
+      "learning_rate": 0.00014998694777462997,
+      "loss": 2.2877,
+      "step": 92
+    },
+    {
+      "epoch": 0.04011214147077852,
+      "grad_norm": 0.15917402505874634,
+      "learning_rate": 0.0001499863033357656,
+      "loss": 2.5187,
+      "step": 93
+    },
+    {
+      "epoch": 0.040543454819926675,
+      "grad_norm": 0.15689204633235931,
+      "learning_rate": 0.00014998564337016706,
+      "loss": 2.3903,
+      "step": 94
+    },
+    {
+      "epoch": 0.04097476816907483,
+      "grad_norm": 0.1784036010503769,
+      "learning_rate": 0.00014998496787797103,
+      "loss": 2.3838,
+      "step": 95
+    },
+    {
+      "epoch": 0.04140608151822299,
+      "grad_norm": 0.15815366804599762,
+      "learning_rate": 0.00014998427685931738,
+      "loss": 2.3389,
+      "step": 96
+    },
+    {
+      "epoch": 0.04183739486737115,
+      "grad_norm": 0.14294639229774475,
+      "learning_rate": 0.00014998357031434918,
+      "loss": 2.1199,
+      "step": 97
+    },
+    {
+      "epoch": 0.042268708216519305,
+      "grad_norm": 0.13665415346622467,
+      "learning_rate": 0.00014998284824321276,
+      "loss": 2.0823,
+      "step": 98
+    },
+    {
+      "epoch": 0.042700021565667455,
+      "grad_norm": 0.1319860965013504,
+      "learning_rate": 0.0001499821106460576,
+      "loss": 2.2892,
+      "step": 99
+    },
+    {
+      "epoch": 0.04313133491481561,
+      "grad_norm": 0.14062654972076416,
+      "learning_rate": 0.00014998135752303647,
+      "loss": 2.1406,
+      "step": 100
+    },
+    {
+      "epoch": 0.04313133491481561,
+      "eval_loss": 2.180278778076172,
+      "eval_runtime": 194.9967,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 100
+    },
+    {
+      "epoch": 0.04356264826396377,
+      "grad_norm": 0.15747618675231934,
+      "learning_rate": 0.00014998058887430527,
+      "loss": 2.329,
+      "step": 101
+    },
+    {
+      "epoch": 0.04399396161311193,
+      "grad_norm": 0.14773201942443848,
+      "learning_rate": 0.00014997980470002318,
+      "loss": 2.3304,
+      "step": 102
+    },
+    {
+      "epoch": 0.044425274962260085,
+      "grad_norm": 0.2709938883781433,
+      "learning_rate": 0.00014997900500035254,
+      "loss": 2.268,
+      "step": 103
+    },
+    {
+      "epoch": 0.044856588311408235,
+      "grad_norm": 0.31209737062454224,
+      "learning_rate": 0.00014997818977545896,
+      "loss": 1.9698,
+      "step": 104
+    },
+    {
+      "epoch": 0.04528790166055639,
+      "grad_norm": 0.20316815376281738,
+      "learning_rate": 0.00014997735902551123,
+      "loss": 2.3328,
+      "step": 105
+    },
+    {
+      "epoch": 0.04571921500970455,
+      "grad_norm": 0.16161882877349854,
+      "learning_rate": 0.0001499765127506814,
+      "loss": 2.401,
+      "step": 106
+    },
+    {
+      "epoch": 0.04615052835885271,
+      "grad_norm": 0.15170305967330933,
+      "learning_rate": 0.00014997565095114466,
+      "loss": 2.4336,
+      "step": 107
+    },
+    {
+      "epoch": 0.046581841708000865,
+      "grad_norm": 0.1538029909133911,
+      "learning_rate": 0.00014997477362707948,
+      "loss": 2.0162,
+      "step": 108
+    },
+    {
+      "epoch": 0.047013155057149016,
+      "grad_norm": 0.1348794847726822,
+      "learning_rate": 0.0001499738807786675,
+      "loss": 2.3923,
+      "step": 109
+    },
+    {
+      "epoch": 0.04744446840629717,
+      "grad_norm": 0.13398659229278564,
+      "learning_rate": 0.00014997297240609361,
+      "loss": 2.2099,
+      "step": 110
+    },
+    {
+      "epoch": 0.04787578175544533,
+      "grad_norm": 0.13340270519256592,
+      "learning_rate": 0.0001499720485095459,
+      "loss": 2.4006,
+      "step": 111
+    },
+    {
+      "epoch": 0.04830709510459349,
+      "grad_norm": 0.15017318725585938,
+      "learning_rate": 0.00014997110908921565,
+      "loss": 2.3831,
+      "step": 112
+    },
+    {
+      "epoch": 0.048738408453741645,
+      "grad_norm": 0.14307159185409546,
+      "learning_rate": 0.0001499701541452974,
+      "loss": 2.3606,
+      "step": 113
+    },
+    {
+      "epoch": 0.0491697218028898,
+      "grad_norm": 0.16115598380565643,
+      "learning_rate": 0.00014996918367798888,
+      "loss": 2.2783,
+      "step": 114
+    },
+    {
+      "epoch": 0.04960103515203795,
+      "grad_norm": 0.1477428674697876,
+      "learning_rate": 0.00014996819768749102,
+      "loss": 2.3053,
+      "step": 115
+    },
+    {
+      "epoch": 0.05003234850118611,
+      "grad_norm": 0.13940371572971344,
+      "learning_rate": 0.00014996719617400798,
+      "loss": 2.4399,
+      "step": 116
+    },
+    {
+      "epoch": 0.05046366185033427,
+      "grad_norm": 10.187359809875488,
+      "learning_rate": 0.00014996617913774716,
+      "loss": 2.1557,
+      "step": 117
+    },
+    {
+      "epoch": 0.050894975199482426,
+      "grad_norm": 0.16093291342258453,
+      "learning_rate": 0.00014996514657891915,
+      "loss": 2.3654,
+      "step": 118
+    },
+    {
+      "epoch": 0.05132628854863058,
+      "grad_norm": 0.39099806547164917,
+      "learning_rate": 0.0001499640984977377,
+      "loss": 2.2144,
+      "step": 119
+    },
+    {
+      "epoch": 0.051757601897778734,
+      "grad_norm": 0.154411181807518,
+      "learning_rate": 0.0001499630348944199,
+      "loss": 2.2602,
+      "step": 120
+    },
+    {
+      "epoch": 0.05218891524692689,
+      "grad_norm": 0.15475720167160034,
+      "learning_rate": 0.00014996195576918593,
+      "loss": 2.492,
+      "step": 121
+    },
+    {
+      "epoch": 0.05262022859607505,
+      "grad_norm": 0.20788845419883728,
+      "learning_rate": 0.00014996086112225923,
+      "loss": 2.1501,
+      "step": 122
+    },
+    {
+      "epoch": 0.053051541945223206,
+      "grad_norm": 0.18076249957084656,
+      "learning_rate": 0.00014995975095386652,
+      "loss": 2.2561,
+      "step": 123
+    },
+    {
+      "epoch": 0.05348285529437136,
+      "grad_norm": 0.14736950397491455,
+      "learning_rate": 0.00014995862526423755,
+      "loss": 2.0482,
+      "step": 124
+    },
+    {
+      "epoch": 0.053914168643519514,
+      "grad_norm": 0.5525259375572205,
+      "learning_rate": 0.00014995748405360556,
+      "loss": 2.1995,
+      "step": 125
+    },
+    {
+      "epoch": 0.053914168643519514,
+      "eval_loss": 2.171874761581421,
+      "eval_runtime": 196.3857,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 125
+    },
+    {
+      "epoch": 0.05434548199266767,
+      "grad_norm": 0.14461691677570343,
+      "learning_rate": 0.00014995632732220675,
+      "loss": 2.2688,
+      "step": 126
+    },
+    {
+      "epoch": 0.05477679534181583,
+      "grad_norm": 0.1665572226047516,
+      "learning_rate": 0.00014995515507028066,
+      "loss": 2.4054,
+      "step": 127
+    },
+    {
+      "epoch": 0.055208108690963986,
+      "grad_norm": 0.1451123058795929,
+      "learning_rate": 0.00014995396729807002,
+      "loss": 2.2555,
+      "step": 128
+    },
+    {
+      "epoch": 0.055639422040112144,
+      "grad_norm": 0.1509980857372284,
+      "learning_rate": 0.00014995276400582076,
+      "loss": 2.0205,
+      "step": 129
+    },
+    {
+      "epoch": 0.056070735389260294,
+      "grad_norm": 0.19234485924243927,
+      "learning_rate": 0.00014995154519378205,
+      "loss": 2.2185,
+      "step": 130
+    },
+    {
+      "epoch": 0.05650204873840845,
+      "grad_norm": 0.16396138072013855,
+      "learning_rate": 0.00014995031086220624,
+      "loss": 2.3354,
+      "step": 131
+    },
+    {
+      "epoch": 0.05693336208755661,
+      "grad_norm": 0.15060104429721832,
+      "learning_rate": 0.00014994906101134894,
+      "loss": 2.4476,
+      "step": 132
+    },
+    {
+      "epoch": 0.057364675436704766,
+      "grad_norm": 0.149602472782135,
+      "learning_rate": 0.0001499477956414689,
+      "loss": 2.2837,
+      "step": 133
+    },
+    {
+      "epoch": 0.057795988785852924,
+      "grad_norm": 0.17236609756946564,
+      "learning_rate": 0.0001499465147528282,
+      "loss": 2.2653,
+      "step": 134
+    },
+    {
+      "epoch": 0.05822730213500108,
+      "grad_norm": 0.18239477276802063,
+      "learning_rate": 0.000149945218345692,
+      "loss": 2.3734,
+      "step": 135
+    },
+    {
+      "epoch": 0.05865861548414923,
+      "grad_norm": 0.17764970660209656,
+      "learning_rate": 0.00014994390642032875,
+      "loss": 2.2622,
+      "step": 136
+    },
+    {
+      "epoch": 0.05908992883329739,
+      "grad_norm": 0.148588165640831,
+      "learning_rate": 0.0001499425789770101,
+      "loss": 2.3332,
+      "step": 137
+    },
+    {
+      "epoch": 0.05952124218244555,
+      "grad_norm": 0.1784641295671463,
+      "learning_rate": 0.00014994123601601091,
+      "loss": 2.4611,
+      "step": 138
+    },
+    {
+      "epoch": 0.059952555531593704,
+      "grad_norm": 0.170943483710289,
+      "learning_rate": 0.00014993987753760926,
+      "loss": 2.2352,
+      "step": 139
+    },
+    {
+      "epoch": 0.06038386888074186,
+      "grad_norm": 0.13959744572639465,
+      "learning_rate": 0.00014993850354208646,
+      "loss": 2.1047,
+      "step": 140
+    },
+    {
+      "epoch": 0.06081518222989001,
+      "grad_norm": 0.14827445149421692,
+      "learning_rate": 0.00014993711402972696,
+      "loss": 2.3221,
+      "step": 141
+    },
+    {
+      "epoch": 0.06124649557903817,
+      "grad_norm": 0.1614200323820114,
+      "learning_rate": 0.00014993570900081852,
+      "loss": 2.2172,
+      "step": 142
+    },
+    {
+      "epoch": 0.06167780892818633,
+      "grad_norm": 0.14604084193706512,
+      "learning_rate": 0.00014993428845565206,
+      "loss": 2.3326,
+      "step": 143
+    },
+    {
+      "epoch": 0.062109122277334484,
+      "grad_norm": 0.15171091258525848,
+      "learning_rate": 0.00014993285239452167,
+      "loss": 2.1391,
+      "step": 144
+    },
+    {
+      "epoch": 0.06254043562648263,
+      "grad_norm": 0.13508319854736328,
+      "learning_rate": 0.00014993140081772478,
+      "loss": 2.2099,
+      "step": 145
+    },
+    {
+      "epoch": 0.06297174897563079,
+      "grad_norm": 0.16098541021347046,
+      "learning_rate": 0.00014992993372556188,
+      "loss": 2.2943,
+      "step": 146
+    },
+    {
+      "epoch": 0.06340306232477895,
+      "grad_norm": 0.1379564106464386,
+      "learning_rate": 0.0001499284511183368,
+      "loss": 2.3795,
+      "step": 147
+    },
+    {
+      "epoch": 0.06383437567392711,
+      "grad_norm": 0.16001056134700775,
+      "learning_rate": 0.0001499269529963565,
+      "loss": 2.0379,
+      "step": 148
+    },
+    {
+      "epoch": 0.06426568902307526,
+      "grad_norm": 0.14426477253437042,
+      "learning_rate": 0.00014992543935993122,
+      "loss": 2.402,
+      "step": 149
+    },
+    {
+      "epoch": 0.06469700237222342,
+      "grad_norm": 0.1733705997467041,
+      "learning_rate": 0.00014992391020937434,
+      "loss": 2.5628,
+      "step": 150
+    },
+    {
+      "epoch": 0.06469700237222342,
+      "eval_loss": 2.165189266204834,
+      "eval_runtime": 195.8058,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 150
+    },
+    {
+      "epoch": 0.06512831572137158,
+      "grad_norm": 0.15047837793827057,
+      "learning_rate": 0.00014992236554500248,
+      "loss": 2.3213,
+      "step": 151
+    },
+    {
+      "epoch": 0.06555962907051974,
+      "grad_norm": 0.15518510341644287,
+      "learning_rate": 0.0001499208053671355,
+      "loss": 2.3589,
+      "step": 152
+    },
+    {
+      "epoch": 0.0659909424196679,
+      "grad_norm": 0.14743168652057648,
+      "learning_rate": 0.00014991922967609646,
+      "loss": 2.393,
+      "step": 153
+    },
+    {
+      "epoch": 0.06642225576881604,
+      "grad_norm": 0.14655107259750366,
+      "learning_rate": 0.00014991763847221163,
+      "loss": 2.2903,
+      "step": 154
+    },
+    {
+      "epoch": 0.0668535691179642,
+      "grad_norm": 0.1595122069120407,
+      "learning_rate": 0.00014991603175581046,
+      "loss": 2.4246,
+      "step": 155
+    },
+    {
+      "epoch": 0.06728488246711235,
+      "grad_norm": 0.14971016347408295,
+      "learning_rate": 0.00014991440952722562,
+      "loss": 2.2262,
+      "step": 156
+    },
+    {
+      "epoch": 0.06771619581626051,
+      "grad_norm": 2.5569145679473877,
+      "learning_rate": 0.00014991277178679304,
+      "loss": 2.1999,
+      "step": 157
+    },
+    {
+      "epoch": 0.06814750916540867,
+      "grad_norm": 0.14273586869239807,
+      "learning_rate": 0.00014991111853485185,
+      "loss": 2.2801,
+      "step": 158
+    },
+    {
+      "epoch": 0.06857882251455683,
+      "grad_norm": 0.15529994666576385,
+      "learning_rate": 0.00014990944977174433,
+      "loss": 2.3245,
+      "step": 159
+    },
+    {
+      "epoch": 0.06901013586370498,
+      "grad_norm": 0.1808357834815979,
+      "learning_rate": 0.00014990776549781608,
+      "loss": 2.3552,
+      "step": 160
+    },
+    {
+      "epoch": 0.06944144921285314,
+      "grad_norm": 0.15095044672489166,
+      "learning_rate": 0.00014990606571341577,
+      "loss": 2.1439,
+      "step": 161
+    },
+    {
+      "epoch": 0.0698727625620013,
+      "grad_norm": 0.1695532202720642,
+      "learning_rate": 0.00014990435041889542,
+      "loss": 2.331,
+      "step": 162
+    },
+    {
+      "epoch": 0.07030407591114946,
+      "grad_norm": 0.17096754908561707,
+      "learning_rate": 0.00014990261961461015,
+      "loss": 2.2373,
+      "step": 163
+    },
+    {
+      "epoch": 0.07073538926029761,
+      "grad_norm": 0.7461947798728943,
+      "learning_rate": 0.00014990087330091838,
+      "loss": 2.1116,
+      "step": 164
+    },
+    {
+      "epoch": 0.07116670260944576,
+      "grad_norm": 0.16269400715827942,
+      "learning_rate": 0.0001498991114781817,
+      "loss": 2.29,
+      "step": 165
+    },
+    {
+      "epoch": 0.07159801595859391,
+      "grad_norm": 0.1433313637971878,
+      "learning_rate": 0.00014989733414676488,
+      "loss": 2.3007,
+      "step": 166
+    },
+    {
+      "epoch": 0.07202932930774207,
+      "grad_norm": 0.15983504056930542,
+      "learning_rate": 0.000149895541307036,
+      "loss": 2.1564,
+      "step": 167
+    },
+    {
+      "epoch": 0.07246064265689023,
+      "grad_norm": 0.15153203904628754,
+      "learning_rate": 0.00014989373295936622,
+      "loss": 2.1013,
+      "step": 168
+    },
+    {
+      "epoch": 0.07289195600603839,
+      "grad_norm": 0.19010236859321594,
+      "learning_rate": 0.00014989190910413,
+      "loss": 2.3796,
+      "step": 169
+    },
+    {
+      "epoch": 0.07332326935518654,
+      "grad_norm": 0.15856477618217468,
+      "learning_rate": 0.00014989006974170502,
+      "loss": 2.2968,
+      "step": 170
+    },
+    {
+      "epoch": 0.0737545827043347,
+      "grad_norm": 0.17133678495883942,
+      "learning_rate": 0.0001498882148724721,
+      "loss": 2.1509,
+      "step": 171
+    },
+    {
+      "epoch": 0.07418589605348286,
+      "grad_norm": 0.15978550910949707,
+      "learning_rate": 0.00014988634449681533,
+      "loss": 2.2124,
+      "step": 172
+    },
+    {
+      "epoch": 0.07461720940263102,
+      "grad_norm": 0.17379288375377655,
+      "learning_rate": 0.00014988445861512198,
+      "loss": 2.4121,
+      "step": 173
+    },
+    {
+      "epoch": 0.07504852275177917,
+      "grad_norm": 0.1456250548362732,
+      "learning_rate": 0.00014988255722778255,
+      "loss": 2.0439,
+      "step": 174
+    },
+    {
+      "epoch": 0.07547983610092733,
+      "grad_norm": 0.14056707918643951,
+      "learning_rate": 0.00014988064033519076,
+      "loss": 2.07,
+      "step": 175
+    },
+    {
+      "epoch": 0.07547983610092733,
+      "eval_loss": 2.163438320159912,
+      "eval_runtime": 199.1167,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 175
+    },
+    {
+      "epoch": 0.07591114945007547,
+      "grad_norm": 0.16266904771327972,
+      "learning_rate": 0.00014987870793774352,
+      "loss": 2.332,
+      "step": 176
+    },
+    {
+      "epoch": 0.07634246279922363,
+      "grad_norm": 0.14274229109287262,
+      "learning_rate": 0.00014987676003584094,
+      "loss": 2.4217,
+      "step": 177
+    },
+    {
+      "epoch": 0.07677377614837179,
+      "grad_norm": 0.19253656268119812,
+      "learning_rate": 0.0001498747966298863,
+      "loss": 2.265,
+      "step": 178
+    },
+    {
+      "epoch": 0.07720508949751995,
+      "grad_norm": 0.15794239938259125,
+      "learning_rate": 0.00014987281772028626,
+      "loss": 2.1961,
+      "step": 179
+    },
+    {
+      "epoch": 0.0776364028466681,
+      "grad_norm": 0.19452308118343353,
+      "learning_rate": 0.0001498708233074505,
+      "loss": 2.1581,
+      "step": 180
+    },
+    {
+      "epoch": 0.07806771619581626,
+      "grad_norm": 0.17844100296497345,
+      "learning_rate": 0.000149868813391792,
+      "loss": 2.2836,
+      "step": 181
+    },
+    {
+      "epoch": 0.07849902954496442,
+      "grad_norm": 0.13737443089485168,
+      "learning_rate": 0.00014986678797372695,
+      "loss": 2.0338,
+      "step": 182
+    },
+    {
+      "epoch": 0.07893034289411258,
+      "grad_norm": 0.2049536406993866,
+      "learning_rate": 0.0001498647470536747,
+      "loss": 2.3868,
+      "step": 183
+    },
+    {
+      "epoch": 0.07936165624326073,
+      "grad_norm": 0.1511652022600174,
+      "learning_rate": 0.00014986269063205785,
+      "loss": 2.3927,
+      "step": 184
+    },
+    {
+      "epoch": 0.07979296959240889,
+      "grad_norm": 0.15115490555763245,
+      "learning_rate": 0.00014986061870930225,
+      "loss": 2.2891,
+      "step": 185
+    },
+    {
+      "epoch": 0.08022428294155703,
+      "grad_norm": 0.2328520268201828,
+      "learning_rate": 0.00014985853128583689,
+      "loss": 1.9137,
+      "step": 186
+    },
+    {
+      "epoch": 0.08065559629070519,
+      "grad_norm": 0.14687751233577728,
+      "learning_rate": 0.00014985642836209397,
+      "loss": 2.3559,
+      "step": 187
+    },
+    {
+      "epoch": 0.08108690963985335,
+      "grad_norm": 0.1597253829240799,
+      "learning_rate": 0.00014985430993850893,
+      "loss": 2.4656,
+      "step": 188
+    },
+    {
+      "epoch": 0.08151822298900151,
+      "grad_norm": 0.1330839842557907,
+      "learning_rate": 0.00014985217601552045,
+      "loss": 2.069,
+      "step": 189
+    },
+    {
+      "epoch": 0.08194953633814966,
+      "grad_norm": 4.166354656219482,
+      "learning_rate": 0.00014985002659357035,
+      "loss": 2.1938,
+      "step": 190
+    },
+    {
+      "epoch": 0.08238084968729782,
+      "grad_norm": 0.14522095024585724,
+      "learning_rate": 0.00014984786167310368,
+      "loss": 2.1597,
+      "step": 191
+    },
+    {
+      "epoch": 0.08281216303644598,
+      "grad_norm": 0.17430412769317627,
+      "learning_rate": 0.00014984568125456875,
+      "loss": 2.39,
+      "step": 192
+    },
+    {
+      "epoch": 0.08324347638559414,
+      "grad_norm": 0.18026231229305267,
+      "learning_rate": 0.000149843485338417,
+      "loss": 2.2799,
+      "step": 193
+    },
+    {
+      "epoch": 0.0836747897347423,
+      "grad_norm": 0.18112143874168396,
+      "learning_rate": 0.00014984127392510314,
+      "loss": 2.122,
+      "step": 194
+    },
+    {
+      "epoch": 0.08410610308389045,
+      "grad_norm": 0.18193262815475464,
+      "learning_rate": 0.0001498390470150851,
+      "loss": 2.3805,
+      "step": 195
+    },
+    {
+      "epoch": 0.08453741643303861,
+      "grad_norm": 0.1746852993965149,
+      "learning_rate": 0.0001498368046088239,
+      "loss": 2.24,
+      "step": 196
+    },
+    {
+      "epoch": 0.08496872978218675,
+      "grad_norm": 0.1676950603723526,
+      "learning_rate": 0.00014983454670678393,
+      "loss": 2.1863,
+      "step": 197
+    },
+    {
+      "epoch": 0.08540004313133491,
+      "grad_norm": 0.21171925961971283,
+      "learning_rate": 0.00014983227330943267,
+      "loss": 2.2388,
+      "step": 198
+    },
+    {
+      "epoch": 0.08583135648048307,
+      "grad_norm": 10.072641372680664,
+      "learning_rate": 0.0001498299844172409,
+      "loss": 2.5266,
+      "step": 199
+    },
+    {
+      "epoch": 0.08626266982963122,
+      "grad_norm": 0.1808120310306549,
+      "learning_rate": 0.00014982768003068248,
+      "loss": 2.2083,
+      "step": 200
+    },
+    {
+      "epoch": 0.08626266982963122,
+      "eval_loss": 2.1606836318969727,
+      "eval_runtime": 197.9014,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 200
+    },
+    {
+      "epoch": 0.08669398317877938,
+      "grad_norm": 0.1853431612253189,
+      "learning_rate": 0.00014982536015023464,
+      "loss": 2.3255,
+      "step": 201
+    },
+    {
+      "epoch": 0.08712529652792754,
+      "grad_norm": 0.16815230250358582,
+      "learning_rate": 0.00014982302477637772,
+      "loss": 2.2362,
+      "step": 202
+    },
+    {
+      "epoch": 0.0875566098770757,
+      "grad_norm": 0.16719475388526917,
+      "learning_rate": 0.00014982067390959523,
+      "loss": 2.2921,
+      "step": 203
+    },
+    {
+      "epoch": 0.08798792322622385,
+      "grad_norm": 0.14663250744342804,
+      "learning_rate": 0.00014981830755037402,
+      "loss": 2.2575,
+      "step": 204
+    },
+    {
+      "epoch": 0.08841923657537201,
+      "grad_norm": 0.17926819622516632,
+      "learning_rate": 0.000149815925699204,
+      "loss": 2.2459,
+      "step": 205
+    },
+    {
+      "epoch": 0.08885054992452017,
+      "grad_norm": 0.20000754296779633,
+      "learning_rate": 0.00014981352835657843,
+      "loss": 2.4623,
+      "step": 206
+    },
+    {
+      "epoch": 0.08928186327366831,
+      "grad_norm": 0.252912700176239,
+      "learning_rate": 0.00014981111552299364,
+      "loss": 2.1574,
+      "step": 207
+    },
+    {
+      "epoch": 0.08971317662281647,
+      "grad_norm": 0.16110049188137054,
+      "learning_rate": 0.00014980868719894925,
+      "loss": 2.5126,
+      "step": 208
+    },
+    {
+      "epoch": 0.09014448997196463,
+      "grad_norm": 0.14292703568935394,
+      "learning_rate": 0.00014980624338494812,
+      "loss": 2.2023,
+      "step": 209
+    },
+    {
+      "epoch": 0.09057580332111279,
+      "grad_norm": 0.14326538145542145,
+      "learning_rate": 0.00014980378408149618,
+      "loss": 2.2351,
+      "step": 210
+    },
+    {
+      "epoch": 0.09100711667026094,
+      "grad_norm": 0.1685512661933899,
+      "learning_rate": 0.00014980130928910274,
+      "loss": 2.2906,
+      "step": 211
+    },
+    {
+      "epoch": 0.0914384300194091,
+      "grad_norm": 0.15357233583927155,
+      "learning_rate": 0.0001497988190082802,
+      "loss": 2.1548,
+      "step": 212
+    },
+    {
+      "epoch": 0.09186974336855726,
+      "grad_norm": 0.14555716514587402,
+      "learning_rate": 0.0001497963132395442,
+      "loss": 2.3084,
+      "step": 213
+    },
+    {
+      "epoch": 0.09230105671770542,
+      "grad_norm": 0.1568971574306488,
+      "learning_rate": 0.00014979379198341354,
+      "loss": 2.4084,
+      "step": 214
+    },
+    {
+      "epoch": 0.09273237006685357,
+      "grad_norm": 0.15939919650554657,
+      "learning_rate": 0.00014979125524041035,
+      "loss": 2.0452,
+      "step": 215
+    },
+    {
+      "epoch": 0.09316368341600173,
+      "grad_norm": 0.17139932513237,
+      "learning_rate": 0.00014978870301105986,
+      "loss": 2.3126,
+      "step": 216
+    },
+    {
+      "epoch": 0.09359499676514989,
+      "grad_norm": 0.16442374885082245,
+      "learning_rate": 0.00014978613529589054,
+      "loss": 2.2247,
+      "step": 217
+    },
+    {
+      "epoch": 0.09402631011429803,
+      "grad_norm": 0.15870434045791626,
+      "learning_rate": 0.00014978355209543408,
+      "loss": 2.2533,
+      "step": 218
+    },
+    {
+      "epoch": 0.09445762346344619,
+      "grad_norm": 0.14200221002101898,
+      "learning_rate": 0.0001497809534102253,
+      "loss": 2.1155,
+      "step": 219
+    },
+    {
+      "epoch": 0.09488893681259435,
+      "grad_norm": 0.15516963601112366,
+      "learning_rate": 0.00014977833924080235,
+      "loss": 2.1345,
+      "step": 220
+    },
+    {
+      "epoch": 0.0953202501617425,
+      "grad_norm": 0.16696950793266296,
+      "learning_rate": 0.00014977570958770646,
+      "loss": 2.2495,
+      "step": 221
+    },
+    {
+      "epoch": 0.09575156351089066,
+      "grad_norm": 0.14794184267520905,
+      "learning_rate": 0.00014977306445148223,
+      "loss": 2.2028,
+      "step": 222
+    },
+    {
+      "epoch": 0.09618287686003882,
+      "grad_norm": 0.15544624626636505,
+      "learning_rate": 0.00014977040383267726,
+      "loss": 2.3332,
+      "step": 223
+    },
+    {
+      "epoch": 0.09661419020918698,
+      "grad_norm": 0.1553119421005249,
+      "learning_rate": 0.0001497677277318425,
+      "loss": 2.2463,
+      "step": 224
+    },
+    {
+      "epoch": 0.09704550355833513,
+      "grad_norm": 0.1486513763666153,
+      "learning_rate": 0.00014976503614953207,
+      "loss": 2.2552,
+      "step": 225
+    },
+    {
+      "epoch": 0.09704550355833513,
+      "eval_loss": 2.1567177772521973,
+      "eval_runtime": 206.3196,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 225
+    },
+    {
+      "epoch": 0.09747681690748329,
+      "grad_norm": 0.8157986402511597,
+      "learning_rate": 0.0001497623290863033,
+      "loss": 2.3277,
+      "step": 226
+    },
+    {
+      "epoch": 0.09790813025663145,
+      "grad_norm": 0.14612847566604614,
+      "learning_rate": 0.0001497596065427167,
+      "loss": 2.2677,
+      "step": 227
+    },
+    {
+      "epoch": 0.0983394436057796,
+      "grad_norm": 0.16154086589813232,
+      "learning_rate": 0.000149756868519336,
+      "loss": 2.3785,
+      "step": 228
+    },
+    {
+      "epoch": 0.09877075695492775,
+      "grad_norm": 0.14477179944515228,
+      "learning_rate": 0.00014975411501672815,
+      "loss": 2.1928,
+      "step": 229
+    },
+    {
+      "epoch": 0.0992020703040759,
+      "grad_norm": 0.1605735719203949,
+      "learning_rate": 0.0001497513460354633,
+      "loss": 2.4549,
+      "step": 230
+    },
+    {
+      "epoch": 0.09963338365322406,
+      "grad_norm": 0.17547203600406647,
+      "learning_rate": 0.00014974856157611477,
+      "loss": 2.4296,
+      "step": 231
+    },
+    {
+      "epoch": 0.10006469700237222,
+      "grad_norm": 0.15618549287319183,
+      "learning_rate": 0.00014974576163925913,
+      "loss": 2.2927,
+      "step": 232
+    },
+    {
+      "epoch": 0.10049601035152038,
+      "grad_norm": 0.14896760880947113,
+      "learning_rate": 0.00014974294622547616,
+      "loss": 2.1635,
+      "step": 233
+    },
+    {
+      "epoch": 0.10092732370066854,
+      "grad_norm": 0.20270375907421112,
+      "learning_rate": 0.0001497401153353488,
+      "loss": 2.3012,
+      "step": 234
+    },
+    {
+      "epoch": 0.1013586370498167,
+      "grad_norm": 0.16877515614032745,
+      "learning_rate": 0.0001497372689694632,
+      "loss": 1.9421,
+      "step": 235
+    },
+    {
+      "epoch": 0.10178995039896485,
+      "grad_norm": 0.1533048003911972,
+      "learning_rate": 0.00014973440712840873,
+      "loss": 2.4441,
+      "step": 236
+    },
+    {
+      "epoch": 0.10222126374811301,
+      "grad_norm": 0.17349474132061005,
+      "learning_rate": 0.000149731529812778,
+      "loss": 2.3928,
+      "step": 237
+    },
+    {
+      "epoch": 0.10265257709726117,
+      "grad_norm": 0.1848156601190567,
+      "learning_rate": 0.00014972863702316676,
+      "loss": 2.1148,
+      "step": 238
+    },
+    {
+      "epoch": 0.10308389044640931,
+      "grad_norm": 0.15170809626579285,
+      "learning_rate": 0.00014972572876017403,
+      "loss": 2.3089,
+      "step": 239
+    },
+    {
+      "epoch": 0.10351520379555747,
+      "grad_norm": 0.14373555779457092,
+      "learning_rate": 0.00014972280502440196,
+      "loss": 2.2718,
+      "step": 240
+    },
+    {
+      "epoch": 0.10394651714470562,
+      "grad_norm": 0.15860696136951447,
+      "learning_rate": 0.00014971986581645595,
+      "loss": 2.3588,
+      "step": 241
+    },
+    {
+      "epoch": 0.10437783049385378,
+      "grad_norm": 0.2343427538871765,
+      "learning_rate": 0.0001497169111369446,
+      "loss": 2.2485,
+      "step": 242
+    },
+    {
+      "epoch": 0.10480914384300194,
+      "grad_norm": 0.14802317321300507,
+      "learning_rate": 0.00014971394098647973,
+      "loss": 2.2126,
+      "step": 243
+    },
+    {
+      "epoch": 0.1052404571921501,
+      "grad_norm": 0.15597142279148102,
+      "learning_rate": 0.0001497109553656763,
+      "loss": 2.5404,
+      "step": 244
+    },
+    {
+      "epoch": 0.10567177054129825,
+      "grad_norm": 0.15421420335769653,
+      "learning_rate": 0.00014970795427515253,
+      "loss": 2.2657,
+      "step": 245
+    },
+    {
+      "epoch": 0.10610308389044641,
+      "grad_norm": 0.18327467143535614,
+      "learning_rate": 0.00014970493771552986,
+      "loss": 2.2102,
+      "step": 246
+    },
+    {
+      "epoch": 0.10653439723959457,
+      "grad_norm": 0.15270157158374786,
+      "learning_rate": 0.00014970190568743284,
+      "loss": 2.3935,
+      "step": 247
+    },
+    {
+      "epoch": 0.10696571058874273,
+      "grad_norm": 0.15383994579315186,
+      "learning_rate": 0.00014969885819148934,
+      "loss": 2.347,
+      "step": 248
+    },
+    {
+      "epoch": 0.10739702393789088,
+      "grad_norm": 0.16574591398239136,
+      "learning_rate": 0.00014969579522833038,
+      "loss": 2.3328,
+      "step": 249
+    },
+    {
+      "epoch": 0.10782833728703903,
+      "grad_norm": 0.1711103320121765,
+      "learning_rate": 0.00014969271679859014,
+      "loss": 2.1721,
+      "step": 250
+    },
+    {
+      "epoch": 0.10782833728703903,
+      "eval_loss": 2.1529393196105957,
+      "eval_runtime": 199.1581,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 250
+    },
+    {
+      "epoch": 0.10825965063618719,
+      "grad_norm": 0.1534319669008255,
+      "learning_rate": 0.00014968962290290606,
+      "loss": 2.2902,
+      "step": 251
+    },
+    {
+      "epoch": 0.10869096398533534,
+      "grad_norm": 0.16753673553466797,
+      "learning_rate": 0.00014968651354191876,
+      "loss": 2.3074,
+      "step": 252
+    },
+    {
+      "epoch": 0.1091222773344835,
+      "grad_norm": 0.1768234521150589,
+      "learning_rate": 0.00014968338871627207,
+      "loss": 2.036,
+      "step": 253
+    },
+    {
+      "epoch": 0.10955359068363166,
+      "grad_norm": 0.1682131290435791,
+      "learning_rate": 0.00014968024842661307,
+      "loss": 2.1679,
+      "step": 254
+    },
+    {
+      "epoch": 0.10998490403277981,
+      "grad_norm": 0.15855084359645844,
+      "learning_rate": 0.00014967709267359191,
+      "loss": 2.2204,
+      "step": 255
+    },
+    {
+      "epoch": 0.11041621738192797,
+      "grad_norm": 0.20829591155052185,
+      "learning_rate": 0.00014967392145786204,
+      "loss": 2.29,
+      "step": 256
+    },
+    {
+      "epoch": 0.11084753073107613,
+      "grad_norm": 0.17669428884983063,
+      "learning_rate": 0.00014967073478008012,
+      "loss": 2.1349,
+      "step": 257
+    },
+    {
+      "epoch": 0.11127884408022429,
+      "grad_norm": 0.1606130748987198,
+      "learning_rate": 0.000149667532640906,
+      "loss": 1.9809,
+      "step": 258
+    },
+    {
+      "epoch": 0.11171015742937244,
+      "grad_norm": 0.16479335725307465,
+      "learning_rate": 0.00014966431504100268,
+      "loss": 2.3708,
+      "step": 259
+    },
+    {
+      "epoch": 0.11214147077852059,
+      "grad_norm": 0.15960106253623962,
+      "learning_rate": 0.0001496610819810364,
+      "loss": 2.1645,
+      "step": 260
+    },
+    {
+      "epoch": 0.11257278412766875,
+      "grad_norm": 0.1663437783718109,
+      "learning_rate": 0.00014965783346167664,
+      "loss": 2.4815,
+      "step": 261
+    },
+    {
+      "epoch": 0.1130040974768169,
+      "grad_norm": 0.2073906511068344,
+      "learning_rate": 0.000149654569483596,
+      "loss": 2.3176,
+      "step": 262
+    },
+    {
+      "epoch": 0.11343541082596506,
+      "grad_norm": 0.16055135428905487,
+      "learning_rate": 0.00014965129004747036,
+      "loss": 2.2771,
+      "step": 263
+    },
+    {
+      "epoch": 0.11386672417511322,
+      "grad_norm": 0.1713322103023529,
+      "learning_rate": 0.00014964799515397874,
+      "loss": 2.1259,
+      "step": 264
+    },
+    {
+      "epoch": 0.11429803752426138,
+      "grad_norm": 0.17094393074512482,
+      "learning_rate": 0.0001496446848038034,
+      "loss": 2.308,
+      "step": 265
+    },
+    {
+      "epoch": 0.11472935087340953,
+      "grad_norm": 0.18342117965221405,
+      "learning_rate": 0.00014964135899762972,
+      "loss": 2.3167,
+      "step": 266
+    },
+    {
+      "epoch": 0.11516066422255769,
+      "grad_norm": 0.17521141469478607,
+      "learning_rate": 0.00014963801773614643,
+      "loss": 2.2702,
+      "step": 267
+    },
+    {
+      "epoch": 0.11559197757170585,
+      "grad_norm": 0.18539130687713623,
+      "learning_rate": 0.00014963466102004534,
+      "loss": 2.2543,
+      "step": 268
+    },
+    {
+      "epoch": 0.116023290920854,
+      "grad_norm": 0.16789484024047852,
+      "learning_rate": 0.0001496312888500215,
+      "loss": 2.2379,
+      "step": 269
+    },
+    {
+      "epoch": 0.11645460427000216,
+      "grad_norm": 0.18903698027133942,
+      "learning_rate": 0.00014962790122677314,
+      "loss": 2.1078,
+      "step": 270
+    },
+    {
+      "epoch": 0.1168859176191503,
+      "grad_norm": 0.15733234584331512,
+      "learning_rate": 0.0001496244981510017,
+      "loss": 2.3288,
+      "step": 271
+    },
+    {
+      "epoch": 0.11731723096829846,
+      "grad_norm": 0.14740827679634094,
+      "learning_rate": 0.00014962107962341188,
+      "loss": 2.2901,
+      "step": 272
+    },
+    {
+      "epoch": 0.11774854431744662,
+      "grad_norm": 0.16103410720825195,
+      "learning_rate": 0.00014961764564471144,
+      "loss": 2.2015,
+      "step": 273
+    },
+    {
+      "epoch": 0.11817985766659478,
+      "grad_norm": 0.18729737401008606,
+      "learning_rate": 0.00014961419621561147,
+      "loss": 2.3337,
+      "step": 274
+    },
+    {
+      "epoch": 0.11861117101574294,
+      "grad_norm": 0.14679095149040222,
+      "learning_rate": 0.00014961073133682622,
+      "loss": 2.2346,
+      "step": 275
+    },
+    {
+      "epoch": 0.11861117101574294,
+      "eval_loss": 2.150817632675171,
+      "eval_runtime": 198.5771,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 275
+    },
+    {
+      "epoch": 0.1190424843648911,
+      "grad_norm": 0.18481458723545074,
+      "learning_rate": 0.0001496072510090731,
+      "loss": 2.3638,
+      "step": 276
+    },
+    {
+      "epoch": 0.11947379771403925,
+      "grad_norm": 0.1526854783296585,
+      "learning_rate": 0.0001496037552330728,
+      "loss": 2.4251,
+      "step": 277
+    },
+    {
+      "epoch": 0.11990511106318741,
+      "grad_norm": 0.16679935157299042,
+      "learning_rate": 0.0001496002440095491,
+      "loss": 2.2725,
+      "step": 278
+    },
+    {
+      "epoch": 0.12033642441233557,
+      "grad_norm": 0.1559242308139801,
+      "learning_rate": 0.00014959671733922908,
+      "loss": 2.2053,
+      "step": 279
+    },
+    {
+      "epoch": 0.12076773776148372,
+      "grad_norm": 0.15493714809417725,
+      "learning_rate": 0.00014959317522284294,
+      "loss": 2.3275,
+      "step": 280
+    },
+    {
+      "epoch": 0.12119905111063188,
+      "grad_norm": 0.17583470046520233,
+      "learning_rate": 0.00014958961766112415,
+      "loss": 2.2412,
+      "step": 281
+    },
+    {
+      "epoch": 0.12163036445978002,
+      "grad_norm": 0.18624302744865417,
+      "learning_rate": 0.0001495860446548093,
+      "loss": 2.257,
+      "step": 282
+    },
+    {
+      "epoch": 0.12206167780892818,
+      "grad_norm": 0.16880694031715393,
+      "learning_rate": 0.00014958245620463827,
+      "loss": 2.3616,
+      "step": 283
+    },
+    {
+      "epoch": 0.12249299115807634,
+      "grad_norm": 0.16436102986335754,
+      "learning_rate": 0.00014957885231135405,
+      "loss": 2.2771,
+      "step": 284
+    },
+    {
+      "epoch": 0.1229243045072245,
+      "grad_norm": 0.1797478049993515,
+      "learning_rate": 0.0001495752329757029,
+      "loss": 2.2618,
+      "step": 285
+    },
+    {
+      "epoch": 0.12335561785637265,
+      "grad_norm": 0.16185274720191956,
+      "learning_rate": 0.00014957159819843418,
+      "loss": 2.326,
+      "step": 286
+    },
+    {
+      "epoch": 0.12378693120552081,
+      "grad_norm": 0.16708776354789734,
+      "learning_rate": 0.0001495679479803006,
+      "loss": 2.4173,
+      "step": 287
+    },
+    {
+      "epoch": 0.12421824455466897,
+      "grad_norm": 0.1608465015888214,
+      "learning_rate": 0.00014956428232205788,
+      "loss": 2.1053,
+      "step": 288
+    },
+    {
+      "epoch": 0.12464955790381713,
+      "grad_norm": 0.15215721726417542,
+      "learning_rate": 0.00014956060122446512,
+      "loss": 2.361,
+      "step": 289
+    },
+    {
+      "epoch": 0.12508087125296527,
+      "grad_norm": 0.18214476108551025,
+      "learning_rate": 0.0001495569046882845,
+      "loss": 2.2829,
+      "step": 290
+    },
+    {
+      "epoch": 0.12551218460211344,
+      "grad_norm": 0.1762077957391739,
+      "learning_rate": 0.0001495531927142814,
+      "loss": 2.22,
+      "step": 291
+    },
+    {
+      "epoch": 0.12594349795126158,
+      "grad_norm": 0.16295194625854492,
+      "learning_rate": 0.00014954946530322446,
+      "loss": 2.2119,
+      "step": 292
+    },
+    {
+      "epoch": 0.12637481130040976,
+      "grad_norm": 0.15086841583251953,
+      "learning_rate": 0.00014954572245588546,
+      "loss": 2.2676,
+      "step": 293
+    },
+    {
+      "epoch": 0.1268061246495579,
+      "grad_norm": 0.1604457050561905,
+      "learning_rate": 0.00014954196417303942,
+      "loss": 2.0791,
+      "step": 294
+    },
+    {
+      "epoch": 0.12723743799870607,
+      "grad_norm": 0.15493498742580414,
+      "learning_rate": 0.00014953819045546453,
+      "loss": 2.1768,
+      "step": 295
+    },
+    {
+      "epoch": 0.12766875134785421,
+      "grad_norm": 0.16815432906150818,
+      "learning_rate": 0.00014953440130394214,
+      "loss": 2.4107,
+      "step": 296
+    },
+    {
+      "epoch": 0.12810006469700239,
+      "grad_norm": 0.15603506565093994,
+      "learning_rate": 0.0001495305967192569,
+      "loss": 2.1845,
+      "step": 297
+    },
+    {
+      "epoch": 0.12853137804615053,
+      "grad_norm": 0.1547204703092575,
+      "learning_rate": 0.00014952677670219654,
+      "loss": 2.3356,
+      "step": 298
+    },
+    {
+      "epoch": 0.12896269139529867,
+      "grad_norm": 0.16853530704975128,
+      "learning_rate": 0.00014952294125355207,
+      "loss": 2.0881,
+      "step": 299
+    },
+    {
+      "epoch": 0.12939400474444684,
+      "grad_norm": 0.1595228761434555,
+      "learning_rate": 0.0001495190903741176,
+      "loss": 2.4599,
+      "step": 300
+    },
+    {
+      "epoch": 0.12939400474444684,
+      "eval_loss": 2.1485748291015625,
+      "eval_runtime": 199.1444,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 300
+    },
+    {
+      "epoch": 0.129825318093595,
+      "grad_norm": 0.22638855874538422,
+      "learning_rate": 0.00014951522406469057,
+      "loss": 2.2354,
+      "step": 301
+    },
+    {
+      "epoch": 0.13025663144274316,
+      "grad_norm": 3.426063299179077,
+      "learning_rate": 0.0001495113423260715,
+      "loss": 2.1358,
+      "step": 302
+    },
+    {
+      "epoch": 0.1306879447918913,
+      "grad_norm": 0.15848495066165924,
+      "learning_rate": 0.00014950744515906416,
+      "loss": 2.1514,
+      "step": 303
+    },
+    {
+      "epoch": 0.13111925814103947,
+      "grad_norm": 0.15022383630275726,
+      "learning_rate": 0.0001495035325644755,
+      "loss": 2.2748,
+      "step": 304
+    },
+    {
+      "epoch": 0.13155057149018762,
+      "grad_norm": 0.21391138434410095,
+      "learning_rate": 0.0001494996045431157,
+      "loss": 2.4523,
+      "step": 305
+    },
+    {
+      "epoch": 0.1319818848393358,
+      "grad_norm": 0.15094725787639618,
+      "learning_rate": 0.00014949566109579804,
+      "loss": 2.1579,
+      "step": 306
+    },
+    {
+      "epoch": 0.13241319818848393,
+      "grad_norm": 0.1510794758796692,
+      "learning_rate": 0.00014949170222333907,
+      "loss": 2.3597,
+      "step": 307
+    },
+    {
+      "epoch": 0.13284451153763208,
+      "grad_norm": 0.18151414394378662,
+      "learning_rate": 0.00014948772792655854,
+      "loss": 2.1657,
+      "step": 308
+    },
+    {
+      "epoch": 0.13327582488678025,
+      "grad_norm": 0.17762115597724915,
+      "learning_rate": 0.00014948373820627934,
+      "loss": 2.2904,
+      "step": 309
+    },
+    {
+      "epoch": 0.1337071382359284,
+      "grad_norm": 0.15702734887599945,
+      "learning_rate": 0.00014947973306332765,
+      "loss": 2.2888,
+      "step": 310
+    },
+    {
+      "epoch": 0.13413845158507656,
+      "grad_norm": 0.16596752405166626,
+      "learning_rate": 0.00014947571249853268,
+      "loss": 2.3807,
+      "step": 311
+    },
+    {
+      "epoch": 0.1345697649342247,
+      "grad_norm": 0.16935662925243378,
+      "learning_rate": 0.00014947167651272703,
+      "loss": 2.3313,
+      "step": 312
+    },
+    {
+      "epoch": 0.13500107828337288,
+      "grad_norm": 0.16634508967399597,
+      "learning_rate": 0.0001494676251067463,
+      "loss": 2.137,
+      "step": 313
+    },
+    {
+      "epoch": 0.13543239163252102,
+      "grad_norm": 4.746288299560547,
+      "learning_rate": 0.00014946355828142946,
+      "loss": 2.2238,
+      "step": 314
+    },
+    {
+      "epoch": 0.1358637049816692,
+      "grad_norm": 0.16148056089878082,
+      "learning_rate": 0.00014945947603761854,
+      "loss": 1.9769,
+      "step": 315
+    },
+    {
+      "epoch": 0.13629501833081734,
+      "grad_norm": 0.1666569709777832,
+      "learning_rate": 0.00014945537837615885,
+      "loss": 2.227,
+      "step": 316
+    },
+    {
+      "epoch": 0.1367263316799655,
+      "grad_norm": 0.16864323616027832,
+      "learning_rate": 0.00014945126529789885,
+      "loss": 2.292,
+      "step": 317
+    },
+    {
+      "epoch": 0.13715764502911365,
+      "grad_norm": 0.15381769835948944,
+      "learning_rate": 0.00014944713680369013,
+      "loss": 2.1039,
+      "step": 318
+    },
+    {
+      "epoch": 0.1375889583782618,
+      "grad_norm": 0.16596980392932892,
+      "learning_rate": 0.00014944299289438765,
+      "loss": 2.5114,
+      "step": 319
+    },
+    {
+      "epoch": 0.13802027172740997,
+      "grad_norm": 0.156742125749588,
+      "learning_rate": 0.0001494388335708494,
+      "loss": 2.4077,
+      "step": 320
+    },
+    {
+      "epoch": 0.1384515850765581,
+      "grad_norm": 0.16346225142478943,
+      "learning_rate": 0.00014943465883393654,
+      "loss": 2.4141,
+      "step": 321
+    },
+    {
+      "epoch": 0.13888289842570628,
+      "grad_norm": 0.17249517142772675,
+      "learning_rate": 0.00014943046868451363,
+      "loss": 2.4374,
+      "step": 322
+    },
+    {
+      "epoch": 0.13931421177485442,
+      "grad_norm": 0.15804658830165863,
+      "learning_rate": 0.00014942626312344822,
+      "loss": 2.3544,
+      "step": 323
+    },
+    {
+      "epoch": 0.1397455251240026,
+      "grad_norm": 0.15350793302059174,
+      "learning_rate": 0.00014942204215161108,
+      "loss": 2.3532,
+      "step": 324
+    },
+    {
+      "epoch": 0.14017683847315074,
+      "grad_norm": 0.15994253754615784,
+      "learning_rate": 0.00014941780576987627,
+      "loss": 2.1783,
+      "step": 325
+    },
+    {
+      "epoch": 0.14017683847315074,
+      "eval_loss": 2.149773597717285,
+      "eval_runtime": 200.0419,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 325
+    },
+    {
+      "epoch": 0.1406081518222989,
+      "grad_norm": 0.16531150043010712,
+      "learning_rate": 0.00014941355397912095,
+      "loss": 2.2161,
+      "step": 326
+    },
+    {
+      "epoch": 0.14103946517144705,
+      "grad_norm": 0.24105919897556305,
+      "learning_rate": 0.00014940928678022552,
+      "loss": 2.1379,
+      "step": 327
+    },
+    {
+      "epoch": 0.14147077852059523,
+      "grad_norm": 0.17379091680049896,
+      "learning_rate": 0.00014940500417407352,
+      "loss": 2.3302,
+      "step": 328
+    },
+    {
+      "epoch": 0.14190209186974337,
+      "grad_norm": 0.155650332570076,
+      "learning_rate": 0.00014940070616155173,
+      "loss": 2.1339,
+      "step": 329
+    },
+    {
+      "epoch": 0.1423334052188915,
+      "grad_norm": 0.16053761541843414,
+      "learning_rate": 0.0001493963927435501,
+      "loss": 2.232,
+      "step": 330
+    },
+    {
+      "epoch": 0.14276471856803968,
+      "grad_norm": 0.22595985233783722,
+      "learning_rate": 0.00014939206392096177,
+      "loss": 2.2803,
+      "step": 331
+    },
+    {
+      "epoch": 0.14319603191718783,
+      "grad_norm": 0.17255187034606934,
+      "learning_rate": 0.00014938771969468304,
+      "loss": 2.0613,
+      "step": 332
+    },
+    {
+      "epoch": 0.143627345266336,
+      "grad_norm": 0.18829742074012756,
+      "learning_rate": 0.00014938336006561348,
+      "loss": 2.4404,
+      "step": 333
+    },
+    {
+      "epoch": 0.14405865861548414,
+      "grad_norm": 0.7800279855728149,
+      "learning_rate": 0.0001493789850346558,
+      "loss": 2.2717,
+      "step": 334
+    },
+    {
+      "epoch": 0.1444899719646323,
+      "grad_norm": 0.1824026256799698,
+      "learning_rate": 0.00014937459460271583,
+      "loss": 2.2762,
+      "step": 335
+    },
+    {
+      "epoch": 0.14492128531378046,
+      "grad_norm": 0.1767878383398056,
+      "learning_rate": 0.00014937018877070272,
+      "loss": 2.3848,
+      "step": 336
+    },
+    {
+      "epoch": 0.14535259866292863,
+      "grad_norm": 0.1794254034757614,
+      "learning_rate": 0.00014936576753952872,
+      "loss": 2.1204,
+      "step": 337
+    },
+    {
+      "epoch": 0.14578391201207677,
+      "grad_norm": 0.17064186930656433,
+      "learning_rate": 0.0001493613309101093,
+      "loss": 2.2617,
+      "step": 338
+    },
+    {
+      "epoch": 0.14621522536122494,
+      "grad_norm": 0.1971042901277542,
+      "learning_rate": 0.00014935687888336313,
+      "loss": 2.3569,
+      "step": 339
+    },
+    {
+      "epoch": 0.1466465387103731,
+      "grad_norm": 0.16145062446594238,
+      "learning_rate": 0.00014935241146021201,
+      "loss": 2.1903,
+      "step": 340
+    },
+    {
+      "epoch": 0.14707785205952123,
+      "grad_norm": 0.18148300051689148,
+      "learning_rate": 0.000149347928641581,
+      "loss": 2.2269,
+      "step": 341
+    },
+    {
+      "epoch": 0.1475091654086694,
+      "grad_norm": 0.19072268903255463,
+      "learning_rate": 0.00014934343042839832,
+      "loss": 2.2133,
+      "step": 342
+    },
+    {
+      "epoch": 0.14794047875781755,
+      "grad_norm": 0.1789669245481491,
+      "learning_rate": 0.00014933891682159537,
+      "loss": 2.2024,
+      "step": 343
+    },
+    {
+      "epoch": 0.14837179210696572,
+      "grad_norm": 0.1523238867521286,
+      "learning_rate": 0.00014933438782210672,
+      "loss": 2.3505,
+      "step": 344
+    },
+    {
+      "epoch": 0.14880310545611386,
+      "grad_norm": 0.15985886752605438,
+      "learning_rate": 0.00014932984343087017,
+      "loss": 2.2357,
+      "step": 345
+    },
+    {
+      "epoch": 0.14923441880526203,
+      "grad_norm": 0.16432873904705048,
+      "learning_rate": 0.0001493252836488267,
+      "loss": 2.2678,
+      "step": 346
+    },
+    {
+      "epoch": 0.14966573215441017,
+      "grad_norm": 0.14971576631069183,
+      "learning_rate": 0.0001493207084769204,
+      "loss": 2.0104,
+      "step": 347
+    },
+    {
+      "epoch": 0.15009704550355835,
+      "grad_norm": 0.16536620259284973,
+      "learning_rate": 0.00014931611791609868,
+      "loss": 2.2598,
+      "step": 348
+    },
+    {
+      "epoch": 0.1505283588527065,
+      "grad_norm": 0.16472128033638,
+      "learning_rate": 0.00014931151196731206,
+      "loss": 2.2177,
+      "step": 349
+    },
+    {
+      "epoch": 0.15095967220185466,
+      "grad_norm": 0.1672239452600479,
+      "learning_rate": 0.0001493068906315142,
+      "loss": 2.5143,
+      "step": 350
+    },
+    {
+      "epoch": 0.15095967220185466,
+      "eval_loss": 2.147500991821289,
+      "eval_runtime": 199.3339,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 350
+    },
+    {
+      "epoch": 0.1513909855510028,
+      "grad_norm": 0.15225522220134735,
+      "learning_rate": 0.00014930225390966201,
+      "loss": 2.1453,
+      "step": 351
+    },
+    {
+      "epoch": 0.15182229890015095,
+      "grad_norm": 0.20649577677249908,
+      "learning_rate": 0.00014929760180271563,
+      "loss": 2.1401,
+      "step": 352
+    },
+    {
+      "epoch": 0.15225361224929912,
+      "grad_norm": 0.20784424245357513,
+      "learning_rate": 0.00014929293431163826,
+      "loss": 2.2101,
+      "step": 353
+    },
+    {
+      "epoch": 0.15268492559844726,
+      "grad_norm": 0.20000191032886505,
+      "learning_rate": 0.0001492882514373964,
+      "loss": 2.091,
+      "step": 354
+    },
+    {
+      "epoch": 0.15311623894759543,
+      "grad_norm": 0.17898254096508026,
+      "learning_rate": 0.0001492835531809597,
+      "loss": 2.1752,
+      "step": 355
+    },
+    {
+      "epoch": 0.15354755229674358,
+      "grad_norm": 1.4756970405578613,
+      "learning_rate": 0.00014927883954330093,
+      "loss": 2.2416,
+      "step": 356
+    },
+    {
+      "epoch": 0.15397886564589175,
+      "grad_norm": 0.17481818795204163,
+      "learning_rate": 0.00014927411052539615,
+      "loss": 2.3631,
+      "step": 357
+    },
+    {
+      "epoch": 0.1544101789950399,
+      "grad_norm": 0.1990717649459839,
+      "learning_rate": 0.00014926936612822452,
+      "loss": 2.4406,
+      "step": 358
+    },
+    {
+      "epoch": 0.15484149234418806,
+      "grad_norm": 0.1901736706495285,
+      "learning_rate": 0.00014926460635276845,
+      "loss": 2.2884,
+      "step": 359
+    },
+    {
+      "epoch": 0.1552728056933362,
+      "grad_norm": 0.16656960546970367,
+      "learning_rate": 0.00014925983120001352,
+      "loss": 2.3036,
+      "step": 360
+    },
+    {
+      "epoch": 0.15570411904248435,
+      "grad_norm": 0.16077920794487,
+      "learning_rate": 0.0001492550406709484,
+      "loss": 2.2038,
+      "step": 361
+    },
+    {
+      "epoch": 0.15613543239163252,
+      "grad_norm": 0.18244610726833344,
+      "learning_rate": 0.0001492502347665651,
+      "loss": 2.2564,
+      "step": 362
+    },
+    {
+      "epoch": 0.15656674574078067,
+      "grad_norm": 0.16418373584747314,
+      "learning_rate": 0.0001492454134878587,
+      "loss": 2.216,
+      "step": 363
+    },
+    {
+      "epoch": 0.15699805908992884,
+      "grad_norm": 0.16991998255252838,
+      "learning_rate": 0.00014924057683582748,
+      "loss": 2.194,
+      "step": 364
+    },
+    {
+      "epoch": 0.15742937243907698,
+      "grad_norm": 0.17290419340133667,
+      "learning_rate": 0.00014923572481147297,
+      "loss": 2.1836,
+      "step": 365
+    },
+    {
+      "epoch": 0.15786068578822515,
+      "grad_norm": 0.17075559496879578,
+      "learning_rate": 0.0001492308574157998,
+      "loss": 2.3898,
+      "step": 366
+    },
+    {
+      "epoch": 0.1582919991373733,
+      "grad_norm": 0.15228039026260376,
+      "learning_rate": 0.00014922597464981583,
+      "loss": 2.1941,
+      "step": 367
+    },
+    {
+      "epoch": 0.15872331248652147,
+      "grad_norm": 0.15793053805828094,
+      "learning_rate": 0.0001492210765145321,
+      "loss": 2.2475,
+      "step": 368
+    },
+    {
+      "epoch": 0.1591546258356696,
+      "grad_norm": 0.1701696217060089,
+      "learning_rate": 0.00014921616301096278,
+      "loss": 2.1133,
+      "step": 369
+    },
+    {
+      "epoch": 0.15958593918481778,
+      "grad_norm": 0.1666203737258911,
+      "learning_rate": 0.0001492112341401253,
+      "loss": 2.2035,
+      "step": 370
+    },
+    {
+      "epoch": 0.16001725253396593,
+      "grad_norm": 0.15982025861740112,
+      "learning_rate": 0.00014920628990304025,
+      "loss": 2.1841,
+      "step": 371
+    },
+    {
+      "epoch": 0.16044856588311407,
+      "grad_norm": 0.18124015629291534,
+      "learning_rate": 0.00014920133030073136,
+      "loss": 2.3534,
+      "step": 372
+    },
+    {
+      "epoch": 0.16087987923226224,
+      "grad_norm": 0.1591426134109497,
+      "learning_rate": 0.00014919635533422557,
+      "loss": 2.1313,
+      "step": 373
+    },
+    {
+      "epoch": 0.16131119258141038,
+      "grad_norm": 0.15558771789073944,
+      "learning_rate": 0.00014919136500455302,
+      "loss": 2.2474,
+      "step": 374
+    },
+    {
+      "epoch": 0.16174250593055856,
+      "grad_norm": 0.15386322140693665,
+      "learning_rate": 0.000149186359312747,
+      "loss": 2.0939,
+      "step": 375
+    },
+    {
+      "epoch": 0.16174250593055856,
+      "eval_loss": 2.145970344543457,
+      "eval_runtime": 199.181,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 375
+    },
+    {
+      "epoch": 0.1621738192797067,
+      "grad_norm": 0.1613401472568512,
+      "learning_rate": 0.000149181338259844,
+      "loss": 2.2186,
+      "step": 376
+    },
+    {
+      "epoch": 0.16260513262885487,
+      "grad_norm": 0.16436243057250977,
+      "learning_rate": 0.0001491763018468837,
+      "loss": 2.1255,
+      "step": 377
+    },
+    {
+      "epoch": 0.16303644597800301,
+      "grad_norm": 0.15056097507476807,
+      "learning_rate": 0.00014917125007490889,
+      "loss": 2.2909,
+      "step": 378
+    },
+    {
+      "epoch": 0.16346775932715119,
+      "grad_norm": 1.3172454833984375,
+      "learning_rate": 0.00014916618294496566,
+      "loss": 2.3205,
+      "step": 379
+    },
+    {
+      "epoch": 0.16389907267629933,
+      "grad_norm": 0.1644919365644455,
+      "learning_rate": 0.00014916110045810316,
+      "loss": 2.2956,
+      "step": 380
+    },
+    {
+      "epoch": 0.1643303860254475,
+      "grad_norm": 0.16997544467449188,
+      "learning_rate": 0.0001491560026153738,
+      "loss": 2.4651,
+      "step": 381
+    },
+    {
+      "epoch": 0.16476169937459564,
+      "grad_norm": 0.17540326714515686,
+      "learning_rate": 0.00014915088941783313,
+      "loss": 2.2358,
+      "step": 382
+    },
+    {
+      "epoch": 0.1651930127237438,
+      "grad_norm": 0.1699107587337494,
+      "learning_rate": 0.0001491457608665399,
+      "loss": 2.4559,
+      "step": 383
+    },
+    {
+      "epoch": 0.16562432607289196,
+      "grad_norm": 0.1600213497877121,
+      "learning_rate": 0.0001491406169625561,
+      "loss": 2.1716,
+      "step": 384
+    },
+    {
+      "epoch": 0.1660556394220401,
+      "grad_norm": 0.5548557639122009,
+      "learning_rate": 0.00014913545770694673,
+      "loss": 2.2096,
+      "step": 385
+    },
+    {
+      "epoch": 0.16648695277118827,
+      "grad_norm": 0.1656799465417862,
+      "learning_rate": 0.0001491302831007801,
+      "loss": 2.18,
+      "step": 386
+    },
+    {
+      "epoch": 0.16691826612033642,
+      "grad_norm": 0.18784065544605255,
+      "learning_rate": 0.00014912509314512768,
+      "loss": 2.3623,
+      "step": 387
+    },
+    {
+      "epoch": 0.1673495794694846,
+      "grad_norm": 0.17505837976932526,
+      "learning_rate": 0.00014911988784106412,
+      "loss": 2.3084,
+      "step": 388
+    },
+    {
+      "epoch": 0.16778089281863273,
+      "grad_norm": 0.1727839559316635,
+      "learning_rate": 0.00014911466718966722,
+      "loss": 2.2042,
+      "step": 389
+    },
+    {
+      "epoch": 0.1682122061677809,
+      "grad_norm": 0.16786758601665497,
+      "learning_rate": 0.00014910943119201796,
+      "loss": 1.9998,
+      "step": 390
+    },
+    {
+      "epoch": 0.16864351951692905,
+      "grad_norm": 0.18020984530448914,
+      "learning_rate": 0.00014910417984920056,
+      "loss": 2.2189,
+      "step": 391
+    },
+    {
+      "epoch": 0.16907483286607722,
+      "grad_norm": 0.16372406482696533,
+      "learning_rate": 0.00014909891316230227,
+      "loss": 2.2852,
+      "step": 392
+    },
+    {
+      "epoch": 0.16950614621522536,
+      "grad_norm": 0.15674197673797607,
+      "learning_rate": 0.00014909363113241374,
+      "loss": 2.0479,
+      "step": 393
+    },
+    {
+      "epoch": 0.1699374595643735,
+      "grad_norm": 0.17500346899032593,
+      "learning_rate": 0.00014908833376062858,
+      "loss": 2.4546,
+      "step": 394
+    },
+    {
+      "epoch": 0.17036877291352168,
+      "grad_norm": 0.20425276458263397,
+      "learning_rate": 0.0001490830210480437,
+      "loss": 2.318,
+      "step": 395
+    },
+    {
+      "epoch": 0.17080008626266982,
+      "grad_norm": 0.16255128383636475,
+      "learning_rate": 0.00014907769299575915,
+      "loss": 2.3926,
+      "step": 396
+    },
+    {
+      "epoch": 0.171231399611818,
+      "grad_norm": 0.18405508995056152,
+      "learning_rate": 0.00014907234960487818,
+      "loss": 2.2712,
+      "step": 397
+    },
+    {
+      "epoch": 0.17166271296096614,
+      "grad_norm": 0.1642206460237503,
+      "learning_rate": 0.00014906699087650717,
+      "loss": 2.1432,
+      "step": 398
+    },
+    {
+      "epoch": 0.1720940263101143,
+      "grad_norm": 0.14774279296398163,
+      "learning_rate": 0.00014906161681175573,
+      "loss": 2.1958,
+      "step": 399
+    },
+    {
+      "epoch": 0.17252533965926245,
+      "grad_norm": 0.1816944181919098,
+      "learning_rate": 0.00014905622741173658,
+      "loss": 2.3528,
+      "step": 400
+    },
+    {
+      "epoch": 0.17252533965926245,
+      "eval_loss": 2.145236015319824,
+      "eval_runtime": 199.4998,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 400
+    },
+    {
+      "epoch": 0.17295665300841062,
+      "grad_norm": 0.15424853563308716,
+      "learning_rate": 0.00014905082267756568,
+      "loss": 2.1429,
+      "step": 401
+    },
+    {
+      "epoch": 0.17338796635755876,
+      "grad_norm": 0.1658765971660614,
+      "learning_rate": 0.00014904540261036217,
+      "loss": 2.2798,
+      "step": 402
+    },
+    {
+      "epoch": 0.17381927970670694,
+      "grad_norm": 0.183359295129776,
+      "learning_rate": 0.0001490399672112483,
+      "loss": 2.3538,
+      "step": 403
+    },
+    {
+      "epoch": 0.17425059305585508,
+      "grad_norm": 0.16044598817825317,
+      "learning_rate": 0.0001490345164813495,
+      "loss": 2.2135,
+      "step": 404
+    },
+    {
+      "epoch": 0.17468190640500322,
+      "grad_norm": 0.1611524522304535,
+      "learning_rate": 0.00014902905042179446,
+      "loss": 2.2036,
+      "step": 405
+    },
+    {
+      "epoch": 0.1751132197541514,
+      "grad_norm": 0.16904768347740173,
+      "learning_rate": 0.00014902356903371495,
+      "loss": 2.3068,
+      "step": 406
+    },
+    {
+      "epoch": 0.17554453310329954,
+      "grad_norm": 0.1789490431547165,
+      "learning_rate": 0.00014901807231824598,
+      "loss": 2.2595,
+      "step": 407
+    },
+    {
+      "epoch": 0.1759758464524477,
+      "grad_norm": 0.21464680135250092,
+      "learning_rate": 0.00014901256027652568,
+      "loss": 2.2024,
+      "step": 408
+    },
+    {
+      "epoch": 0.17640715980159585,
+      "grad_norm": 0.16914889216423035,
+      "learning_rate": 0.0001490070329096954,
+      "loss": 2.1231,
+      "step": 409
+    },
+    {
+      "epoch": 0.17683847315074402,
+      "grad_norm": 0.17402493953704834,
+      "learning_rate": 0.0001490014902188996,
+      "loss": 1.9689,
+      "step": 410
+    },
+    {
+      "epoch": 0.17726978649989217,
+      "grad_norm": 0.15200650691986084,
+      "learning_rate": 0.000148995932205286,
+      "loss": 2.34,
+      "step": 411
+    },
+    {
+      "epoch": 0.17770109984904034,
+      "grad_norm": 0.16262347996234894,
+      "learning_rate": 0.0001489903588700054,
+      "loss": 2.216,
+      "step": 412
+    },
+    {
+      "epoch": 0.17813241319818848,
+      "grad_norm": 0.16141052544116974,
+      "learning_rate": 0.0001489847702142119,
+      "loss": 2.3693,
+      "step": 413
+    },
+    {
+      "epoch": 0.17856372654733663,
+      "grad_norm": 0.1716516762971878,
+      "learning_rate": 0.00014897916623906262,
+      "loss": 2.2875,
+      "step": 414
+    },
+    {
+      "epoch": 0.1789950398964848,
+      "grad_norm": 0.16397427022457123,
+      "learning_rate": 0.00014897354694571794,
+      "loss": 2.3241,
+      "step": 415
+    },
+    {
+      "epoch": 0.17942635324563294,
+      "grad_norm": 0.1554698348045349,
+      "learning_rate": 0.00014896791233534143,
+      "loss": 2.2666,
+      "step": 416
+    },
+    {
+      "epoch": 0.1798576665947811,
+      "grad_norm": 0.16670356690883636,
+      "learning_rate": 0.00014896226240909975,
+      "loss": 2.3704,
+      "step": 417
+    },
+    {
+      "epoch": 0.18028897994392926,
+      "grad_norm": 0.14467710256576538,
+      "learning_rate": 0.0001489565971681628,
+      "loss": 2.2538,
+      "step": 418
+    },
+    {
+      "epoch": 0.18072029329307743,
+      "grad_norm": 0.1782013177871704,
+      "learning_rate": 0.00014895091661370363,
+      "loss": 2.263,
+      "step": 419
+    },
+    {
+      "epoch": 0.18115160664222557,
+      "grad_norm": 0.16968831419944763,
+      "learning_rate": 0.00014894522074689845,
+      "loss": 2.4698,
+      "step": 420
+    },
+    {
+      "epoch": 0.18158291999137374,
+      "grad_norm": 0.1799907386302948,
+      "learning_rate": 0.0001489395095689267,
+      "loss": 2.2043,
+      "step": 421
+    },
+    {
+      "epoch": 0.18201423334052189,
+      "grad_norm": 0.14666646718978882,
+      "learning_rate": 0.00014893378308097088,
+      "loss": 2.1512,
+      "step": 422
+    },
+    {
+      "epoch": 0.18244554668967006,
+      "grad_norm": 0.15499548614025116,
+      "learning_rate": 0.00014892804128421676,
+      "loss": 2.0666,
+      "step": 423
+    },
+    {
+      "epoch": 0.1828768600388182,
+      "grad_norm": 0.17115478217601776,
+      "learning_rate": 0.00014892228417985324,
+      "loss": 2.4236,
+      "step": 424
+    },
+    {
+      "epoch": 0.18330817338796634,
+      "grad_norm": 0.16812491416931152,
+      "learning_rate": 0.00014891651176907235,
+      "loss": 2.2477,
+      "step": 425
+    },
+    {
+      "epoch": 0.18330817338796634,
+      "eval_loss": 2.1428017616271973,
+      "eval_runtime": 202.8329,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 425
+    },
+    {
+      "epoch": 0.18373948673711452,
+      "grad_norm": 0.16948385536670685,
+      "learning_rate": 0.00014891072405306938,
+      "loss": 2.3145,
+      "step": 426
+    },
+    {
+      "epoch": 0.18417080008626266,
+      "grad_norm": 0.14548641443252563,
+      "learning_rate": 0.0001489049210330427,
+      "loss": 2.2195,
+      "step": 427
+    },
+    {
+      "epoch": 0.18460211343541083,
+      "grad_norm": 0.14978563785552979,
+      "learning_rate": 0.00014889910271019392,
+      "loss": 2.0593,
+      "step": 428
+    },
+    {
+      "epoch": 0.18503342678455897,
+      "grad_norm": 0.16009832918643951,
+      "learning_rate": 0.0001488932690857278,
+      "loss": 2.3423,
+      "step": 429
+    },
+    {
+      "epoch": 0.18546474013370715,
+      "grad_norm": 0.17507681250572205,
+      "learning_rate": 0.0001488874201608522,
+      "loss": 2.1819,
+      "step": 430
+    },
+    {
+      "epoch": 0.1858960534828553,
+      "grad_norm": 0.18338724970817566,
+      "learning_rate": 0.00014888155593677823,
+      "loss": 2.1853,
+      "step": 431
+    },
+    {
+      "epoch": 0.18632736683200346,
+      "grad_norm": 0.15442681312561035,
+      "learning_rate": 0.0001488756764147202,
+      "loss": 2.2035,
+      "step": 432
+    },
+    {
+      "epoch": 0.1867586801811516,
+      "grad_norm": 0.5245089530944824,
+      "learning_rate": 0.00014886978159589541,
+      "loss": 2.2897,
+      "step": 433
+    },
+    {
+      "epoch": 0.18718999353029978,
+      "grad_norm": 0.13658595085144043,
+      "learning_rate": 0.0001488638714815246,
+      "loss": 2.1846,
+      "step": 434
+    },
+    {
+      "epoch": 0.18762130687944792,
+      "grad_norm": 0.17327919602394104,
+      "learning_rate": 0.00014885794607283136,
+      "loss": 2.1647,
+      "step": 435
+    },
+    {
+      "epoch": 0.18805262022859606,
+      "grad_norm": 0.17164286971092224,
+      "learning_rate": 0.0001488520053710427,
+      "loss": 2.3012,
+      "step": 436
+    },
+    {
+      "epoch": 0.18848393357774423,
+      "grad_norm": 0.15758675336837769,
+      "learning_rate": 0.00014884604937738874,
+      "loss": 2.3051,
+      "step": 437
+    },
+    {
+      "epoch": 0.18891524692689238,
+      "grad_norm": 0.1563560664653778,
+      "learning_rate": 0.00014884007809310264,
+      "loss": 2.1007,
+      "step": 438
+    },
+    {
+      "epoch": 0.18934656027604055,
+      "grad_norm": 0.17045928537845612,
+      "learning_rate": 0.00014883409151942087,
+      "loss": 2.2063,
+      "step": 439
+    },
+    {
+      "epoch": 0.1897778736251887,
+      "grad_norm": 0.1447650045156479,
+      "learning_rate": 0.00014882808965758305,
+      "loss": 2.2638,
+      "step": 440
+    },
+    {
+      "epoch": 0.19020918697433686,
+      "grad_norm": 0.1615784913301468,
+      "learning_rate": 0.0001488220725088319,
+      "loss": 2.1472,
+      "step": 441
+    },
+    {
+      "epoch": 0.190640500323485,
+      "grad_norm": 0.16287849843502045,
+      "learning_rate": 0.00014881604007441332,
+      "loss": 1.9561,
+      "step": 442
+    },
+    {
+      "epoch": 0.19107181367263318,
+      "grad_norm": 0.2232157289981842,
+      "learning_rate": 0.0001488099923555764,
+      "loss": 2.3082,
+      "step": 443
+    },
+    {
+      "epoch": 0.19150312702178132,
+      "grad_norm": 0.17425796389579773,
+      "learning_rate": 0.00014880392935357338,
+      "loss": 2.3123,
+      "step": 444
+    },
+    {
+      "epoch": 0.1919344403709295,
+      "grad_norm": 0.1674915999174118,
+      "learning_rate": 0.00014879785106965967,
+      "loss": 2.1359,
+      "step": 445
+    },
+    {
+      "epoch": 0.19236575372007764,
+      "grad_norm": 0.16911302506923676,
+      "learning_rate": 0.0001487917575050939,
+      "loss": 2.124,
+      "step": 446
+    },
+    {
+      "epoch": 0.19279706706922578,
+      "grad_norm": 0.13820582628250122,
+      "learning_rate": 0.00014878564866113774,
+      "loss": 2.1094,
+      "step": 447
+    },
+    {
+      "epoch": 0.19322838041837395,
+      "grad_norm": 0.19064851105213165,
+      "learning_rate": 0.0001487795245390561,
+      "loss": 2.1212,
+      "step": 448
+    },
+    {
+      "epoch": 0.1936596937675221,
+      "grad_norm": 10.566106796264648,
+      "learning_rate": 0.0001487733851401171,
+      "loss": 2.2606,
+      "step": 449
+    },
+    {
+      "epoch": 0.19409100711667027,
+      "grad_norm": 0.17199164628982544,
+      "learning_rate": 0.0001487672304655919,
+      "loss": 2.3024,
+      "step": 450
+    },
+    {
+      "epoch": 0.19409100711667027,
+      "eval_loss": 2.142256259918213,
+      "eval_runtime": 198.5198,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 450
+    },
+    {
+      "epoch": 0.1945223204658184,
+      "grad_norm": 0.15751536190509796,
+      "learning_rate": 0.00014876106051675496,
+      "loss": 2.3373,
+      "step": 451
+    },
+    {
+      "epoch": 0.19495363381496658,
+      "grad_norm": 0.1832340806722641,
+      "learning_rate": 0.0001487548752948838,
+      "loss": 2.2288,
+      "step": 452
+    },
+    {
+      "epoch": 0.19538494716411473,
+      "grad_norm": 0.7646205425262451,
+      "learning_rate": 0.00014874867480125912,
+      "loss": 2.4942,
+      "step": 453
+    },
+    {
+      "epoch": 0.1958162605132629,
+      "grad_norm": 0.1642490178346634,
+      "learning_rate": 0.00014874245903716483,
+      "loss": 2.3353,
+      "step": 454
+    },
+    {
+      "epoch": 0.19624757386241104,
+      "grad_norm": 0.16240373253822327,
+      "learning_rate": 0.00014873622800388797,
+      "loss": 2.1596,
+      "step": 455
+    },
+    {
+      "epoch": 0.1966788872115592,
+      "grad_norm": 0.14881034195423126,
+      "learning_rate": 0.00014872998170271874,
+      "loss": 1.8476,
+      "step": 456
+    },
+    {
+      "epoch": 0.19711020056070735,
+      "grad_norm": 0.1751919984817505,
+      "learning_rate": 0.0001487237201349505,
+      "loss": 2.3051,
+      "step": 457
+    },
+    {
+      "epoch": 0.1975415139098555,
+      "grad_norm": 0.19177457690238953,
+      "learning_rate": 0.00014871744330187975,
+      "loss": 2.3034,
+      "step": 458
+    },
+    {
+      "epoch": 0.19797282725900367,
+      "grad_norm": 0.17044596374034882,
+      "learning_rate": 0.0001487111512048062,
+      "loss": 2.1733,
+      "step": 459
+    },
+    {
+      "epoch": 0.1984041406081518,
+      "grad_norm": 0.1570066660642624,
+      "learning_rate": 0.00014870484384503274,
+      "loss": 2.2867,
+      "step": 460
+    },
+    {
+      "epoch": 0.19883545395729998,
+      "grad_norm": 0.18342995643615723,
+      "learning_rate": 0.00014869852122386532,
+      "loss": 2.3922,
+      "step": 461
+    },
+    {
+      "epoch": 0.19926676730644813,
+      "grad_norm": 0.18954606354236603,
+      "learning_rate": 0.00014869218334261314,
+      "loss": 2.2386,
+      "step": 462
+    },
+    {
+      "epoch": 0.1996980806555963,
+      "grad_norm": 0.1563069224357605,
+      "learning_rate": 0.0001486858302025885,
+      "loss": 1.9606,
+      "step": 463
+    },
+    {
+      "epoch": 0.20012939400474444,
+      "grad_norm": 0.15879355370998383,
+      "learning_rate": 0.0001486794618051069,
+      "loss": 2.3816,
+      "step": 464
+    },
+    {
+      "epoch": 0.20056070735389261,
+      "grad_norm": 0.17720089852809906,
+      "learning_rate": 0.000148673078151487,
+      "loss": 2.3701,
+      "step": 465
+    },
+    {
+      "epoch": 0.20099202070304076,
+      "grad_norm": 0.1530313342809677,
+      "learning_rate": 0.0001486666792430506,
+      "loss": 2.2818,
+      "step": 466
+    },
+    {
+      "epoch": 0.2014233340521889,
+      "grad_norm": 0.1817789524793625,
+      "learning_rate": 0.00014866026508112264,
+      "loss": 2.333,
+      "step": 467
+    },
+    {
+      "epoch": 0.20185464740133707,
+      "grad_norm": 0.16977185010910034,
+      "learning_rate": 0.00014865383566703126,
+      "loss": 2.1365,
+      "step": 468
+    },
+    {
+      "epoch": 0.20228596075048522,
+      "grad_norm": 0.15945596992969513,
+      "learning_rate": 0.00014864739100210774,
+      "loss": 2.2966,
+      "step": 469
+    },
+    {
+      "epoch": 0.2027172740996334,
+      "grad_norm": 0.1694299727678299,
+      "learning_rate": 0.00014864093108768654,
+      "loss": 2.1316,
+      "step": 470
+    },
+    {
+      "epoch": 0.20314858744878153,
+      "grad_norm": 0.15608814358711243,
+      "learning_rate": 0.0001486344559251052,
+      "loss": 2.1924,
+      "step": 471
+    },
+    {
+      "epoch": 0.2035799007979297,
+      "grad_norm": 0.15335038304328918,
+      "learning_rate": 0.00014862796551570455,
+      "loss": 2.1299,
+      "step": 472
+    },
+    {
+      "epoch": 0.20401121414707785,
+      "grad_norm": 0.1774802803993225,
+      "learning_rate": 0.00014862145986082844,
+      "loss": 2.0524,
+      "step": 473
+    },
+    {
+      "epoch": 0.20444252749622602,
+      "grad_norm": 0.1639878749847412,
+      "learning_rate": 0.00014861493896182394,
+      "loss": 2.1435,
+      "step": 474
+    },
+    {
+      "epoch": 0.20487384084537416,
+      "grad_norm": 0.17186541855335236,
+      "learning_rate": 0.0001486084028200413,
+      "loss": 2.2342,
+      "step": 475
+    },
+    {
+      "epoch": 0.20487384084537416,
+      "eval_loss": 2.142070770263672,
+      "eval_runtime": 198.599,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 475
+    },
+    {
+      "epoch": 0.20530515419452233,
+      "grad_norm": 0.17359185218811035,
+      "learning_rate": 0.00014860185143683387,
+      "loss": 2.23,
+      "step": 476
+    },
+    {
+      "epoch": 0.20573646754367048,
+      "grad_norm": 0.16410641372203827,
+      "learning_rate": 0.00014859528481355825,
+      "loss": 1.7696,
+      "step": 477
+    },
+    {
+      "epoch": 0.20616778089281862,
+      "grad_norm": 0.19071485102176666,
+      "learning_rate": 0.00014858870295157405,
+      "loss": 2.1906,
+      "step": 478
+    },
+    {
+      "epoch": 0.2065990942419668,
+      "grad_norm": 0.1821126639842987,
+      "learning_rate": 0.00014858210585224415,
+      "loss": 2.1561,
+      "step": 479
+    },
+    {
+      "epoch": 0.20703040759111493,
+      "grad_norm": 0.20578120648860931,
+      "learning_rate": 0.00014857549351693456,
+      "loss": 2.0733,
+      "step": 480
+    },
+    {
+      "epoch": 0.2074617209402631,
+      "grad_norm": 0.15326827764511108,
+      "learning_rate": 0.00014856886594701442,
+      "loss": 2.2958,
+      "step": 481
+    },
+    {
+      "epoch": 0.20789303428941125,
+      "grad_norm": 0.2035706788301468,
+      "learning_rate": 0.00014856222314385607,
+      "loss": 2.2846,
+      "step": 482
+    },
+    {
+      "epoch": 0.20832434763855942,
+      "grad_norm": 0.17650489509105682,
+      "learning_rate": 0.00014855556510883495,
+      "loss": 2.2103,
+      "step": 483
+    },
+    {
+      "epoch": 0.20875566098770756,
+      "grad_norm": 0.199966162443161,
+      "learning_rate": 0.00014854889184332968,
+      "loss": 1.9659,
+      "step": 484
+    },
+    {
+      "epoch": 0.20918697433685574,
+      "grad_norm": 0.17148557305335999,
+      "learning_rate": 0.00014854220334872205,
+      "loss": 2.2062,
+      "step": 485
+    },
+    {
+      "epoch": 0.20961828768600388,
+      "grad_norm": 0.5571771264076233,
+      "learning_rate": 0.00014853549962639694,
+      "loss": 2.191,
+      "step": 486
+    },
+    {
+      "epoch": 0.21004960103515205,
+      "grad_norm": 0.16572660207748413,
+      "learning_rate": 0.0001485287806777425,
+      "loss": 2.1958,
+      "step": 487
+    },
+    {
+      "epoch": 0.2104809143843002,
+      "grad_norm": 0.16806870698928833,
+      "learning_rate": 0.0001485220465041499,
+      "loss": 2.3831,
+      "step": 488
+    },
+    {
+      "epoch": 0.21091222773344834,
+      "grad_norm": 0.1517253816127777,
+      "learning_rate": 0.00014851529710701355,
+      "loss": 1.9694,
+      "step": 489
+    },
+    {
+      "epoch": 0.2113435410825965,
+      "grad_norm": 1.1633365154266357,
+      "learning_rate": 0.000148508532487731,
+      "loss": 2.1299,
+      "step": 490
+    },
+    {
+      "epoch": 0.21177485443174465,
+      "grad_norm": 0.16392192244529724,
+      "learning_rate": 0.0001485017526477029,
+      "loss": 2.1838,
+      "step": 491
+    },
+    {
+      "epoch": 0.21220616778089282,
+      "grad_norm": 0.16261541843414307,
+      "learning_rate": 0.0001484949575883331,
+      "loss": 2.3216,
+      "step": 492
+    },
+    {
+      "epoch": 0.21263748113004097,
+      "grad_norm": 0.18670570850372314,
+      "learning_rate": 0.00014848814731102864,
+      "loss": 2.195,
+      "step": 493
+    },
+    {
+      "epoch": 0.21306879447918914,
+      "grad_norm": 0.18141891062259674,
+      "learning_rate": 0.00014848132181719963,
+      "loss": 2.3045,
+      "step": 494
+    },
+    {
+      "epoch": 0.21350010782833728,
+      "grad_norm": 0.19345717132091522,
+      "learning_rate": 0.00014847448110825933,
+      "loss": 2.238,
+      "step": 495
+    },
+    {
+      "epoch": 0.21393142117748545,
+      "grad_norm": 0.1645248383283615,
+      "learning_rate": 0.00014846762518562422,
+      "loss": 1.9521,
+      "step": 496
+    },
+    {
+      "epoch": 0.2143627345266336,
+      "grad_norm": 0.18492425978183746,
+      "learning_rate": 0.0001484607540507139,
+      "loss": 2.3596,
+      "step": 497
+    },
+    {
+      "epoch": 0.21479404787578177,
+      "grad_norm": 0.2229699045419693,
+      "learning_rate": 0.00014845386770495107,
+      "loss": 2.2964,
+      "step": 498
+    },
+    {
+      "epoch": 0.2152253612249299,
+      "grad_norm": 0.161407932639122,
+      "learning_rate": 0.0001484469661497617,
+      "loss": 2.1899,
+      "step": 499
+    },
+    {
+      "epoch": 0.21565667457407806,
+      "grad_norm": 0.1566958725452423,
+      "learning_rate": 0.00014844004938657475,
+      "loss": 2.1988,
+      "step": 500
+    },
+    {
+      "epoch": 0.21565667457407806,
+      "eval_loss": 2.1414709091186523,
+      "eval_runtime": 195.9589,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 500
+    },
+    {
+      "epoch": 0.21608798792322623,
+      "grad_norm": 0.17611227929592133,
+      "learning_rate": 0.00014843311741682247,
+      "loss": 2.1853,
+      "step": 501
+    },
+    {
+      "epoch": 0.21651930127237437,
+      "grad_norm": 0.16318093240261078,
+      "learning_rate": 0.00014842617024194015,
+      "loss": 1.9843,
+      "step": 502
+    },
+    {
+      "epoch": 0.21695061462152254,
+      "grad_norm": 0.18686994910240173,
+      "learning_rate": 0.00014841920786336633,
+      "loss": 2.1425,
+      "step": 503
+    },
+    {
+      "epoch": 0.21738192797067069,
+      "grad_norm": 0.21011656522750854,
+      "learning_rate": 0.00014841223028254263,
+      "loss": 2.2464,
+      "step": 504
+    },
+    {
+      "epoch": 0.21781324131981886,
+      "grad_norm": 0.1605004221200943,
+      "learning_rate": 0.00014840523750091382,
+      "loss": 2.4065,
+      "step": 505
+    },
+    {
+      "epoch": 0.218244554668967,
+      "grad_norm": 0.16983674466609955,
+      "learning_rate": 0.00014839822951992783,
+      "loss": 2.25,
+      "step": 506
+    },
+    {
+      "epoch": 0.21867586801811517,
+      "grad_norm": 0.18663226068019867,
+      "learning_rate": 0.00014839120634103575,
+      "loss": 2.432,
+      "step": 507
+    },
+    {
+      "epoch": 0.21910718136726332,
+      "grad_norm": 0.17589536309242249,
+      "learning_rate": 0.0001483841679656918,
+      "loss": 2.3051,
+      "step": 508
+    },
+    {
+      "epoch": 0.2195384947164115,
+      "grad_norm": 0.162202388048172,
+      "learning_rate": 0.00014837711439535337,
+      "loss": 2.2593,
+      "step": 509
+    },
+    {
+      "epoch": 0.21996980806555963,
+      "grad_norm": 0.17092770338058472,
+      "learning_rate": 0.00014837004563148095,
+      "loss": 2.1711,
+      "step": 510
+    },
+    {
+      "epoch": 0.22040112141470777,
+      "grad_norm": 0.17335964739322662,
+      "learning_rate": 0.00014836296167553825,
+      "loss": 2.2748,
+      "step": 511
+    },
+    {
+      "epoch": 0.22083243476385594,
+      "grad_norm": 0.15895693004131317,
+      "learning_rate": 0.00014835586252899202,
+      "loss": 2.2698,
+      "step": 512
+    },
+    {
+      "epoch": 0.2212637481130041,
+      "grad_norm": 0.17679838836193085,
+      "learning_rate": 0.00014834874819331226,
+      "loss": 2.3605,
+      "step": 513
+    },
+    {
+      "epoch": 0.22169506146215226,
+      "grad_norm": 13.117202758789062,
+      "learning_rate": 0.00014834161866997207,
+      "loss": 2.3758,
+      "step": 514
+    },
+    {
+      "epoch": 0.2221263748113004,
+      "grad_norm": 0.15782105922698975,
+      "learning_rate": 0.00014833447396044768,
+      "loss": 2.0502,
+      "step": 515
+    },
+    {
+      "epoch": 0.22255768816044857,
+      "grad_norm": 0.19260673224925995,
+      "learning_rate": 0.00014832731406621844,
+      "loss": 2.3573,
+      "step": 516
+    },
+    {
+      "epoch": 0.22298900150959672,
+      "grad_norm": 0.16335803270339966,
+      "learning_rate": 0.000148320138988767,
+      "loss": 2.4363,
+      "step": 517
+    },
+    {
+      "epoch": 0.2234203148587449,
+      "grad_norm": 0.16421473026275635,
+      "learning_rate": 0.00014831294872957892,
+      "loss": 2.3026,
+      "step": 518
+    },
+    {
+      "epoch": 0.22385162820789303,
+      "grad_norm": 0.1660144180059433,
+      "learning_rate": 0.00014830574329014308,
+      "loss": 2.2679,
+      "step": 519
+    },
+    {
+      "epoch": 0.22428294155704118,
+      "grad_norm": 0.17254631221294403,
+      "learning_rate": 0.00014829852267195142,
+      "loss": 2.1773,
+      "step": 520
+    },
+    {
+      "epoch": 0.22471425490618935,
+      "grad_norm": 0.19264696538448334,
+      "learning_rate": 0.00014829128687649908,
+      "loss": 1.8575,
+      "step": 521
+    },
+    {
+      "epoch": 0.2251455682553375,
+      "grad_norm": 0.16468121111392975,
+      "learning_rate": 0.0001482840359052843,
+      "loss": 2.2782,
+      "step": 522
+    },
+    {
+      "epoch": 0.22557688160448566,
+      "grad_norm": 0.17862504720687866,
+      "learning_rate": 0.00014827676975980844,
+      "loss": 2.2626,
+      "step": 523
+    },
+    {
+      "epoch": 0.2260081949536338,
+      "grad_norm": 0.16050106287002563,
+      "learning_rate": 0.00014826948844157608,
+      "loss": 2.1099,
+      "step": 524
+    },
+    {
+      "epoch": 0.22643950830278198,
+      "grad_norm": 0.17345581948757172,
+      "learning_rate": 0.00014826219195209487,
+      "loss": 2.2409,
+      "step": 525
+    },
+    {
+      "epoch": 0.22643950830278198,
+      "eval_loss": 2.1415786743164062,
+      "eval_runtime": 195.7665,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 525
+    },
+    {
+      "epoch": 0.22687082165193012,
+      "grad_norm": 0.15715128183364868,
+      "learning_rate": 0.00014825488029287563,
+      "loss": 2.2605,
+      "step": 526
+    },
+    {
+      "epoch": 0.2273021350010783,
+      "grad_norm": 0.16525809466838837,
+      "learning_rate": 0.00014824755346543234,
+      "loss": 2.3405,
+      "step": 527
+    },
+    {
+      "epoch": 0.22773344835022644,
+      "grad_norm": 0.17459775507450104,
+      "learning_rate": 0.00014824021147128206,
+      "loss": 2.3439,
+      "step": 528
+    },
+    {
+      "epoch": 0.2281647616993746,
+      "grad_norm": 0.14913460612297058,
+      "learning_rate": 0.0001482328543119451,
+      "loss": 2.2562,
+      "step": 529
+    },
+    {
+      "epoch": 0.22859607504852275,
+      "grad_norm": 0.15568678081035614,
+      "learning_rate": 0.00014822548198894473,
+      "loss": 2.3318,
+      "step": 530
+    },
+    {
+      "epoch": 0.2290273883976709,
+      "grad_norm": 0.21238815784454346,
+      "learning_rate": 0.00014821809450380754,
+      "loss": 2.4029,
+      "step": 531
+    },
+    {
+      "epoch": 0.22945870174681907,
+      "grad_norm": 0.16378119587898254,
+      "learning_rate": 0.00014821069185806324,
+      "loss": 2.2611,
+      "step": 532
+    },
+    {
+      "epoch": 0.2298900150959672,
+      "grad_norm": 0.1750461906194687,
+      "learning_rate": 0.00014820327405324452,
+      "loss": 2.3605,
+      "step": 533
+    },
+    {
+      "epoch": 0.23032132844511538,
+      "grad_norm": 0.17745572328567505,
+      "learning_rate": 0.0001481958410908874,
+      "loss": 2.3389,
+      "step": 534
+    },
+    {
+      "epoch": 0.23075264179426352,
+      "grad_norm": 0.18830518424510956,
+      "learning_rate": 0.0001481883929725309,
+      "loss": 2.4119,
+      "step": 535
+    },
+    {
+      "epoch": 0.2311839551434117,
+      "grad_norm": 0.18070408701896667,
+      "learning_rate": 0.00014818092969971728,
+      "loss": 2.1729,
+      "step": 536
+    },
+    {
+      "epoch": 0.23161526849255984,
+      "grad_norm": 0.15680450201034546,
+      "learning_rate": 0.00014817345127399187,
+      "loss": 2.3044,
+      "step": 537
+    },
+    {
+      "epoch": 0.232046581841708,
+      "grad_norm": 0.1724395602941513,
+      "learning_rate": 0.00014816595769690316,
+      "loss": 2.202,
+      "step": 538
+    },
+    {
+      "epoch": 0.23247789519085615,
+      "grad_norm": 0.14011593163013458,
+      "learning_rate": 0.0001481584489700028,
+      "loss": 2.1531,
+      "step": 539
+    },
+    {
+      "epoch": 0.23290920854000433,
+      "grad_norm": 0.2459859997034073,
+      "learning_rate": 0.00014815092509484548,
+      "loss": 2.2046,
+      "step": 540
+    },
+    {
+      "epoch": 0.23334052188915247,
+      "grad_norm": 0.17075932025909424,
+      "learning_rate": 0.0001481433860729892,
+      "loss": 2.1311,
+      "step": 541
+    },
+    {
+      "epoch": 0.2337718352383006,
+      "grad_norm": 0.1609731912612915,
+      "learning_rate": 0.00014813583190599491,
+      "loss": 2.2081,
+      "step": 542
+    },
+    {
+      "epoch": 0.23420314858744878,
+      "grad_norm": 0.16049246490001678,
+      "learning_rate": 0.00014812826259542688,
+      "loss": 2.256,
+      "step": 543
+    },
+    {
+      "epoch": 0.23463446193659693,
+      "grad_norm": 0.1557115763425827,
+      "learning_rate": 0.00014812067814285232,
+      "loss": 2.2185,
+      "step": 544
+    },
+    {
+      "epoch": 0.2350657752857451,
+      "grad_norm": 0.15727244317531586,
+      "learning_rate": 0.00014811307854984172,
+      "loss": 2.3666,
+      "step": 545
+    },
+    {
+      "epoch": 0.23549708863489324,
+      "grad_norm": 0.7236835360527039,
+      "learning_rate": 0.00014810546381796867,
+      "loss": 2.2764,
+      "step": 546
+    },
+    {
+      "epoch": 0.2359284019840414,
+      "grad_norm": 0.16988235712051392,
+      "learning_rate": 0.00014809783394880986,
+      "loss": 2.138,
+      "step": 547
+    },
+    {
+      "epoch": 0.23635971533318956,
+      "grad_norm": 0.1607985943555832,
+      "learning_rate": 0.00014809018894394515,
+      "loss": 2.0603,
+      "step": 548
+    },
+    {
+      "epoch": 0.23679102868233773,
+      "grad_norm": 0.15483079850673676,
+      "learning_rate": 0.00014808252880495754,
+      "loss": 2.2166,
+      "step": 549
+    },
+    {
+      "epoch": 0.23722234203148587,
+      "grad_norm": 0.16665418446063995,
+      "learning_rate": 0.00014807485353343308,
+      "loss": 2.4235,
+      "step": 550
+    },
+    {
+      "epoch": 0.23722234203148587,
+      "eval_loss": 2.139251708984375,
+      "eval_runtime": 195.574,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 550
+    },
+    {
+      "epoch": 0.23765365538063404,
+      "grad_norm": 0.15788012742996216,
+      "learning_rate": 0.0001480671631309611,
+      "loss": 2.2751,
+      "step": 551
+    },
+    {
+      "epoch": 0.2380849687297822,
+      "grad_norm": 0.16924196481704712,
+      "learning_rate": 0.00014805945759913393,
+      "loss": 2.152,
+      "step": 552
+    },
+    {
+      "epoch": 0.23851628207893033,
+      "grad_norm": 0.16306330263614655,
+      "learning_rate": 0.0001480517369395471,
+      "loss": 2.2638,
+      "step": 553
+    },
+    {
+      "epoch": 0.2389475954280785,
+      "grad_norm": 0.20340776443481445,
+      "learning_rate": 0.00014804400115379926,
+      "loss": 2.4568,
+      "step": 554
+    },
+    {
+      "epoch": 0.23937890877722665,
+      "grad_norm": 0.15684518218040466,
+      "learning_rate": 0.00014803625024349218,
+      "loss": 2.3376,
+      "step": 555
+    },
+    {
+      "epoch": 0.23981022212637482,
+      "grad_norm": 0.17012250423431396,
+      "learning_rate": 0.00014802848421023078,
+      "loss": 2.3626,
+      "step": 556
+    },
+    {
+      "epoch": 0.24024153547552296,
+      "grad_norm": 0.150764599442482,
+      "learning_rate": 0.0001480207030556231,
+      "loss": 2.2943,
+      "step": 557
+    },
+    {
+      "epoch": 0.24067284882467113,
+      "grad_norm": 0.17265529930591583,
+      "learning_rate": 0.0001480129067812803,
+      "loss": 2.2501,
+      "step": 558
+    },
+    {
+      "epoch": 0.24110416217381928,
+      "grad_norm": 0.18661445379257202,
+      "learning_rate": 0.00014800509538881668,
+      "loss": 2.4074,
+      "step": 559
+    },
+    {
+      "epoch": 0.24153547552296745,
+      "grad_norm": 0.16257712244987488,
+      "learning_rate": 0.00014799726887984973,
+      "loss": 2.2272,
+      "step": 560
+    },
+    {
+      "epoch": 0.2419667888721156,
+      "grad_norm": 0.16717654466629028,
+      "learning_rate": 0.00014798942725599996,
+      "loss": 2.2114,
+      "step": 561
+    },
+    {
+      "epoch": 0.24239810222126376,
+      "grad_norm": 0.17024226486682892,
+      "learning_rate": 0.00014798157051889107,
+      "loss": 2.2616,
+      "step": 562
+    },
+    {
+      "epoch": 0.2428294155704119,
+      "grad_norm": 0.18863873183727264,
+      "learning_rate": 0.0001479736986701499,
+      "loss": 2.3047,
+      "step": 563
+    },
+    {
+      "epoch": 0.24326072891956005,
+      "grad_norm": 0.15407370030879974,
+      "learning_rate": 0.00014796581171140638,
+      "loss": 2.3092,
+      "step": 564
+    },
+    {
+      "epoch": 0.24369204226870822,
+      "grad_norm": 0.1805611550807953,
+      "learning_rate": 0.0001479579096442936,
+      "loss": 2.4571,
+      "step": 565
+    },
+    {
+      "epoch": 0.24412335561785636,
+      "grad_norm": 0.16363826394081116,
+      "learning_rate": 0.0001479499924704478,
+      "loss": 2.3163,
+      "step": 566
+    },
+    {
+      "epoch": 0.24455466896700453,
+      "grad_norm": 0.15935292840003967,
+      "learning_rate": 0.00014794206019150827,
+      "loss": 2.0979,
+      "step": 567
+    },
+    {
+      "epoch": 0.24498598231615268,
+      "grad_norm": 0.18893010914325714,
+      "learning_rate": 0.0001479341128091175,
+      "loss": 2.1063,
+      "step": 568
+    },
+    {
+      "epoch": 0.24541729566530085,
+      "grad_norm": 0.16460704803466797,
+      "learning_rate": 0.0001479261503249211,
+      "loss": 2.2173,
+      "step": 569
+    },
+    {
+      "epoch": 0.245848609014449,
+      "grad_norm": 0.1792575865983963,
+      "learning_rate": 0.00014791817274056776,
+      "loss": 2.2617,
+      "step": 570
+    },
+    {
+      "epoch": 0.24627992236359716,
+      "grad_norm": 0.2035822719335556,
+      "learning_rate": 0.00014791018005770931,
+      "loss": 2.3372,
+      "step": 571
+    },
+    {
+      "epoch": 0.2467112357127453,
+      "grad_norm": 0.17065055668354034,
+      "learning_rate": 0.00014790217227800075,
+      "loss": 2.1626,
+      "step": 572
+    },
+    {
+      "epoch": 0.24714254906189345,
+      "grad_norm": 0.21874402463436127,
+      "learning_rate": 0.0001478941494031002,
+      "loss": 2.2004,
+      "step": 573
+    },
+    {
+      "epoch": 0.24757386241104162,
+      "grad_norm": 0.190892294049263,
+      "learning_rate": 0.00014788611143466885,
+      "loss": 2.3801,
+      "step": 574
+    },
+    {
+      "epoch": 0.24800517576018977,
+      "grad_norm": 0.17411410808563232,
+      "learning_rate": 0.00014787805837437103,
+      "loss": 2.3002,
+      "step": 575
+    },
+    {
+      "epoch": 0.24800517576018977,
+      "eval_loss": 2.138429641723633,
+      "eval_runtime": 196.0983,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 575
+    },
+    {
+      "epoch": 0.24843648910933794,
+      "grad_norm": 0.15937082469463348,
+      "learning_rate": 0.00014786999022387427,
+      "loss": 2.1931,
+      "step": 576
+    },
+    {
+      "epoch": 0.24886780245848608,
+      "grad_norm": 0.3105604648590088,
+      "learning_rate": 0.00014786190698484915,
+      "loss": 2.1752,
+      "step": 577
+    },
+    {
+      "epoch": 0.24929911580763425,
+      "grad_norm": 0.17199891805648804,
+      "learning_rate": 0.00014785380865896939,
+      "loss": 2.1575,
+      "step": 578
+    },
+    {
+      "epoch": 0.2497304291567824,
+      "grad_norm": 0.15182305872440338,
+      "learning_rate": 0.0001478456952479118,
+      "loss": 2.1933,
+      "step": 579
+    },
+    {
+      "epoch": 0.25016174250593054,
+      "grad_norm": 0.20546135306358337,
+      "learning_rate": 0.0001478375667533564,
+      "loss": 2.2331,
+      "step": 580
+    },
+    {
+      "epoch": 0.25059305585507874,
+      "grad_norm": 0.1734628528356552,
+      "learning_rate": 0.00014782942317698627,
+      "loss": 1.8555,
+      "step": 581
+    },
+    {
+      "epoch": 0.2510243692042269,
+      "grad_norm": 0.22307294607162476,
+      "learning_rate": 0.00014782126452048762,
+      "loss": 2.1859,
+      "step": 582
+    },
+    {
+      "epoch": 0.251455682553375,
+      "grad_norm": 0.1479979008436203,
+      "learning_rate": 0.0001478130907855498,
+      "loss": 2.0667,
+      "step": 583
+    },
+    {
+      "epoch": 0.25188699590252317,
+      "grad_norm": 0.15659348666667938,
+      "learning_rate": 0.00014780490197386522,
+      "loss": 2.2639,
+      "step": 584
+    },
+    {
+      "epoch": 0.2523183092516713,
+      "grad_norm": 0.16493523120880127,
+      "learning_rate": 0.00014779669808712953,
+      "loss": 2.2895,
+      "step": 585
+    },
+    {
+      "epoch": 0.2527496226008195,
+      "grad_norm": 0.14518970251083374,
+      "learning_rate": 0.00014778847912704142,
+      "loss": 2.1188,
+      "step": 586
+    },
+    {
+      "epoch": 0.25318093594996766,
+      "grad_norm": 0.15645526349544525,
+      "learning_rate": 0.0001477802450953027,
+      "loss": 2.4878,
+      "step": 587
+    },
+    {
+      "epoch": 0.2536122492991158,
+      "grad_norm": 0.1723669171333313,
+      "learning_rate": 0.00014777199599361833,
+      "loss": 2.164,
+      "step": 588
+    },
+    {
+      "epoch": 0.25404356264826394,
+      "grad_norm": 0.17445899546146393,
+      "learning_rate": 0.00014776373182369634,
+      "loss": 2.2338,
+      "step": 589
+    },
+    {
+      "epoch": 0.25447487599741214,
+      "grad_norm": 0.15474815666675568,
+      "learning_rate": 0.000147755452587248,
+      "loss": 2.2222,
+      "step": 590
+    },
+    {
+      "epoch": 0.2549061893465603,
+      "grad_norm": 0.16550877690315247,
+      "learning_rate": 0.0001477471582859875,
+      "loss": 2.199,
+      "step": 591
+    },
+    {
+      "epoch": 0.25533750269570843,
+      "grad_norm": 1.0499933958053589,
+      "learning_rate": 0.00014773884892163236,
+      "loss": 2.1105,
+      "step": 592
+    },
+    {
+      "epoch": 0.2557688160448566,
+      "grad_norm": 0.7829015254974365,
+      "learning_rate": 0.00014773052449590309,
+      "loss": 1.9241,
+      "step": 593
+    },
+    {
+      "epoch": 0.25620012939400477,
+      "grad_norm": 0.17413672804832458,
+      "learning_rate": 0.00014772218501052335,
+      "loss": 2.3395,
+      "step": 594
+    },
+    {
+      "epoch": 0.2566314427431529,
+      "grad_norm": 0.1572626531124115,
+      "learning_rate": 0.00014771383046721992,
+      "loss": 2.1767,
+      "step": 595
+    },
+    {
+      "epoch": 0.25706275609230106,
+      "grad_norm": 0.17021185159683228,
+      "learning_rate": 0.0001477054608677227,
+      "loss": 2.2199,
+      "step": 596
+    },
+    {
+      "epoch": 0.2574940694414492,
+      "grad_norm": 0.1993943303823471,
+      "learning_rate": 0.00014769707621376473,
+      "loss": 2.3534,
+      "step": 597
+    },
+    {
+      "epoch": 0.25792538279059735,
+      "grad_norm": 0.17457185685634613,
+      "learning_rate": 0.00014768867650708214,
+      "loss": 2.1193,
+      "step": 598
+    },
+    {
+      "epoch": 0.25835669613974555,
+      "grad_norm": 0.16530904173851013,
+      "learning_rate": 0.00014768026174941416,
+      "loss": 2.2143,
+      "step": 599
+    },
+    {
+      "epoch": 0.2587880094888937,
+      "grad_norm": 0.16474269330501556,
+      "learning_rate": 0.00014767183194250316,
+      "loss": 2.2017,
+      "step": 600
+    },
+    {
+      "epoch": 0.2587880094888937,
+      "eval_loss": 2.1378190517425537,
+      "eval_runtime": 196.4733,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 600
+    },
+    {
+      "epoch": 0.25921932283804183,
+      "grad_norm": 0.1671011596918106,
+      "learning_rate": 0.00014766338708809465,
+      "loss": 2.0947,
+      "step": 601
+    },
+    {
+      "epoch": 0.25965063618719,
+      "grad_norm": 0.15906308591365814,
+      "learning_rate": 0.00014765492718793718,
+      "loss": 2.128,
+      "step": 602
+    },
+    {
+      "epoch": 0.2600819495363382,
+      "grad_norm": 0.15791428089141846,
+      "learning_rate": 0.00014764645224378252,
+      "loss": 2.1973,
+      "step": 603
+    },
+    {
+      "epoch": 0.2605132628854863,
+      "grad_norm": 0.2012777179479599,
+      "learning_rate": 0.00014763796225738542,
+      "loss": 2.2386,
+      "step": 604
+    },
+    {
+      "epoch": 0.26094457623463446,
+      "grad_norm": 0.18990804255008698,
+      "learning_rate": 0.00014762945723050395,
+      "loss": 2.4138,
+      "step": 605
+    },
+    {
+      "epoch": 0.2613758895837826,
+      "grad_norm": 0.17072105407714844,
+      "learning_rate": 0.00014762093716489907,
+      "loss": 2.1979,
+      "step": 606
+    },
+    {
+      "epoch": 0.26180720293293075,
+      "grad_norm": 0.1727512776851654,
+      "learning_rate": 0.00014761240206233499,
+      "loss": 2.3602,
+      "step": 607
+    },
+    {
+      "epoch": 0.26223851628207895,
+      "grad_norm": 0.1772853136062622,
+      "learning_rate": 0.00014760385192457898,
+      "loss": 2.2454,
+      "step": 608
+    },
+    {
+      "epoch": 0.2626698296312271,
+      "grad_norm": 0.91371750831604,
+      "learning_rate": 0.00014759528675340145,
+      "loss": 2.3401,
+      "step": 609
+    },
+    {
+      "epoch": 0.26310114298037524,
+      "grad_norm": 0.1926833838224411,
+      "learning_rate": 0.0001475867065505759,
+      "loss": 2.3298,
+      "step": 610
+    },
+    {
+      "epoch": 0.2635324563295234,
+      "grad_norm": 0.17731206119060516,
+      "learning_rate": 0.00014757811131787895,
+      "loss": 2.3576,
+      "step": 611
+    },
+    {
+      "epoch": 0.2639637696786716,
+      "grad_norm": 0.1678379476070404,
+      "learning_rate": 0.00014756950105709036,
+      "loss": 2.1995,
+      "step": 612
+    },
+    {
+      "epoch": 0.2643950830278197,
+      "grad_norm": 0.1588241308927536,
+      "learning_rate": 0.000147560875769993,
+      "loss": 2.3786,
+      "step": 613
+    },
+    {
+      "epoch": 0.26482639637696787,
+      "grad_norm": 0.16079789400100708,
+      "learning_rate": 0.00014755223545837278,
+      "loss": 2.3308,
+      "step": 614
+    },
+    {
+      "epoch": 0.265257709726116,
+      "grad_norm": 0.18283496797084808,
+      "learning_rate": 0.0001475435801240188,
+      "loss": 2.2975,
+      "step": 615
+    },
+    {
+      "epoch": 0.26568902307526415,
+      "grad_norm": 0.18275459110736847,
+      "learning_rate": 0.00014753490976872322,
+      "loss": 2.4674,
+      "step": 616
+    },
+    {
+      "epoch": 0.26612033642441235,
+      "grad_norm": 0.18903575837612152,
+      "learning_rate": 0.00014752622439428132,
+      "loss": 2.2237,
+      "step": 617
+    },
+    {
+      "epoch": 0.2665516497735605,
+      "grad_norm": 0.1476614773273468,
+      "learning_rate": 0.00014751752400249155,
+      "loss": 2.0617,
+      "step": 618
+    },
+    {
+      "epoch": 0.26698296312270864,
+      "grad_norm": 0.18297837674617767,
+      "learning_rate": 0.0001475088085951554,
+      "loss": 2.163,
+      "step": 619
+    },
+    {
+      "epoch": 0.2674142764718568,
+      "grad_norm": 0.18463227152824402,
+      "learning_rate": 0.0001475000781740775,
+      "loss": 2.4564,
+      "step": 620
+    },
+    {
+      "epoch": 0.267845589821005,
+      "grad_norm": 0.18771764636039734,
+      "learning_rate": 0.00014749133274106555,
+      "loss": 2.13,
+      "step": 621
+    },
+    {
+      "epoch": 0.2682769031701531,
+      "grad_norm": 0.15326857566833496,
+      "learning_rate": 0.0001474825722979304,
+      "loss": 2.3268,
+      "step": 622
+    },
+    {
+      "epoch": 0.26870821651930127,
+      "grad_norm": 0.16682656109333038,
+      "learning_rate": 0.000147473796846486,
+      "loss": 2.396,
+      "step": 623
+    },
+    {
+      "epoch": 0.2691395298684494,
+      "grad_norm": 0.21941545605659485,
+      "learning_rate": 0.0001474650063885494,
+      "loss": 2.206,
+      "step": 624
+    },
+    {
+      "epoch": 0.2695708432175976,
+      "grad_norm": 0.15367084741592407,
+      "learning_rate": 0.00014745620092594078,
+      "loss": 2.2849,
+      "step": 625
+    },
+    {
+      "epoch": 0.2695708432175976,
+      "eval_loss": 2.136373519897461,
+      "eval_runtime": 196.1623,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 625
+    },
+    {
+      "epoch": 0.27000215656674575,
+      "grad_norm": 0.1778152734041214,
+      "learning_rate": 0.0001474473804604834,
+      "loss": 2.2033,
+      "step": 626
+    },
+    {
+      "epoch": 0.2704334699158939,
+      "grad_norm": 0.16819840669631958,
+      "learning_rate": 0.00014743854499400357,
+      "loss": 2.1546,
+      "step": 627
+    },
+    {
+      "epoch": 0.27086478326504204,
+      "grad_norm": 0.191078320145607,
+      "learning_rate": 0.00014742969452833087,
+      "loss": 2.2636,
+      "step": 628
+    },
+    {
+      "epoch": 0.2712960966141902,
+      "grad_norm": 0.18553073704242706,
+      "learning_rate": 0.00014742082906529785,
+      "loss": 2.2205,
+      "step": 629
+    },
+    {
+      "epoch": 0.2717274099633384,
+      "grad_norm": 0.15881304442882538,
+      "learning_rate": 0.00014741194860674016,
+      "loss": 2.2768,
+      "step": 630
+    },
+    {
+      "epoch": 0.27215872331248653,
+      "grad_norm": 0.1653841882944107,
+      "learning_rate": 0.00014740305315449664,
+      "loss": 2.1803,
+      "step": 631
+    },
+    {
+      "epoch": 0.27259003666163467,
+      "grad_norm": 0.20279088616371155,
+      "learning_rate": 0.00014739414271040918,
+      "loss": 2.3493,
+      "step": 632
+    },
+    {
+      "epoch": 0.2730213500107828,
+      "grad_norm": 0.16524416208267212,
+      "learning_rate": 0.0001473852172763228,
+      "loss": 2.2841,
+      "step": 633
+    },
+    {
+      "epoch": 0.273452663359931,
+      "grad_norm": 0.1860021948814392,
+      "learning_rate": 0.00014737627685408556,
+      "loss": 2.178,
+      "step": 634
+    },
+    {
+      "epoch": 0.27388397670907916,
+      "grad_norm": 0.16222679615020752,
+      "learning_rate": 0.00014736732144554873,
+      "loss": 2.2774,
+      "step": 635
+    },
+    {
+      "epoch": 0.2743152900582273,
+      "grad_norm": 0.1624583899974823,
+      "learning_rate": 0.00014735835105256657,
+      "loss": 2.329,
+      "step": 636
+    },
+    {
+      "epoch": 0.27474660340737544,
+      "grad_norm": 0.17073509097099304,
+      "learning_rate": 0.00014734936567699657,
+      "loss": 2.1383,
+      "step": 637
+    },
+    {
+      "epoch": 0.2751779167565236,
+      "grad_norm": 0.15490107238292694,
+      "learning_rate": 0.00014734036532069916,
+      "loss": 2.1866,
+      "step": 638
+    },
+    {
+      "epoch": 0.2756092301056718,
+      "grad_norm": 0.17504987120628357,
+      "learning_rate": 0.00014733134998553803,
+      "loss": 2.3105,
+      "step": 639
+    },
+    {
+      "epoch": 0.27604054345481993,
+      "grad_norm": 0.16620095074176788,
+      "learning_rate": 0.00014732231967337989,
+      "loss": 2.1842,
+      "step": 640
+    },
+    {
+      "epoch": 0.2764718568039681,
+      "grad_norm": 0.15766702592372894,
+      "learning_rate": 0.00014731327438609454,
+      "loss": 2.2546,
+      "step": 641
+    },
+    {
+      "epoch": 0.2769031701531162,
+      "grad_norm": 0.18117624521255493,
+      "learning_rate": 0.00014730421412555492,
+      "loss": 2.2438,
+      "step": 642
+    },
+    {
+      "epoch": 0.2773344835022644,
+      "grad_norm": 0.1685836762189865,
+      "learning_rate": 0.00014729513889363708,
+      "loss": 2.1841,
+      "step": 643
+    },
+    {
+      "epoch": 0.27776579685141256,
+      "grad_norm": 0.1785241663455963,
+      "learning_rate": 0.00014728604869222012,
+      "loss": 2.3096,
+      "step": 644
+    },
+    {
+      "epoch": 0.2781971102005607,
+      "grad_norm": 0.1662873923778534,
+      "learning_rate": 0.00014727694352318626,
+      "loss": 2.3142,
+      "step": 645
+    },
+    {
+      "epoch": 0.27862842354970885,
+      "grad_norm": 0.15653882920742035,
+      "learning_rate": 0.00014726782338842087,
+      "loss": 2.1199,
+      "step": 646
+    },
+    {
+      "epoch": 0.27905973689885705,
+      "grad_norm": 0.19263307750225067,
+      "learning_rate": 0.00014725868828981232,
+      "loss": 2.3715,
+      "step": 647
+    },
+    {
+      "epoch": 0.2794910502480052,
+      "grad_norm": 0.18505576252937317,
+      "learning_rate": 0.00014724953822925214,
+      "loss": 2.3807,
+      "step": 648
+    },
+    {
+      "epoch": 0.27992236359715333,
+      "grad_norm": 0.16494807600975037,
+      "learning_rate": 0.00014724037320863498,
+      "loss": 2.3978,
+      "step": 649
+    },
+    {
+      "epoch": 0.2803536769463015,
+      "grad_norm": 0.16470667719841003,
+      "learning_rate": 0.00014723119322985852,
+      "loss": 2.4136,
+      "step": 650
+    },
+    {
+      "epoch": 0.2803536769463015,
+      "eval_loss": 2.134341239929199,
+      "eval_runtime": 203.163,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 650
+    },
+    {
+      "epoch": 0.2807849902954496,
+      "grad_norm": 0.16153836250305176,
+      "learning_rate": 0.0001472219982948236,
+      "loss": 2.1043,
+      "step": 651
+    },
+    {
+      "epoch": 0.2812163036445978,
+      "grad_norm": 0.17460724711418152,
+      "learning_rate": 0.00014721278840543414,
+      "loss": 2.0528,
+      "step": 652
+    },
+    {
+      "epoch": 0.28164761699374596,
+      "grad_norm": 0.1704031229019165,
+      "learning_rate": 0.00014720356356359715,
+      "loss": 2.5944,
+      "step": 653
+    },
+    {
+      "epoch": 0.2820789303428941,
+      "grad_norm": 0.17145408689975739,
+      "learning_rate": 0.00014719432377122268,
+      "loss": 2.4141,
+      "step": 654
+    },
+    {
+      "epoch": 0.28251024369204225,
+      "grad_norm": 0.16281622648239136,
+      "learning_rate": 0.00014718506903022398,
+      "loss": 2.4925,
+      "step": 655
+    },
+    {
+      "epoch": 0.28294155704119045,
+      "grad_norm": 0.17861981689929962,
+      "learning_rate": 0.00014717579934251735,
+      "loss": 2.1748,
+      "step": 656
+    },
+    {
+      "epoch": 0.2833728703903386,
+      "grad_norm": 0.15419550240039825,
+      "learning_rate": 0.00014716651471002214,
+      "loss": 2.0366,
+      "step": 657
+    },
+    {
+      "epoch": 0.28380418373948674,
+      "grad_norm": 0.18950116634368896,
+      "learning_rate": 0.0001471572151346609,
+      "loss": 2.0928,
+      "step": 658
+    },
+    {
+      "epoch": 0.2842354970886349,
+      "grad_norm": 0.1749858558177948,
+      "learning_rate": 0.00014714790061835914,
+      "loss": 2.164,
+      "step": 659
+    },
+    {
+      "epoch": 0.284666810437783,
+      "grad_norm": 0.18648329377174377,
+      "learning_rate": 0.00014713857116304554,
+      "loss": 2.2707,
+      "step": 660
+    },
+    {
+      "epoch": 0.2850981237869312,
+      "grad_norm": 0.15878723561763763,
+      "learning_rate": 0.00014712922677065192,
+      "loss": 2.3001,
+      "step": 661
+    },
+    {
+      "epoch": 0.28552943713607937,
+      "grad_norm": 0.16697610914707184,
+      "learning_rate": 0.0001471198674431131,
+      "loss": 2.4064,
+      "step": 662
+    },
+    {
+      "epoch": 0.2859607504852275,
+      "grad_norm": 0.16395072638988495,
+      "learning_rate": 0.00014711049318236706,
+      "loss": 2.3017,
+      "step": 663
+    },
+    {
+      "epoch": 0.28639206383437565,
+      "grad_norm": 0.16236403584480286,
+      "learning_rate": 0.00014710110399035478,
+      "loss": 2.2722,
+      "step": 664
+    },
+    {
+      "epoch": 0.28682337718352385,
+      "grad_norm": 0.16682536900043488,
+      "learning_rate": 0.00014709169986902042,
+      "loss": 2.3292,
+      "step": 665
+    },
+    {
+      "epoch": 0.287254690532672,
+      "grad_norm": 0.16841153800487518,
+      "learning_rate": 0.00014708228082031127,
+      "loss": 2.3933,
+      "step": 666
+    },
+    {
+      "epoch": 0.28768600388182014,
+      "grad_norm": 0.16072210669517517,
+      "learning_rate": 0.00014707284684617756,
+      "loss": 2.2569,
+      "step": 667
+    },
+    {
+      "epoch": 0.2881173172309683,
+      "grad_norm": 0.15559151768684387,
+      "learning_rate": 0.00014706339794857276,
+      "loss": 2.3392,
+      "step": 668
+    },
+    {
+      "epoch": 0.2885486305801164,
+      "grad_norm": 0.159805566072464,
+      "learning_rate": 0.00014705393412945333,
+      "loss": 2.2788,
+      "step": 669
+    },
+    {
+      "epoch": 0.2889799439292646,
+      "grad_norm": 0.1495869755744934,
+      "learning_rate": 0.0001470444553907789,
+      "loss": 2.0518,
+      "step": 670
+    },
+    {
+      "epoch": 0.28941125727841277,
+      "grad_norm": 0.17200689017772675,
+      "learning_rate": 0.00014703496173451206,
+      "loss": 2.3155,
+      "step": 671
+    },
+    {
+      "epoch": 0.2898425706275609,
+      "grad_norm": 0.15233959257602692,
+      "learning_rate": 0.00014702545316261869,
+      "loss": 2.1359,
+      "step": 672
+    },
+    {
+      "epoch": 0.29027388397670906,
+      "grad_norm": 0.16399161517620087,
+      "learning_rate": 0.00014701592967706755,
+      "loss": 2.1428,
+      "step": 673
+    },
+    {
+      "epoch": 0.29070519732585726,
+      "grad_norm": 0.165985107421875,
+      "learning_rate": 0.00014700639127983062,
+      "loss": 2.1885,
+      "step": 674
+    },
+    {
+      "epoch": 0.2911365106750054,
+      "grad_norm": 0.16751165688037872,
+      "learning_rate": 0.00014699683797288294,
+      "loss": 2.3933,
+      "step": 675
+    },
+    {
+      "epoch": 0.2911365106750054,
+      "eval_loss": 2.132812023162842,
+      "eval_runtime": 200.6349,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 675
+    },
+    {
+      "epoch": 0.29156782402415354,
+      "grad_norm": 0.19334676861763,
+      "learning_rate": 0.0001469872697582026,
+      "loss": 2.2254,
+      "step": 676
+    },
+    {
+      "epoch": 0.2919991373733017,
+      "grad_norm": 0.15076959133148193,
+      "learning_rate": 0.0001469776866377708,
+      "loss": 2.3313,
+      "step": 677
+    },
+    {
+      "epoch": 0.2924304507224499,
+      "grad_norm": 0.17511612176895142,
+      "learning_rate": 0.0001469680886135719,
+      "loss": 2.4024,
+      "step": 678
+    },
+    {
+      "epoch": 0.29286176407159803,
+      "grad_norm": 0.1722462773323059,
+      "learning_rate": 0.00014695847568759317,
+      "loss": 2.2334,
+      "step": 679
+    },
+    {
+      "epoch": 0.2932930774207462,
+      "grad_norm": 0.15810130536556244,
+      "learning_rate": 0.00014694884786182517,
+      "loss": 2.405,
+      "step": 680
+    },
+    {
+      "epoch": 0.2937243907698943,
+      "grad_norm": 0.19121390581130981,
+      "learning_rate": 0.00014693920513826137,
+      "loss": 2.1286,
+      "step": 681
+    },
+    {
+      "epoch": 0.29415570411904246,
+      "grad_norm": 0.15669195353984833,
+      "learning_rate": 0.00014692954751889843,
+      "loss": 2.1573,
+      "step": 682
+    },
+    {
+      "epoch": 0.29458701746819066,
+      "grad_norm": 0.16549846529960632,
+      "learning_rate": 0.00014691987500573607,
+      "loss": 2.2062,
+      "step": 683
+    },
+    {
+      "epoch": 0.2950183308173388,
+      "grad_norm": 0.1586558222770691,
+      "learning_rate": 0.0001469101876007771,
+      "loss": 2.127,
+      "step": 684
+    },
+    {
+      "epoch": 0.29544964416648695,
+      "grad_norm": 0.16695398092269897,
+      "learning_rate": 0.0001469004853060274,
+      "loss": 2.4235,
+      "step": 685
+    },
+    {
+      "epoch": 0.2958809575156351,
+      "grad_norm": 0.17500896751880646,
+      "learning_rate": 0.0001468907681234959,
+      "loss": 2.3542,
+      "step": 686
+    },
+    {
+      "epoch": 0.2963122708647833,
+      "grad_norm": 0.1749137043952942,
+      "learning_rate": 0.0001468810360551947,
+      "loss": 2.1788,
+      "step": 687
+    },
+    {
+      "epoch": 0.29674358421393143,
+      "grad_norm": 0.15084341168403625,
+      "learning_rate": 0.0001468712891031389,
+      "loss": 2.2913,
+      "step": 688
+    },
+    {
+      "epoch": 0.2971748975630796,
+      "grad_norm": 0.1857728809118271,
+      "learning_rate": 0.0001468615272693467,
+      "loss": 2.3329,
+      "step": 689
+    },
+    {
+      "epoch": 0.2976062109122277,
+      "grad_norm": 0.1701570600271225,
+      "learning_rate": 0.00014685175055583944,
+      "loss": 2.0987,
+      "step": 690
+    },
+    {
+      "epoch": 0.29803752426137586,
+      "grad_norm": 0.18066200613975525,
+      "learning_rate": 0.00014684195896464146,
+      "loss": 2.199,
+      "step": 691
+    },
+    {
+      "epoch": 0.29846883761052406,
+      "grad_norm": 0.15877336263656616,
+      "learning_rate": 0.00014683215249778022,
+      "loss": 2.1216,
+      "step": 692
+    },
+    {
+      "epoch": 0.2989001509596722,
+      "grad_norm": 0.16353558003902435,
+      "learning_rate": 0.00014682233115728628,
+      "loss": 2.1863,
+      "step": 693
+    },
+    {
+      "epoch": 0.29933146430882035,
+      "grad_norm": 0.16829341650009155,
+      "learning_rate": 0.00014681249494519322,
+      "loss": 2.2876,
+      "step": 694
+    },
+    {
+      "epoch": 0.2997627776579685,
+      "grad_norm": 0.1578504890203476,
+      "learning_rate": 0.00014680264386353776,
+      "loss": 2.3461,
+      "step": 695
+    },
+    {
+      "epoch": 0.3001940910071167,
+      "grad_norm": 0.20184138417243958,
+      "learning_rate": 0.00014679277791435966,
+      "loss": 2.2431,
+      "step": 696
+    },
+    {
+      "epoch": 0.30062540435626484,
+      "grad_norm": 0.1779334396123886,
+      "learning_rate": 0.0001467828970997018,
+      "loss": 2.3081,
+      "step": 697
+    },
+    {
+      "epoch": 0.301056717705413,
+      "grad_norm": 0.17701883614063263,
+      "learning_rate": 0.00014677300142161006,
+      "loss": 2.2477,
+      "step": 698
+    },
+    {
+      "epoch": 0.3014880310545611,
+      "grad_norm": 0.17183318734169006,
+      "learning_rate": 0.00014676309088213353,
+      "loss": 2.0537,
+      "step": 699
+    },
+    {
+      "epoch": 0.3019193444037093,
+      "grad_norm": 0.17507003247737885,
+      "learning_rate": 0.00014675316548332418,
+      "loss": 2.2578,
+      "step": 700
+    },
+    {
+      "epoch": 0.3019193444037093,
+      "eval_loss": 2.131542682647705,
+      "eval_runtime": 204.1156,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 700
+    },
+    {
+      "epoch": 0.30235065775285747,
+      "grad_norm": 0.18731343746185303,
+      "learning_rate": 0.0001467432252272373,
+      "loss": 2.3435,
+      "step": 701
+    },
+    {
+      "epoch": 0.3027819711020056,
+      "grad_norm": 0.1506258249282837,
+      "learning_rate": 0.00014673327011593103,
+      "loss": 2.1025,
+      "step": 702
+    },
+    {
+      "epoch": 0.30321328445115375,
+      "grad_norm": 0.15649913251399994,
+      "learning_rate": 0.00014672330015146672,
+      "loss": 2.0297,
+      "step": 703
+    },
+    {
+      "epoch": 0.3036445978003019,
+      "grad_norm": 0.17421182990074158,
+      "learning_rate": 0.00014671331533590878,
+      "loss": 2.1474,
+      "step": 704
+    },
+    {
+      "epoch": 0.3040759111494501,
+      "grad_norm": 0.18708555400371552,
+      "learning_rate": 0.00014670331567132465,
+      "loss": 2.0628,
+      "step": 705
+    },
+    {
+      "epoch": 0.30450722449859824,
+      "grad_norm": 0.17888276278972626,
+      "learning_rate": 0.00014669330115978484,
+      "loss": 2.2094,
+      "step": 706
+    },
+    {
+      "epoch": 0.3049385378477464,
+      "grad_norm": 0.16658446192741394,
+      "learning_rate": 0.00014668327180336304,
+      "loss": 2.2866,
+      "step": 707
+    },
+    {
+      "epoch": 0.3053698511968945,
+      "grad_norm": 0.18156693875789642,
+      "learning_rate": 0.0001466732276041359,
+      "loss": 2.3104,
+      "step": 708
+    },
+    {
+      "epoch": 0.3058011645460427,
+      "grad_norm": 0.16488181054592133,
+      "learning_rate": 0.00014666316856418315,
+      "loss": 2.1646,
+      "step": 709
+    },
+    {
+      "epoch": 0.30623247789519087,
+      "grad_norm": 0.15647055208683014,
+      "learning_rate": 0.0001466530946855877,
+      "loss": 2.3348,
+      "step": 710
+    },
+    {
+      "epoch": 0.306663791244339,
+      "grad_norm": 0.16166602075099945,
+      "learning_rate": 0.00014664300597043537,
+      "loss": 2.3414,
+      "step": 711
+    },
+    {
+      "epoch": 0.30709510459348716,
+      "grad_norm": 0.160395085811615,
+      "learning_rate": 0.00014663290242081519,
+      "loss": 2.3181,
+      "step": 712
+    },
+    {
+      "epoch": 0.3075264179426353,
+      "grad_norm": 0.21888132393360138,
+      "learning_rate": 0.0001466227840388192,
+      "loss": 2.1063,
+      "step": 713
+    },
+    {
+      "epoch": 0.3079577312917835,
+      "grad_norm": 0.17477422952651978,
+      "learning_rate": 0.00014661265082654255,
+      "loss": 2.1944,
+      "step": 714
+    },
+    {
+      "epoch": 0.30838904464093164,
+      "grad_norm": 0.15590433776378632,
+      "learning_rate": 0.00014660250278608335,
+      "loss": 2.3067,
+      "step": 715
+    },
+    {
+      "epoch": 0.3088203579900798,
+      "grad_norm": 0.17154192924499512,
+      "learning_rate": 0.00014659233991954295,
+      "loss": 2.2371,
+      "step": 716
+    },
+    {
+      "epoch": 0.30925167133922793,
+      "grad_norm": 0.18001079559326172,
+      "learning_rate": 0.00014658216222902562,
+      "loss": 2.3827,
+      "step": 717
+    },
+    {
+      "epoch": 0.30968298468837613,
+      "grad_norm": 0.19673855602741241,
+      "learning_rate": 0.00014657196971663882,
+      "loss": 2.2921,
+      "step": 718
+    },
+    {
+      "epoch": 0.31011429803752427,
+      "grad_norm": 0.2262069284915924,
+      "learning_rate": 0.000146561762384493,
+      "loss": 2.2864,
+      "step": 719
+    },
+    {
+      "epoch": 0.3105456113866724,
+      "grad_norm": 0.17984148859977722,
+      "learning_rate": 0.00014655154023470167,
+      "loss": 1.9371,
+      "step": 720
+    },
+    {
+      "epoch": 0.31097692473582056,
+      "grad_norm": 0.17629437148571014,
+      "learning_rate": 0.00014654130326938146,
+      "loss": 2.1894,
+      "step": 721
+    },
+    {
+      "epoch": 0.3114082380849687,
+      "grad_norm": 0.17931878566741943,
+      "learning_rate": 0.00014653105149065206,
+      "loss": 2.0503,
+      "step": 722
+    },
+    {
+      "epoch": 0.3118395514341169,
+      "grad_norm": 0.15631668269634247,
+      "learning_rate": 0.00014652078490063619,
+      "loss": 2.1298,
+      "step": 723
+    },
+    {
+      "epoch": 0.31227086478326505,
+      "grad_norm": 0.15321604907512665,
+      "learning_rate": 0.00014651050350145966,
+      "loss": 2.245,
+      "step": 724
+    },
+    {
+      "epoch": 0.3127021781324132,
+      "grad_norm": 0.16721682250499725,
+      "learning_rate": 0.00014650020729525137,
+      "loss": 2.2935,
+      "step": 725
+    },
+    {
+      "epoch": 0.3127021781324132,
+      "eval_loss": 2.13155460357666,
+      "eval_runtime": 205.1729,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 725
+    },
+    {
+      "epoch": 0.31313349148156133,
+      "grad_norm": 0.17930914461612701,
+      "learning_rate": 0.00014648989628414323,
+      "loss": 2.1548,
+      "step": 726
+    },
+    {
+      "epoch": 0.31356480483070953,
+      "grad_norm": 0.16474506258964539,
+      "learning_rate": 0.00014647957047027028,
+      "loss": 2.3057,
+      "step": 727
+    },
+    {
+      "epoch": 0.3139961181798577,
+      "grad_norm": 0.1486273854970932,
+      "learning_rate": 0.00014646922985577058,
+      "loss": 2.1771,
+      "step": 728
+    },
+    {
+      "epoch": 0.3144274315290058,
+      "grad_norm": 0.1768963783979416,
+      "learning_rate": 0.00014645887444278524,
+      "loss": 2.0942,
+      "step": 729
+    },
+    {
+      "epoch": 0.31485874487815396,
+      "grad_norm": 0.16324689984321594,
+      "learning_rate": 0.0001464485042334585,
+      "loss": 2.0724,
+      "step": 730
+    },
+    {
+      "epoch": 0.31529005822730216,
+      "grad_norm": 0.15467900037765503,
+      "learning_rate": 0.00014643811922993762,
+      "loss": 2.1993,
+      "step": 731
+    },
+    {
+      "epoch": 0.3157213715764503,
+      "grad_norm": 0.15890255570411682,
+      "learning_rate": 0.00014642771943437291,
+      "loss": 2.2111,
+      "step": 732
+    },
+    {
+      "epoch": 0.31615268492559845,
+      "grad_norm": 0.1673458218574524,
+      "learning_rate": 0.00014641730484891776,
+      "loss": 2.0386,
+      "step": 733
+    },
+    {
+      "epoch": 0.3165839982747466,
+      "grad_norm": 0.21124054491519928,
+      "learning_rate": 0.00014640687547572866,
+      "loss": 2.2538,
+      "step": 734
+    },
+    {
+      "epoch": 0.31701531162389474,
+      "grad_norm": 0.17169217765331268,
+      "learning_rate": 0.0001463964313169651,
+      "loss": 2.1802,
+      "step": 735
+    },
+    {
+      "epoch": 0.31744662497304293,
+      "grad_norm": 0.18922211229801178,
+      "learning_rate": 0.00014638597237478964,
+      "loss": 2.3256,
+      "step": 736
+    },
+    {
+      "epoch": 0.3178779383221911,
+      "grad_norm": 0.16592665016651154,
+      "learning_rate": 0.00014637549865136794,
+      "loss": 2.4164,
+      "step": 737
+    },
+    {
+      "epoch": 0.3183092516713392,
+      "grad_norm": 0.19938214123249054,
+      "learning_rate": 0.0001463650101488687,
+      "loss": 2.2966,
+      "step": 738
+    },
+    {
+      "epoch": 0.31874056502048737,
+      "grad_norm": 0.18568481504917145,
+      "learning_rate": 0.00014635450686946365,
+      "loss": 2.2379,
+      "step": 739
+    },
+    {
+      "epoch": 0.31917187836963556,
+      "grad_norm": 0.15624523162841797,
+      "learning_rate": 0.00014634398881532766,
+      "loss": 2.0803,
+      "step": 740
+    },
+    {
+      "epoch": 0.3196031917187837,
+      "grad_norm": 0.14758580923080444,
+      "learning_rate": 0.00014633345598863857,
+      "loss": 1.9887,
+      "step": 741
+    },
+    {
+      "epoch": 0.32003450506793185,
+      "grad_norm": 0.1929500699043274,
+      "learning_rate": 0.00014632290839157734,
+      "loss": 2.4384,
+      "step": 742
+    },
+    {
+      "epoch": 0.32046581841708,
+      "grad_norm": 0.20792317390441895,
+      "learning_rate": 0.00014631234602632796,
+      "loss": 2.1976,
+      "step": 743
+    },
+    {
+      "epoch": 0.32089713176622814,
+      "grad_norm": 0.1632717400789261,
+      "learning_rate": 0.00014630176889507743,
+      "loss": 2.3856,
+      "step": 744
+    },
+    {
+      "epoch": 0.32132844511537634,
+      "grad_norm": 0.1877809315919876,
+      "learning_rate": 0.00014629117700001593,
+      "loss": 2.2074,
+      "step": 745
+    },
+    {
+      "epoch": 0.3217597584645245,
+      "grad_norm": 0.16195634007453918,
+      "learning_rate": 0.00014628057034333665,
+      "loss": 2.1698,
+      "step": 746
+    },
+    {
+      "epoch": 0.3221910718136726,
+      "grad_norm": 0.17172454297542572,
+      "learning_rate": 0.0001462699489272357,
+      "loss": 2.2282,
+      "step": 747
+    },
+    {
+      "epoch": 0.32262238516282077,
+      "grad_norm": 0.1560070663690567,
+      "learning_rate": 0.0001462593127539125,
+      "loss": 2.0006,
+      "step": 748
+    },
+    {
+      "epoch": 0.32305369851196897,
+      "grad_norm": 0.15756119787693024,
+      "learning_rate": 0.00014624866182556926,
+      "loss": 2.27,
+      "step": 749
+    },
+    {
+      "epoch": 0.3234850118611171,
+      "grad_norm": 0.16225910186767578,
+      "learning_rate": 0.00014623799614441144,
+      "loss": 2.3256,
+      "step": 750
+    },
+    {
+      "epoch": 0.3234850118611171,
+      "eval_loss": 2.131229877471924,
+      "eval_runtime": 204.3866,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 750
+    },
+    {
+      "epoch": 0.32391632521026525,
+      "grad_norm": 0.1817995011806488,
+      "learning_rate": 0.00014622731571264748,
+      "loss": 2.3521,
+      "step": 751
+    },
+    {
+      "epoch": 0.3243476385594134,
+      "grad_norm": 0.18910230696201324,
+      "learning_rate": 0.00014621662053248888,
+      "loss": 2.4786,
+      "step": 752
+    },
+    {
+      "epoch": 0.3247789519085616,
+      "grad_norm": 0.16400721669197083,
+      "learning_rate": 0.00014620591060615016,
+      "loss": 2.3265,
+      "step": 753
+    },
+    {
+      "epoch": 0.32521026525770974,
+      "grad_norm": 0.15586580336093903,
+      "learning_rate": 0.000146195185935849,
+      "loss": 2.2058,
+      "step": 754
+    },
+    {
+      "epoch": 0.3256415786068579,
+      "grad_norm": 1.2911137342453003,
+      "learning_rate": 0.000146184446523806,
+      "loss": 2.1915,
+      "step": 755
+    },
+    {
+      "epoch": 0.32607289195600603,
+      "grad_norm": 0.18798713386058807,
+      "learning_rate": 0.00014617369237224486,
+      "loss": 2.3488,
+      "step": 756
+    },
+    {
+      "epoch": 0.32650420530515417,
+      "grad_norm": 0.16819198429584503,
+      "learning_rate": 0.00014616292348339238,
+      "loss": 2.3031,
+      "step": 757
+    },
+    {
+      "epoch": 0.32693551865430237,
+      "grad_norm": 0.16455453634262085,
+      "learning_rate": 0.00014615213985947836,
+      "loss": 2.2154,
+      "step": 758
+    },
+    {
+      "epoch": 0.3273668320034505,
+      "grad_norm": 0.18589816987514496,
+      "learning_rate": 0.0001461413415027357,
+      "loss": 2.2736,
+      "step": 759
+    },
+    {
+      "epoch": 0.32779814535259866,
+      "grad_norm": 0.17672494053840637,
+      "learning_rate": 0.0001461305284154003,
+      "loss": 2.2528,
+      "step": 760
+    },
+    {
+      "epoch": 0.3282294587017468,
+      "grad_norm": 0.15951240062713623,
+      "learning_rate": 0.0001461197005997111,
+      "loss": 1.9911,
+      "step": 761
+    },
+    {
+      "epoch": 0.328660772050895,
+      "grad_norm": 0.1629369705915451,
+      "learning_rate": 0.00014610885805791012,
+      "loss": 2.4015,
+      "step": 762
+    },
+    {
+      "epoch": 0.32909208540004314,
+      "grad_norm": 0.2176440805196762,
+      "learning_rate": 0.0001460980007922425,
+      "loss": 2.2203,
+      "step": 763
+    },
+    {
+      "epoch": 0.3295233987491913,
+      "grad_norm": 0.18058481812477112,
+      "learning_rate": 0.00014608712880495627,
+      "loss": 2.3013,
+      "step": 764
+    },
+    {
+      "epoch": 0.32995471209833943,
+      "grad_norm": 0.1542855054140091,
+      "learning_rate": 0.0001460762420983026,
+      "loss": 2.4098,
+      "step": 765
+    },
+    {
+      "epoch": 0.3303860254474876,
+      "grad_norm": 0.17161045968532562,
+      "learning_rate": 0.00014606534067453577,
+      "loss": 2.117,
+      "step": 766
+    },
+    {
+      "epoch": 0.3308173387966358,
+      "grad_norm": 0.1644153594970703,
+      "learning_rate": 0.00014605442453591297,
+      "loss": 2.084,
+      "step": 767
+    },
+    {
+      "epoch": 0.3312486521457839,
+      "grad_norm": 0.15911726653575897,
+      "learning_rate": 0.00014604349368469452,
+      "loss": 2.1107,
+      "step": 768
+    },
+    {
+      "epoch": 0.33167996549493206,
+      "grad_norm": 0.1879819631576538,
+      "learning_rate": 0.00014603254812314384,
+      "loss": 2.303,
+      "step": 769
+    },
+    {
+      "epoch": 0.3321112788440802,
+      "grad_norm": 0.1810748130083084,
+      "learning_rate": 0.00014602158785352723,
+      "loss": 2.3159,
+      "step": 770
+    },
+    {
+      "epoch": 0.3325425921932284,
+      "grad_norm": 0.16361068189144135,
+      "learning_rate": 0.0001460106128781142,
+      "loss": 2.2551,
+      "step": 771
+    },
+    {
+      "epoch": 0.33297390554237655,
+      "grad_norm": 0.17128886282444,
+      "learning_rate": 0.0001459996231991772,
+      "loss": 2.2071,
+      "step": 772
+    },
+    {
+      "epoch": 0.3334052188915247,
+      "grad_norm": 0.1600296050310135,
+      "learning_rate": 0.00014598861881899176,
+      "loss": 2.0581,
+      "step": 773
+    },
+    {
+      "epoch": 0.33383653224067283,
+      "grad_norm": 0.16864512860774994,
+      "learning_rate": 0.00014597759973983648,
+      "loss": 2.1672,
+      "step": 774
+    },
+    {
+      "epoch": 0.334267845589821,
+      "grad_norm": 0.19236478209495544,
+      "learning_rate": 0.000145966565963993,
+      "loss": 2.4652,
+      "step": 775
+    },
+    {
+      "epoch": 0.334267845589821,
+      "eval_loss": 2.1317968368530273,
+      "eval_runtime": 204.7723,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 775
+    },
+    {
+      "epoch": 0.3346991589389692,
+      "grad_norm": 0.17529065907001495,
+      "learning_rate": 0.00014595551749374593,
+      "loss": 2.4414,
+      "step": 776
+    },
+    {
+      "epoch": 0.3351304722881173,
+      "grad_norm": 0.16876325011253357,
+      "learning_rate": 0.00014594445433138302,
+      "loss": 2.1084,
+      "step": 777
+    },
+    {
+      "epoch": 0.33556178563726546,
+      "grad_norm": 0.16747407615184784,
+      "learning_rate": 0.00014593337647919501,
+      "loss": 2.2267,
+      "step": 778
+    },
+    {
+      "epoch": 0.3359930989864136,
+      "grad_norm": 0.17824633419513702,
+      "learning_rate": 0.00014592228393947568,
+      "loss": 2.2853,
+      "step": 779
+    },
+    {
+      "epoch": 0.3364244123355618,
+      "grad_norm": 0.1787227839231491,
+      "learning_rate": 0.00014591117671452187,
+      "loss": 2.2528,
+      "step": 780
+    },
+    {
+      "epoch": 0.33685572568470995,
+      "grad_norm": 0.1657239943742752,
+      "learning_rate": 0.00014590005480663348,
+      "loss": 2.2671,
+      "step": 781
+    },
+    {
+      "epoch": 0.3372870390338581,
+      "grad_norm": 0.1929490864276886,
+      "learning_rate": 0.00014588891821811333,
+      "loss": 2.2231,
+      "step": 782
+    },
+    {
+      "epoch": 0.33771835238300624,
+      "grad_norm": 0.17597073316574097,
+      "learning_rate": 0.00014587776695126748,
+      "loss": 2.2238,
+      "step": 783
+    },
+    {
+      "epoch": 0.33814966573215444,
+      "grad_norm": 0.17662476003170013,
+      "learning_rate": 0.00014586660100840486,
+      "loss": 2.1492,
+      "step": 784
+    },
+    {
+      "epoch": 0.3385809790813026,
+      "grad_norm": 0.16856659948825836,
+      "learning_rate": 0.00014585542039183752,
+      "loss": 2.2535,
+      "step": 785
+    },
+    {
+      "epoch": 0.3390122924304507,
+      "grad_norm": 0.17280738055706024,
+      "learning_rate": 0.00014584422510388053,
+      "loss": 2.178,
+      "step": 786
+    },
+    {
+      "epoch": 0.33944360577959887,
+      "grad_norm": 0.17797470092773438,
+      "learning_rate": 0.000145833015146852,
+      "loss": 2.3755,
+      "step": 787
+    },
+    {
+      "epoch": 0.339874919128747,
+      "grad_norm": 0.1805947721004486,
+      "learning_rate": 0.00014582179052307305,
+      "loss": 2.195,
+      "step": 788
+    },
+    {
+      "epoch": 0.3403062324778952,
+      "grad_norm": 0.18467366695404053,
+      "learning_rate": 0.0001458105512348679,
+      "loss": 2.2721,
+      "step": 789
+    },
+    {
+      "epoch": 0.34073754582704335,
+      "grad_norm": 0.15401802957057953,
+      "learning_rate": 0.0001457992972845637,
+      "loss": 2.1344,
+      "step": 790
+    },
+    {
+      "epoch": 0.3411688591761915,
+      "grad_norm": 0.19222162663936615,
+      "learning_rate": 0.0001457880286744908,
+      "loss": 2.2172,
+      "step": 791
+    },
+    {
+      "epoch": 0.34160017252533964,
+      "grad_norm": 0.1638176292181015,
+      "learning_rate": 0.00014577674540698244,
+      "loss": 2.0332,
+      "step": 792
+    },
+    {
+      "epoch": 0.34203148587448784,
+      "grad_norm": 0.2038831114768982,
+      "learning_rate": 0.0001457654474843749,
+      "loss": 1.9864,
+      "step": 793
+    },
+    {
+      "epoch": 0.342462799223636,
+      "grad_norm": 0.2040897160768509,
+      "learning_rate": 0.0001457541349090076,
+      "loss": 2.1794,
+      "step": 794
+    },
+    {
+      "epoch": 0.3428941125727841,
+      "grad_norm": 0.17992299795150757,
+      "learning_rate": 0.00014574280768322293,
+      "loss": 2.3073,
+      "step": 795
+    },
+    {
+      "epoch": 0.34332542592193227,
+      "grad_norm": 0.1908801645040512,
+      "learning_rate": 0.00014573146580936628,
+      "loss": 2.3596,
+      "step": 796
+    },
+    {
+      "epoch": 0.3437567392710804,
+      "grad_norm": 0.1825549155473709,
+      "learning_rate": 0.0001457201092897861,
+      "loss": 2.1582,
+      "step": 797
+    },
+    {
+      "epoch": 0.3441880526202286,
+      "grad_norm": 0.1776050329208374,
+      "learning_rate": 0.00014570873812683397,
+      "loss": 2.1406,
+      "step": 798
+    },
+    {
+      "epoch": 0.34461936596937676,
+      "grad_norm": 0.16008296608924866,
+      "learning_rate": 0.00014569735232286432,
+      "loss": 2.3458,
+      "step": 799
+    },
+    {
+      "epoch": 0.3450506793185249,
+      "grad_norm": 0.1878412365913391,
+      "learning_rate": 0.00014568595188023473,
+      "loss": 2.0734,
+      "step": 800
+    },
+    {
+      "epoch": 0.3450506793185249,
+      "eval_loss": 2.130296230316162,
+      "eval_runtime": 204.5115,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 800
+    },
+    {
+      "epoch": 0.34548199266767304,
+      "grad_norm": 0.15881088376045227,
+      "learning_rate": 0.00014567453680130584,
+      "loss": 2.2351,
+      "step": 801
+    },
+    {
+      "epoch": 0.34591330601682124,
+      "grad_norm": 0.2246032953262329,
+      "learning_rate": 0.00014566310708844117,
+      "loss": 2.4797,
+      "step": 802
+    },
+    {
+      "epoch": 0.3463446193659694,
+      "grad_norm": 0.16534580290317535,
+      "learning_rate": 0.00014565166274400744,
+      "loss": 2.3781,
+      "step": 803
+    },
+    {
+      "epoch": 0.34677593271511753,
+      "grad_norm": 0.17714694142341614,
+      "learning_rate": 0.00014564020377037432,
+      "loss": 2.2876,
+      "step": 804
+    },
+    {
+      "epoch": 0.3472072460642657,
+      "grad_norm": 0.16253145039081573,
+      "learning_rate": 0.00014562873016991446,
+      "loss": 2.3543,
+      "step": 805
+    },
+    {
+      "epoch": 0.3476385594134139,
+      "grad_norm": 0.15457387268543243,
+      "learning_rate": 0.00014561724194500368,
+      "loss": 2.0988,
+      "step": 806
+    },
+    {
+      "epoch": 0.348069872762562,
+      "grad_norm": 0.1657828837633133,
+      "learning_rate": 0.00014560573909802065,
+      "loss": 2.1691,
+      "step": 807
+    },
+    {
+      "epoch": 0.34850118611171016,
+      "grad_norm": 0.15933750569820404,
+      "learning_rate": 0.00014559422163134723,
+      "loss": 2.1606,
+      "step": 808
+    },
+    {
+      "epoch": 0.3489324994608583,
+      "grad_norm": 0.18061064183712006,
+      "learning_rate": 0.00014558268954736824,
+      "loss": 2.3919,
+      "step": 809
+    },
+    {
+      "epoch": 0.34936381281000645,
+      "grad_norm": 0.16154713928699493,
+      "learning_rate": 0.0001455711428484715,
+      "loss": 2.1546,
+      "step": 810
+    },
+    {
+      "epoch": 0.34979512615915465,
+      "grad_norm": 0.1660330444574356,
+      "learning_rate": 0.00014555958153704787,
+      "loss": 2.1342,
+      "step": 811
+    },
+    {
+      "epoch": 0.3502264395083028,
+      "grad_norm": 0.1714458018541336,
+      "learning_rate": 0.0001455480056154912,
+      "loss": 2.1792,
+      "step": 812
+    },
+    {
+      "epoch": 0.35065775285745093,
+      "grad_norm": 0.15950365364551544,
+      "learning_rate": 0.00014553641508619852,
+      "loss": 2.458,
+      "step": 813
+    },
+    {
+      "epoch": 0.3510890662065991,
+      "grad_norm": 0.15440607070922852,
+      "learning_rate": 0.00014552480995156975,
+      "loss": 2.388,
+      "step": 814
+    },
+    {
+      "epoch": 0.3515203795557473,
+      "grad_norm": 0.15359917283058167,
+      "learning_rate": 0.00014551319021400777,
+      "loss": 2.2415,
+      "step": 815
+    },
+    {
+      "epoch": 0.3519516929048954,
+      "grad_norm": 0.1715104579925537,
+      "learning_rate": 0.00014550155587591867,
+      "loss": 2.2896,
+      "step": 816
+    },
+    {
+      "epoch": 0.35238300625404356,
+      "grad_norm": 0.16861777007579803,
+      "learning_rate": 0.00014548990693971143,
+      "loss": 2.2502,
+      "step": 817
+    },
+    {
+      "epoch": 0.3528143196031917,
+      "grad_norm": 0.18368162214756012,
+      "learning_rate": 0.00014547824340779811,
+      "loss": 2.2043,
+      "step": 818
+    },
+    {
+      "epoch": 0.35324563295233985,
+      "grad_norm": 0.15260176360607147,
+      "learning_rate": 0.00014546656528259376,
+      "loss": 2.1346,
+      "step": 819
+    },
+    {
+      "epoch": 0.35367694630148805,
+      "grad_norm": 0.15488527715206146,
+      "learning_rate": 0.00014545487256651643,
+      "loss": 2.146,
+      "step": 820
+    },
+    {
+      "epoch": 0.3541082596506362,
+      "grad_norm": 0.16105197370052338,
+      "learning_rate": 0.00014544316526198726,
+      "loss": 2.2272,
+      "step": 821
+    },
+    {
+      "epoch": 0.35453957299978434,
+      "grad_norm": 0.17276594042778015,
+      "learning_rate": 0.00014543144337143037,
+      "loss": 2.1993,
+      "step": 822
+    },
+    {
+      "epoch": 0.3549708863489325,
+      "grad_norm": 0.15881919860839844,
+      "learning_rate": 0.0001454197068972729,
+      "loss": 2.2004,
+      "step": 823
+    },
+    {
+      "epoch": 0.3554021996980807,
+      "grad_norm": 0.16273415088653564,
+      "learning_rate": 0.00014540795584194505,
+      "loss": 2.0913,
+      "step": 824
+    },
+    {
+      "epoch": 0.3558335130472288,
+      "grad_norm": 0.17026178538799286,
+      "learning_rate": 0.00014539619020787994,
+      "loss": 2.2855,
+      "step": 825
+    },
+    {
+      "epoch": 0.3558335130472288,
+      "eval_loss": 2.1297106742858887,
+      "eval_runtime": 204.6653,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 825
+    },
+    {
+      "epoch": 0.35626482639637697,
+      "grad_norm": 0.1643732637166977,
+      "learning_rate": 0.00014538440999751383,
+      "loss": 2.1876,
+      "step": 826
+    },
+    {
+      "epoch": 0.3566961397455251,
+      "grad_norm": 0.16995814442634583,
+      "learning_rate": 0.00014537261521328592,
+      "loss": 2.1945,
+      "step": 827
+    },
+    {
+      "epoch": 0.35712745309467325,
+      "grad_norm": 0.19104905426502228,
+      "learning_rate": 0.00014536080585763844,
+      "loss": 2.0627,
+      "step": 828
+    },
+    {
+      "epoch": 0.35755876644382145,
+      "grad_norm": 0.18704453110694885,
+      "learning_rate": 0.00014534898193301665,
+      "loss": 2.2512,
+      "step": 829
+    },
+    {
+      "epoch": 0.3579900797929696,
+      "grad_norm": 0.17079386115074158,
+      "learning_rate": 0.00014533714344186884,
+      "loss": 2.1159,
+      "step": 830
+    },
+    {
+      "epoch": 0.35842139314211774,
+      "grad_norm": 0.1565588116645813,
+      "learning_rate": 0.00014532529038664626,
+      "loss": 2.2767,
+      "step": 831
+    },
+    {
+      "epoch": 0.3588527064912659,
+      "grad_norm": 1.9668205976486206,
+      "learning_rate": 0.00014531342276980327,
+      "loss": 2.2867,
+      "step": 832
+    },
+    {
+      "epoch": 0.3592840198404141,
+      "grad_norm": 0.14429862797260284,
+      "learning_rate": 0.00014530154059379713,
+      "loss": 2.1932,
+      "step": 833
+    },
+    {
+      "epoch": 0.3597153331895622,
+      "grad_norm": 0.1885858178138733,
+      "learning_rate": 0.00014528964386108825,
+      "loss": 2.3821,
+      "step": 834
+    },
+    {
+      "epoch": 0.36014664653871037,
+      "grad_norm": 0.16591544449329376,
+      "learning_rate": 0.00014527773257413988,
+      "loss": 2.3467,
+      "step": 835
+    },
+    {
+      "epoch": 0.3605779598878585,
+      "grad_norm": 0.17843496799468994,
+      "learning_rate": 0.00014526580673541848,
+      "loss": 2.2347,
+      "step": 836
+    },
+    {
+      "epoch": 0.3610092732370067,
+      "grad_norm": 1.2142467498779297,
+      "learning_rate": 0.00014525386634739335,
+      "loss": 2.0654,
+      "step": 837
+    },
+    {
+      "epoch": 0.36144058658615486,
+      "grad_norm": 0.18093134462833405,
+      "learning_rate": 0.0001452419114125369,
+      "loss": 2.1252,
+      "step": 838
+    },
+    {
+      "epoch": 0.361871899935303,
+      "grad_norm": 0.17511476576328278,
+      "learning_rate": 0.0001452299419333246,
+      "loss": 2.2497,
+      "step": 839
+    },
+    {
+      "epoch": 0.36230321328445114,
+      "grad_norm": 0.18539828062057495,
+      "learning_rate": 0.00014521795791223475,
+      "loss": 2.3388,
+      "step": 840
+    },
+    {
+      "epoch": 0.3627345266335993,
+      "grad_norm": 0.22852657735347748,
+      "learning_rate": 0.00014520595935174887,
+      "loss": 2.4679,
+      "step": 841
+    },
+    {
+      "epoch": 0.3631658399827475,
+      "grad_norm": 0.18394552171230316,
+      "learning_rate": 0.0001451939462543513,
+      "loss": 2.1722,
+      "step": 842
+    },
+    {
+      "epoch": 0.36359715333189563,
+      "grad_norm": 0.1762983798980713,
+      "learning_rate": 0.00014518191862252953,
+      "loss": 1.9698,
+      "step": 843
+    },
+    {
+      "epoch": 0.36402846668104377,
+      "grad_norm": 0.2538037598133087,
+      "learning_rate": 0.00014516987645877403,
+      "loss": 2.2843,
+      "step": 844
+    },
+    {
+      "epoch": 0.3644597800301919,
+      "grad_norm": 0.1793690174818039,
+      "learning_rate": 0.00014515781976557826,
+      "loss": 2.2725,
+      "step": 845
+    },
+    {
+      "epoch": 0.3648910933793401,
+      "grad_norm": 0.17635536193847656,
+      "learning_rate": 0.00014514574854543867,
+      "loss": 2.2042,
+      "step": 846
+    },
+    {
+      "epoch": 0.36532240672848826,
+      "grad_norm": 0.1704099327325821,
+      "learning_rate": 0.0001451336628008547,
+      "loss": 2.1619,
+      "step": 847
+    },
+    {
+      "epoch": 0.3657537200776364,
+      "grad_norm": 0.1625613272190094,
+      "learning_rate": 0.00014512156253432891,
+      "loss": 2.2149,
+      "step": 848
+    },
+    {
+      "epoch": 0.36618503342678455,
+      "grad_norm": 0.1626460701227188,
+      "learning_rate": 0.00014510944774836677,
+      "loss": 2.2051,
+      "step": 849
+    },
+    {
+      "epoch": 0.3666163467759327,
+      "grad_norm": 0.16857227683067322,
+      "learning_rate": 0.00014509731844547676,
+      "loss": 2.1927,
+      "step": 850
+    },
+    {
+      "epoch": 0.3666163467759327,
+      "eval_loss": 2.133108139038086,
+      "eval_runtime": 204.1873,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 850
+    },
+    {
+      "epoch": 0.3670476601250809,
+      "grad_norm": 0.18687738478183746,
+      "learning_rate": 0.0001450851746281704,
+      "loss": 2.2809,
+      "step": 851
+    },
+    {
+      "epoch": 0.36747897347422903,
+      "grad_norm": 0.19265635311603546,
+      "learning_rate": 0.0001450730162989622,
+      "loss": 2.3029,
+      "step": 852
+    },
+    {
+      "epoch": 0.3679102868233772,
+      "grad_norm": 0.16877678036689758,
+      "learning_rate": 0.00014506084346036963,
+      "loss": 2.2134,
+      "step": 853
+    },
+    {
+      "epoch": 0.3683416001725253,
+      "grad_norm": 0.17525163292884827,
+      "learning_rate": 0.0001450486561149133,
+      "loss": 2.1285,
+      "step": 854
+    },
+    {
+      "epoch": 0.3687729135216735,
+      "grad_norm": 0.1960926353931427,
+      "learning_rate": 0.00014503645426511665,
+      "loss": 2.2813,
+      "step": 855
+    },
+    {
+      "epoch": 0.36920422687082166,
+      "grad_norm": 0.1757328361272812,
+      "learning_rate": 0.00014502423791350625,
+      "loss": 2.2732,
+      "step": 856
+    },
+    {
+      "epoch": 0.3696355402199698,
+      "grad_norm": 0.1861843466758728,
+      "learning_rate": 0.00014501200706261162,
+      "loss": 2.1508,
+      "step": 857
+    },
+    {
+      "epoch": 0.37006685356911795,
+      "grad_norm": 0.19854430854320526,
+      "learning_rate": 0.00014499976171496528,
+      "loss": 2.2179,
+      "step": 858
+    },
+    {
+      "epoch": 0.37049816691826615,
+      "grad_norm": 0.16625559329986572,
+      "learning_rate": 0.00014498750187310277,
+      "loss": 1.99,
+      "step": 859
+    },
+    {
+      "epoch": 0.3709294802674143,
+      "grad_norm": 0.15606555342674255,
+      "learning_rate": 0.00014497522753956264,
+      "loss": 2.3534,
+      "step": 860
+    },
+    {
+      "epoch": 0.37136079361656243,
+      "grad_norm": 0.17310819029808044,
+      "learning_rate": 0.00014496293871688642,
+      "loss": 2.1748,
+      "step": 861
+    },
+    {
+      "epoch": 0.3717921069657106,
+      "grad_norm": 0.17876191437244415,
+      "learning_rate": 0.0001449506354076186,
+      "loss": 2.2285,
+      "step": 862
+    },
+    {
+      "epoch": 0.3722234203148587,
+      "grad_norm": 0.18471559882164001,
+      "learning_rate": 0.0001449383176143068,
+      "loss": 2.2541,
+      "step": 863
+    },
+    {
+      "epoch": 0.3726547336640069,
+      "grad_norm": 0.171518474817276,
+      "learning_rate": 0.00014492598533950146,
+      "loss": 2.357,
+      "step": 864
+    },
+    {
+      "epoch": 0.37308604701315506,
+      "grad_norm": 0.1647776961326599,
+      "learning_rate": 0.0001449136385857562,
+      "loss": 2.2007,
+      "step": 865
+    },
+    {
+      "epoch": 0.3735173603623032,
+      "grad_norm": 0.1820192039012909,
+      "learning_rate": 0.0001449012773556275,
+      "loss": 2.2049,
+      "step": 866
+    },
+    {
+      "epoch": 0.37394867371145135,
+      "grad_norm": 0.16099920868873596,
+      "learning_rate": 0.00014488890165167487,
+      "loss": 2.258,
+      "step": 867
+    },
+    {
+      "epoch": 0.37437998706059955,
+      "grad_norm": 0.1925518810749054,
+      "learning_rate": 0.00014487651147646088,
+      "loss": 2.3798,
+      "step": 868
+    },
+    {
+      "epoch": 0.3748113004097477,
+      "grad_norm": 0.17227499186992645,
+      "learning_rate": 0.00014486410683255103,
+      "loss": 2.3747,
+      "step": 869
+    },
+    {
+      "epoch": 0.37524261375889584,
+      "grad_norm": 0.16866856813430786,
+      "learning_rate": 0.00014485168772251382,
+      "loss": 2.2091,
+      "step": 870
+    },
+    {
+      "epoch": 0.375673927108044,
+      "grad_norm": 0.16172897815704346,
+      "learning_rate": 0.00014483925414892078,
+      "loss": 2.3496,
+      "step": 871
+    },
+    {
+      "epoch": 0.3761052404571921,
+      "grad_norm": 0.23066756129264832,
+      "learning_rate": 0.00014482680611434644,
+      "loss": 1.9826,
+      "step": 872
+    },
+    {
+      "epoch": 0.3765365538063403,
+      "grad_norm": 0.18224367499351501,
+      "learning_rate": 0.00014481434362136828,
+      "loss": 2.2193,
+      "step": 873
+    },
+    {
+      "epoch": 0.37696786715548847,
+      "grad_norm": 1.6786094903945923,
+      "learning_rate": 0.0001448018666725668,
+      "loss": 2.1686,
+      "step": 874
+    },
+    {
+      "epoch": 0.3773991805046366,
+      "grad_norm": 0.1985001266002655,
+      "learning_rate": 0.00014478937527052547,
+      "loss": 2.2994,
+      "step": 875
+    },
+    {
+      "epoch": 0.3773991805046366,
+      "eval_loss": 2.131563425064087,
+      "eval_runtime": 198.5072,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 875
+    },
+    {
+      "epoch": 0.37783049385378475,
+      "grad_norm": 0.19998972117900848,
+      "learning_rate": 0.0001447768694178308,
+      "loss": 2.3498,
+      "step": 876
+    },
+    {
+      "epoch": 0.37826180720293295,
+      "grad_norm": 0.3351645767688751,
+      "learning_rate": 0.00014476434911707225,
+      "loss": 2.0949,
+      "step": 877
+    },
+    {
+      "epoch": 0.3786931205520811,
+      "grad_norm": 0.19838584959506989,
+      "learning_rate": 0.00014475181437084228,
+      "loss": 2.3821,
+      "step": 878
+    },
+    {
+      "epoch": 0.37912443390122924,
+      "grad_norm": 0.17798852920532227,
+      "learning_rate": 0.00014473926518173636,
+      "loss": 2.3823,
+      "step": 879
+    },
+    {
+      "epoch": 0.3795557472503774,
+      "grad_norm": 0.26538652181625366,
+      "learning_rate": 0.0001447267015523529,
+      "loss": 2.2315,
+      "step": 880
+    },
+    {
+      "epoch": 0.37998706059952553,
+      "grad_norm": 0.20238934457302094,
+      "learning_rate": 0.0001447141234852934,
+      "loss": 2.3112,
+      "step": 881
+    },
+    {
+      "epoch": 0.3804183739486737,
+      "grad_norm": 0.16403162479400635,
+      "learning_rate": 0.00014470153098316226,
+      "loss": 1.8282,
+      "step": 882
+    },
+    {
+      "epoch": 0.38084968729782187,
+      "grad_norm": 0.17718282341957092,
+      "learning_rate": 0.00014468892404856685,
+      "loss": 2.0251,
+      "step": 883
+    },
+    {
+      "epoch": 0.38128100064697,
+      "grad_norm": 0.21754078567028046,
+      "learning_rate": 0.00014467630268411762,
+      "loss": 2.1585,
+      "step": 884
+    },
+    {
+      "epoch": 0.38171231399611816,
+      "grad_norm": 0.21408778429031372,
+      "learning_rate": 0.00014466366689242795,
+      "loss": 1.9296,
+      "step": 885
+    },
+    {
+      "epoch": 0.38214362734526636,
+      "grad_norm": 0.1843462884426117,
+      "learning_rate": 0.00014465101667611425,
+      "loss": 2.1953,
+      "step": 886
+    },
+    {
+      "epoch": 0.3825749406944145,
+      "grad_norm": 0.21462063491344452,
+      "learning_rate": 0.00014463835203779583,
+      "loss": 2.1426,
+      "step": 887
+    },
+    {
+      "epoch": 0.38300625404356264,
+      "grad_norm": 0.20921021699905396,
+      "learning_rate": 0.00014462567298009508,
+      "loss": 2.3491,
+      "step": 888
+    },
+    {
+      "epoch": 0.3834375673927108,
+      "grad_norm": 0.19191640615463257,
+      "learning_rate": 0.0001446129795056373,
+      "loss": 2.1436,
+      "step": 889
+    },
+    {
+      "epoch": 0.383868880741859,
+      "grad_norm": 0.1827997863292694,
+      "learning_rate": 0.00014460027161705086,
+      "loss": 2.2466,
+      "step": 890
+    },
+    {
+      "epoch": 0.38430019409100713,
+      "grad_norm": 0.16525106132030487,
+      "learning_rate": 0.00014458754931696705,
+      "loss": 2.2714,
+      "step": 891
+    },
+    {
+      "epoch": 0.3847315074401553,
+      "grad_norm": 0.1852799654006958,
+      "learning_rate": 0.00014457481260802014,
+      "loss": 2.0365,
+      "step": 892
+    },
+    {
+      "epoch": 0.3851628207893034,
+      "grad_norm": 0.1985684037208557,
+      "learning_rate": 0.00014456206149284745,
+      "loss": 2.0933,
+      "step": 893
+    },
+    {
+      "epoch": 0.38559413413845156,
+      "grad_norm": 0.19180408120155334,
+      "learning_rate": 0.00014454929597408918,
+      "loss": 2.2093,
+      "step": 894
+    },
+    {
+      "epoch": 0.38602544748759976,
+      "grad_norm": 0.17508214712142944,
+      "learning_rate": 0.0001445365160543886,
+      "loss": 2.2649,
+      "step": 895
+    },
+    {
+      "epoch": 0.3864567608367479,
+      "grad_norm": 0.1650542914867401,
+      "learning_rate": 0.00014452372173639195,
+      "loss": 2.3021,
+      "step": 896
+    },
+    {
+      "epoch": 0.38688807418589605,
+      "grad_norm": 0.37150564789772034,
+      "learning_rate": 0.00014451091302274843,
+      "loss": 2.0946,
+      "step": 897
+    },
+    {
+      "epoch": 0.3873193875350442,
+      "grad_norm": 0.17382101714611053,
+      "learning_rate": 0.00014449808991611019,
+      "loss": 2.2065,
+      "step": 898
+    },
+    {
+      "epoch": 0.3877507008841924,
+      "grad_norm": 0.16627924144268036,
+      "learning_rate": 0.0001444852524191324,
+      "loss": 2.244,
+      "step": 899
+    },
+    {
+      "epoch": 0.38818201423334053,
+      "grad_norm": 0.17999930679798126,
+      "learning_rate": 0.00014447240053447327,
+      "loss": 2.2263,
+      "step": 900
+    },
+    {
+      "epoch": 0.38818201423334053,
+      "eval_loss": 2.128645896911621,
+      "eval_runtime": 196.0098,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 900
+    },
+    {
+      "epoch": 0.3886133275824887,
+      "grad_norm": 3.2199740409851074,
+      "learning_rate": 0.00014445953426479383,
+      "loss": 2.1536,
+      "step": 901
+    },
+    {
+      "epoch": 0.3890446409316368,
+      "grad_norm": 0.19231465458869934,
+      "learning_rate": 0.00014444665361275822,
+      "loss": 1.9445,
+      "step": 902
+    },
+    {
+      "epoch": 0.38947595428078496,
+      "grad_norm": 0.21291469037532806,
+      "learning_rate": 0.00014443375858103354,
+      "loss": 2.267,
+      "step": 903
+    },
+    {
+      "epoch": 0.38990726762993316,
+      "grad_norm": 0.1833806037902832,
+      "learning_rate": 0.00014442084917228983,
+      "loss": 2.2679,
+      "step": 904
+    },
+    {
+      "epoch": 0.3903385809790813,
+      "grad_norm": 0.16631187498569489,
+      "learning_rate": 0.00014440792538920012,
+      "loss": 2.1788,
+      "step": 905
+    },
+    {
+      "epoch": 0.39076989432822945,
+      "grad_norm": 0.2123652994632721,
+      "learning_rate": 0.00014439498723444044,
+      "loss": 2.3326,
+      "step": 906
+    },
+    {
+      "epoch": 0.3912012076773776,
+      "grad_norm": 0.18742474913597107,
+      "learning_rate": 0.00014438203471068975,
+      "loss": 2.2317,
+      "step": 907
+    },
+    {
+      "epoch": 0.3916325210265258,
+      "grad_norm": 0.1959163248538971,
+      "learning_rate": 0.00014436906782063,
+      "loss": 2.1793,
+      "step": 908
+    },
+    {
+      "epoch": 0.39206383437567394,
+      "grad_norm": 0.176411971449852,
+      "learning_rate": 0.00014435608656694618,
+      "loss": 2.2729,
+      "step": 909
+    },
+    {
+      "epoch": 0.3924951477248221,
+      "grad_norm": 0.1642094999551773,
+      "learning_rate": 0.00014434309095232617,
+      "loss": 2.1317,
+      "step": 910
+    },
+    {
+      "epoch": 0.3929264610739702,
+      "grad_norm": 0.19887696206569672,
+      "learning_rate": 0.00014433008097946084,
+      "loss": 2.2663,
+      "step": 911
+    },
+    {
+      "epoch": 0.3933577744231184,
+      "grad_norm": 0.18333375453948975,
+      "learning_rate": 0.00014431705665104408,
+      "loss": 2.3066,
+      "step": 912
+    },
+    {
+      "epoch": 0.39378908777226657,
+      "grad_norm": 0.17824384570121765,
+      "learning_rate": 0.0001443040179697727,
+      "loss": 2.2588,
+      "step": 913
+    },
+    {
+      "epoch": 0.3942204011214147,
+      "grad_norm": 0.1979813426733017,
+      "learning_rate": 0.0001442909649383465,
+      "loss": 2.1181,
+      "step": 914
+    },
+    {
+      "epoch": 0.39465171447056285,
+      "grad_norm": 0.19859793782234192,
+      "learning_rate": 0.00014427789755946824,
+      "loss": 2.3726,
+      "step": 915
+    },
+    {
+      "epoch": 0.395083027819711,
+      "grad_norm": 0.14289377629756927,
+      "learning_rate": 0.0001442648158358437,
+      "loss": 2.0499,
+      "step": 916
+    },
+    {
+      "epoch": 0.3955143411688592,
+      "grad_norm": 0.1597006618976593,
+      "learning_rate": 0.00014425171977018158,
+      "loss": 2.2563,
+      "step": 917
+    },
+    {
+      "epoch": 0.39594565451800734,
+      "grad_norm": 0.1751205325126648,
+      "learning_rate": 0.00014423860936519354,
+      "loss": 2.2895,
+      "step": 918
+    },
+    {
+      "epoch": 0.3963769678671555,
+      "grad_norm": 0.21087686717510223,
+      "learning_rate": 0.00014422548462359423,
+      "loss": 2.2271,
+      "step": 919
+    },
+    {
+      "epoch": 0.3968082812163036,
+      "grad_norm": 0.17048583924770355,
+      "learning_rate": 0.00014421234554810134,
+      "loss": 2.2043,
+      "step": 920
+    },
+    {
+      "epoch": 0.3972395945654518,
+      "grad_norm": 0.1800399273633957,
+      "learning_rate": 0.00014419919214143538,
+      "loss": 2.2716,
+      "step": 921
+    },
+    {
+      "epoch": 0.39767090791459997,
+      "grad_norm": 0.17856170237064362,
+      "learning_rate": 0.00014418602440631996,
+      "loss": 2.3445,
+      "step": 922
+    },
+    {
+      "epoch": 0.3981022212637481,
+      "grad_norm": 0.15118065476417542,
+      "learning_rate": 0.0001441728423454816,
+      "loss": 2.1092,
+      "step": 923
+    },
+    {
+      "epoch": 0.39853353461289626,
+      "grad_norm": 0.18618765473365784,
+      "learning_rate": 0.0001441596459616497,
+      "loss": 2.3339,
+      "step": 924
+    },
+    {
+      "epoch": 0.3989648479620444,
+      "grad_norm": 0.17451119422912598,
+      "learning_rate": 0.00014414643525755683,
+      "loss": 2.2336,
+      "step": 925
+    },
+    {
+      "epoch": 0.3989648479620444,
+      "eval_loss": 2.128000259399414,
+      "eval_runtime": 197.4817,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 925
+    },
+    {
+      "epoch": 0.3993961613111926,
+      "grad_norm": 0.168556809425354,
+      "learning_rate": 0.00014413321023593837,
+      "loss": 2.4635,
+      "step": 926
+    },
+    {
+      "epoch": 0.39982747466034074,
+      "grad_norm": 0.1529935747385025,
+      "learning_rate": 0.00014411997089953268,
+      "loss": 2.0522,
+      "step": 927
+    },
+    {
+      "epoch": 0.4002587880094889,
+      "grad_norm": 0.15923142433166504,
+      "learning_rate": 0.00014410671725108117,
+      "loss": 2.3012,
+      "step": 928
+    },
+    {
+      "epoch": 0.40069010135863703,
+      "grad_norm": 0.17462576925754547,
+      "learning_rate": 0.00014409344929332808,
+      "loss": 2.2029,
+      "step": 929
+    },
+    {
+      "epoch": 0.40112141470778523,
+      "grad_norm": 0.17241305112838745,
+      "learning_rate": 0.00014408016702902073,
+      "loss": 2.379,
+      "step": 930
+    },
+    {
+      "epoch": 0.4015527280569334,
+      "grad_norm": 0.15656477212905884,
+      "learning_rate": 0.00014406687046090934,
+      "loss": 2.2175,
+      "step": 931
+    },
+    {
+      "epoch": 0.4019840414060815,
+      "grad_norm": 0.1552082747220993,
+      "learning_rate": 0.00014405355959174712,
+      "loss": 2.106,
+      "step": 932
+    },
+    {
+      "epoch": 0.40241535475522966,
+      "grad_norm": 0.18489037454128265,
+      "learning_rate": 0.00014404023442429027,
+      "loss": 2.2944,
+      "step": 933
+    },
+    {
+      "epoch": 0.4028466681043778,
+      "grad_norm": 0.1694725602865219,
+      "learning_rate": 0.0001440268949612978,
+      "loss": 2.014,
+      "step": 934
+    },
+    {
+      "epoch": 0.403277981453526,
+      "grad_norm": 0.1695718765258789,
+      "learning_rate": 0.00014401354120553193,
+      "loss": 2.1954,
+      "step": 935
+    },
+    {
+      "epoch": 0.40370929480267415,
+      "grad_norm": 0.16405755281448364,
+      "learning_rate": 0.00014400017315975761,
+      "loss": 2.3878,
+      "step": 936
+    },
+    {
+      "epoch": 0.4041406081518223,
+      "grad_norm": 0.17652428150177002,
+      "learning_rate": 0.00014398679082674288,
+      "loss": 2.31,
+      "step": 937
+    },
+    {
+      "epoch": 0.40457192150097043,
+      "grad_norm": 0.1729700267314911,
+      "learning_rate": 0.00014397339420925865,
+      "loss": 2.1835,
+      "step": 938
+    },
+    {
+      "epoch": 0.40500323485011863,
+      "grad_norm": 0.1890537589788437,
+      "learning_rate": 0.00014395998331007888,
+      "loss": 2.2332,
+      "step": 939
+    },
+    {
+      "epoch": 0.4054345481992668,
+      "grad_norm": 0.1828519105911255,
+      "learning_rate": 0.00014394655813198043,
+      "loss": 2.1929,
+      "step": 940
+    },
+    {
+      "epoch": 0.4058658615484149,
+      "grad_norm": 0.16532336175441742,
+      "learning_rate": 0.00014393311867774315,
+      "loss": 2.2461,
+      "step": 941
+    },
+    {
+      "epoch": 0.40629717489756306,
+      "grad_norm": 3.688812494277954,
+      "learning_rate": 0.0001439196649501498,
+      "loss": 2.1308,
+      "step": 942
+    },
+    {
+      "epoch": 0.40672848824671126,
+      "grad_norm": 0.1970452517271042,
+      "learning_rate": 0.00014390619695198612,
+      "loss": 2.3738,
+      "step": 943
+    },
+    {
+      "epoch": 0.4071598015958594,
+      "grad_norm": 0.1757008284330368,
+      "learning_rate": 0.0001438927146860408,
+      "loss": 2.3448,
+      "step": 944
+    },
+    {
+      "epoch": 0.40759111494500755,
+      "grad_norm": 0.1817890703678131,
+      "learning_rate": 0.00014387921815510555,
+      "loss": 2.1039,
+      "step": 945
+    },
+    {
+      "epoch": 0.4080224282941557,
+      "grad_norm": 0.21462790668010712,
+      "learning_rate": 0.0001438657073619749,
+      "loss": 2.2177,
+      "step": 946
+    },
+    {
+      "epoch": 0.40845374164330384,
+      "grad_norm": 0.18851101398468018,
+      "learning_rate": 0.00014385218230944643,
+      "loss": 2.1774,
+      "step": 947
+    },
+    {
+      "epoch": 0.40888505499245204,
+      "grad_norm": 0.18490348756313324,
+      "learning_rate": 0.00014383864300032069,
+      "loss": 2.3663,
+      "step": 948
+    },
+    {
+      "epoch": 0.4093163683416002,
+      "grad_norm": 0.1923254281282425,
+      "learning_rate": 0.00014382508943740107,
+      "loss": 2.3492,
+      "step": 949
+    },
+    {
+      "epoch": 0.4097476816907483,
+      "grad_norm": 0.16575244069099426,
+      "learning_rate": 0.00014381152162349406,
+      "loss": 2.2695,
+      "step": 950
+    },
+    {
+      "epoch": 0.4097476816907483,
+      "eval_loss": 2.1288790702819824,
+      "eval_runtime": 196.1803,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 950
+    },
+    {
+      "epoch": 0.41017899503989647,
+      "grad_norm": 0.1713244915008545,
+      "learning_rate": 0.00014379793956140899,
+      "loss": 2.2441,
+      "step": 951
+    },
+    {
+      "epoch": 0.41061030838904466,
+      "grad_norm": 0.15283671021461487,
+      "learning_rate": 0.00014378434325395815,
+      "loss": 2.2407,
+      "step": 952
+    },
+    {
+      "epoch": 0.4110416217381928,
+      "grad_norm": 0.2225114405155182,
+      "learning_rate": 0.00014377073270395682,
+      "loss": 2.2592,
+      "step": 953
+    },
+    {
+      "epoch": 0.41147293508734095,
+      "grad_norm": 0.1541251540184021,
+      "learning_rate": 0.00014375710791422324,
+      "loss": 2.0837,
+      "step": 954
+    },
+    {
+      "epoch": 0.4119042484364891,
+      "grad_norm": 0.1644444763660431,
+      "learning_rate": 0.00014374346888757853,
+      "loss": 2.2987,
+      "step": 955
+    },
+    {
+      "epoch": 0.41233556178563724,
+      "grad_norm": 0.16322757303714752,
+      "learning_rate": 0.00014372981562684684,
+      "loss": 2.2194,
+      "step": 956
+    },
+    {
+      "epoch": 0.41276687513478544,
+      "grad_norm": 0.22558414936065674,
+      "learning_rate": 0.0001437161481348552,
+      "loss": 2.264,
+      "step": 957
+    },
+    {
+      "epoch": 0.4131981884839336,
+      "grad_norm": 0.20409363508224487,
+      "learning_rate": 0.00014370246641443363,
+      "loss": 2.5098,
+      "step": 958
+    },
+    {
+      "epoch": 0.4136295018330817,
+      "grad_norm": 0.17902424931526184,
+      "learning_rate": 0.00014368877046841506,
+      "loss": 2.3053,
+      "step": 959
+    },
+    {
+      "epoch": 0.41406081518222987,
+      "grad_norm": 0.16873684525489807,
+      "learning_rate": 0.0001436750602996354,
+      "loss": 2.2177,
+      "step": 960
+    },
+    {
+      "epoch": 0.41449212853137807,
+      "grad_norm": 0.1802307814359665,
+      "learning_rate": 0.0001436613359109335,
+      "loss": 2.2051,
+      "step": 961
+    },
+    {
+      "epoch": 0.4149234418805262,
+      "grad_norm": 0.2217787653207779,
+      "learning_rate": 0.00014364759730515112,
+      "loss": 2.4363,
+      "step": 962
+    },
+    {
+      "epoch": 0.41535475522967436,
+      "grad_norm": 0.17444701492786407,
+      "learning_rate": 0.000143633844485133,
+      "loss": 2.2822,
+      "step": 963
+    },
+    {
+      "epoch": 0.4157860685788225,
+      "grad_norm": 0.14976367354393005,
+      "learning_rate": 0.00014362007745372682,
+      "loss": 2.2694,
+      "step": 964
+    },
+    {
+      "epoch": 0.4162173819279707,
+      "grad_norm": 0.17860938608646393,
+      "learning_rate": 0.00014360629621378316,
+      "loss": 2.291,
+      "step": 965
+    },
+    {
+      "epoch": 0.41664869527711884,
+      "grad_norm": 0.18789131939411163,
+      "learning_rate": 0.00014359250076815565,
+      "loss": 2.2155,
+      "step": 966
+    },
+    {
+      "epoch": 0.417080008626267,
+      "grad_norm": 0.16874562203884125,
+      "learning_rate": 0.00014357869111970072,
+      "loss": 2.1584,
+      "step": 967
+    },
+    {
+      "epoch": 0.41751132197541513,
+      "grad_norm": 0.14934919774532318,
+      "learning_rate": 0.00014356486727127782,
+      "loss": 2.1014,
+      "step": 968
+    },
+    {
+      "epoch": 0.41794263532456327,
+      "grad_norm": 0.17155322432518005,
+      "learning_rate": 0.00014355102922574935,
+      "loss": 2.3703,
+      "step": 969
+    },
+    {
+      "epoch": 0.41837394867371147,
+      "grad_norm": 0.16445334255695343,
+      "learning_rate": 0.0001435371769859806,
+      "loss": 2.1432,
+      "step": 970
+    },
+    {
+      "epoch": 0.4188052620228596,
+      "grad_norm": 0.15587712824344635,
+      "learning_rate": 0.00014352331055483987,
+      "loss": 2.2831,
+      "step": 971
+    },
+    {
+      "epoch": 0.41923657537200776,
+      "grad_norm": 0.17527459561824799,
+      "learning_rate": 0.0001435094299351983,
+      "loss": 2.0739,
+      "step": 972
+    },
+    {
+      "epoch": 0.4196678887211559,
+      "grad_norm": 0.15222637355327606,
+      "learning_rate": 0.00014349553512993008,
+      "loss": 2.152,
+      "step": 973
+    },
+    {
+      "epoch": 0.4200992020703041,
+      "grad_norm": 0.18465347588062286,
+      "learning_rate": 0.00014348162614191224,
+      "loss": 2.3775,
+      "step": 974
+    },
+    {
+      "epoch": 0.42053051541945224,
+      "grad_norm": 0.1829536408185959,
+      "learning_rate": 0.00014346770297402482,
+      "loss": 2.2017,
+      "step": 975
+    },
+    {
+      "epoch": 0.42053051541945224,
+      "eval_loss": 2.1273937225341797,
+      "eval_runtime": 199.8206,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 975
+    },
+    {
+      "epoch": 0.4209618287686004,
+      "grad_norm": 0.18385931849479675,
+      "learning_rate": 0.00014345376562915075,
+      "loss": 2.3043,
+      "step": 976
+    },
+    {
+      "epoch": 0.42139314211774853,
+      "grad_norm": 0.16265417635440826,
+      "learning_rate": 0.0001434398141101759,
+      "loss": 2.2154,
+      "step": 977
+    },
+    {
+      "epoch": 0.4218244554668967,
+      "grad_norm": 0.1492079794406891,
+      "learning_rate": 0.0001434258484199891,
+      "loss": 2.2189,
+      "step": 978
+    },
+    {
+      "epoch": 0.4222557688160449,
+      "grad_norm": 0.16092488169670105,
+      "learning_rate": 0.0001434118685614821,
+      "loss": 2.2781,
+      "step": 979
+    },
+    {
+      "epoch": 0.422687082165193,
+      "grad_norm": 0.16646727919578552,
+      "learning_rate": 0.00014339787453754955,
+      "loss": 2.1738,
+      "step": 980
+    },
+    {
+      "epoch": 0.42311839551434116,
+      "grad_norm": 0.18572886288166046,
+      "learning_rate": 0.00014338386635108911,
+      "loss": 2.1226,
+      "step": 981
+    },
+    {
+      "epoch": 0.4235497088634893,
+      "grad_norm": 0.17068977653980255,
+      "learning_rate": 0.0001433698440050013,
+      "loss": 2.3263,
+      "step": 982
+    },
+    {
+      "epoch": 0.4239810222126375,
+      "grad_norm": 0.15894320607185364,
+      "learning_rate": 0.0001433558075021896,
+      "loss": 2.2735,
+      "step": 983
+    },
+    {
+      "epoch": 0.42441233556178565,
+      "grad_norm": 0.16953733563423157,
+      "learning_rate": 0.00014334175684556046,
+      "loss": 2.3021,
+      "step": 984
+    },
+    {
+      "epoch": 0.4248436489109338,
+      "grad_norm": 0.16865341365337372,
+      "learning_rate": 0.00014332769203802316,
+      "loss": 1.8232,
+      "step": 985
+    },
+    {
+      "epoch": 0.42527496226008193,
+      "grad_norm": 0.17496295273303986,
+      "learning_rate": 0.00014331361308249002,
+      "loss": 2.0722,
+      "step": 986
+    },
+    {
+      "epoch": 0.4257062756092301,
+      "grad_norm": 0.17187760770320892,
+      "learning_rate": 0.00014329951998187624,
+      "loss": 2.1005,
+      "step": 987
+    },
+    {
+      "epoch": 0.4261375889583783,
+      "grad_norm": 0.17174141108989716,
+      "learning_rate": 0.00014328541273909992,
+      "loss": 2.3126,
+      "step": 988
+    },
+    {
+      "epoch": 0.4265689023075264,
+      "grad_norm": 0.16444823145866394,
+      "learning_rate": 0.00014327129135708214,
+      "loss": 2.2856,
+      "step": 989
+    },
+    {
+      "epoch": 0.42700021565667456,
+      "grad_norm": 0.20330820977687836,
+      "learning_rate": 0.00014325715583874687,
+      "loss": 2.0886,
+      "step": 990
+    },
+    {
+      "epoch": 0.4274315290058227,
+      "grad_norm": 0.1728164255619049,
+      "learning_rate": 0.00014324300618702104,
+      "loss": 2.0495,
+      "step": 991
+    },
+    {
+      "epoch": 0.4278628423549709,
+      "grad_norm": 0.17656084895133972,
+      "learning_rate": 0.0001432288424048345,
+      "loss": 2.3372,
+      "step": 992
+    },
+    {
+      "epoch": 0.42829415570411905,
+      "grad_norm": 0.17885622382164001,
+      "learning_rate": 0.00014321466449512,
+      "loss": 2.2132,
+      "step": 993
+    },
+    {
+      "epoch": 0.4287254690532672,
+      "grad_norm": 0.17777030169963837,
+      "learning_rate": 0.00014320047246081321,
+      "loss": 2.4089,
+      "step": 994
+    },
+    {
+      "epoch": 0.42915678240241534,
+      "grad_norm": 0.17604289948940277,
+      "learning_rate": 0.00014318626630485277,
+      "loss": 2.1123,
+      "step": 995
+    },
+    {
+      "epoch": 0.42958809575156354,
+      "grad_norm": 0.15696115791797638,
+      "learning_rate": 0.00014317204603018024,
+      "loss": 2.1366,
+      "step": 996
+    },
+    {
+      "epoch": 0.4300194091007117,
+      "grad_norm": 0.14866608381271362,
+      "learning_rate": 0.00014315781163974008,
+      "loss": 2.2554,
+      "step": 997
+    },
+    {
+      "epoch": 0.4304507224498598,
+      "grad_norm": 0.15694919228553772,
+      "learning_rate": 0.00014314356313647962,
+      "loss": 2.2106,
+      "step": 998
+    },
+    {
+      "epoch": 0.43088203579900797,
+      "grad_norm": 0.1933506727218628,
+      "learning_rate": 0.00014312930052334924,
+      "loss": 2.1608,
+      "step": 999
+    },
+    {
+      "epoch": 0.4313133491481561,
+      "grad_norm": 0.16768963634967804,
+      "learning_rate": 0.0001431150238033021,
+      "loss": 1.8865,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4313133491481561,
+      "eval_loss": 2.1255507469177246,
+      "eval_runtime": 200.2665,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4317446624973043,
+      "grad_norm": 0.17734010517597198,
+      "learning_rate": 0.00014310073297929443,
+      "loss": 2.3392,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43217597584645245,
+      "grad_norm": 0.16111144423484802,
+      "learning_rate": 0.00014308642805428525,
+      "loss": 2.1516,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4326072891956006,
+      "grad_norm": 0.16525249183177948,
+      "learning_rate": 0.00014307210903123654,
+      "loss": 2.1927,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43303860254474874,
+      "grad_norm": 0.20979925990104675,
+      "learning_rate": 0.00014305777591311323,
+      "loss": 2.3105,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43346991589389694,
+      "grad_norm": 0.15769192576408386,
+      "learning_rate": 0.00014304342870288318,
+      "loss": 2.2937,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4339012292430451,
+      "grad_norm": 0.1562070995569229,
+      "learning_rate": 0.00014302906740351707,
+      "loss": 2.2987,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4343325425921932,
+      "grad_norm": 0.17766329646110535,
+      "learning_rate": 0.00014301469201798863,
+      "loss": 2.2952,
+      "step": 1007
+    },
+    {
+      "epoch": 0.43476385594134137,
+      "grad_norm": 0.2226317971944809,
+      "learning_rate": 0.0001430003025492744,
+      "loss": 2.3549,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4351951692904895,
+      "grad_norm": 0.19673429429531097,
+      "learning_rate": 0.00014298589900035388,
+      "loss": 2.3077,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4356264826396377,
+      "grad_norm": 0.16575779020786285,
+      "learning_rate": 0.0001429714813742095,
+      "loss": 2.3121,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43605779598878586,
+      "grad_norm": 0.17023572325706482,
+      "learning_rate": 0.00014295704967382656,
+      "loss": 2.1495,
+      "step": 1011
+    },
+    {
+      "epoch": 0.436489109337934,
+      "grad_norm": 0.17324426770210266,
+      "learning_rate": 0.00014294260390219335,
+      "loss": 2.2113,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43692042268708214,
+      "grad_norm": 0.1725987046957016,
+      "learning_rate": 0.00014292814406230097,
+      "loss": 2.172,
+      "step": 1013
+    },
+    {
+      "epoch": 0.43735173603623034,
+      "grad_norm": 0.16376245021820068,
+      "learning_rate": 0.00014291367015714353,
+      "loss": 2.1641,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4377830493853785,
+      "grad_norm": 0.19950352609157562,
+      "learning_rate": 0.000142899182189718,
+      "loss": 2.2105,
+      "step": 1015
+    },
+    {
+      "epoch": 0.43821436273452663,
+      "grad_norm": 0.16930456459522247,
+      "learning_rate": 0.00014288468016302423,
+      "loss": 2.2622,
+      "step": 1016
+    },
+    {
+      "epoch": 0.4386456760836748,
+      "grad_norm": 0.1521873027086258,
+      "learning_rate": 0.0001428701640800651,
+      "loss": 2.2575,
+      "step": 1017
+    },
+    {
+      "epoch": 0.439076989432823,
+      "grad_norm": 0.1518697440624237,
+      "learning_rate": 0.00014285563394384623,
+      "loss": 2.162,
+      "step": 1018
+    },
+    {
+      "epoch": 0.4395083027819711,
+      "grad_norm": 0.1935448795557022,
+      "learning_rate": 0.00014284108975737636,
+      "loss": 2.3691,
+      "step": 1019
+    },
+    {
+      "epoch": 0.43993961613111926,
+      "grad_norm": 0.18544535338878632,
+      "learning_rate": 0.0001428265315236669,
+      "loss": 2.2723,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4403709294802674,
+      "grad_norm": 0.18568457663059235,
+      "learning_rate": 0.00014281195924573236,
+      "loss": 2.3239,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44080224282941555,
+      "grad_norm": 0.1548377275466919,
+      "learning_rate": 0.00014279737292659012,
+      "loss": 2.0073,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44123355617856375,
+      "grad_norm": 0.2214689552783966,
+      "learning_rate": 0.00014278277256926037,
+      "loss": 2.3241,
+      "step": 1023
+    },
+    {
+      "epoch": 0.4416648695277119,
+      "grad_norm": 0.1754104644060135,
+      "learning_rate": 0.0001427681581767663,
+      "loss": 2.2573,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44209618287686003,
+      "grad_norm": 0.15047457814216614,
+      "learning_rate": 0.00014275352975213397,
+      "loss": 2.1531,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44209618287686003,
+      "eval_loss": 2.1244282722473145,
+      "eval_runtime": 200.1144,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 1025
+    },
+    {
+      "epoch": 0.4425274962260082,
+      "grad_norm": 0.1728946715593338,
+      "learning_rate": 0.00014273888729839237,
+      "loss": 2.3304,
+      "step": 1026
+    },
+    {
+      "epoch": 0.4429588095751564,
+      "grad_norm": 0.16061633825302124,
+      "learning_rate": 0.00014272423081857337,
+      "loss": 2.2651,
+      "step": 1027
+    },
+    {
+      "epoch": 0.4433901229243045,
+      "grad_norm": 0.1639309525489807,
+      "learning_rate": 0.00014270956031571178,
+      "loss": 1.9692,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44382143627345266,
+      "grad_norm": 0.18484780192375183,
+      "learning_rate": 0.00014269487579284524,
+      "loss": 2.4299,
+      "step": 1029
+    },
+    {
+      "epoch": 0.4442527496226008,
+      "grad_norm": 0.17049303650856018,
+      "learning_rate": 0.00014268017725301439,
+      "loss": 2.1617,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44468406297174895,
+      "grad_norm": 0.148739755153656,
+      "learning_rate": 0.0001426654646992627,
+      "loss": 2.2107,
+      "step": 1031
+    },
+    {
+      "epoch": 0.44511537632089715,
+      "grad_norm": 0.1590275764465332,
+      "learning_rate": 0.00014265073813463655,
+      "loss": 2.2585,
+      "step": 1032
+    },
+    {
+      "epoch": 0.4455466896700453,
+      "grad_norm": 0.20391826331615448,
+      "learning_rate": 0.00014263599756218526,
+      "loss": 2.3769,
+      "step": 1033
+    },
+    {
+      "epoch": 0.44597800301919344,
+      "grad_norm": 0.22355666756629944,
+      "learning_rate": 0.00014262124298496104,
+      "loss": 2.1261,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4464093163683416,
+      "grad_norm": 0.1847553700208664,
+      "learning_rate": 0.00014260647440601895,
+      "loss": 2.2571,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4468406297174898,
+      "grad_norm": 0.16822926700115204,
+      "learning_rate": 0.000142591691828417,
+      "loss": 2.0749,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4472719430666379,
+      "grad_norm": 0.2041289359331131,
+      "learning_rate": 0.00014257689525521607,
+      "loss": 2.3264,
+      "step": 1037
+    },
+    {
+      "epoch": 0.44770325641578607,
+      "grad_norm": 0.20407123863697052,
+      "learning_rate": 0.00014256208468948,
+      "loss": 2.2982,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4481345697649342,
+      "grad_norm": 0.18789510428905487,
+      "learning_rate": 0.00014254726013427545,
+      "loss": 2.1637,
+      "step": 1039
+    },
+    {
+      "epoch": 0.44856588311408235,
+      "grad_norm": 0.25421515107154846,
+      "learning_rate": 0.00014253242159267197,
+      "loss": 2.423,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44899719646323055,
+      "grad_norm": 3.8317348957061768,
+      "learning_rate": 0.0001425175690677421,
+      "loss": 2.121,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4494285098123787,
+      "grad_norm": 0.20088981091976166,
+      "learning_rate": 0.00014250270256256119,
+      "loss": 2.272,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44985982316152684,
+      "grad_norm": 3.788172721862793,
+      "learning_rate": 0.00014248782208020754,
+      "loss": 2.293,
+      "step": 1043
+    },
+    {
+      "epoch": 0.450291136510675,
+      "grad_norm": 0.1680663675069809,
+      "learning_rate": 0.00014247292762376227,
+      "loss": 2.073,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4507224498598232,
+      "grad_norm": 0.1550556868314743,
+      "learning_rate": 0.00014245801919630946,
+      "loss": 2.2315,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4511537632089713,
+      "grad_norm": 0.17220419645309448,
+      "learning_rate": 0.00014244309680093607,
+      "loss": 2.3914,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45158507655811947,
+      "grad_norm": 0.7895448207855225,
+      "learning_rate": 0.00014242816044073196,
+      "loss": 2.0289,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4520163899072676,
+      "grad_norm": 0.16854602098464966,
+      "learning_rate": 0.00014241321011878983,
+      "loss": 2.2614,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4524477032564158,
+      "grad_norm": 0.16735167801380157,
+      "learning_rate": 0.00014239824583820533,
+      "loss": 2.1894,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45287901660556396,
+      "grad_norm": 0.1836962252855301,
+      "learning_rate": 0.00014238326760207697,
+      "loss": 2.2541,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45287901660556396,
+      "eval_loss": 2.1253392696380615,
+      "eval_runtime": 199.256,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4533103299547121,
+      "grad_norm": 0.16606397926807404,
+      "learning_rate": 0.00014236827541350619,
+      "loss": 2.1566,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45374164330386024,
+      "grad_norm": 0.15843674540519714,
+      "learning_rate": 0.00014235326927559725,
+      "loss": 2.1098,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4541729566530084,
+      "grad_norm": 0.160728320479393,
+      "learning_rate": 0.0001423382491914573,
+      "loss": 2.0559,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4546042700021566,
+      "grad_norm": 0.16055908799171448,
+      "learning_rate": 0.00014232321516419652,
+      "loss": 2.2103,
+      "step": 1054
+    },
+    {
+      "epoch": 0.45503558335130473,
+      "grad_norm": 0.16064168512821198,
+      "learning_rate": 0.00014230816719692775,
+      "loss": 2.3539,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4554668967004529,
+      "grad_norm": 0.19479875266551971,
+      "learning_rate": 0.00014229310529276692,
+      "loss": 2.179,
+      "step": 1056
+    },
+    {
+      "epoch": 0.455898210049601,
+      "grad_norm": 0.15349072217941284,
+      "learning_rate": 0.00014227802945483272,
+      "loss": 2.0798,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4563295233987492,
+      "grad_norm": 0.15733902156352997,
+      "learning_rate": 0.00014226293968624683,
+      "loss": 2.187,
+      "step": 1058
+    },
+    {
+      "epoch": 0.45676083674789736,
+      "grad_norm": 0.1859322190284729,
+      "learning_rate": 0.00014224783599013364,
+      "loss": 2.1991,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4571921500970455,
+      "grad_norm": 0.22436240315437317,
+      "learning_rate": 0.00014223271836962063,
+      "loss": 2.127,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45762346344619365,
+      "grad_norm": 0.16210895776748657,
+      "learning_rate": 0.00014221758682783804,
+      "loss": 2.2366,
+      "step": 1061
+    },
+    {
+      "epoch": 0.4580547767953418,
+      "grad_norm": 0.16493889689445496,
+      "learning_rate": 0.000142202441367919,
+      "loss": 2.1817,
+      "step": 1062
+    },
+    {
+      "epoch": 0.45848609014449,
+      "grad_norm": 0.16778267920017242,
+      "learning_rate": 0.0001421872819929996,
+      "loss": 2.2076,
+      "step": 1063
+    },
+    {
+      "epoch": 0.45891740349363813,
+      "grad_norm": 0.15672996640205383,
+      "learning_rate": 0.00014217210870621865,
+      "loss": 2.0867,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4593487168427863,
+      "grad_norm": 0.18593546748161316,
+      "learning_rate": 0.00014215692151071805,
+      "loss": 2.0031,
+      "step": 1065
+    },
+    {
+      "epoch": 0.4597800301919344,
+      "grad_norm": 0.17268811166286469,
+      "learning_rate": 0.00014214172040964243,
+      "loss": 2.204,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4602113435410826,
+      "grad_norm": 0.1761752963066101,
+      "learning_rate": 0.00014212650540613937,
+      "loss": 2.2918,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46064265689023076,
+      "grad_norm": 0.18261955678462982,
+      "learning_rate": 0.00014211127650335927,
+      "loss": 2.1724,
+      "step": 1068
+    },
+    {
+      "epoch": 0.4610739702393789,
+      "grad_norm": 0.19669634103775024,
+      "learning_rate": 0.00014209603370445544,
+      "loss": 1.9778,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46150528358852705,
+      "grad_norm": 0.17395718395709991,
+      "learning_rate": 0.0001420807770125841,
+      "loss": 2.1735,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46193659693767525,
+      "grad_norm": 0.1703193187713623,
+      "learning_rate": 0.00014206550643090427,
+      "loss": 2.2564,
+      "step": 1071
+    },
+    {
+      "epoch": 0.4623679102868234,
+      "grad_norm": 0.4253257215023041,
+      "learning_rate": 0.00014205022196257794,
+      "loss": 2.2818,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46279922363597154,
+      "grad_norm": 0.16150128841400146,
+      "learning_rate": 0.0001420349236107699,
+      "loss": 2.1974,
+      "step": 1073
+    },
+    {
+      "epoch": 0.4632305369851197,
+      "grad_norm": 0.16813993453979492,
+      "learning_rate": 0.00014201961137864783,
+      "loss": 2.0688,
+      "step": 1074
+    },
+    {
+      "epoch": 0.4636618503342678,
+      "grad_norm": 0.21386326849460602,
+      "learning_rate": 0.00014200428526938233,
+      "loss": 2.3743,
+      "step": 1075
+    },
+    {
+      "epoch": 0.4636618503342678,
+      "eval_loss": 2.1256213188171387,
+      "eval_runtime": 205.8902,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 1075
+    },
+    {
+      "epoch": 0.464093163683416,
+      "grad_norm": 0.18708860874176025,
+      "learning_rate": 0.0001419889452861468,
+      "loss": 2.2642,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46452447703256416,
+      "grad_norm": 0.18455776572227478,
+      "learning_rate": 0.0001419735914321176,
+      "loss": 2.2839,
+      "step": 1077
+    },
+    {
+      "epoch": 0.4649557903817123,
+      "grad_norm": 0.15820108354091644,
+      "learning_rate": 0.00014195822371047384,
+      "loss": 2.162,
+      "step": 1078
+    },
+    {
+      "epoch": 0.46538710373086045,
+      "grad_norm": 0.17975345253944397,
+      "learning_rate": 0.00014194284212439763,
+      "loss": 2.0403,
+      "step": 1079
+    },
+    {
+      "epoch": 0.46581841708000865,
+      "grad_norm": 0.18395720422267914,
+      "learning_rate": 0.0001419274466770739,
+      "loss": 2.1124,
+      "step": 1080
+    },
+    {
+      "epoch": 0.4662497304291568,
+      "grad_norm": 0.158233180642128,
+      "learning_rate": 0.00014191203737169044,
+      "loss": 1.8722,
+      "step": 1081
+    },
+    {
+      "epoch": 0.46668104377830494,
+      "grad_norm": 0.16250333189964294,
+      "learning_rate": 0.0001418966142114379,
+      "loss": 2.0023,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4671123571274531,
+      "grad_norm": 0.16339930891990662,
+      "learning_rate": 0.0001418811771995098,
+      "loss": 2.2289,
+      "step": 1083
+    },
+    {
+      "epoch": 0.4675436704766012,
+      "grad_norm": 0.1458890587091446,
+      "learning_rate": 0.00014186572633910258,
+      "loss": 1.9933,
+      "step": 1084
+    },
+    {
+      "epoch": 0.4679749838257494,
+      "grad_norm": 0.16976583003997803,
+      "learning_rate": 0.00014185026163341546,
+      "loss": 2.0317,
+      "step": 1085
+    },
+    {
+      "epoch": 0.46840629717489757,
+      "grad_norm": 0.1867656409740448,
+      "learning_rate": 0.00014183478308565063,
+      "loss": 2.2477,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4688376105240457,
+      "grad_norm": 0.16612374782562256,
+      "learning_rate": 0.00014181929069901307,
+      "loss": 2.2293,
+      "step": 1087
+    },
+    {
+      "epoch": 0.46926892387319386,
+      "grad_norm": 0.16322407126426697,
+      "learning_rate": 0.00014180378447671064,
+      "loss": 2.2751,
+      "step": 1088
+    },
+    {
+      "epoch": 0.46970023722234205,
+      "grad_norm": 0.20379450917243958,
+      "learning_rate": 0.0001417882644219541,
+      "loss": 2.1489,
+      "step": 1089
+    },
+    {
+      "epoch": 0.4701315505714902,
+      "grad_norm": 0.18934619426727295,
+      "learning_rate": 0.00014177273053795697,
+      "loss": 2.5549,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47056286392063834,
+      "grad_norm": 0.17536357045173645,
+      "learning_rate": 0.00014175718282793585,
+      "loss": 2.2638,
+      "step": 1091
+    },
+    {
+      "epoch": 0.4709941772697865,
+      "grad_norm": 0.17303945124149323,
+      "learning_rate": 0.00014174162129510992,
+      "loss": 2.0555,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47142549061893463,
+      "grad_norm": 0.16105377674102783,
+      "learning_rate": 0.00014172604594270144,
+      "loss": 2.2979,
+      "step": 1093
+    },
+    {
+      "epoch": 0.4718568039680828,
+      "grad_norm": 0.16773848235607147,
+      "learning_rate": 0.00014171045677393543,
+      "loss": 2.1319,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47228811731723097,
+      "grad_norm": 0.16544897854328156,
+      "learning_rate": 0.00014169485379203985,
+      "loss": 2.2134,
+      "step": 1095
+    },
+    {
+      "epoch": 0.4727194306663791,
+      "grad_norm": 0.3179582953453064,
+      "learning_rate": 0.0001416792370002454,
+      "loss": 2.174,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47315074401552726,
+      "grad_norm": 0.16636958718299866,
+      "learning_rate": 0.00014166360640178576,
+      "loss": 2.188,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47358205736467546,
+      "grad_norm": 0.15930014848709106,
+      "learning_rate": 0.0001416479619998974,
+      "loss": 1.9799,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4740133707138236,
+      "grad_norm": 0.17669852077960968,
+      "learning_rate": 0.00014163230379781963,
+      "loss": 2.0805,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47444468406297174,
+      "grad_norm": 0.1416047066450119,
+      "learning_rate": 0.00014161663179879469,
+      "loss": 2.3382,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47444468406297174,
+      "eval_loss": 2.1245031356811523,
+      "eval_runtime": 196.9672,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4748759974121199,
+      "grad_norm": 0.17482198774814606,
+      "learning_rate": 0.00014160094600606762,
+      "loss": 2.2444,
+      "step": 1101
+    },
+    {
+      "epoch": 0.4753073107612681,
+      "grad_norm": 0.176433727145195,
+      "learning_rate": 0.00014158524642288634,
+      "loss": 2.2817,
+      "step": 1102
+    },
+    {
+      "epoch": 0.47573862411041623,
+      "grad_norm": 0.16114936769008636,
+      "learning_rate": 0.00014156953305250166,
+      "loss": 2.2974,
+      "step": 1103
+    },
+    {
+      "epoch": 0.4761699374595644,
+      "grad_norm": 0.16793087124824524,
+      "learning_rate": 0.00014155380589816715,
+      "loss": 2.1037,
+      "step": 1104
+    },
+    {
+      "epoch": 0.4766012508087125,
+      "grad_norm": 0.1911628097295761,
+      "learning_rate": 0.0001415380649631393,
+      "loss": 2.2625,
+      "step": 1105
+    },
+    {
+      "epoch": 0.47703256415786066,
+      "grad_norm": 0.25380977988243103,
+      "learning_rate": 0.00014152231025067746,
+      "loss": 2.1,
+      "step": 1106
+    },
+    {
+      "epoch": 0.47746387750700886,
+      "grad_norm": 0.1790010780096054,
+      "learning_rate": 0.0001415065417640438,
+      "loss": 2.0709,
+      "step": 1107
+    },
+    {
+      "epoch": 0.477895190856157,
+      "grad_norm": 0.15084779262542725,
+      "learning_rate": 0.00014149075950650335,
+      "loss": 2.0395,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47832650420530515,
+      "grad_norm": 0.17813463509082794,
+      "learning_rate": 0.00014147496348132404,
+      "loss": 2.4109,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4787578175544533,
+      "grad_norm": 0.17824919521808624,
+      "learning_rate": 0.00014145915369177657,
+      "loss": 2.058,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4791891309036015,
+      "grad_norm": 0.17649079859256744,
+      "learning_rate": 0.00014144333014113456,
+      "loss": 2.0866,
+      "step": 1111
+    },
+    {
+      "epoch": 0.47962044425274963,
+      "grad_norm": 0.1977768838405609,
+      "learning_rate": 0.00014142749283267442,
+      "loss": 2.1816,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4800517576018978,
+      "grad_norm": 0.17968136072158813,
+      "learning_rate": 0.00014141164176967544,
+      "loss": 2.2179,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4804830709510459,
+      "grad_norm": 0.16740866005420685,
+      "learning_rate": 0.00014139577695541976,
+      "loss": 2.2603,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48091438430019406,
+      "grad_norm": 0.16543352603912354,
+      "learning_rate": 0.0001413798983931924,
+      "loss": 2.271,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48134569764934226,
+      "grad_norm": 0.1757402867078781,
+      "learning_rate": 0.00014136400608628113,
+      "loss": 2.3212,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4817770109984904,
+      "grad_norm": 0.20364037156105042,
+      "learning_rate": 0.0001413481000379767,
+      "loss": 2.2111,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48220832434763855,
+      "grad_norm": 0.18235564231872559,
+      "learning_rate": 0.00014133218025157255,
+      "loss": 2.6066,
+      "step": 1118
+    },
+    {
+      "epoch": 0.4826396376967867,
+      "grad_norm": 0.16168133914470673,
+      "learning_rate": 0.0001413162467303651,
+      "loss": 2.189,
+      "step": 1119
+    },
+    {
+      "epoch": 0.4830709510459349,
+      "grad_norm": 0.17553839087486267,
+      "learning_rate": 0.00014130029947765352,
+      "loss": 2.1359,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48350226439508304,
+      "grad_norm": 0.17774251103401184,
+      "learning_rate": 0.0001412843384967399,
+      "loss": 2.4902,
+      "step": 1121
+    },
+    {
+      "epoch": 0.4839335777442312,
+      "grad_norm": 0.17376278340816498,
+      "learning_rate": 0.0001412683637909291,
+      "loss": 2.2022,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4843648910933793,
+      "grad_norm": 0.18041923642158508,
+      "learning_rate": 0.0001412523753635289,
+      "loss": 2.3228,
+      "step": 1123
+    },
+    {
+      "epoch": 0.4847962044425275,
+      "grad_norm": 0.2041550576686859,
+      "learning_rate": 0.00014123637321784987,
+      "loss": 2.4281,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48522751779167567,
+      "grad_norm": 0.1704571694135666,
+      "learning_rate": 0.00014122035735720538,
+      "loss": 2.1997,
+      "step": 1125
+    },
+    {
+      "epoch": 0.48522751779167567,
+      "eval_loss": 2.1237313747406006,
+      "eval_runtime": 197.9828,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1125
+    },
+    {
+      "epoch": 0.4856588311408238,
+      "grad_norm": 0.18956632912158966,
+      "learning_rate": 0.00014120432778491176,
+      "loss": 2.2244,
+      "step": 1126
+    },
+    {
+      "epoch": 0.48609014448997195,
+      "grad_norm": 0.16811709105968475,
+      "learning_rate": 0.00014118828450428807,
+      "loss": 2.1037,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4865214578391201,
+      "grad_norm": 0.20404136180877686,
+      "learning_rate": 0.00014117222751865621,
+      "loss": 2.3572,
+      "step": 1128
+    },
+    {
+      "epoch": 0.4869527711882683,
+      "grad_norm": 0.1897653490304947,
+      "learning_rate": 0.00014115615683134105,
+      "loss": 2.0607,
+      "step": 1129
+    },
+    {
+      "epoch": 0.48738408453741644,
+      "grad_norm": 0.20207540690898895,
+      "learning_rate": 0.00014114007244567012,
+      "loss": 2.3401,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4878153978865646,
+      "grad_norm": 0.16764988005161285,
+      "learning_rate": 0.0001411239743649739,
+      "loss": 2.1391,
+      "step": 1131
+    },
+    {
+      "epoch": 0.4882467112357127,
+      "grad_norm": 0.15346284210681915,
+      "learning_rate": 0.00014110786259258565,
+      "loss": 2.1538,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4886780245848609,
+      "grad_norm": 0.17945681512355804,
+      "learning_rate": 0.00014109173713184155,
+      "loss": 2.2123,
+      "step": 1133
+    },
+    {
+      "epoch": 0.48910933793400907,
+      "grad_norm": 0.1814691722393036,
+      "learning_rate": 0.00014107559798608047,
+      "loss": 2.1695,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4895406512831572,
+      "grad_norm": 0.17287328839302063,
+      "learning_rate": 0.00014105944515864426,
+      "loss": 2.2417,
+      "step": 1135
+    },
+    {
+      "epoch": 0.48997196463230536,
+      "grad_norm": 0.17122970521450043,
+      "learning_rate": 0.00014104327865287748,
+      "loss": 2.0969,
+      "step": 1136
+    },
+    {
+      "epoch": 0.4904032779814535,
+      "grad_norm": 0.19306020438671112,
+      "learning_rate": 0.00014102709847212765,
+      "loss": 2.4507,
+      "step": 1137
+    },
+    {
+      "epoch": 0.4908345913306017,
+      "grad_norm": 0.17338979244232178,
+      "learning_rate": 0.000141010904619745,
+      "loss": 2.2043,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49126590467974984,
+      "grad_norm": 0.19331465661525726,
+      "learning_rate": 0.0001409946970990827,
+      "loss": 2.2461,
+      "step": 1139
+    },
+    {
+      "epoch": 0.491697218028898,
+      "grad_norm": 0.1610681712627411,
+      "learning_rate": 0.00014097847591349662,
+      "loss": 2.091,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49212853137804613,
+      "grad_norm": 0.17557750642299652,
+      "learning_rate": 0.0001409622410663456,
+      "loss": 2.1339,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49255984472719433,
+      "grad_norm": 0.16115573048591614,
+      "learning_rate": 0.00014094599256099123,
+      "loss": 2.1913,
+      "step": 1142
+    },
+    {
+      "epoch": 0.4929911580763425,
+      "grad_norm": 0.15505258738994598,
+      "learning_rate": 0.0001409297304007979,
+      "loss": 2.1831,
+      "step": 1143
+    },
+    {
+      "epoch": 0.4934224714254906,
+      "grad_norm": 0.1573905646800995,
+      "learning_rate": 0.00014091345458913288,
+      "loss": 2.2225,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49385378477463876,
+      "grad_norm": 0.15298591554164886,
+      "learning_rate": 0.00014089716512936628,
+      "loss": 2.1078,
+      "step": 1145
+    },
+    {
+      "epoch": 0.4942850981237869,
+      "grad_norm": 0.14728635549545288,
+      "learning_rate": 0.00014088086202487102,
+      "loss": 2.034,
+      "step": 1146
+    },
+    {
+      "epoch": 0.4947164114729351,
+      "grad_norm": 0.18687205016613007,
+      "learning_rate": 0.0001408645452790228,
+      "loss": 2.2212,
+      "step": 1147
+    },
+    {
+      "epoch": 0.49514772482208325,
+      "grad_norm": 0.22122670710086823,
+      "learning_rate": 0.00014084821489520022,
+      "loss": 2.1835,
+      "step": 1148
+    },
+    {
+      "epoch": 0.4955790381712314,
+      "grad_norm": 0.1600857973098755,
+      "learning_rate": 0.00014083187087678462,
+      "loss": 2.3065,
+      "step": 1149
+    },
+    {
+      "epoch": 0.49601035152037953,
+      "grad_norm": 0.15622857213020325,
+      "learning_rate": 0.00014081551322716023,
+      "loss": 2.1417,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49601035152037953,
+      "eval_loss": 2.122781753540039,
+      "eval_runtime": 198.1311,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49644166486952773,
+      "grad_norm": 0.159952774643898,
+      "learning_rate": 0.0001407991419497141,
+      "loss": 2.3553,
+      "step": 1151
+    },
+    {
+      "epoch": 0.4968729782186759,
+      "grad_norm": 0.17937713861465454,
+      "learning_rate": 0.00014078275704783604,
+      "loss": 2.2021,
+      "step": 1152
+    },
+    {
+      "epoch": 0.497304291567824,
+      "grad_norm": 0.16701209545135498,
+      "learning_rate": 0.00014076635852491878,
+      "loss": 2.0779,
+      "step": 1153
+    },
+    {
+      "epoch": 0.49773560491697216,
+      "grad_norm": 0.19412510097026825,
+      "learning_rate": 0.00014074994638435775,
+      "loss": 1.895,
+      "step": 1154
+    },
+    {
+      "epoch": 0.49816691826612036,
+      "grad_norm": 0.17660222947597504,
+      "learning_rate": 0.00014073352062955128,
+      "loss": 2.2594,
+      "step": 1155
+    },
+    {
+      "epoch": 0.4985982316152685,
+      "grad_norm": 0.15943753719329834,
+      "learning_rate": 0.00014071708126390053,
+      "loss": 2.4051,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49902954496441665,
+      "grad_norm": 0.165870800614357,
+      "learning_rate": 0.0001407006282908094,
+      "loss": 2.338,
+      "step": 1157
+    },
+    {
+      "epoch": 0.4994608583135648,
+      "grad_norm": 0.16600917279720306,
+      "learning_rate": 0.00014068416171368476,
+      "loss": 2.0591,
+      "step": 1158
+    },
+    {
+      "epoch": 0.49989217166271294,
+      "grad_norm": 0.18140505254268646,
+      "learning_rate": 0.00014066768153593609,
+      "loss": 2.4228,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5003234850118611,
+      "grad_norm": 0.17261415719985962,
+      "learning_rate": 0.0001406511877609758,
+      "loss": 2.1168,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5007547983610092,
+      "grad_norm": 0.17106929421424866,
+      "learning_rate": 0.0001406346803922192,
+      "loss": 2.1904,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5011861117101575,
+      "grad_norm": 0.16084757447242737,
+      "learning_rate": 0.00014061815943308424,
+      "loss": 2.2569,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5016174250593056,
+      "grad_norm": 0.17182020843029022,
+      "learning_rate": 0.00014060162488699175,
+      "loss": 2.0081,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5020487384084538,
+      "grad_norm": 0.16303688287734985,
+      "learning_rate": 0.00014058507675736542,
+      "loss": 2.141,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5024800517576019,
+      "grad_norm": 0.17593824863433838,
+      "learning_rate": 0.00014056851504763175,
+      "loss": 2.2078,
+      "step": 1165
+    },
+    {
+      "epoch": 0.50291136510675,
+      "grad_norm": 0.17629596590995789,
+      "learning_rate": 0.00014055193976121998,
+      "loss": 2.2364,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5033426784558982,
+      "grad_norm": 0.19598311185836792,
+      "learning_rate": 0.0001405353509015622,
+      "loss": 2.2881,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5037739918050463,
+      "grad_norm": 0.1822650283575058,
+      "learning_rate": 0.00014051874847209334,
+      "loss": 2.3105,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5042053051541945,
+      "grad_norm": 0.2322334349155426,
+      "learning_rate": 0.00014050213247625114,
+      "loss": 2.1919,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5046366185033426,
+      "grad_norm": 0.173886239528656,
+      "learning_rate": 0.00014048550291747606,
+      "loss": 2.2836,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5050679318524909,
+      "grad_norm": 0.23277299106121063,
+      "learning_rate": 0.0001404688597992115,
+      "loss": 2.2851,
+      "step": 1171
+    },
+    {
+      "epoch": 0.505499245201639,
+      "grad_norm": 0.1585807353258133,
+      "learning_rate": 0.0001404522031249035,
+      "loss": 2.3763,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5059305585507872,
+      "grad_norm": 0.17657655477523804,
+      "learning_rate": 0.00014043553289800112,
+      "loss": 2.287,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5063618718999353,
+      "grad_norm": 0.1734786033630371,
+      "learning_rate": 0.00014041884912195608,
+      "loss": 2.1696,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5067931852490835,
+      "grad_norm": 0.1727166771888733,
+      "learning_rate": 0.0001404021518002229,
+      "loss": 2.1269,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5067931852490835,
+      "eval_loss": 2.1227898597717285,
+      "eval_runtime": 198.1515,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5072244985982316,
+      "grad_norm": 0.19915609061717987,
+      "learning_rate": 0.00014038544093625892,
+      "loss": 2.1747,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5076558119473797,
+      "grad_norm": 0.15546812117099762,
+      "learning_rate": 0.0001403687165335244,
+      "loss": 2.0218,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5080871252965279,
+      "grad_norm": 0.17874981462955475,
+      "learning_rate": 0.00014035197859548225,
+      "loss": 2.4689,
+      "step": 1178
+    },
+    {
+      "epoch": 0.508518438645676,
+      "grad_norm": 0.15583276748657227,
+      "learning_rate": 0.00014033522712559825,
+      "loss": 2.0931,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5089497519948243,
+      "grad_norm": 0.19189707934856415,
+      "learning_rate": 0.00014031846212734096,
+      "loss": 2.2619,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5093810653439724,
+      "grad_norm": 0.20342782139778137,
+      "learning_rate": 0.00014030168360418182,
+      "loss": 2.0796,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5098123786931206,
+      "grad_norm": 0.17643365263938904,
+      "learning_rate": 0.00014028489155959494,
+      "loss": 2.274,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5102436920422687,
+      "grad_norm": 0.1894954890012741,
+      "learning_rate": 0.00014026808599705733,
+      "loss": 2.2807,
+      "step": 1183
+    },
+    {
+      "epoch": 0.5106750053914169,
+      "grad_norm": 0.17265945672988892,
+      "learning_rate": 0.00014025126692004874,
+      "loss": 2.2272,
+      "step": 1184
+    },
+    {
+      "epoch": 0.511106318740565,
+      "grad_norm": 0.2056184709072113,
+      "learning_rate": 0.00014023443433205176,
+      "loss": 2.2147,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5115376320897131,
+      "grad_norm": 0.16164915263652802,
+      "learning_rate": 0.00014021758823655174,
+      "loss": 1.9599,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5119689454388613,
+      "grad_norm": 0.1566849946975708,
+      "learning_rate": 0.0001402007286370369,
+      "loss": 2.0889,
+      "step": 1187
+    },
+    {
+      "epoch": 0.5124002587880095,
+      "grad_norm": 0.17637118697166443,
+      "learning_rate": 0.00014018385553699818,
+      "loss": 2.2879,
+      "step": 1188
+    },
+    {
+      "epoch": 0.5128315721371577,
+      "grad_norm": 0.18087397515773773,
+      "learning_rate": 0.0001401669689399293,
+      "loss": 2.2892,
+      "step": 1189
+    },
+    {
+      "epoch": 0.5132628854863058,
+      "grad_norm": 0.19656838476657867,
+      "learning_rate": 0.00014015006884932687,
+      "loss": 2.3285,
+      "step": 1190
+    },
+    {
+      "epoch": 0.513694198835454,
+      "grad_norm": 0.1986808478832245,
+      "learning_rate": 0.0001401331552686902,
+      "loss": 2.2076,
+      "step": 1191
+    },
+    {
+      "epoch": 0.5141255121846021,
+      "grad_norm": 0.15159070491790771,
+      "learning_rate": 0.00014011622820152145,
+      "loss": 2.0549,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5145568255337503,
+      "grad_norm": 0.17124521732330322,
+      "learning_rate": 0.00014009928765132556,
+      "loss": 2.2253,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5149881388828984,
+      "grad_norm": 0.19993092119693756,
+      "learning_rate": 0.00014008233362161024,
+      "loss": 2.1887,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5154194522320465,
+      "grad_norm": 0.17296721041202545,
+      "learning_rate": 0.00014006536611588602,
+      "loss": 2.2939,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5158507655811947,
+      "grad_norm": 0.15552176535129547,
+      "learning_rate": 0.0001400483851376662,
+      "loss": 2.1718,
+      "step": 1196
+    },
+    {
+      "epoch": 0.516282078930343,
+      "grad_norm": 0.17558616399765015,
+      "learning_rate": 0.00014003139069046692,
+      "loss": 2.367,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5167133922794911,
+      "grad_norm": 0.15603621304035187,
+      "learning_rate": 0.00014001438277780697,
+      "loss": 1.8832,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5171447056286392,
+      "grad_norm": 0.16005119681358337,
+      "learning_rate": 0.0001399973614032081,
+      "loss": 1.8694,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5175760189777874,
+      "grad_norm": 0.17181624472141266,
+      "learning_rate": 0.00013998032657019474,
+      "loss": 2.3409,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5175760189777874,
+      "eval_loss": 2.1217408180236816,
+      "eval_runtime": 199.1599,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5180073323269355,
+      "grad_norm": 0.15452653169631958,
+      "learning_rate": 0.00013996327828229418,
+      "loss": 2.1593,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5184386456760837,
+      "grad_norm": 0.16115358471870422,
+      "learning_rate": 0.0001399462165430364,
+      "loss": 2.1336,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5188699590252318,
+      "grad_norm": 0.16711373627185822,
+      "learning_rate": 0.00013992914135595427,
+      "loss": 2.3766,
+      "step": 1203
+    },
+    {
+      "epoch": 0.51930127237438,
+      "grad_norm": 0.1763235479593277,
+      "learning_rate": 0.00013991205272458334,
+      "loss": 2.4137,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5197325857235281,
+      "grad_norm": 0.16084948182106018,
+      "learning_rate": 0.00013989495065246205,
+      "loss": 2.1188,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5201638990726764,
+      "grad_norm": 0.17245417833328247,
+      "learning_rate": 0.00013987783514313155,
+      "loss": 2.2178,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5205952124218245,
+      "grad_norm": 0.1630506068468094,
+      "learning_rate": 0.00013986070620013576,
+      "loss": 2.073,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5210265257709726,
+      "grad_norm": 0.17694434523582458,
+      "learning_rate": 0.00013984356382702146,
+      "loss": 2.1957,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5214578391201208,
+      "grad_norm": 0.18180590867996216,
+      "learning_rate": 0.00013982640802733817,
+      "loss": 2.4528,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5218891524692689,
+      "grad_norm": 0.24107345938682556,
+      "learning_rate": 0.00013980923880463816,
+      "loss": 2.2918,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5223204658184171,
+      "grad_norm": 0.15491077303886414,
+      "learning_rate": 0.0001397920561624765,
+      "loss": 2.0963,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5227517791675652,
+      "grad_norm": 0.1684110015630722,
+      "learning_rate": 0.00013977486010441107,
+      "loss": 2.2447,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5231830925167134,
+      "grad_norm": 0.16640383005142212,
+      "learning_rate": 0.0001397576506340025,
+      "loss": 2.1924,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5236144058658615,
+      "grad_norm": 0.18121518194675446,
+      "learning_rate": 0.00013974042775481417,
+      "loss": 2.2443,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5240457192150098,
+      "grad_norm": 0.17246213555335999,
+      "learning_rate": 0.0001397231914704123,
+      "loss": 2.2232,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5244770325641579,
+      "grad_norm": 0.18331019580364227,
+      "learning_rate": 0.0001397059417843658,
+      "loss": 2.0938,
+      "step": 1216
+    },
+    {
+      "epoch": 0.524908345913306,
+      "grad_norm": 0.17655853927135468,
+      "learning_rate": 0.00013968867870024648,
+      "loss": 2.2938,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5253396592624542,
+      "grad_norm": 0.17127923667430878,
+      "learning_rate": 0.0001396714022216288,
+      "loss": 2.2469,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5257709726116023,
+      "grad_norm": 0.1631791591644287,
+      "learning_rate": 0.00013965411235209005,
+      "loss": 2.2233,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5262022859607505,
+      "grad_norm": 0.15704067051410675,
+      "learning_rate": 0.0001396368090952103,
+      "loss": 2.1231,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5266335993098986,
+      "grad_norm": 0.17695368826389313,
+      "learning_rate": 0.00013961949245457235,
+      "loss": 2.2012,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5270649126590468,
+      "grad_norm": 0.1563882827758789,
+      "learning_rate": 0.00013960216243376184,
+      "loss": 2.0977,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5274962260081949,
+      "grad_norm": 0.16035982966423035,
+      "learning_rate": 0.00013958481903636713,
+      "loss": 2.1171,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5279275393573432,
+      "grad_norm": 0.16886377334594727,
+      "learning_rate": 0.00013956746226597935,
+      "loss": 2.3124,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5283588527064913,
+      "grad_norm": 0.1622340977191925,
+      "learning_rate": 0.0001395500921261924,
+      "loss": 2.339,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5283588527064913,
+      "eval_loss": 2.120978832244873,
+      "eval_runtime": 197.7462,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5287901660556394,
+      "grad_norm": 0.16553667187690735,
+      "learning_rate": 0.00013953270862060294,
+      "loss": 2.2944,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5292214794047876,
+      "grad_norm": 0.15495768189430237,
+      "learning_rate": 0.00013951531175281046,
+      "loss": 2.2237,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5296527927539357,
+      "grad_norm": 0.16204552352428436,
+      "learning_rate": 0.0001394979015264172,
+      "loss": 2.3077,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5300841061030839,
+      "grad_norm": 0.17178687453269958,
+      "learning_rate": 0.00013948047794502802,
+      "loss": 2.2002,
+      "step": 1229
+    },
+    {
+      "epoch": 0.530515419452232,
+      "grad_norm": 1.4567164182662964,
+      "learning_rate": 0.0001394630410122508,
+      "loss": 2.2502,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5309467328013802,
+      "grad_norm": 0.1543223112821579,
+      "learning_rate": 0.00013944559073169593,
+      "loss": 2.1162,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5313780461505283,
+      "grad_norm": 0.18608568608760834,
+      "learning_rate": 0.00013942812710697678,
+      "loss": 2.3083,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5318093594996766,
+      "grad_norm": 0.1594236046075821,
+      "learning_rate": 0.00013941065014170932,
+      "loss": 2.1773,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5322406728488247,
+      "grad_norm": 0.18374276161193848,
+      "learning_rate": 0.00013939315983951234,
+      "loss": 2.3547,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5326719861979728,
+      "grad_norm": 0.14752689003944397,
+      "learning_rate": 0.00013937565620400746,
+      "loss": 2.2261,
+      "step": 1235
+    },
+    {
+      "epoch": 0.533103299547121,
+      "grad_norm": 0.1832878440618515,
+      "learning_rate": 0.00013935813923881895,
+      "loss": 2.1805,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5335346128962691,
+      "grad_norm": 0.18583160638809204,
+      "learning_rate": 0.0001393406089475739,
+      "loss": 2.0176,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5339659262454173,
+      "grad_norm": 0.1821640282869339,
+      "learning_rate": 0.00013932306533390214,
+      "loss": 2.2021,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5343972395945654,
+      "grad_norm": 0.16310039162635803,
+      "learning_rate": 0.0001393055084014363,
+      "loss": 2.0919,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5348285529437136,
+      "grad_norm": 0.18390604853630066,
+      "learning_rate": 0.00013928793815381167,
+      "loss": 2.0216,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5352598662928618,
+      "grad_norm": 0.15094532072544098,
+      "learning_rate": 0.00013927035459466644,
+      "loss": 2.336,
+      "step": 1241
+    },
+    {
+      "epoch": 0.53569117964201,
+      "grad_norm": 0.16046181321144104,
+      "learning_rate": 0.00013925275772764143,
+      "loss": 2.3603,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5361224929911581,
+      "grad_norm": 0.18129126727581024,
+      "learning_rate": 0.00013923514755638029,
+      "loss": 2.3465,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5365538063403062,
+      "grad_norm": 0.17180679738521576,
+      "learning_rate": 0.00013921752408452937,
+      "loss": 2.1805,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5369851196894544,
+      "grad_norm": 0.15546143054962158,
+      "learning_rate": 0.00013919988731573782,
+      "loss": 2.1498,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5374164330386025,
+      "grad_norm": 0.16976916790008545,
+      "learning_rate": 0.00013918223725365754,
+      "loss": 2.1929,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5378477463877507,
+      "grad_norm": 0.16079477965831757,
+      "learning_rate": 0.00013916457390194314,
+      "loss": 2.3685,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5382790597368988,
+      "grad_norm": 0.16682679951190948,
+      "learning_rate": 0.000139146897264252,
+      "loss": 2.147,
+      "step": 1248
+    },
+    {
+      "epoch": 0.538710373086047,
+      "grad_norm": 0.20510974526405334,
+      "learning_rate": 0.00013912920734424433,
+      "loss": 2.2688,
+      "step": 1249
+    },
+    {
+      "epoch": 0.5391416864351952,
+      "grad_norm": 0.1665913313627243,
+      "learning_rate": 0.00013911150414558292,
+      "loss": 2.2088,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5391416864351952,
+      "eval_loss": 2.122371196746826,
+      "eval_runtime": 198.043,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5395729997843434,
+      "grad_norm": 0.17404893040657043,
+      "learning_rate": 0.0001390937876719335,
+      "loss": 2.2236,
+      "step": 1251
+    },
+    {
+      "epoch": 0.5400043131334915,
+      "grad_norm": 0.21290186047554016,
+      "learning_rate": 0.00013907605792696444,
+      "loss": 2.1855,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5404356264826397,
+      "grad_norm": 0.18014317750930786,
+      "learning_rate": 0.0001390583149143468,
+      "loss": 2.3149,
+      "step": 1253
+    },
+    {
+      "epoch": 0.5408669398317878,
+      "grad_norm": 0.1657971888780594,
+      "learning_rate": 0.0001390405586377546,
+      "loss": 1.9182,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5412982531809359,
+      "grad_norm": 0.17486490309238434,
+      "learning_rate": 0.00013902278910086433,
+      "loss": 2.1968,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5417295665300841,
+      "grad_norm": 0.1682259440422058,
+      "learning_rate": 0.00013900500630735548,
+      "loss": 2.3308,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5421608798792322,
+      "grad_norm": 0.19142164289951324,
+      "learning_rate": 0.00013898721026091014,
+      "loss": 2.2679,
+      "step": 1257
+    },
+    {
+      "epoch": 0.5425921932283804,
+      "grad_norm": 0.15226717293262482,
+      "learning_rate": 0.00013896940096521312,
+      "loss": 2.0877,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5430235065775286,
+      "grad_norm": 0.17772473394870758,
+      "learning_rate": 0.00013895157842395208,
+      "loss": 2.4032,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5434548199266768,
+      "grad_norm": 0.17212212085723877,
+      "learning_rate": 0.00013893374264081738,
+      "loss": 2.298,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5438861332758249,
+      "grad_norm": 0.1911373734474182,
+      "learning_rate": 0.0001389158936195021,
+      "loss": 2.3969,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5443174466249731,
+      "grad_norm": 0.1889583170413971,
+      "learning_rate": 0.00013889803136370207,
+      "loss": 2.136,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5447487599741212,
+      "grad_norm": 0.16074030101299286,
+      "learning_rate": 0.00013888015587711587,
+      "loss": 2.2086,
+      "step": 1263
+    },
+    {
+      "epoch": 0.5451800733232693,
+      "grad_norm": 0.15296761691570282,
+      "learning_rate": 0.0001388622671634448,
+      "loss": 2.3906,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5456113866724175,
+      "grad_norm": 0.18919295072555542,
+      "learning_rate": 0.00013884436522639297,
+      "loss": 2.3383,
+      "step": 1265
+    },
+    {
+      "epoch": 0.5460427000215656,
+      "grad_norm": 0.16433317959308624,
+      "learning_rate": 0.0001388264500696671,
+      "loss": 2.1688,
+      "step": 1266
+    },
+    {
+      "epoch": 0.5464740133707138,
+      "grad_norm": 0.16185767948627472,
+      "learning_rate": 0.00013880852169697672,
+      "loss": 2.5015,
+      "step": 1267
+    },
+    {
+      "epoch": 0.546905326719862,
+      "grad_norm": 0.18482503294944763,
+      "learning_rate": 0.00013879058011203417,
+      "loss": 2.2468,
+      "step": 1268
+    },
+    {
+      "epoch": 0.5473366400690102,
+      "grad_norm": 0.1860150247812271,
+      "learning_rate": 0.00013877262531855438,
+      "loss": 2.204,
+      "step": 1269
+    },
+    {
+      "epoch": 0.5477679534181583,
+      "grad_norm": 0.1531609743833542,
+      "learning_rate": 0.0001387546573202551,
+      "loss": 2.2641,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5481992667673065,
+      "grad_norm": 0.19258692860603333,
+      "learning_rate": 0.00013873667612085687,
+      "loss": 2.1963,
+      "step": 1271
+    },
+    {
+      "epoch": 0.5486305801164546,
+      "grad_norm": 0.20093025267124176,
+      "learning_rate": 0.0001387186817240828,
+      "loss": 2.0085,
+      "step": 1272
+    },
+    {
+      "epoch": 0.5490618934656027,
+      "grad_norm": 0.16600941121578217,
+      "learning_rate": 0.00013870067413365887,
+      "loss": 2.3559,
+      "step": 1273
+    },
+    {
+      "epoch": 0.5494932068147509,
+      "grad_norm": 0.15207615494728088,
+      "learning_rate": 0.00013868265335331372,
+      "loss": 1.9825,
+      "step": 1274
+    },
+    {
+      "epoch": 0.549924520163899,
+      "grad_norm": 0.16786298155784607,
+      "learning_rate": 0.0001386646193867788,
+      "loss": 2.3603,
+      "step": 1275
+    },
+    {
+      "epoch": 0.549924520163899,
+      "eval_loss": 2.120718479156494,
+      "eval_runtime": 197.0115,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 1275
+    },
+    {
+      "epoch": 0.5503558335130472,
+      "grad_norm": 0.8668006658554077,
+      "learning_rate": 0.0001386465722377882,
+      "loss": 2.1801,
+      "step": 1276
+    },
+    {
+      "epoch": 0.5507871468621954,
+      "grad_norm": 0.16311630606651306,
+      "learning_rate": 0.00013862851191007877,
+      "loss": 2.0761,
+      "step": 1277
+    },
+    {
+      "epoch": 0.5512184602113436,
+      "grad_norm": 0.1599023938179016,
+      "learning_rate": 0.00013861043840739015,
+      "loss": 2.1566,
+      "step": 1278
+    },
+    {
+      "epoch": 0.5516497735604917,
+      "grad_norm": 0.1699908971786499,
+      "learning_rate": 0.0001385923517334646,
+      "loss": 2.2602,
+      "step": 1279
+    },
+    {
+      "epoch": 0.5520810869096399,
+      "grad_norm": 0.16559387743473053,
+      "learning_rate": 0.00013857425189204724,
+      "loss": 2.3211,
+      "step": 1280
+    },
+    {
+      "epoch": 0.552512400258788,
+      "grad_norm": 0.1768280565738678,
+      "learning_rate": 0.00013855613888688574,
+      "loss": 2.2501,
+      "step": 1281
+    },
+    {
+      "epoch": 0.5529437136079361,
+      "grad_norm": 0.16502228379249573,
+      "learning_rate": 0.00013853801272173063,
+      "loss": 1.9774,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5533750269570843,
+      "grad_norm": 0.16906069219112396,
+      "learning_rate": 0.0001385198734003352,
+      "loss": 2.2409,
+      "step": 1283
+    },
+    {
+      "epoch": 0.5538063403062324,
+      "grad_norm": 0.15726394951343536,
+      "learning_rate": 0.0001385017209264553,
+      "loss": 2.3878,
+      "step": 1284
+    },
+    {
+      "epoch": 0.5542376536553806,
+      "grad_norm": 0.17712968587875366,
+      "learning_rate": 0.00013848355530384965,
+      "loss": 2.1526,
+      "step": 1285
+    },
+    {
+      "epoch": 0.5546689670045288,
+      "grad_norm": 0.19086754322052002,
+      "learning_rate": 0.00013846537653627964,
+      "loss": 2.2802,
+      "step": 1286
+    },
+    {
+      "epoch": 0.555100280353677,
+      "grad_norm": 0.16991381347179413,
+      "learning_rate": 0.0001384471846275094,
+      "loss": 2.332,
+      "step": 1287
+    },
+    {
+      "epoch": 0.5555315937028251,
+      "grad_norm": 0.19330179691314697,
+      "learning_rate": 0.00013842897958130568,
+      "loss": 2.4018,
+      "step": 1288
+    },
+    {
+      "epoch": 0.5559629070519733,
+      "grad_norm": 0.16834238171577454,
+      "learning_rate": 0.00013841076140143813,
+      "loss": 2.2892,
+      "step": 1289
+    },
+    {
+      "epoch": 0.5563942204011214,
+      "grad_norm": 0.17593060433864594,
+      "learning_rate": 0.00013839253009167898,
+      "loss": 2.269,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5568255337502696,
+      "grad_norm": 0.15262264013290405,
+      "learning_rate": 0.00013837428565580318,
+      "loss": 2.2531,
+      "step": 1291
+    },
+    {
+      "epoch": 0.5572568470994177,
+      "grad_norm": 0.16939087212085724,
+      "learning_rate": 0.00013835602809758853,
+      "loss": 2.1666,
+      "step": 1292
+    },
+    {
+      "epoch": 0.5576881604485658,
+      "grad_norm": 0.183957040309906,
+      "learning_rate": 0.0001383377574208154,
+      "loss": 2.1925,
+      "step": 1293
+    },
+    {
+      "epoch": 0.5581194737977141,
+      "grad_norm": 0.17081241309642792,
+      "learning_rate": 0.00013831947362926694,
+      "loss": 2.1878,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5585507871468622,
+      "grad_norm": 0.17042112350463867,
+      "learning_rate": 0.00013830117672672902,
+      "loss": 2.4236,
+      "step": 1295
+    },
+    {
+      "epoch": 0.5589821004960104,
+      "grad_norm": 0.17426973581314087,
+      "learning_rate": 0.0001382828667169902,
+      "loss": 2.3423,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5594134138451585,
+      "grad_norm": 0.1627834439277649,
+      "learning_rate": 0.00013826454360384182,
+      "loss": 2.3386,
+      "step": 1297
+    },
+    {
+      "epoch": 0.5598447271943067,
+      "grad_norm": 0.15972758829593658,
+      "learning_rate": 0.0001382462073910778,
+      "loss": 2.0709,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5602760405434548,
+      "grad_norm": 0.17867346107959747,
+      "learning_rate": 0.0001382278580824949,
+      "loss": 2.1761,
+      "step": 1299
+    },
+    {
+      "epoch": 0.560707353892603,
+      "grad_norm": 0.20676644146442413,
+      "learning_rate": 0.0001382094956818925,
+      "loss": 2.3478,
+      "step": 1300
+    },
+    {
+      "epoch": 0.560707353892603,
+      "eval_loss": 2.121127128601074,
+      "eval_runtime": 207.5162,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5611386672417511,
+      "grad_norm": 0.1815398633480072,
+      "learning_rate": 0.00013819112019307283,
+      "loss": 2.3379,
+      "step": 1301
+    },
+    {
+      "epoch": 0.5615699805908992,
+      "grad_norm": 0.1549919694662094,
+      "learning_rate": 0.00013817273161984066,
+      "loss": 2.0807,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5620012939400475,
+      "grad_norm": 0.17683330178260803,
+      "learning_rate": 0.00013815432996600353,
+      "loss": 2.2306,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5624326072891956,
+      "grad_norm": 0.17177826166152954,
+      "learning_rate": 0.00013813591523537176,
+      "loss": 2.2926,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5628639206383438,
+      "grad_norm": 0.16859027743339539,
+      "learning_rate": 0.0001381174874317583,
+      "loss": 2.2788,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5632952339874919,
+      "grad_norm": 0.17350350320339203,
+      "learning_rate": 0.00013809904655897882,
+      "loss": 2.255,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5637265473366401,
+      "grad_norm": 0.17333823442459106,
+      "learning_rate": 0.00013808059262085173,
+      "loss": 2.1696,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5641578606857882,
+      "grad_norm": 0.1684228777885437,
+      "learning_rate": 0.00013806212562119806,
+      "loss": 2.2234,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5645891740349364,
+      "grad_norm": 0.16519196331501007,
+      "learning_rate": 0.0001380436455638417,
+      "loss": 2.1744,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5650204873840845,
+      "grad_norm": 0.159685418009758,
+      "learning_rate": 0.00013802515245260906,
+      "loss": 2.3333,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5654518007332326,
+      "grad_norm": 0.21496149897575378,
+      "learning_rate": 0.00013800664629132935,
+      "loss": 2.187,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5658831140823809,
+      "grad_norm": 0.19159434735774994,
+      "learning_rate": 0.00013798812708383453,
+      "loss": 2.2162,
+      "step": 1312
+    },
+    {
+      "epoch": 0.566314427431529,
+      "grad_norm": 0.1915857344865799,
+      "learning_rate": 0.00013796959483395915,
+      "loss": 2.195,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5667457407806772,
+      "grad_norm": 0.18786486983299255,
+      "learning_rate": 0.00013795104954554056,
+      "loss": 2.1581,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5671770541298253,
+      "grad_norm": 0.1549830585718155,
+      "learning_rate": 0.00013793249122241873,
+      "loss": 2.2125,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5676083674789735,
+      "grad_norm": 0.18538445234298706,
+      "learning_rate": 0.00013791391986843636,
+      "loss": 2.327,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5680396808281216,
+      "grad_norm": 0.15483881533145905,
+      "learning_rate": 0.00013789533548743888,
+      "loss": 2.0952,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5684709941772698,
+      "grad_norm": 0.14627200365066528,
+      "learning_rate": 0.00013787673808327437,
+      "loss": 2.1907,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5689023075264179,
+      "grad_norm": 0.1589733511209488,
+      "learning_rate": 0.00013785812765979366,
+      "loss": 2.1557,
+      "step": 1319
+    },
+    {
+      "epoch": 0.569333620875566,
+      "grad_norm": 0.17344968020915985,
+      "learning_rate": 0.00013783950422085016,
+      "loss": 2.2803,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5697649342247143,
+      "grad_norm": 0.15491415560245514,
+      "learning_rate": 0.00013782086777030015,
+      "loss": 1.9636,
+      "step": 1321
+    },
+    {
+      "epoch": 0.5701962475738624,
+      "grad_norm": 0.16137534379959106,
+      "learning_rate": 0.00013780221831200245,
+      "loss": 2.2102,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5706275609230106,
+      "grad_norm": 0.15676361322402954,
+      "learning_rate": 0.00013778355584981867,
+      "loss": 2.2007,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5710588742721587,
+      "grad_norm": 0.16707196831703186,
+      "learning_rate": 0.00013776488038761306,
+      "loss": 2.1038,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5714901876213069,
+      "grad_norm": 0.18508413434028625,
+      "learning_rate": 0.00013774619192925258,
+      "loss": 2.2576,
+      "step": 1325
+    },
+    {
+      "epoch": 0.5714901876213069,
+      "eval_loss": 2.1195273399353027,
+      "eval_runtime": 201.6363,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1325
+    },
+    {
+      "epoch": 0.571921500970455,
+      "grad_norm": 0.1671893149614334,
+      "learning_rate": 0.00013772749047860687,
+      "loss": 2.1635,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5723528143196032,
+      "grad_norm": 0.15377773344516754,
+      "learning_rate": 0.00013770877603954827,
+      "loss": 2.273,
+      "step": 1327
+    },
+    {
+      "epoch": 0.5727841276687513,
+      "grad_norm": 0.1514381468296051,
+      "learning_rate": 0.00013769004861595189,
+      "loss": 2.1513,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5732154410178995,
+      "grad_norm": 0.16755861043930054,
+      "learning_rate": 0.0001376713082116953,
+      "loss": 2.0788,
+      "step": 1329
+    },
+    {
+      "epoch": 0.5736467543670477,
+      "grad_norm": 0.1689564436674118,
+      "learning_rate": 0.000137652554830659,
+      "loss": 2.0674,
+      "step": 1330
+    },
+    {
+      "epoch": 0.5740780677161959,
+      "grad_norm": 0.16019587218761444,
+      "learning_rate": 0.0001376337884767261,
+      "loss": 2.1102,
+      "step": 1331
+    },
+    {
+      "epoch": 0.574509381065344,
+      "grad_norm": 0.16313177347183228,
+      "learning_rate": 0.00013761500915378228,
+      "loss": 2.0397,
+      "step": 1332
+    },
+    {
+      "epoch": 0.5749406944144921,
+      "grad_norm": 0.183137446641922,
+      "learning_rate": 0.00013759621686571612,
+      "loss": 2.2121,
+      "step": 1333
+    },
+    {
+      "epoch": 0.5753720077636403,
+      "grad_norm": 0.18410666286945343,
+      "learning_rate": 0.0001375774116164187,
+      "loss": 2.2207,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5758033211127884,
+      "grad_norm": 0.16913245618343353,
+      "learning_rate": 0.00013755859340978388,
+      "loss": 2.2753,
+      "step": 1335
+    },
+    {
+      "epoch": 0.5762346344619366,
+      "grad_norm": 0.18499286472797394,
+      "learning_rate": 0.00013753976224970815,
+      "loss": 2.0903,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5766659478110847,
+      "grad_norm": 0.19554150104522705,
+      "learning_rate": 0.00013752091814009073,
+      "loss": 2.3823,
+      "step": 1337
+    },
+    {
+      "epoch": 0.5770972611602329,
+      "grad_norm": 0.2066139131784439,
+      "learning_rate": 0.0001375020610848335,
+      "loss": 2.3225,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5775285745093811,
+      "grad_norm": 0.1761176884174347,
+      "learning_rate": 0.000137483191087841,
+      "loss": 2.1969,
+      "step": 1339
+    },
+    {
+      "epoch": 0.5779598878585293,
+      "grad_norm": 0.1563565880060196,
+      "learning_rate": 0.00013746430815302047,
+      "loss": 1.9912,
+      "step": 1340
+    },
+    {
+      "epoch": 0.5783912012076774,
+      "grad_norm": 0.1895601898431778,
+      "learning_rate": 0.00013744541228428187,
+      "loss": 2.1682,
+      "step": 1341
+    },
+    {
+      "epoch": 0.5788225145568255,
+      "grad_norm": 0.1506771296262741,
+      "learning_rate": 0.00013742650348553775,
+      "loss": 1.7265,
+      "step": 1342
+    },
+    {
+      "epoch": 0.5792538279059737,
+      "grad_norm": 0.1695070117712021,
+      "learning_rate": 0.0001374075817607034,
+      "loss": 2.1685,
+      "step": 1343
+    },
+    {
+      "epoch": 0.5796851412551218,
+      "grad_norm": 0.15319640934467316,
+      "learning_rate": 0.00013738864711369675,
+      "loss": 2.0115,
+      "step": 1344
+    },
+    {
+      "epoch": 0.58011645460427,
+      "grad_norm": 0.15707378089427948,
+      "learning_rate": 0.00013736969954843847,
+      "loss": 1.9165,
+      "step": 1345
+    },
+    {
+      "epoch": 0.5805477679534181,
+      "grad_norm": 0.18513138592243195,
+      "learning_rate": 0.00013735073906885183,
+      "loss": 2.0196,
+      "step": 1346
+    },
+    {
+      "epoch": 0.5809790813025664,
+      "grad_norm": 0.1941574513912201,
+      "learning_rate": 0.0001373317656788628,
+      "loss": 2.1579,
+      "step": 1347
+    },
+    {
+      "epoch": 0.5814103946517145,
+      "grad_norm": 0.1600506603717804,
+      "learning_rate": 0.00013731277938240004,
+      "loss": 2.1314,
+      "step": 1348
+    },
+    {
+      "epoch": 0.5818417080008627,
+      "grad_norm": 0.16690665483474731,
+      "learning_rate": 0.00013729378018339486,
+      "loss": 2.5433,
+      "step": 1349
+    },
+    {
+      "epoch": 0.5822730213500108,
+      "grad_norm": 0.1693698763847351,
+      "learning_rate": 0.00013727476808578128,
+      "loss": 2.2022,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5822730213500108,
+      "eval_loss": 2.118407964706421,
+      "eval_runtime": 201.4914,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5827043346991589,
+      "grad_norm": 0.1695820689201355,
+      "learning_rate": 0.00013725574309349592,
+      "loss": 2.162,
+      "step": 1351
+    },
+    {
+      "epoch": 0.5831356480483071,
+      "grad_norm": 0.17078326642513275,
+      "learning_rate": 0.00013723670521047817,
+      "loss": 2.1169,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5835669613974552,
+      "grad_norm": 0.18800406157970428,
+      "learning_rate": 0.00013721765444066998,
+      "loss": 2.0955,
+      "step": 1353
+    },
+    {
+      "epoch": 0.5839982747466034,
+      "grad_norm": 0.1687876284122467,
+      "learning_rate": 0.00013719859078801603,
+      "loss": 2.1847,
+      "step": 1354
+    },
+    {
+      "epoch": 0.5844295880957515,
+      "grad_norm": 0.17234942317008972,
+      "learning_rate": 0.00013717951425646366,
+      "loss": 2.0536,
+      "step": 1355
+    },
+    {
+      "epoch": 0.5848609014448998,
+      "grad_norm": 0.17885074019432068,
+      "learning_rate": 0.00013716042484996288,
+      "loss": 2.283,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5852922147940479,
+      "grad_norm": 0.1754123568534851,
+      "learning_rate": 0.00013714132257246637,
+      "loss": 2.3275,
+      "step": 1357
+    },
+    {
+      "epoch": 0.5857235281431961,
+      "grad_norm": 0.19185900688171387,
+      "learning_rate": 0.00013712220742792945,
+      "loss": 2.2428,
+      "step": 1358
+    },
+    {
+      "epoch": 0.5861548414923442,
+      "grad_norm": 0.16430750489234924,
+      "learning_rate": 0.00013710307942031014,
+      "loss": 2.1534,
+      "step": 1359
+    },
+    {
+      "epoch": 0.5865861548414923,
+      "grad_norm": 0.15442483127117157,
+      "learning_rate": 0.00013708393855356907,
+      "loss": 2.2003,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5870174681906405,
+      "grad_norm": 0.16345712542533875,
+      "learning_rate": 0.0001370647848316696,
+      "loss": 2.2425,
+      "step": 1361
+    },
+    {
+      "epoch": 0.5874487815397886,
+      "grad_norm": 0.22671642899513245,
+      "learning_rate": 0.0001370456182585777,
+      "loss": 2.1916,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5878800948889368,
+      "grad_norm": 0.17455963790416718,
+      "learning_rate": 0.00013702643883826198,
+      "loss": 2.0417,
+      "step": 1363
+    },
+    {
+      "epoch": 0.5883114082380849,
+      "grad_norm": 0.16011476516723633,
+      "learning_rate": 0.00013700724657469382,
+      "loss": 2.1057,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5887427215872332,
+      "grad_norm": 0.14858493208885193,
+      "learning_rate": 0.00013698804147184717,
+      "loss": 2.1275,
+      "step": 1365
+    },
+    {
+      "epoch": 0.5891740349363813,
+      "grad_norm": 0.17408862709999084,
+      "learning_rate": 0.00013696882353369863,
+      "loss": 2.1891,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5896053482855295,
+      "grad_norm": 0.18313269317150116,
+      "learning_rate": 0.0001369495927642275,
+      "loss": 2.2143,
+      "step": 1367
+    },
+    {
+      "epoch": 0.5900366616346776,
+      "grad_norm": 0.1936151534318924,
+      "learning_rate": 0.00013693034916741572,
+      "loss": 2.0642,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5904679749838257,
+      "grad_norm": 0.2034844607114792,
+      "learning_rate": 0.00013691109274724787,
+      "loss": 2.1864,
+      "step": 1369
+    },
+    {
+      "epoch": 0.5908992883329739,
+      "grad_norm": 0.1795257031917572,
+      "learning_rate": 0.00013689182350771123,
+      "loss": 2.0779,
+      "step": 1370
+    },
+    {
+      "epoch": 0.591330601682122,
+      "grad_norm": 0.17389395833015442,
+      "learning_rate": 0.0001368725414527957,
+      "loss": 2.2393,
+      "step": 1371
+    },
+    {
+      "epoch": 0.5917619150312702,
+      "grad_norm": 0.21569593250751495,
+      "learning_rate": 0.00013685324658649383,
+      "loss": 2.07,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5921932283804183,
+      "grad_norm": 0.18595005571842194,
+      "learning_rate": 0.00013683393891280086,
+      "loss": 2.2535,
+      "step": 1373
+    },
+    {
+      "epoch": 0.5926245417295666,
+      "grad_norm": 0.16178612411022186,
+      "learning_rate": 0.0001368146184357146,
+      "loss": 2.248,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5930558550787147,
+      "grad_norm": 0.16927756369113922,
+      "learning_rate": 0.00013679528515923563,
+      "loss": 2.0816,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5930558550787147,
+      "eval_loss": 2.118720054626465,
+      "eval_runtime": 202.3739,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 1375
+    },
+    {
+      "epoch": 0.5934871684278629,
+      "grad_norm": 0.19910386204719543,
+      "learning_rate": 0.00013677593908736704,
+      "loss": 2.2726,
+      "step": 1376
+    },
+    {
+      "epoch": 0.593918481777011,
+      "grad_norm": 0.22931334376335144,
+      "learning_rate": 0.00013675658022411474,
+      "loss": 2.0352,
+      "step": 1377
+    },
+    {
+      "epoch": 0.5943497951261592,
+      "grad_norm": 0.18184387683868408,
+      "learning_rate": 0.00013673720857348712,
+      "loss": 2.1897,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5947811084753073,
+      "grad_norm": 0.1753135472536087,
+      "learning_rate": 0.00013671782413949532,
+      "loss": 2.3276,
+      "step": 1379
+    },
+    {
+      "epoch": 0.5952124218244554,
+      "grad_norm": 0.17682583630084991,
+      "learning_rate": 0.0001366984269261531,
+      "loss": 2.2037,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5956437351736036,
+      "grad_norm": 0.21315497159957886,
+      "learning_rate": 0.00013667901693747688,
+      "loss": 2.2358,
+      "step": 1381
+    },
+    {
+      "epoch": 0.5960750485227517,
+      "grad_norm": 0.20123091340065002,
+      "learning_rate": 0.0001366595941774857,
+      "loss": 2.1988,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5965063618719,
+      "grad_norm": 0.18664579093456268,
+      "learning_rate": 0.0001366401586502012,
+      "loss": 2.0231,
+      "step": 1383
+    },
+    {
+      "epoch": 0.5969376752210481,
+      "grad_norm": 0.1737365573644638,
+      "learning_rate": 0.0001366207103596478,
+      "loss": 2.1643,
+      "step": 1384
+    },
+    {
+      "epoch": 0.5973689885701963,
+      "grad_norm": 0.22827516496181488,
+      "learning_rate": 0.00013660124930985244,
+      "loss": 2.2214,
+      "step": 1385
+    },
+    {
+      "epoch": 0.5978003019193444,
+      "grad_norm": 0.18609853088855743,
+      "learning_rate": 0.00013658177550484476,
+      "loss": 2.1498,
+      "step": 1386
+    },
+    {
+      "epoch": 0.5982316152684926,
+      "grad_norm": 0.42302677035331726,
+      "learning_rate": 0.00013656228894865698,
+      "loss": 2.3019,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5986629286176407,
+      "grad_norm": 0.1694529801607132,
+      "learning_rate": 0.00013654278964532409,
+      "loss": 1.8489,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5990942419667888,
+      "grad_norm": 0.16748681664466858,
+      "learning_rate": 0.00013652327759888354,
+      "loss": 2.1027,
+      "step": 1389
+    },
+    {
+      "epoch": 0.599525555315937,
+      "grad_norm": 0.19268131256103516,
+      "learning_rate": 0.00013650375281337558,
+      "loss": 2.4342,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5999568686650851,
+      "grad_norm": 0.17943158745765686,
+      "learning_rate": 0.000136484215292843,
+      "loss": 2.3367,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6003881820142334,
+      "grad_norm": 0.20505180954933167,
+      "learning_rate": 0.00013646466504133124,
+      "loss": 2.0027,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6008194953633815,
+      "grad_norm": 0.14545932412147522,
+      "learning_rate": 0.00013644510206288845,
+      "loss": 2.1271,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6012508087125297,
+      "grad_norm": 0.1819576621055603,
+      "learning_rate": 0.0001364255263615653,
+      "loss": 2.3828,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6016821220616778,
+      "grad_norm": 0.1994846612215042,
+      "learning_rate": 0.00013640593794141522,
+      "loss": 2.0621,
+      "step": 1395
+    },
+    {
+      "epoch": 0.602113435410826,
+      "grad_norm": 0.16838571429252625,
+      "learning_rate": 0.00013638633680649413,
+      "loss": 2.3161,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6025447487599741,
+      "grad_norm": 0.16921788454055786,
+      "learning_rate": 0.0001363667229608607,
+      "loss": 2.239,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6029760621091222,
+      "grad_norm": 0.18538618087768555,
+      "learning_rate": 0.0001363470964085762,
+      "loss": 2.1616,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6034073754582704,
+      "grad_norm": 0.19689178466796875,
+      "learning_rate": 0.0001363274571537045,
+      "loss": 2.0951,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6038386888074186,
+      "grad_norm": 0.17458391189575195,
+      "learning_rate": 0.00013630780520031214,
+      "loss": 2.2259,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6038386888074186,
+      "eval_loss": 2.1179585456848145,
+      "eval_runtime": 202.3652,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6042700021565668,
+      "grad_norm": 0.17060507833957672,
+      "learning_rate": 0.00013628814055246824,
+      "loss": 2.228,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6047013155057149,
+      "grad_norm": 0.1892073005437851,
+      "learning_rate": 0.00013626846321424462,
+      "loss": 2.0956,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6051326288548631,
+      "grad_norm": 0.33646124601364136,
+      "learning_rate": 0.00013624877318971565,
+      "loss": 2.3609,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6055639422040112,
+      "grad_norm": 0.17053109407424927,
+      "learning_rate": 0.0001362290704829584,
+      "loss": 2.0822,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6059952555531594,
+      "grad_norm": 0.19565366208553314,
+      "learning_rate": 0.00013620935509805253,
+      "loss": 2.2828,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6064265689023075,
+      "grad_norm": 0.34117087721824646,
+      "learning_rate": 0.0001361896270390803,
+      "loss": 2.3242,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6068578822514556,
+      "grad_norm": 0.1925416886806488,
+      "learning_rate": 0.00013616988631012664,
+      "loss": 2.2093,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6072891956006038,
+      "grad_norm": 0.17825543880462646,
+      "learning_rate": 0.00013615013291527907,
+      "loss": 2.3198,
+      "step": 1408
+    },
+    {
+      "epoch": 0.607720508949752,
+      "grad_norm": 0.175543874502182,
+      "learning_rate": 0.00013613036685862774,
+      "loss": 2.432,
+      "step": 1409
+    },
+    {
+      "epoch": 0.6081518222989002,
+      "grad_norm": 0.16355106234550476,
+      "learning_rate": 0.00013611058814426546,
+      "loss": 2.2179,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6085831356480483,
+      "grad_norm": 0.16528934240341187,
+      "learning_rate": 0.00013609079677628761,
+      "loss": 2.2397,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6090144489971965,
+      "grad_norm": 0.17749467492103577,
+      "learning_rate": 0.0001360709927587922,
+      "loss": 2.0724,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6094457623463446,
+      "grad_norm": 0.18812529742717743,
+      "learning_rate": 0.00013605117609587986,
+      "loss": 2.1415,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6098770756954928,
+      "grad_norm": 0.16908733546733856,
+      "learning_rate": 0.00013603134679165393,
+      "loss": 2.2852,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6103083890446409,
+      "grad_norm": 0.18603071570396423,
+      "learning_rate": 0.00013601150485022018,
+      "loss": 2.2209,
+      "step": 1415
+    },
+    {
+      "epoch": 0.610739702393789,
+      "grad_norm": 0.18312253057956696,
+      "learning_rate": 0.00013599165027568713,
+      "loss": 2.203,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6111710157429372,
+      "grad_norm": 0.16831669211387634,
+      "learning_rate": 0.0001359717830721659,
+      "loss": 1.9639,
+      "step": 1417
+    },
+    {
+      "epoch": 0.6116023290920855,
+      "grad_norm": 0.18003444373607635,
+      "learning_rate": 0.0001359519032437702,
+      "loss": 2.4113,
+      "step": 1418
+    },
+    {
+      "epoch": 0.6120336424412336,
+      "grad_norm": 0.18845191597938538,
+      "learning_rate": 0.0001359320107946164,
+      "loss": 2.5032,
+      "step": 1419
+    },
+    {
+      "epoch": 0.6124649557903817,
+      "grad_norm": 0.19460377097129822,
+      "learning_rate": 0.00013591210572882342,
+      "loss": 2.2331,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6128962691395299,
+      "grad_norm": 0.17799149453639984,
+      "learning_rate": 0.00013589218805051283,
+      "loss": 2.2636,
+      "step": 1421
+    },
+    {
+      "epoch": 0.613327582488678,
+      "grad_norm": 0.17384687066078186,
+      "learning_rate": 0.00013587225776380877,
+      "loss": 2.0293,
+      "step": 1422
+    },
+    {
+      "epoch": 0.6137588958378262,
+      "grad_norm": 0.1670762002468109,
+      "learning_rate": 0.00013585231487283806,
+      "loss": 2.2104,
+      "step": 1423
+    },
+    {
+      "epoch": 0.6141902091869743,
+      "grad_norm": 0.1775355190038681,
+      "learning_rate": 0.00013583235938173011,
+      "loss": 2.2815,
+      "step": 1424
+    },
+    {
+      "epoch": 0.6146215225361225,
+      "grad_norm": 0.2014247626066208,
+      "learning_rate": 0.00013581239129461687,
+      "loss": 2.0719,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6146215225361225,
+      "eval_loss": 2.1178596019744873,
+      "eval_runtime": 202.7019,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6150528358852706,
+      "grad_norm": 0.16708634793758392,
+      "learning_rate": 0.000135792410615633,
+      "loss": 2.1602,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6154841492344189,
+      "grad_norm": 0.16631104052066803,
+      "learning_rate": 0.0001357724173489157,
+      "loss": 2.3131,
+      "step": 1427
+    },
+    {
+      "epoch": 0.615915462583567,
+      "grad_norm": 0.1849827915430069,
+      "learning_rate": 0.00013575241149860477,
+      "loss": 2.2917,
+      "step": 1428
+    },
+    {
+      "epoch": 0.6163467759327151,
+      "grad_norm": 0.18862168490886688,
+      "learning_rate": 0.00013573239306884267,
+      "loss": 2.1622,
+      "step": 1429
+    },
+    {
+      "epoch": 0.6167780892818633,
+      "grad_norm": 0.18775883316993713,
+      "learning_rate": 0.00013571236206377443,
+      "loss": 2.232,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6172094026310114,
+      "grad_norm": 0.17551693320274353,
+      "learning_rate": 0.00013569231848754767,
+      "loss": 2.1698,
+      "step": 1431
+    },
+    {
+      "epoch": 0.6176407159801596,
+      "grad_norm": 0.19101572036743164,
+      "learning_rate": 0.00013567226234431268,
+      "loss": 2.3313,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6180720293293077,
+      "grad_norm": 0.22161993384361267,
+      "learning_rate": 0.00013565219363822226,
+      "loss": 1.9818,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6185033426784559,
+      "grad_norm": 0.15873490273952484,
+      "learning_rate": 0.00013563211237343185,
+      "loss": 2.1122,
+      "step": 1434
+    },
+    {
+      "epoch": 0.618934656027604,
+      "grad_norm": 0.20279757678508759,
+      "learning_rate": 0.00013561201855409952,
+      "loss": 2.2819,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6193659693767523,
+      "grad_norm": 0.18889349699020386,
+      "learning_rate": 0.0001355919121843859,
+      "loss": 2.4155,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6197972827259004,
+      "grad_norm": 0.1641744077205658,
+      "learning_rate": 0.00013557179326845428,
+      "loss": 2.2855,
+      "step": 1437
+    },
+    {
+      "epoch": 0.6202285960750485,
+      "grad_norm": 0.20285926759243011,
+      "learning_rate": 0.00013555166181047042,
+      "loss": 2.2571,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6206599094241967,
+      "grad_norm": 0.1879100352525711,
+      "learning_rate": 0.00013553151781460282,
+      "loss": 2.0196,
+      "step": 1439
+    },
+    {
+      "epoch": 0.6210912227733448,
+      "grad_norm": 0.20289397239685059,
+      "learning_rate": 0.00013551136128502249,
+      "loss": 2.3061,
+      "step": 1440
+    },
+    {
+      "epoch": 0.621522536122493,
+      "grad_norm": 0.15986093878746033,
+      "learning_rate": 0.00013549119222590306,
+      "loss": 2.4075,
+      "step": 1441
+    },
+    {
+      "epoch": 0.6219538494716411,
+      "grad_norm": 0.17021022737026215,
+      "learning_rate": 0.00013547101064142073,
+      "loss": 2.1968,
+      "step": 1442
+    },
+    {
+      "epoch": 0.6223851628207893,
+      "grad_norm": 0.18939296901226044,
+      "learning_rate": 0.00013545081653575442,
+      "loss": 2.1435,
+      "step": 1443
+    },
+    {
+      "epoch": 0.6228164761699374,
+      "grad_norm": 0.1589539647102356,
+      "learning_rate": 0.0001354306099130854,
+      "loss": 2.0662,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6232477895190857,
+      "grad_norm": 0.17256878316402435,
+      "learning_rate": 0.00013541039077759776,
+      "loss": 2.1539,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6236791028682338,
+      "grad_norm": 0.16598886251449585,
+      "learning_rate": 0.00013539015913347809,
+      "loss": 2.1643,
+      "step": 1446
+    },
+    {
+      "epoch": 0.624110416217382,
+      "grad_norm": 0.1731709986925125,
+      "learning_rate": 0.00013536991498491554,
+      "loss": 2.2187,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6245417295665301,
+      "grad_norm": 0.21180196106433868,
+      "learning_rate": 0.0001353496583361019,
+      "loss": 2.2559,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6249730429156782,
+      "grad_norm": 0.18084551393985748,
+      "learning_rate": 0.00013532938919123152,
+      "loss": 2.3106,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6254043562648264,
+      "grad_norm": 0.18181219696998596,
+      "learning_rate": 0.00013530910755450137,
+      "loss": 2.1849,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6254043562648264,
+      "eval_loss": 2.1173229217529297,
+      "eval_runtime": 201.5491,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6258356696139745,
+      "grad_norm": 0.2053252011537552,
+      "learning_rate": 0.00013528881343011096,
+      "loss": 2.3693,
+      "step": 1451
+    },
+    {
+      "epoch": 0.6262669829631227,
+      "grad_norm": 0.20996972918510437,
+      "learning_rate": 0.00013526850682226243,
+      "loss": 2.2502,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6266982963122709,
+      "grad_norm": 0.1840354949235916,
+      "learning_rate": 0.00013524818773516048,
+      "loss": 2.2374,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6271296096614191,
+      "grad_norm": 0.1774517297744751,
+      "learning_rate": 0.0001352278561730124,
+      "loss": 2.1423,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6275609230105672,
+      "grad_norm": 0.20158281922340393,
+      "learning_rate": 0.00013520751214002804,
+      "loss": 2.2852,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6279922363597154,
+      "grad_norm": 0.15398597717285156,
+      "learning_rate": 0.0001351871556404199,
+      "loss": 2.2491,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6284235497088635,
+      "grad_norm": 0.17070600390434265,
+      "learning_rate": 0.00013516678667840296,
+      "loss": 2.1026,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6288548630580116,
+      "grad_norm": 0.1867612600326538,
+      "learning_rate": 0.0001351464052581949,
+      "loss": 2.1568,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6292861764071598,
+      "grad_norm": 0.16975905001163483,
+      "learning_rate": 0.00013512601138401583,
+      "loss": 2.0452,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6297174897563079,
+      "grad_norm": 0.17002518475055695,
+      "learning_rate": 0.00013510560506008862,
+      "loss": 2.1918,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6301488031054561,
+      "grad_norm": 0.17760762572288513,
+      "learning_rate": 0.00013508518629063858,
+      "loss": 2.1783,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6305801164546043,
+      "grad_norm": 0.259189248085022,
+      "learning_rate": 0.0001350647550798936,
+      "loss": 2.3244,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6310114298037525,
+      "grad_norm": 0.1676439344882965,
+      "learning_rate": 0.00013504431143208424,
+      "loss": 2.22,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6314427431529006,
+      "grad_norm": 0.15792736411094666,
+      "learning_rate": 0.0001350238553514436,
+      "loss": 2.2445,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6318740565020488,
+      "grad_norm": 0.1680126190185547,
+      "learning_rate": 0.00013500338684220727,
+      "loss": 2.3718,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6323053698511969,
+      "grad_norm": 0.16542601585388184,
+      "learning_rate": 0.00013498290590861352,
+      "loss": 2.3419,
+      "step": 1466
+    },
+    {
+      "epoch": 0.632736683200345,
+      "grad_norm": 0.17210273444652557,
+      "learning_rate": 0.0001349624125549031,
+      "loss": 2.2629,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6331679965494932,
+      "grad_norm": 0.1726984828710556,
+      "learning_rate": 0.00013494190678531946,
+      "loss": 2.4776,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6335993098986413,
+      "grad_norm": 0.16882048547267914,
+      "learning_rate": 0.0001349213886041085,
+      "loss": 2.1406,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6340306232477895,
+      "grad_norm": 0.1476915329694748,
+      "learning_rate": 0.00013490085801551874,
+      "loss": 1.9109,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6344619365969377,
+      "grad_norm": 0.1599586308002472,
+      "learning_rate": 0.00013488031502380127,
+      "loss": 2.2298,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6348932499460859,
+      "grad_norm": 0.22215233743190765,
+      "learning_rate": 0.00013485975963320972,
+      "loss": 2.2152,
+      "step": 1472
+    },
+    {
+      "epoch": 0.635324563295234,
+      "grad_norm": 0.17221570014953613,
+      "learning_rate": 0.00013483919184800034,
+      "loss": 2.2218,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6357558766443822,
+      "grad_norm": 0.17482635378837585,
+      "learning_rate": 0.0001348186116724319,
+      "loss": 2.4065,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6361871899935303,
+      "grad_norm": 0.17080114781856537,
+      "learning_rate": 0.00013479801911076572,
+      "loss": 2.2043,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6361871899935303,
+      "eval_loss": 2.1166903972625732,
+      "eval_runtime": 202.3179,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6366185033426784,
+      "grad_norm": 0.18259508907794952,
+      "learning_rate": 0.00013477741416726574,
+      "loss": 2.2346,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6370498166918266,
+      "grad_norm": 0.17076414823532104,
+      "learning_rate": 0.00013475679684619846,
+      "loss": 2.1289,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6374811300409747,
+      "grad_norm": 0.16617844998836517,
+      "learning_rate": 0.0001347361671518329,
+      "loss": 2.2685,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6379124433901229,
+      "grad_norm": 0.1720925122499466,
+      "learning_rate": 0.00013471552508844068,
+      "loss": 2.2616,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6383437567392711,
+      "grad_norm": 0.18244807422161102,
+      "learning_rate": 0.00013469487066029596,
+      "loss": 2.2319,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6387750700884193,
+      "grad_norm": 0.16281767189502716,
+      "learning_rate": 0.00013467420387167542,
+      "loss": 2.2527,
+      "step": 1481
+    },
+    {
+      "epoch": 0.6392063834375674,
+      "grad_norm": 0.1821095198392868,
+      "learning_rate": 0.00013465352472685839,
+      "loss": 2.2209,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6396376967867156,
+      "grad_norm": 0.21405041217803955,
+      "learning_rate": 0.00013463283323012672,
+      "loss": 2.169,
+      "step": 1483
+    },
+    {
+      "epoch": 0.6400690101358637,
+      "grad_norm": 0.17871929705142975,
+      "learning_rate": 0.0001346121293857648,
+      "loss": 2.2038,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6405003234850118,
+      "grad_norm": 0.15631656348705292,
+      "learning_rate": 0.0001345914131980596,
+      "loss": 2.2222,
+      "step": 1485
+    },
+    {
+      "epoch": 0.64093163683416,
+      "grad_norm": 0.15821655094623566,
+      "learning_rate": 0.00013457068467130057,
+      "loss": 2.2539,
+      "step": 1486
+    },
+    {
+      "epoch": 0.6413629501833081,
+      "grad_norm": 0.16861869394779205,
+      "learning_rate": 0.00013454994380977987,
+      "loss": 1.9289,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6417942635324563,
+      "grad_norm": 0.1654800921678543,
+      "learning_rate": 0.00013452919061779204,
+      "loss": 2.3163,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6422255768816045,
+      "grad_norm": 0.1614343225955963,
+      "learning_rate": 0.00013450842509963432,
+      "loss": 2.3947,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6426568902307527,
+      "grad_norm": 0.17356038093566895,
+      "learning_rate": 0.0001344876472596064,
+      "loss": 2.174,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6430882035799008,
+      "grad_norm": 0.1758730709552765,
+      "learning_rate": 0.00013446685710201056,
+      "loss": 2.1513,
+      "step": 1491
+    },
+    {
+      "epoch": 0.643519516929049,
+      "grad_norm": 0.1510227918624878,
+      "learning_rate": 0.00013444605463115163,
+      "loss": 1.8994,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6439508302781971,
+      "grad_norm": 0.17163372039794922,
+      "learning_rate": 0.00013442523985133704,
+      "loss": 1.8301,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6443821436273452,
+      "grad_norm": 0.1685454398393631,
+      "learning_rate": 0.00013440441276687664,
+      "loss": 2.2378,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6448134569764934,
+      "grad_norm": 0.16419187188148499,
+      "learning_rate": 0.00013438357338208293,
+      "loss": 2.3511,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6452447703256415,
+      "grad_norm": 0.17738017439842224,
+      "learning_rate": 0.000134362721701271,
+      "loss": 2.3097,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6456760836747897,
+      "grad_norm": 0.1585911363363266,
+      "learning_rate": 0.0001343418577287583,
+      "loss": 2.2102,
+      "step": 1497
+    },
+    {
+      "epoch": 0.6461073970239379,
+      "grad_norm": 0.18356698751449585,
+      "learning_rate": 0.00013432098146886506,
+      "loss": 2.1842,
+      "step": 1498
+    },
+    {
+      "epoch": 0.6465387103730861,
+      "grad_norm": 0.17388349771499634,
+      "learning_rate": 0.00013430009292591386,
+      "loss": 2.1864,
+      "step": 1499
+    },
+    {
+      "epoch": 0.6469700237222342,
+      "grad_norm": 0.17987026274204254,
+      "learning_rate": 0.0001342791921042299,
+      "loss": 2.0218,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6469700237222342,
+      "eval_loss": 2.1155967712402344,
+      "eval_runtime": 201.8593,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6474013370713824,
+      "grad_norm": 0.1733091026544571,
+      "learning_rate": 0.00013425827900814097,
+      "loss": 2.4174,
+      "step": 1501
+    },
+    {
+      "epoch": 0.6478326504205305,
+      "grad_norm": 0.19367888569831848,
+      "learning_rate": 0.00013423735364197735,
+      "loss": 2.1148,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6482639637696787,
+      "grad_norm": 0.1763000339269638,
+      "learning_rate": 0.0001342164160100718,
+      "loss": 2.4165,
+      "step": 1503
+    },
+    {
+      "epoch": 0.6486952771188268,
+      "grad_norm": 0.17568454146385193,
+      "learning_rate": 0.00013419546611675979,
+      "loss": 2.2303,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6491265904679749,
+      "grad_norm": 0.18122456967830658,
+      "learning_rate": 0.00013417450396637913,
+      "loss": 2.2891,
+      "step": 1505
+    },
+    {
+      "epoch": 0.6495579038171232,
+      "grad_norm": 0.17672553658485413,
+      "learning_rate": 0.00013415352956327033,
+      "loss": 2.1847,
+      "step": 1506
+    },
+    {
+      "epoch": 0.6499892171662713,
+      "grad_norm": 0.1894833892583847,
+      "learning_rate": 0.00013413254291177632,
+      "loss": 2.3146,
+      "step": 1507
+    },
+    {
+      "epoch": 0.6504205305154195,
+      "grad_norm": 0.16244618594646454,
+      "learning_rate": 0.00013411154401624266,
+      "loss": 2.2783,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6508518438645676,
+      "grad_norm": 0.1534096598625183,
+      "learning_rate": 0.00013409053288101735,
+      "loss": 2.0793,
+      "step": 1509
+    },
+    {
+      "epoch": 0.6512831572137158,
+      "grad_norm": 0.16889546811580658,
+      "learning_rate": 0.00013406950951045104,
+      "loss": 2.1439,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6517144705628639,
+      "grad_norm": 0.21417348086833954,
+      "learning_rate": 0.0001340484739088968,
+      "loss": 2.2297,
+      "step": 1511
+    },
+    {
+      "epoch": 0.6521457839120121,
+      "grad_norm": 0.16018380224704742,
+      "learning_rate": 0.0001340274260807103,
+      "loss": 2.1571,
+      "step": 1512
+    },
+    {
+      "epoch": 0.6525770972611602,
+      "grad_norm": 0.18208736181259155,
+      "learning_rate": 0.0001340063660302497,
+      "loss": 2.1953,
+      "step": 1513
+    },
+    {
+      "epoch": 0.6530084106103083,
+      "grad_norm": 0.1658926159143448,
+      "learning_rate": 0.0001339852937618757,
+      "loss": 2.2122,
+      "step": 1514
+    },
+    {
+      "epoch": 0.6534397239594566,
+      "grad_norm": 0.14056910574436188,
+      "learning_rate": 0.0001339642092799516,
+      "loss": 2.18,
+      "step": 1515
+    },
+    {
+      "epoch": 0.6538710373086047,
+      "grad_norm": 0.1788419485092163,
+      "learning_rate": 0.0001339431125888432,
+      "loss": 2.2319,
+      "step": 1516
+    },
+    {
+      "epoch": 0.6543023506577529,
+      "grad_norm": 0.1722828596830368,
+      "learning_rate": 0.0001339220036929187,
+      "loss": 2.3425,
+      "step": 1517
+    },
+    {
+      "epoch": 0.654733664006901,
+      "grad_norm": 0.2073151171207428,
+      "learning_rate": 0.00013390088259654896,
+      "loss": 2.1697,
+      "step": 1518
+    },
+    {
+      "epoch": 0.6551649773560492,
+      "grad_norm": 0.1691809594631195,
+      "learning_rate": 0.0001338797493041074,
+      "loss": 2.4019,
+      "step": 1519
+    },
+    {
+      "epoch": 0.6555962907051973,
+      "grad_norm": 0.16326852142810822,
+      "learning_rate": 0.0001338586038199698,
+      "loss": 2.1642,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6560276040543455,
+      "grad_norm": 0.16764166951179504,
+      "learning_rate": 0.0001338374461485146,
+      "loss": 2.2035,
+      "step": 1521
+    },
+    {
+      "epoch": 0.6564589174034936,
+      "grad_norm": 0.16470211744308472,
+      "learning_rate": 0.0001338162762941228,
+      "loss": 2.1875,
+      "step": 1522
+    },
+    {
+      "epoch": 0.6568902307526417,
+      "grad_norm": 0.149167001247406,
+      "learning_rate": 0.00013379509426117775,
+      "loss": 2.0018,
+      "step": 1523
+    },
+    {
+      "epoch": 0.65732154410179,
+      "grad_norm": 0.16698133945465088,
+      "learning_rate": 0.00013377390005406547,
+      "loss": 2.1576,
+      "step": 1524
+    },
+    {
+      "epoch": 0.6577528574509381,
+      "grad_norm": 0.1746443808078766,
+      "learning_rate": 0.00013375269367717443,
+      "loss": 2.2006,
+      "step": 1525
+    },
+    {
+      "epoch": 0.6577528574509381,
+      "eval_loss": 2.115690231323242,
+      "eval_runtime": 212.2087,
+      "eval_samples_per_second": 0.151,
+      "eval_steps_per_second": 0.151,
+      "step": 1525
+    },
+    {
+      "epoch": 0.6581841708000863,
+      "grad_norm": 0.17736956477165222,
+      "learning_rate": 0.00013373147513489567,
+      "loss": 2.1085,
+      "step": 1526
+    },
+    {
+      "epoch": 0.6586154841492344,
+      "grad_norm": 0.19343900680541992,
+      "learning_rate": 0.00013371024443162272,
+      "loss": 2.078,
+      "step": 1527
+    },
+    {
+      "epoch": 0.6590467974983826,
+      "grad_norm": 0.1655140072107315,
+      "learning_rate": 0.0001336890015717516,
+      "loss": 2.1921,
+      "step": 1528
+    },
+    {
+      "epoch": 0.6594781108475307,
+      "grad_norm": 0.1496494710445404,
+      "learning_rate": 0.00013366774655968088,
+      "loss": 2.1519,
+      "step": 1529
+    },
+    {
+      "epoch": 0.6599094241966789,
+      "grad_norm": 0.18701104819774628,
+      "learning_rate": 0.0001336464793998117,
+      "loss": 2.2564,
+      "step": 1530
+    },
+    {
+      "epoch": 0.660340737545827,
+      "grad_norm": 0.16007700562477112,
+      "learning_rate": 0.00013362520009654755,
+      "loss": 2.1806,
+      "step": 1531
+    },
+    {
+      "epoch": 0.6607720508949751,
+      "grad_norm": 0.19236847758293152,
+      "learning_rate": 0.00013360390865429464,
+      "loss": 2.1388,
+      "step": 1532
+    },
+    {
+      "epoch": 0.6612033642441234,
+      "grad_norm": 0.18432816863059998,
+      "learning_rate": 0.00013358260507746156,
+      "loss": 2.3327,
+      "step": 1533
+    },
+    {
+      "epoch": 0.6616346775932715,
+      "grad_norm": 0.16497185826301575,
+      "learning_rate": 0.00013356128937045946,
+      "loss": 2.1085,
+      "step": 1534
+    },
+    {
+      "epoch": 0.6620659909424197,
+      "grad_norm": 0.1854122281074524,
+      "learning_rate": 0.00013353996153770196,
+      "loss": 2.2072,
+      "step": 1535
+    },
+    {
+      "epoch": 0.6624973042915678,
+      "grad_norm": 0.17767350375652313,
+      "learning_rate": 0.00013351862158360526,
+      "loss": 2.2957,
+      "step": 1536
+    },
+    {
+      "epoch": 0.662928617640716,
+      "grad_norm": 0.1622297763824463,
+      "learning_rate": 0.000133497269512588,
+      "loss": 2.2327,
+      "step": 1537
+    },
+    {
+      "epoch": 0.6633599309898641,
+      "grad_norm": 0.17678019404411316,
+      "learning_rate": 0.00013347590532907137,
+      "loss": 2.3593,
+      "step": 1538
+    },
+    {
+      "epoch": 0.6637912443390123,
+      "grad_norm": 0.16748957335948944,
+      "learning_rate": 0.00013345452903747905,
+      "loss": 1.9395,
+      "step": 1539
+    },
+    {
+      "epoch": 0.6642225576881604,
+      "grad_norm": 0.17669428884983063,
+      "learning_rate": 0.00013343314064223727,
+      "loss": 2.1601,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6646538710373086,
+      "grad_norm": 0.17978622019290924,
+      "learning_rate": 0.0001334117401477747,
+      "loss": 2.1725,
+      "step": 1541
+    },
+    {
+      "epoch": 0.6650851843864568,
+      "grad_norm": 0.1864762157201767,
+      "learning_rate": 0.00013339032755852256,
+      "loss": 2.46,
+      "step": 1542
+    },
+    {
+      "epoch": 0.665516497735605,
+      "grad_norm": 0.1701022833585739,
+      "learning_rate": 0.00013336890287891454,
+      "loss": 2.105,
+      "step": 1543
+    },
+    {
+      "epoch": 0.6659478110847531,
+      "grad_norm": 0.14923730492591858,
+      "learning_rate": 0.00013334746611338687,
+      "loss": 2.1351,
+      "step": 1544
+    },
+    {
+      "epoch": 0.6663791244339012,
+      "grad_norm": 0.14656859636306763,
+      "learning_rate": 0.00013332601726637826,
+      "loss": 2.0266,
+      "step": 1545
+    },
+    {
+      "epoch": 0.6668104377830494,
+      "grad_norm": 0.1627119928598404,
+      "learning_rate": 0.00013330455634232992,
+      "loss": 2.1563,
+      "step": 1546
+    },
+    {
+      "epoch": 0.6672417511321975,
+      "grad_norm": 0.20333190262317657,
+      "learning_rate": 0.0001332830833456856,
+      "loss": 2.242,
+      "step": 1547
+    },
+    {
+      "epoch": 0.6676730644813457,
+      "grad_norm": 0.19267724454402924,
+      "learning_rate": 0.00013326159828089151,
+      "loss": 2.2732,
+      "step": 1548
+    },
+    {
+      "epoch": 0.6681043778304938,
+      "grad_norm": 0.19448162615299225,
+      "learning_rate": 0.00013324010115239635,
+      "loss": 2.1141,
+      "step": 1549
+    },
+    {
+      "epoch": 0.668535691179642,
+      "grad_norm": 0.161507710814476,
+      "learning_rate": 0.00013321859196465134,
+      "loss": 2.1465,
+      "step": 1550
+    },
+    {
+      "epoch": 0.668535691179642,
+      "eval_loss": 2.1159982681274414,
+      "eval_runtime": 202.5102,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6689670045287902,
+      "grad_norm": 0.18547660112380981,
+      "learning_rate": 0.00013319707072211018,
+      "loss": 2.0345,
+      "step": 1551
+    },
+    {
+      "epoch": 0.6693983178779384,
+      "grad_norm": 0.1630726009607315,
+      "learning_rate": 0.00013317553742922913,
+      "loss": 2.2126,
+      "step": 1552
+    },
+    {
+      "epoch": 0.6698296312270865,
+      "grad_norm": 0.196413055062294,
+      "learning_rate": 0.00013315399209046683,
+      "loss": 2.2386,
+      "step": 1553
+    },
+    {
+      "epoch": 0.6702609445762346,
+      "grad_norm": 0.17378070950508118,
+      "learning_rate": 0.0001331324347102845,
+      "loss": 2.0881,
+      "step": 1554
+    },
+    {
+      "epoch": 0.6706922579253828,
+      "grad_norm": 0.17834343016147614,
+      "learning_rate": 0.00013311086529314587,
+      "loss": 2.2791,
+      "step": 1555
+    },
+    {
+      "epoch": 0.6711235712745309,
+      "grad_norm": 0.16736628115177155,
+      "learning_rate": 0.00013308928384351708,
+      "loss": 2.323,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6715548846236791,
+      "grad_norm": 0.19940096139907837,
+      "learning_rate": 0.0001330676903658668,
+      "loss": 1.9494,
+      "step": 1557
+    },
+    {
+      "epoch": 0.6719861979728272,
+      "grad_norm": 0.1720159649848938,
+      "learning_rate": 0.00013304608486466624,
+      "loss": 2.3472,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6724175113219755,
+      "grad_norm": 0.18553723394870758,
+      "learning_rate": 0.000133024467344389,
+      "loss": 2.3026,
+      "step": 1559
+    },
+    {
+      "epoch": 0.6728488246711236,
+      "grad_norm": 0.1666479855775833,
+      "learning_rate": 0.00013300283780951123,
+      "loss": 2.3268,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6732801380202718,
+      "grad_norm": 0.16056552529335022,
+      "learning_rate": 0.00013298119626451162,
+      "loss": 2.144,
+      "step": 1561
+    },
+    {
+      "epoch": 0.6737114513694199,
+      "grad_norm": 0.17346884310245514,
+      "learning_rate": 0.00013295954271387128,
+      "loss": 2.0181,
+      "step": 1562
+    },
+    {
+      "epoch": 0.674142764718568,
+      "grad_norm": 0.187408909201622,
+      "learning_rate": 0.00013293787716207373,
+      "loss": 2.3228,
+      "step": 1563
+    },
+    {
+      "epoch": 0.6745740780677162,
+      "grad_norm": 0.16592130064964294,
+      "learning_rate": 0.00013291619961360514,
+      "loss": 2.0589,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6750053914168643,
+      "grad_norm": 0.15113787353038788,
+      "learning_rate": 0.00013289451007295405,
+      "loss": 2.0183,
+      "step": 1565
+    },
+    {
+      "epoch": 0.6754367047660125,
+      "grad_norm": 0.17211243510246277,
+      "learning_rate": 0.00013287280854461152,
+      "loss": 2.0831,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6758680181151606,
+      "grad_norm": 0.17111659049987793,
+      "learning_rate": 0.00013285109503307114,
+      "loss": 2.1257,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6762993314643089,
+      "grad_norm": 0.166707381606102,
+      "learning_rate": 0.00013282936954282884,
+      "loss": 2.1942,
+      "step": 1568
+    },
+    {
+      "epoch": 0.676730644813457,
+      "grad_norm": 0.14764513075351715,
+      "learning_rate": 0.00013280763207838315,
+      "loss": 2.2528,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6771619581626052,
+      "grad_norm": 0.207320898771286,
+      "learning_rate": 0.0001327858826442351,
+      "loss": 2.0984,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6775932715117533,
+      "grad_norm": 0.16856639087200165,
+      "learning_rate": 0.0001327641212448881,
+      "loss": 2.4237,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6780245848609014,
+      "grad_norm": 0.17226271331310272,
+      "learning_rate": 0.00013274234788484814,
+      "loss": 2.0749,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6784558982100496,
+      "grad_norm": 0.17077532410621643,
+      "learning_rate": 0.00013272056256862354,
+      "loss": 2.3413,
+      "step": 1573
+    },
+    {
+      "epoch": 0.6788872115591977,
+      "grad_norm": 0.17257219552993774,
+      "learning_rate": 0.00013269876530072524,
+      "loss": 2.2468,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6793185249083459,
+      "grad_norm": 0.18015973269939423,
+      "learning_rate": 0.0001326769560856666,
+      "loss": 2.028,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6793185249083459,
+      "eval_loss": 2.115028142929077,
+      "eval_runtime": 199.1239,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1575
+    },
+    {
+      "epoch": 0.679749838257494,
+      "grad_norm": 0.15598492324352264,
+      "learning_rate": 0.0001326551349279635,
+      "loss": 2.2803,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6801811516066423,
+      "grad_norm": 0.17086711525917053,
+      "learning_rate": 0.00013263330183213419,
+      "loss": 2.2152,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6806124649557904,
+      "grad_norm": 0.14766903221607208,
+      "learning_rate": 0.00013261145680269943,
+      "loss": 2.1492,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6810437783049386,
+      "grad_norm": 0.20790471136569977,
+      "learning_rate": 0.00013258959984418253,
+      "loss": 2.0664,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6814750916540867,
+      "grad_norm": 0.15962834656238556,
+      "learning_rate": 0.0001325677309611092,
+      "loss": 2.061,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6819064050032349,
+      "grad_norm": 0.2015094757080078,
+      "learning_rate": 0.00013254585015800763,
+      "loss": 2.1781,
+      "step": 1581
+    },
+    {
+      "epoch": 0.682337718352383,
+      "grad_norm": 0.15662479400634766,
+      "learning_rate": 0.0001325239574394085,
+      "loss": 1.9441,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6827690317015311,
+      "grad_norm": 0.17095041275024414,
+      "learning_rate": 0.00013250205280984485,
+      "loss": 2.3391,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6832003450506793,
+      "grad_norm": 0.17483989894390106,
+      "learning_rate": 0.0001324801362738524,
+      "loss": 2.1489,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6836316583998274,
+      "grad_norm": 0.15106312930583954,
+      "learning_rate": 0.00013245820783596907,
+      "loss": 2.1818,
+      "step": 1585
+    },
+    {
+      "epoch": 0.6840629717489757,
+      "grad_norm": 0.1592348963022232,
+      "learning_rate": 0.0001324362675007355,
+      "loss": 2.051,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6844942850981238,
+      "grad_norm": 0.1669055074453354,
+      "learning_rate": 0.00013241431527269464,
+      "loss": 2.3425,
+      "step": 1587
+    },
+    {
+      "epoch": 0.684925598447272,
+      "grad_norm": 0.15657608211040497,
+      "learning_rate": 0.00013239235115639192,
+      "loss": 2.2442,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6853569117964201,
+      "grad_norm": 0.17330090701580048,
+      "learning_rate": 0.00013237037515637524,
+      "loss": 2.1638,
+      "step": 1589
+    },
+    {
+      "epoch": 0.6857882251455683,
+      "grad_norm": 0.16446971893310547,
+      "learning_rate": 0.000132348387277195,
+      "loss": 2.1034,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6862195384947164,
+      "grad_norm": 0.1576288938522339,
+      "learning_rate": 0.00013232638752340404,
+      "loss": 2.0789,
+      "step": 1591
+    },
+    {
+      "epoch": 0.6866508518438645,
+      "grad_norm": 0.18883441388607025,
+      "learning_rate": 0.0001323043758995576,
+      "loss": 2.2768,
+      "step": 1592
+    },
+    {
+      "epoch": 0.6870821651930127,
+      "grad_norm": 0.1743883192539215,
+      "learning_rate": 0.0001322823524102135,
+      "loss": 2.2914,
+      "step": 1593
+    },
+    {
+      "epoch": 0.6875134785421608,
+      "grad_norm": 0.16601242125034332,
+      "learning_rate": 0.0001322603170599319,
+      "loss": 2.2557,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6879447918913091,
+      "grad_norm": 0.19149911403656006,
+      "learning_rate": 0.00013223826985327543,
+      "loss": 2.3684,
+      "step": 1595
+    },
+    {
+      "epoch": 0.6883761052404572,
+      "grad_norm": 0.16588738560676575,
+      "learning_rate": 0.00013221621079480926,
+      "loss": 2.165,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6888074185896054,
+      "grad_norm": 0.1697489619255066,
+      "learning_rate": 0.00013219413988910096,
+      "loss": 2.2294,
+      "step": 1597
+    },
+    {
+      "epoch": 0.6892387319387535,
+      "grad_norm": 0.16599583625793457,
+      "learning_rate": 0.0001321720571407205,
+      "loss": 2.2237,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6896700452879017,
+      "grad_norm": 0.16053009033203125,
+      "learning_rate": 0.00013214996255424039,
+      "loss": 2.1803,
+      "step": 1599
+    },
+    {
+      "epoch": 0.6901013586370498,
+      "grad_norm": 0.18454283475875854,
+      "learning_rate": 0.00013212785613423558,
+      "loss": 2.3049,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6901013586370498,
+      "eval_loss": 2.1139638423919678,
+      "eval_runtime": 198.9309,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6905326719861979,
+      "grad_norm": 0.16557788848876953,
+      "learning_rate": 0.00013210573788528338,
+      "loss": 2.0268,
+      "step": 1601
+    },
+    {
+      "epoch": 0.6909639853353461,
+      "grad_norm": 0.18790724873542786,
+      "learning_rate": 0.00013208360781196367,
+      "loss": 2.3526,
+      "step": 1602
+    },
+    {
+      "epoch": 0.6913952986844942,
+      "grad_norm": 0.19259871542453766,
+      "learning_rate": 0.0001320614659188587,
+      "loss": 2.0988,
+      "step": 1603
+    },
+    {
+      "epoch": 0.6918266120336425,
+      "grad_norm": 0.15778112411499023,
+      "learning_rate": 0.0001320393122105532,
+      "loss": 2.2377,
+      "step": 1604
+    },
+    {
+      "epoch": 0.6922579253827906,
+      "grad_norm": 0.18303173780441284,
+      "learning_rate": 0.00013201714669163435,
+      "loss": 2.2089,
+      "step": 1605
+    },
+    {
+      "epoch": 0.6926892387319388,
+      "grad_norm": 0.17680174112319946,
+      "learning_rate": 0.00013199496936669176,
+      "loss": 2.3218,
+      "step": 1606
+    },
+    {
+      "epoch": 0.6931205520810869,
+      "grad_norm": 0.17215006053447723,
+      "learning_rate": 0.00013197278024031745,
+      "loss": 2.1908,
+      "step": 1607
+    },
+    {
+      "epoch": 0.6935518654302351,
+      "grad_norm": 0.16192790865898132,
+      "learning_rate": 0.00013195057931710593,
+      "loss": 2.1841,
+      "step": 1608
+    },
+    {
+      "epoch": 0.6939831787793832,
+      "grad_norm": 0.14507627487182617,
+      "learning_rate": 0.0001319283666016542,
+      "loss": 2.0118,
+      "step": 1609
+    },
+    {
+      "epoch": 0.6944144921285313,
+      "grad_norm": 0.18748687207698822,
+      "learning_rate": 0.00013190614209856156,
+      "loss": 2.4215,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6948458054776795,
+      "grad_norm": 0.17146620154380798,
+      "learning_rate": 0.00013188390581242993,
+      "loss": 2.1129,
+      "step": 1611
+    },
+    {
+      "epoch": 0.6952771188268277,
+      "grad_norm": 0.16766369342803955,
+      "learning_rate": 0.00013186165774786352,
+      "loss": 2.1973,
+      "step": 1612
+    },
+    {
+      "epoch": 0.6957084321759759,
+      "grad_norm": 0.17402087152004242,
+      "learning_rate": 0.00013183939790946901,
+      "loss": 2.1548,
+      "step": 1613
+    },
+    {
+      "epoch": 0.696139745525124,
+      "grad_norm": 0.1601400375366211,
+      "learning_rate": 0.00013181712630185564,
+      "loss": 2.0221,
+      "step": 1614
+    },
+    {
+      "epoch": 0.6965710588742722,
+      "grad_norm": 0.1470576971769333,
+      "learning_rate": 0.00013179484292963488,
+      "loss": 2.1986,
+      "step": 1615
+    },
+    {
+      "epoch": 0.6970023722234203,
+      "grad_norm": 0.166117861866951,
+      "learning_rate": 0.0001317725477974208,
+      "loss": 2.4442,
+      "step": 1616
+    },
+    {
+      "epoch": 0.6974336855725685,
+      "grad_norm": 0.1759869009256363,
+      "learning_rate": 0.00013175024090982988,
+      "loss": 2.2802,
+      "step": 1617
+    },
+    {
+      "epoch": 0.6978649989217166,
+      "grad_norm": 0.14935262501239777,
+      "learning_rate": 0.00013172792227148094,
+      "loss": 2.0724,
+      "step": 1618
+    },
+    {
+      "epoch": 0.6982963122708647,
+      "grad_norm": 0.19077062606811523,
+      "learning_rate": 0.0001317055918869953,
+      "loss": 2.1706,
+      "step": 1619
+    },
+    {
+      "epoch": 0.6987276256200129,
+      "grad_norm": 0.16506893932819366,
+      "learning_rate": 0.00013168324976099678,
+      "loss": 2.2949,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6991589389691611,
+      "grad_norm": 0.18387648463249207,
+      "learning_rate": 0.0001316608958981115,
+      "loss": 2.1215,
+      "step": 1621
+    },
+    {
+      "epoch": 0.6995902523183093,
+      "grad_norm": 0.18008792400360107,
+      "learning_rate": 0.0001316385303029681,
+      "loss": 2.1253,
+      "step": 1622
+    },
+    {
+      "epoch": 0.7000215656674574,
+      "grad_norm": 0.18667413294315338,
+      "learning_rate": 0.0001316161529801976,
+      "loss": 2.2715,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7004528790166056,
+      "grad_norm": 0.20948120951652527,
+      "learning_rate": 0.0001315937639344335,
+      "loss": 2.1971,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7008841923657537,
+      "grad_norm": 0.16733720898628235,
+      "learning_rate": 0.00013157136317031167,
+      "loss": 2.24,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7008841923657537,
+      "eval_loss": 2.1133079528808594,
+      "eval_runtime": 198.8724,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7013155057149019,
+      "grad_norm": 0.17461319267749786,
+      "learning_rate": 0.00013154895069247046,
+      "loss": 2.2935,
+      "step": 1626
+    },
+    {
+      "epoch": 0.70174681906405,
+      "grad_norm": 0.16576853394508362,
+      "learning_rate": 0.00013152652650555056,
+      "loss": 2.3532,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7021781324131982,
+      "grad_norm": 0.1629101186990738,
+      "learning_rate": 0.00013150409061419522,
+      "loss": 2.2289,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7026094457623463,
+      "grad_norm": 0.19192031025886536,
+      "learning_rate": 0.00013148164302304998,
+      "loss": 2.1729,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7030407591114946,
+      "grad_norm": 0.23054882884025574,
+      "learning_rate": 0.00013145918373676287,
+      "loss": 2.2828,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7034720724606427,
+      "grad_norm": 0.15536463260650635,
+      "learning_rate": 0.00013143671275998438,
+      "loss": 1.9453,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7039033858097908,
+      "grad_norm": 0.1682126522064209,
+      "learning_rate": 0.0001314142300973673,
+      "loss": 2.2695,
+      "step": 1632
+    },
+    {
+      "epoch": 0.704334699158939,
+      "grad_norm": 0.1809571087360382,
+      "learning_rate": 0.00013139173575356695,
+      "loss": 2.2376,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7047660125080871,
+      "grad_norm": 0.14714162051677704,
+      "learning_rate": 0.000131369229733241,
+      "loss": 1.9291,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7051973258572353,
+      "grad_norm": 0.17537806928157806,
+      "learning_rate": 0.00013134671204104962,
+      "loss": 2.1672,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7056286392063834,
+      "grad_norm": 0.19214670360088348,
+      "learning_rate": 0.0001313241826816553,
+      "loss": 2.3329,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7060599525555316,
+      "grad_norm": 0.1673903614282608,
+      "learning_rate": 0.000131301641659723,
+      "loss": 2.1169,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7064912659046797,
+      "grad_norm": 0.16690760850906372,
+      "learning_rate": 0.0001312790889799201,
+      "loss": 2.2248,
+      "step": 1638
+    },
+    {
+      "epoch": 0.706922579253828,
+      "grad_norm": 0.1609828621149063,
+      "learning_rate": 0.00013125652464691638,
+      "loss": 2.2809,
+      "step": 1639
+    },
+    {
+      "epoch": 0.7073538926029761,
+      "grad_norm": 0.20783407986164093,
+      "learning_rate": 0.00013123394866538404,
+      "loss": 2.3455,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7077852059521242,
+      "grad_norm": 0.17028851807117462,
+      "learning_rate": 0.00013121136103999762,
+      "loss": 2.1143,
+      "step": 1641
+    },
+    {
+      "epoch": 0.7082165193012724,
+      "grad_norm": 0.16780084371566772,
+      "learning_rate": 0.00013118876177543417,
+      "loss": 2.1457,
+      "step": 1642
+    },
+    {
+      "epoch": 0.7086478326504205,
+      "grad_norm": 0.18313480913639069,
+      "learning_rate": 0.00013116615087637316,
+      "loss": 2.4553,
+      "step": 1643
+    },
+    {
+      "epoch": 0.7090791459995687,
+      "grad_norm": 0.17945381999015808,
+      "learning_rate": 0.00013114352834749638,
+      "loss": 2.2977,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7095104593487168,
+      "grad_norm": 0.17324642837047577,
+      "learning_rate": 0.00013112089419348809,
+      "loss": 2.3561,
+      "step": 1645
+    },
+    {
+      "epoch": 0.709941772697865,
+      "grad_norm": 0.18440823256969452,
+      "learning_rate": 0.00013109824841903492,
+      "loss": 2.399,
+      "step": 1646
+    },
+    {
+      "epoch": 0.7103730860470131,
+      "grad_norm": 0.17321951687335968,
+      "learning_rate": 0.00013107559102882592,
+      "loss": 2.3652,
+      "step": 1647
+    },
+    {
+      "epoch": 0.7108043993961614,
+      "grad_norm": 0.1459442675113678,
+      "learning_rate": 0.0001310529220275526,
+      "loss": 1.9473,
+      "step": 1648
+    },
+    {
+      "epoch": 0.7112357127453095,
+      "grad_norm": 0.16677115857601166,
+      "learning_rate": 0.00013103024141990875,
+      "loss": 2.2992,
+      "step": 1649
+    },
+    {
+      "epoch": 0.7116670260944576,
+      "grad_norm": 0.16851958632469177,
+      "learning_rate": 0.0001310075492105907,
+      "loss": 2.3005,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7116670260944576,
+      "eval_loss": 2.113213062286377,
+      "eval_runtime": 198.74,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7120983394436058,
+      "grad_norm": 0.17779399454593658,
+      "learning_rate": 0.00013098484540429708,
+      "loss": 2.1,
+      "step": 1651
+    },
+    {
+      "epoch": 0.7125296527927539,
+      "grad_norm": 0.1771998256444931,
+      "learning_rate": 0.000130962130005729,
+      "loss": 2.1188,
+      "step": 1652
+    },
+    {
+      "epoch": 0.7129609661419021,
+      "grad_norm": 0.7633037567138672,
+      "learning_rate": 0.00013093940301958992,
+      "loss": 2.1638,
+      "step": 1653
+    },
+    {
+      "epoch": 0.7133922794910502,
+      "grad_norm": 0.16661760210990906,
+      "learning_rate": 0.0001309166644505857,
+      "loss": 1.8541,
+      "step": 1654
+    },
+    {
+      "epoch": 0.7138235928401984,
+      "grad_norm": 0.16859006881713867,
+      "learning_rate": 0.0001308939143034246,
+      "loss": 2.3292,
+      "step": 1655
+    },
+    {
+      "epoch": 0.7142549061893465,
+      "grad_norm": 0.1498507708311081,
+      "learning_rate": 0.00013087115258281727,
+      "loss": 2.0369,
+      "step": 1656
+    },
+    {
+      "epoch": 0.7146862195384948,
+      "grad_norm": 0.15672807395458221,
+      "learning_rate": 0.0001308483792934768,
+      "loss": 2.283,
+      "step": 1657
+    },
+    {
+      "epoch": 0.7151175328876429,
+      "grad_norm": 0.181760773062706,
+      "learning_rate": 0.0001308255944401187,
+      "loss": 2.2157,
+      "step": 1658
+    },
+    {
+      "epoch": 0.715548846236791,
+      "grad_norm": 0.15118049085140228,
+      "learning_rate": 0.00013080279802746072,
+      "loss": 2.2297,
+      "step": 1659
+    },
+    {
+      "epoch": 0.7159801595859392,
+      "grad_norm": 0.1727578490972519,
+      "learning_rate": 0.0001307799900602232,
+      "loss": 2.3038,
+      "step": 1660
+    },
+    {
+      "epoch": 0.7164114729350873,
+      "grad_norm": 0.20055891573429108,
+      "learning_rate": 0.00013075717054312875,
+      "loss": 2.1896,
+      "step": 1661
+    },
+    {
+      "epoch": 0.7168427862842355,
+      "grad_norm": 0.27386513352394104,
+      "learning_rate": 0.00013073433948090237,
+      "loss": 2.139,
+      "step": 1662
+    },
+    {
+      "epoch": 0.7172740996333836,
+      "grad_norm": 0.16511470079421997,
+      "learning_rate": 0.00013071149687827148,
+      "loss": 2.2663,
+      "step": 1663
+    },
+    {
+      "epoch": 0.7177054129825318,
+      "grad_norm": 0.15780644118785858,
+      "learning_rate": 0.0001306886427399659,
+      "loss": 2.1271,
+      "step": 1664
+    },
+    {
+      "epoch": 0.71813672633168,
+      "grad_norm": 0.15114596486091614,
+      "learning_rate": 0.00013066577707071788,
+      "loss": 1.9493,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7185680396808282,
+      "grad_norm": 0.2407820224761963,
+      "learning_rate": 0.00013064289987526197,
+      "loss": 2.0795,
+      "step": 1666
+    },
+    {
+      "epoch": 0.7189993530299763,
+      "grad_norm": 0.17818287014961243,
+      "learning_rate": 0.00013062001115833513,
+      "loss": 2.311,
+      "step": 1667
+    },
+    {
+      "epoch": 0.7194306663791245,
+      "grad_norm": 0.16417573392391205,
+      "learning_rate": 0.0001305971109246767,
+      "loss": 2.1726,
+      "step": 1668
+    },
+    {
+      "epoch": 0.7198619797282726,
+      "grad_norm": 0.1662161648273468,
+      "learning_rate": 0.00013057419917902847,
+      "loss": 2.2471,
+      "step": 1669
+    },
+    {
+      "epoch": 0.7202932930774207,
+      "grad_norm": 0.20036767423152924,
+      "learning_rate": 0.00013055127592613458,
+      "loss": 2.3349,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7207246064265689,
+      "grad_norm": 0.21882568299770355,
+      "learning_rate": 0.0001305283411707415,
+      "loss": 2.0971,
+      "step": 1671
+    },
+    {
+      "epoch": 0.721155919775717,
+      "grad_norm": 0.15568691492080688,
+      "learning_rate": 0.0001305053949175981,
+      "loss": 2.3335,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7215872331248652,
+      "grad_norm": 0.18450284004211426,
+      "learning_rate": 0.00013048243717145568,
+      "loss": 2.2341,
+      "step": 1673
+    },
+    {
+      "epoch": 0.7220185464740134,
+      "grad_norm": 0.16564737260341644,
+      "learning_rate": 0.00013045946793706792,
+      "loss": 2.3969,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7224498598231616,
+      "grad_norm": 0.17275582253932953,
+      "learning_rate": 0.00013043648721919085,
+      "loss": 2.2686,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7224498598231616,
+      "eval_loss": 2.1133737564086914,
+      "eval_runtime": 207.8226,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7228811731723097,
+      "grad_norm": 0.1843978464603424,
+      "learning_rate": 0.00013041349502258283,
+      "loss": 2.1839,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7233124865214579,
+      "grad_norm": 0.16497057676315308,
+      "learning_rate": 0.0001303904913520047,
+      "loss": 2.3698,
+      "step": 1677
+    },
+    {
+      "epoch": 0.723743799870606,
+      "grad_norm": 0.40322139859199524,
+      "learning_rate": 0.00013036747621221958,
+      "loss": 2.2511,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7241751132197541,
+      "grad_norm": 0.17039461433887482,
+      "learning_rate": 0.00013034444960799302,
+      "loss": 2.6212,
+      "step": 1679
+    },
+    {
+      "epoch": 0.7246064265689023,
+      "grad_norm": 0.20115548372268677,
+      "learning_rate": 0.00013032141154409294,
+      "loss": 2.239,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7250377399180504,
+      "grad_norm": 0.17254777252674103,
+      "learning_rate": 0.00013029836202528962,
+      "loss": 2.204,
+      "step": 1681
+    },
+    {
+      "epoch": 0.7254690532671986,
+      "grad_norm": 0.180454283952713,
+      "learning_rate": 0.00013027530105635566,
+      "loss": 2.1329,
+      "step": 1682
+    },
+    {
+      "epoch": 0.7259003666163468,
+      "grad_norm": 0.16699275374412537,
+      "learning_rate": 0.0001302522286420662,
+      "loss": 2.1671,
+      "step": 1683
+    },
+    {
+      "epoch": 0.726331679965495,
+      "grad_norm": 0.179158553481102,
+      "learning_rate": 0.00013022914478719855,
+      "loss": 2.3858,
+      "step": 1684
+    },
+    {
+      "epoch": 0.7267629933146431,
+      "grad_norm": 0.16012237966060638,
+      "learning_rate": 0.00013020604949653248,
+      "loss": 2.2873,
+      "step": 1685
+    },
+    {
+      "epoch": 0.7271943066637913,
+      "grad_norm": 0.16123633086681366,
+      "learning_rate": 0.00013018294277485017,
+      "loss": 2.2534,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7276256200129394,
+      "grad_norm": 0.1661210060119629,
+      "learning_rate": 0.00013015982462693607,
+      "loss": 2.3161,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7280569333620875,
+      "grad_norm": 0.17415104806423187,
+      "learning_rate": 0.00013013669505757705,
+      "loss": 2.0021,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7284882467112357,
+      "grad_norm": 0.18667307496070862,
+      "learning_rate": 0.00013011355407156237,
+      "loss": 2.0903,
+      "step": 1689
+    },
+    {
+      "epoch": 0.7289195600603838,
+      "grad_norm": 0.19069887697696686,
+      "learning_rate": 0.0001300904016736836,
+      "loss": 2.0884,
+      "step": 1690
+    },
+    {
+      "epoch": 0.729350873409532,
+      "grad_norm": 0.15097585320472717,
+      "learning_rate": 0.0001300672378687347,
+      "loss": 2.3432,
+      "step": 1691
+    },
+    {
+      "epoch": 0.7297821867586802,
+      "grad_norm": 0.18867550790309906,
+      "learning_rate": 0.00013004406266151202,
+      "loss": 2.2952,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7302135001078284,
+      "grad_norm": 0.1902410238981247,
+      "learning_rate": 0.00013002087605681417,
+      "loss": 2.2431,
+      "step": 1693
+    },
+    {
+      "epoch": 0.7306448134569765,
+      "grad_norm": 0.20812374353408813,
+      "learning_rate": 0.00012999767805944228,
+      "loss": 2.1721,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7310761268061247,
+      "grad_norm": 0.1791360080242157,
+      "learning_rate": 0.0001299744686741997,
+      "loss": 2.425,
+      "step": 1695
+    },
+    {
+      "epoch": 0.7315074401552728,
+      "grad_norm": 0.18107859790325165,
+      "learning_rate": 0.00012995124790589216,
+      "loss": 2.3085,
+      "step": 1696
+    },
+    {
+      "epoch": 0.731938753504421,
+      "grad_norm": 0.17820578813552856,
+      "learning_rate": 0.0001299280157593278,
+      "loss": 2.2546,
+      "step": 1697
+    },
+    {
+      "epoch": 0.7323700668535691,
+      "grad_norm": 0.18765179812908173,
+      "learning_rate": 0.0001299047722393171,
+      "loss": 1.9474,
+      "step": 1698
+    },
+    {
+      "epoch": 0.7328013802027172,
+      "grad_norm": 0.16326814889907837,
+      "learning_rate": 0.00012988151735067292,
+      "loss": 2.2706,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7332326935518654,
+      "grad_norm": 0.16765248775482178,
+      "learning_rate": 0.0001298582510982104,
+      "loss": 2.2238,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7332326935518654,
+      "eval_loss": 2.112516403198242,
+      "eval_runtime": 203.7373,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7336640069010136,
+      "grad_norm": 0.1879028081893921,
+      "learning_rate": 0.00012983497348674703,
+      "loss": 2.2476,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7340953202501618,
+      "grad_norm": 0.17669300734996796,
+      "learning_rate": 0.00012981168452110275,
+      "loss": 2.1725,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7345266335993099,
+      "grad_norm": 0.16430909931659698,
+      "learning_rate": 0.00012978838420609978,
+      "loss": 2.1852,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7349579469484581,
+      "grad_norm": 0.20571014285087585,
+      "learning_rate": 0.00012976507254656267,
+      "loss": 2.2288,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7353892602976062,
+      "grad_norm": 0.17860174179077148,
+      "learning_rate": 0.00012974174954731845,
+      "loss": 2.4026,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7358205736467544,
+      "grad_norm": 0.16562241315841675,
+      "learning_rate": 0.0001297184152131963,
+      "loss": 2.0744,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7362518869959025,
+      "grad_norm": 0.17572078108787537,
+      "learning_rate": 0.0001296950695490279,
+      "loss": 2.3155,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7366832003450506,
+      "grad_norm": 0.17776505649089813,
+      "learning_rate": 0.00012967171255964723,
+      "loss": 2.2062,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7371145136941988,
+      "grad_norm": 0.24452145397663116,
+      "learning_rate": 0.00012964834424989056,
+      "loss": 2.0492,
+      "step": 1709
+    },
+    {
+      "epoch": 0.737545827043347,
+      "grad_norm": 0.18888737261295319,
+      "learning_rate": 0.00012962496462459662,
+      "loss": 2.2781,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7379771403924952,
+      "grad_norm": 0.16819074749946594,
+      "learning_rate": 0.00012960157368860635,
+      "loss": 2.2522,
+      "step": 1711
+    },
+    {
+      "epoch": 0.7384084537416433,
+      "grad_norm": 0.1931333839893341,
+      "learning_rate": 0.00012957817144676318,
+      "loss": 2.2074,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7388397670907915,
+      "grad_norm": 0.20771650969982147,
+      "learning_rate": 0.00012955475790391274,
+      "loss": 2.1816,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7392710804399396,
+      "grad_norm": 0.14832836389541626,
+      "learning_rate": 0.00012953133306490305,
+      "loss": 2.1424,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7397023937890878,
+      "grad_norm": 0.16810181736946106,
+      "learning_rate": 0.00012950789693458453,
+      "loss": 2.2168,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7401337071382359,
+      "grad_norm": 0.17525973916053772,
+      "learning_rate": 0.00012948444951780985,
+      "loss": 2.1605,
+      "step": 1716
+    },
+    {
+      "epoch": 0.740565020487384,
+      "grad_norm": 0.17031776905059814,
+      "learning_rate": 0.0001294609908194341,
+      "loss": 2.1088,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7409963338365323,
+      "grad_norm": 0.21623241901397705,
+      "learning_rate": 0.00012943752084431458,
+      "loss": 2.2305,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7414276471856804,
+      "grad_norm": 0.18505755066871643,
+      "learning_rate": 0.00012941403959731106,
+      "loss": 2.273,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7418589605348286,
+      "grad_norm": 0.1552658975124359,
+      "learning_rate": 0.00012939054708328555,
+      "loss": 2.195,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7422902738839767,
+      "grad_norm": 0.17611075937747955,
+      "learning_rate": 0.0001293670433071025,
+      "loss": 2.2172,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7427215872331249,
+      "grad_norm": 0.16624145209789276,
+      "learning_rate": 0.00012934352827362862,
+      "loss": 2.3531,
+      "step": 1722
+    },
+    {
+      "epoch": 0.743152900582273,
+      "grad_norm": 0.18196657299995422,
+      "learning_rate": 0.00012932000198773288,
+      "loss": 2.286,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7435842139314212,
+      "grad_norm": 0.16066540777683258,
+      "learning_rate": 0.00012929646445428668,
+      "loss": 2.4649,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7440155272805693,
+      "grad_norm": 0.17888200283050537,
+      "learning_rate": 0.0001292729156781638,
+      "loss": 2.0565,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7440155272805693,
+      "eval_loss": 2.1118383407592773,
+      "eval_runtime": 200.5771,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7444468406297174,
+      "grad_norm": 0.19620761275291443,
+      "learning_rate": 0.00012924935566424012,
+      "loss": 2.1012,
+      "step": 1726
+    },
+    {
+      "epoch": 0.7448781539788657,
+      "grad_norm": 0.1674032360315323,
+      "learning_rate": 0.00012922578441739414,
+      "loss": 2.151,
+      "step": 1727
+    },
+    {
+      "epoch": 0.7453094673280138,
+      "grad_norm": 0.16218839585781097,
+      "learning_rate": 0.0001292022019425065,
+      "loss": 2.099,
+      "step": 1728
+    },
+    {
+      "epoch": 0.745740780677162,
+      "grad_norm": 0.16986028850078583,
+      "learning_rate": 0.0001291786082444602,
+      "loss": 2.2028,
+      "step": 1729
+    },
+    {
+      "epoch": 0.7461720940263101,
+      "grad_norm": 0.16744236648082733,
+      "learning_rate": 0.00012915500332814054,
+      "loss": 2.2487,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7466034073754583,
+      "grad_norm": 0.15288467705249786,
+      "learning_rate": 0.00012913138719843526,
+      "loss": 2.3417,
+      "step": 1731
+    },
+    {
+      "epoch": 0.7470347207246064,
+      "grad_norm": 0.16407521069049835,
+      "learning_rate": 0.00012910775986023427,
+      "loss": 2.1652,
+      "step": 1732
+    },
+    {
+      "epoch": 0.7474660340737546,
+      "grad_norm": 0.16800500452518463,
+      "learning_rate": 0.0001290841213184299,
+      "loss": 2.3682,
+      "step": 1733
+    },
+    {
+      "epoch": 0.7478973474229027,
+      "grad_norm": 0.1646740883588791,
+      "learning_rate": 0.00012906047157791673,
+      "loss": 1.9321,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7483286607720508,
+      "grad_norm": 0.173130065202713,
+      "learning_rate": 0.00012903681064359174,
+      "loss": 2.0989,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7487599741211991,
+      "grad_norm": 0.16296248137950897,
+      "learning_rate": 0.00012901313852035416,
+      "loss": 2.0963,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7491912874703472,
+      "grad_norm": 0.16170215606689453,
+      "learning_rate": 0.00012898945521310558,
+      "loss": 2.4658,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7496226008194954,
+      "grad_norm": 0.1658870279788971,
+      "learning_rate": 0.00012896576072674989,
+      "loss": 2.2508,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7500539141686435,
+      "grad_norm": 0.1697956621646881,
+      "learning_rate": 0.00012894205506619325,
+      "loss": 2.224,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7504852275177917,
+      "grad_norm": 0.1617509126663208,
+      "learning_rate": 0.00012891833823634426,
+      "loss": 2.2718,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7509165408669398,
+      "grad_norm": 0.17925076186656952,
+      "learning_rate": 0.00012889461024211365,
+      "loss": 2.3989,
+      "step": 1741
+    },
+    {
+      "epoch": 0.751347854216088,
+      "grad_norm": 0.17088088393211365,
+      "learning_rate": 0.00012887087108841462,
+      "loss": 2.2883,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7517791675652361,
+      "grad_norm": 0.1580200046300888,
+      "learning_rate": 0.0001288471207801626,
+      "loss": 2.1735,
+      "step": 1743
+    },
+    {
+      "epoch": 0.7522104809143842,
+      "grad_norm": 0.16144931316375732,
+      "learning_rate": 0.00012882335932227538,
+      "loss": 2.1267,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7526417942635325,
+      "grad_norm": 0.17619143426418304,
+      "learning_rate": 0.000128799586719673,
+      "loss": 2.103,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7530731076126806,
+      "grad_norm": 0.16034960746765137,
+      "learning_rate": 0.00012877580297727788,
+      "loss": 2.0091,
+      "step": 1746
+    },
+    {
+      "epoch": 0.7535044209618288,
+      "grad_norm": 0.17221620678901672,
+      "learning_rate": 0.00012875200810001463,
+      "loss": 2.1892,
+      "step": 1747
+    },
+    {
+      "epoch": 0.7539357343109769,
+      "grad_norm": 0.18462108075618744,
+      "learning_rate": 0.0001287282020928103,
+      "loss": 2.3866,
+      "step": 1748
+    },
+    {
+      "epoch": 0.7543670476601251,
+      "grad_norm": 0.22234316170215607,
+      "learning_rate": 0.00012870438496059416,
+      "loss": 2.254,
+      "step": 1749
+    },
+    {
+      "epoch": 0.7547983610092732,
+      "grad_norm": 0.16872234642505646,
+      "learning_rate": 0.0001286805567082978,
+      "loss": 2.267,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7547983610092732,
+      "eval_loss": 2.111443519592285,
+      "eval_runtime": 201.8678,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7552296743584214,
+      "grad_norm": 0.20871369540691376,
+      "learning_rate": 0.00012865671734085514,
+      "loss": 2.0419,
+      "step": 1751
+    },
+    {
+      "epoch": 0.7556609877075695,
+      "grad_norm": 0.21343334019184113,
+      "learning_rate": 0.00012863286686320237,
+      "loss": 2.1727,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7560923010567177,
+      "grad_norm": 0.15887396037578583,
+      "learning_rate": 0.00012860900528027802,
+      "loss": 2.1621,
+      "step": 1753
+    },
+    {
+      "epoch": 0.7565236144058659,
+      "grad_norm": 0.1738533079624176,
+      "learning_rate": 0.00012858513259702287,
+      "loss": 2.347,
+      "step": 1754
+    },
+    {
+      "epoch": 0.756954927755014,
+      "grad_norm": 0.1632392704486847,
+      "learning_rate": 0.00012856124881837998,
+      "loss": 2.0953,
+      "step": 1755
+    },
+    {
+      "epoch": 0.7573862411041622,
+      "grad_norm": 0.163606658577919,
+      "learning_rate": 0.00012853735394929483,
+      "loss": 2.0839,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7578175544533103,
+      "grad_norm": 0.18141624331474304,
+      "learning_rate": 0.00012851344799471502,
+      "loss": 2.3043,
+      "step": 1757
+    },
+    {
+      "epoch": 0.7582488678024585,
+      "grad_norm": 0.172869473695755,
+      "learning_rate": 0.00012848953095959063,
+      "loss": 2.1286,
+      "step": 1758
+    },
+    {
+      "epoch": 0.7586801811516066,
+      "grad_norm": 0.17472733557224274,
+      "learning_rate": 0.0001284656028488739,
+      "loss": 2.1691,
+      "step": 1759
+    },
+    {
+      "epoch": 0.7591114945007548,
+      "grad_norm": 0.1549355536699295,
+      "learning_rate": 0.0001284416636675194,
+      "loss": 2.053,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7595428078499029,
+      "grad_norm": 0.1682683378458023,
+      "learning_rate": 0.000128417713420484,
+      "loss": 2.2314,
+      "step": 1761
+    },
+    {
+      "epoch": 0.7599741211990511,
+      "grad_norm": 0.1796865016222,
+      "learning_rate": 0.00012839375211272692,
+      "loss": 2.0654,
+      "step": 1762
+    },
+    {
+      "epoch": 0.7604054345481993,
+      "grad_norm": 0.1781945377588272,
+      "learning_rate": 0.0001283697797492095,
+      "loss": 2.3292,
+      "step": 1763
+    },
+    {
+      "epoch": 0.7608367478973475,
+      "grad_norm": 0.17216362059116364,
+      "learning_rate": 0.00012834579633489557,
+      "loss": 2.1891,
+      "step": 1764
+    },
+    {
+      "epoch": 0.7612680612464956,
+      "grad_norm": 0.17948682606220245,
+      "learning_rate": 0.00012832180187475112,
+      "loss": 2.2332,
+      "step": 1765
+    },
+    {
+      "epoch": 0.7616993745956437,
+      "grad_norm": 0.17100459337234497,
+      "learning_rate": 0.00012829779637374446,
+      "loss": 2.291,
+      "step": 1766
+    },
+    {
+      "epoch": 0.7621306879447919,
+      "grad_norm": 0.15035024285316467,
+      "learning_rate": 0.00012827377983684623,
+      "loss": 2.1859,
+      "step": 1767
+    },
+    {
+      "epoch": 0.76256200129394,
+      "grad_norm": 0.17111943662166595,
+      "learning_rate": 0.0001282497522690293,
+      "loss": 2.0255,
+      "step": 1768
+    },
+    {
+      "epoch": 0.7629933146430882,
+      "grad_norm": 0.19134071469306946,
+      "learning_rate": 0.0001282257136752688,
+      "loss": 2.1275,
+      "step": 1769
+    },
+    {
+      "epoch": 0.7634246279922363,
+      "grad_norm": 0.16713270545005798,
+      "learning_rate": 0.00012820166406054225,
+      "loss": 2.2782,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7638559413413846,
+      "grad_norm": 0.16537237167358398,
+      "learning_rate": 0.00012817760342982934,
+      "loss": 2.1584,
+      "step": 1771
+    },
+    {
+      "epoch": 0.7642872546905327,
+      "grad_norm": 0.17069876194000244,
+      "learning_rate": 0.0001281535317881121,
+      "loss": 2.2834,
+      "step": 1772
+    },
+    {
+      "epoch": 0.7647185680396809,
+      "grad_norm": 0.16779346764087677,
+      "learning_rate": 0.00012812944914037481,
+      "loss": 2.3434,
+      "step": 1773
+    },
+    {
+      "epoch": 0.765149881388829,
+      "grad_norm": 0.16906912624835968,
+      "learning_rate": 0.00012810535549160408,
+      "loss": 2.3108,
+      "step": 1774
+    },
+    {
+      "epoch": 0.7655811947379771,
+      "grad_norm": 0.16279946267604828,
+      "learning_rate": 0.00012808125084678876,
+      "loss": 1.9309,
+      "step": 1775
+    },
+    {
+      "epoch": 0.7655811947379771,
+      "eval_loss": 2.111085891723633,
+      "eval_runtime": 200.7241,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1775
+    },
+    {
+      "epoch": 0.7660125080871253,
+      "grad_norm": 0.16854463517665863,
+      "learning_rate": 0.00012805713521091993,
+      "loss": 2.2577,
+      "step": 1776
+    },
+    {
+      "epoch": 0.7664438214362734,
+      "grad_norm": 0.1790621429681778,
+      "learning_rate": 0.00012803300858899104,
+      "loss": 2.0779,
+      "step": 1777
+    },
+    {
+      "epoch": 0.7668751347854216,
+      "grad_norm": 0.17216108739376068,
+      "learning_rate": 0.0001280088709859978,
+      "loss": 2.1478,
+      "step": 1778
+    },
+    {
+      "epoch": 0.7673064481345697,
+      "grad_norm": 0.16908183693885803,
+      "learning_rate": 0.00012798472240693808,
+      "loss": 2.3447,
+      "step": 1779
+    },
+    {
+      "epoch": 0.767737761483718,
+      "grad_norm": 0.16917240619659424,
+      "learning_rate": 0.00012796056285681217,
+      "loss": 2.2356,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7681690748328661,
+      "grad_norm": 0.1798560619354248,
+      "learning_rate": 0.00012793639234062254,
+      "loss": 2.2074,
+      "step": 1781
+    },
+    {
+      "epoch": 0.7686003881820143,
+      "grad_norm": 0.1688271462917328,
+      "learning_rate": 0.000127912210863374,
+      "loss": 2.2913,
+      "step": 1782
+    },
+    {
+      "epoch": 0.7690317015311624,
+      "grad_norm": 0.1530812680721283,
+      "learning_rate": 0.00012788801843007357,
+      "loss": 2.2181,
+      "step": 1783
+    },
+    {
+      "epoch": 0.7694630148803105,
+      "grad_norm": 0.1799415647983551,
+      "learning_rate": 0.0001278638150457305,
+      "loss": 2.0686,
+      "step": 1784
+    },
+    {
+      "epoch": 0.7698943282294587,
+      "grad_norm": 0.16236048936843872,
+      "learning_rate": 0.00012783960071535645,
+      "loss": 2.0682,
+      "step": 1785
+    },
+    {
+      "epoch": 0.7703256415786068,
+      "grad_norm": 0.1817246377468109,
+      "learning_rate": 0.00012781537544396526,
+      "loss": 2.1458,
+      "step": 1786
+    },
+    {
+      "epoch": 0.770756954927755,
+      "grad_norm": 0.16584637761116028,
+      "learning_rate": 0.00012779113923657294,
+      "loss": 2.1422,
+      "step": 1787
+    },
+    {
+      "epoch": 0.7711882682769031,
+      "grad_norm": 0.16539442539215088,
+      "learning_rate": 0.000127766892098198,
+      "loss": 2.1348,
+      "step": 1788
+    },
+    {
+      "epoch": 0.7716195816260514,
+      "grad_norm": 0.16954608261585236,
+      "learning_rate": 0.00012774263403386095,
+      "loss": 2.2259,
+      "step": 1789
+    },
+    {
+      "epoch": 0.7720508949751995,
+      "grad_norm": 0.19718922674655914,
+      "learning_rate": 0.00012771836504858473,
+      "loss": 2.2328,
+      "step": 1790
+    },
+    {
+      "epoch": 0.7724822083243477,
+      "grad_norm": 0.17745161056518555,
+      "learning_rate": 0.00012769408514739453,
+      "loss": 2.2707,
+      "step": 1791
+    },
+    {
+      "epoch": 0.7729135216734958,
+      "grad_norm": 0.1657809466123581,
+      "learning_rate": 0.00012766979433531778,
+      "loss": 2.2097,
+      "step": 1792
+    },
+    {
+      "epoch": 0.773344835022644,
+      "grad_norm": 0.19770976901054382,
+      "learning_rate": 0.00012764549261738407,
+      "loss": 2.2412,
+      "step": 1793
+    },
+    {
+      "epoch": 0.7737761483717921,
+      "grad_norm": 0.1652282029390335,
+      "learning_rate": 0.0001276211799986254,
+      "loss": 2.2458,
+      "step": 1794
+    },
+    {
+      "epoch": 0.7742074617209402,
+      "grad_norm": 0.16925354301929474,
+      "learning_rate": 0.00012759685648407597,
+      "loss": 2.2538,
+      "step": 1795
+    },
+    {
+      "epoch": 0.7746387750700884,
+      "grad_norm": 0.16957129538059235,
+      "learning_rate": 0.00012757252207877223,
+      "loss": 2.275,
+      "step": 1796
+    },
+    {
+      "epoch": 0.7750700884192365,
+      "grad_norm": 0.16301840543746948,
+      "learning_rate": 0.00012754817678775285,
+      "loss": 2.2116,
+      "step": 1797
+    },
+    {
+      "epoch": 0.7755014017683848,
+      "grad_norm": 0.16502845287322998,
+      "learning_rate": 0.0001275238206160588,
+      "loss": 2.3485,
+      "step": 1798
+    },
+    {
+      "epoch": 0.7759327151175329,
+      "grad_norm": 0.16845610737800598,
+      "learning_rate": 0.00012749945356873335,
+      "loss": 2.046,
+      "step": 1799
+    },
+    {
+      "epoch": 0.7763640284666811,
+      "grad_norm": 0.16439995169639587,
+      "learning_rate": 0.00012747507565082184,
+      "loss": 2.1971,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7763640284666811,
+      "eval_loss": 2.110645294189453,
+      "eval_runtime": 201.0521,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7767953418158292,
+      "grad_norm": 0.17457184195518494,
+      "learning_rate": 0.0001274506868673721,
+      "loss": 2.2552,
+      "step": 1801
+    },
+    {
+      "epoch": 0.7772266551649774,
+      "grad_norm": 0.15566860139369965,
+      "learning_rate": 0.00012742628722343407,
+      "loss": 2.248,
+      "step": 1802
+    },
+    {
+      "epoch": 0.7776579685141255,
+      "grad_norm": 0.175991490483284,
+      "learning_rate": 0.0001274018767240599,
+      "loss": 2.2171,
+      "step": 1803
+    },
+    {
+      "epoch": 0.7780892818632736,
+      "grad_norm": 0.17536449432373047,
+      "learning_rate": 0.00012737745537430413,
+      "loss": 2.059,
+      "step": 1804
+    },
+    {
+      "epoch": 0.7785205952124218,
+      "grad_norm": 0.19441308081150055,
+      "learning_rate": 0.00012735302317922343,
+      "loss": 1.9221,
+      "step": 1805
+    },
+    {
+      "epoch": 0.7789519085615699,
+      "grad_norm": 0.17545975744724274,
+      "learning_rate": 0.00012732858014387677,
+      "loss": 2.1557,
+      "step": 1806
+    },
+    {
+      "epoch": 0.7793832219107182,
+      "grad_norm": 0.18923307955265045,
+      "learning_rate": 0.0001273041262733253,
+      "loss": 2.3349,
+      "step": 1807
+    },
+    {
+      "epoch": 0.7798145352598663,
+      "grad_norm": 0.16929149627685547,
+      "learning_rate": 0.00012727966157263253,
+      "loss": 2.2812,
+      "step": 1808
+    },
+    {
+      "epoch": 0.7802458486090145,
+      "grad_norm": 0.16833718121051788,
+      "learning_rate": 0.0001272551860468641,
+      "loss": 2.2105,
+      "step": 1809
+    },
+    {
+      "epoch": 0.7806771619581626,
+      "grad_norm": 0.163452610373497,
+      "learning_rate": 0.00012723069970108794,
+      "loss": 2.2795,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7811084753073108,
+      "grad_norm": 0.16303634643554688,
+      "learning_rate": 0.00012720620254037424,
+      "loss": 2.0641,
+      "step": 1811
+    },
+    {
+      "epoch": 0.7815397886564589,
+      "grad_norm": 0.1870717853307724,
+      "learning_rate": 0.00012718169456979534,
+      "loss": 2.1919,
+      "step": 1812
+    },
+    {
+      "epoch": 0.781971102005607,
+      "grad_norm": 0.1807531863451004,
+      "learning_rate": 0.000127157175794426,
+      "loss": 2.3677,
+      "step": 1813
+    },
+    {
+      "epoch": 0.7824024153547552,
+      "grad_norm": 0.24385060369968414,
+      "learning_rate": 0.00012713264621934297,
+      "loss": 2.2957,
+      "step": 1814
+    },
+    {
+      "epoch": 0.7828337287039033,
+      "grad_norm": 0.16218340396881104,
+      "learning_rate": 0.00012710810584962547,
+      "loss": 2.2153,
+      "step": 1815
+    },
+    {
+      "epoch": 0.7832650420530516,
+      "grad_norm": 0.19149371981620789,
+      "learning_rate": 0.0001270835546903548,
+      "loss": 2.1862,
+      "step": 1816
+    },
+    {
+      "epoch": 0.7836963554021997,
+      "grad_norm": 0.21503080427646637,
+      "learning_rate": 0.00012705899274661453,
+      "loss": 2.3009,
+      "step": 1817
+    },
+    {
+      "epoch": 0.7841276687513479,
+      "grad_norm": 0.16244001686573029,
+      "learning_rate": 0.00012703442002349053,
+      "loss": 2.2839,
+      "step": 1818
+    },
+    {
+      "epoch": 0.784558982100496,
+      "grad_norm": 0.19366413354873657,
+      "learning_rate": 0.0001270098365260708,
+      "loss": 2.1591,
+      "step": 1819
+    },
+    {
+      "epoch": 0.7849902954496442,
+      "grad_norm": 0.16676150262355804,
+      "learning_rate": 0.00012698524225944566,
+      "loss": 2.2976,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7854216087987923,
+      "grad_norm": 0.18334507942199707,
+      "learning_rate": 0.00012696063722870762,
+      "loss": 2.2395,
+      "step": 1821
+    },
+    {
+      "epoch": 0.7858529221479404,
+      "grad_norm": 0.20181263983249664,
+      "learning_rate": 0.0001269360214389514,
+      "loss": 2.3733,
+      "step": 1822
+    },
+    {
+      "epoch": 0.7862842354970886,
+      "grad_norm": 0.1683868169784546,
+      "learning_rate": 0.00012691139489527398,
+      "loss": 2.0121,
+      "step": 1823
+    },
+    {
+      "epoch": 0.7867155488462368,
+      "grad_norm": 0.1778002679347992,
+      "learning_rate": 0.00012688675760277456,
+      "loss": 2.4024,
+      "step": 1824
+    },
+    {
+      "epoch": 0.787146862195385,
+      "grad_norm": 0.173460453748703,
+      "learning_rate": 0.00012686210956655455,
+      "loss": 2.408,
+      "step": 1825
+    },
+    {
+      "epoch": 0.787146862195385,
+      "eval_loss": 2.1109418869018555,
+      "eval_runtime": 201.6399,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1825
+    },
+    {
+      "epoch": 0.7875781755445331,
+      "grad_norm": 0.18741174042224884,
+      "learning_rate": 0.00012683745079171757,
+      "loss": 2.3349,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7880094888936813,
+      "grad_norm": 0.151434987783432,
+      "learning_rate": 0.00012681278128336956,
+      "loss": 2.049,
+      "step": 1827
+    },
+    {
+      "epoch": 0.7884408022428294,
+      "grad_norm": 0.17712783813476562,
+      "learning_rate": 0.00012678810104661858,
+      "loss": 2.1582,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7888721155919776,
+      "grad_norm": 0.18137238919734955,
+      "learning_rate": 0.0001267634100865749,
+      "loss": 2.0288,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7893034289411257,
+      "grad_norm": 0.1671016663312912,
+      "learning_rate": 0.0001267387084083511,
+      "loss": 2.3045,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7897347422902739,
+      "grad_norm": 0.1871432512998581,
+      "learning_rate": 0.0001267139960170619,
+      "loss": 2.115,
+      "step": 1831
+    },
+    {
+      "epoch": 0.790166055639422,
+      "grad_norm": 0.16146980226039886,
+      "learning_rate": 0.0001266892729178243,
+      "loss": 2.2883,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7905973689885702,
+      "grad_norm": 0.18456147611141205,
+      "learning_rate": 0.00012666453911575748,
+      "loss": 2.3554,
+      "step": 1833
+    },
+    {
+      "epoch": 0.7910286823377184,
+      "grad_norm": 0.16101932525634766,
+      "learning_rate": 0.00012663979461598288,
+      "loss": 2.1936,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7914599956868665,
+      "grad_norm": 0.15960389375686646,
+      "learning_rate": 0.00012661503942362404,
+      "loss": 2.0886,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7918913090360147,
+      "grad_norm": 0.1770126223564148,
+      "learning_rate": 0.00012659027354380685,
+      "loss": 2.163,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7923226223851628,
+      "grad_norm": 0.16850055754184723,
+      "learning_rate": 0.0001265654969816593,
+      "loss": 2.3053,
+      "step": 1837
+    },
+    {
+      "epoch": 0.792753935734311,
+      "grad_norm": 0.9188832640647888,
+      "learning_rate": 0.00012654070974231173,
+      "loss": 2.1439,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7931852490834591,
+      "grad_norm": 0.16128169000148773,
+      "learning_rate": 0.00012651591183089657,
+      "loss": 1.889,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7936165624326073,
+      "grad_norm": 0.16812586784362793,
+      "learning_rate": 0.00012649110325254847,
+      "loss": 2.4054,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7940478757817554,
+      "grad_norm": 0.1671290248632431,
+      "learning_rate": 0.00012646628401240438,
+      "loss": 2.2513,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7944791891309037,
+      "grad_norm": 0.17929555475711823,
+      "learning_rate": 0.00012644145411560333,
+      "loss": 2.1672,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7949105024800518,
+      "grad_norm": 0.1603858768939972,
+      "learning_rate": 0.00012641661356728668,
+      "loss": 1.9774,
+      "step": 1843
+    },
+    {
+      "epoch": 0.7953418158291999,
+      "grad_norm": 0.17137469351291656,
+      "learning_rate": 0.0001263917623725979,
+      "loss": 2.0653,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7957731291783481,
+      "grad_norm": 0.17383916676044464,
+      "learning_rate": 0.0001263669005366827,
+      "loss": 2.1493,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7962044425274962,
+      "grad_norm": 0.1622949242591858,
+      "learning_rate": 0.000126342028064689,
+      "loss": 2.2813,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7966357558766444,
+      "grad_norm": 0.3186436891555786,
+      "learning_rate": 0.0001263171449617669,
+      "loss": 2.4686,
+      "step": 1847
+    },
+    {
+      "epoch": 0.7970670692257925,
+      "grad_norm": 0.17199741303920746,
+      "learning_rate": 0.0001262922512330688,
+      "loss": 2.2529,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7974983825749407,
+      "grad_norm": 0.17930395901203156,
+      "learning_rate": 0.0001262673468837491,
+      "loss": 2.6291,
+      "step": 1849
+    },
+    {
+      "epoch": 0.7979296959240888,
+      "grad_norm": 0.176521897315979,
+      "learning_rate": 0.0001262424319189646,
+      "loss": 2.2637,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7979296959240888,
+      "eval_loss": 2.111254930496216,
+      "eval_runtime": 201.5573,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 1850
+    },
+    {
+      "epoch": 0.798361009273237,
+      "grad_norm": 0.196044459939003,
+      "learning_rate": 0.00012621750634387422,
+      "loss": 2.129,
+      "step": 1851
+    },
+    {
+      "epoch": 0.7987923226223852,
+      "grad_norm": 0.2050708383321762,
+      "learning_rate": 0.00012619257016363903,
+      "loss": 2.3564,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7992236359715333,
+      "grad_norm": 0.16014248132705688,
+      "learning_rate": 0.00012616762338342236,
+      "loss": 2.2885,
+      "step": 1853
+    },
+    {
+      "epoch": 0.7996549493206815,
+      "grad_norm": 0.17007140815258026,
+      "learning_rate": 0.00012614266600838967,
+      "loss": 2.1042,
+      "step": 1854
+    },
+    {
+      "epoch": 0.8000862626698296,
+      "grad_norm": 0.17264525592327118,
+      "learning_rate": 0.00012611769804370874,
+      "loss": 2.2499,
+      "step": 1855
+    },
+    {
+      "epoch": 0.8005175760189778,
+      "grad_norm": 0.17515993118286133,
+      "learning_rate": 0.0001260927194945494,
+      "loss": 2.1508,
+      "step": 1856
+    },
+    {
+      "epoch": 0.8009488893681259,
+      "grad_norm": 0.18426083028316498,
+      "learning_rate": 0.00012606773036608374,
+      "loss": 2.2293,
+      "step": 1857
+    },
+    {
+      "epoch": 0.8013802027172741,
+      "grad_norm": 0.1829826682806015,
+      "learning_rate": 0.00012604273066348605,
+      "loss": 2.1917,
+      "step": 1858
+    },
+    {
+      "epoch": 0.8018115160664222,
+      "grad_norm": 0.16429126262664795,
+      "learning_rate": 0.0001260177203919328,
+      "loss": 2.114,
+      "step": 1859
+    },
+    {
+      "epoch": 0.8022428294155705,
+      "grad_norm": 0.3393753468990326,
+      "learning_rate": 0.0001259926995566026,
+      "loss": 2.1461,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8026741427647186,
+      "grad_norm": 0.1876833587884903,
+      "learning_rate": 0.00012596766816267633,
+      "loss": 2.2348,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8031054561138667,
+      "grad_norm": 0.1625937670469284,
+      "learning_rate": 0.00012594262621533697,
+      "loss": 2.4149,
+      "step": 1862
+    },
+    {
+      "epoch": 0.8035367694630149,
+      "grad_norm": 0.17854106426239014,
+      "learning_rate": 0.00012591757371976974,
+      "loss": 2.3269,
+      "step": 1863
+    },
+    {
+      "epoch": 0.803968082812163,
+      "grad_norm": 10.761648178100586,
+      "learning_rate": 0.00012589251068116208,
+      "loss": 2.1866,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8043993961613112,
+      "grad_norm": 0.17221496999263763,
+      "learning_rate": 0.00012586743710470353,
+      "loss": 2.127,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8048307095104593,
+      "grad_norm": 0.1671273559331894,
+      "learning_rate": 0.00012584235299558584,
+      "loss": 2.1688,
+      "step": 1866
+    },
+    {
+      "epoch": 0.8052620228596075,
+      "grad_norm": 0.1608709841966629,
+      "learning_rate": 0.00012581725835900297,
+      "loss": 2.2199,
+      "step": 1867
+    },
+    {
+      "epoch": 0.8056933362087556,
+      "grad_norm": 0.18937575817108154,
+      "learning_rate": 0.000125792153200151,
+      "loss": 1.926,
+      "step": 1868
+    },
+    {
+      "epoch": 0.8061246495579039,
+      "grad_norm": 0.16524960100650787,
+      "learning_rate": 0.0001257670375242283,
+      "loss": 2.2916,
+      "step": 1869
+    },
+    {
+      "epoch": 0.806555962907052,
+      "grad_norm": 0.18175652623176575,
+      "learning_rate": 0.00012574191133643526,
+      "loss": 2.3457,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8069872762562001,
+      "grad_norm": 0.19647067785263062,
+      "learning_rate": 0.00012571677464197462,
+      "loss": 2.1629,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8074185896053483,
+      "grad_norm": 0.2882465124130249,
+      "learning_rate": 0.0001256916274460511,
+      "loss": 2.2529,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8078499029544964,
+      "grad_norm": 0.1850413829088211,
+      "learning_rate": 0.00012566646975387181,
+      "loss": 2.2831,
+      "step": 1873
+    },
+    {
+      "epoch": 0.8082812163036446,
+      "grad_norm": 0.15984345972537994,
+      "learning_rate": 0.00012564130157064588,
+      "loss": 2.1268,
+      "step": 1874
+    },
+    {
+      "epoch": 0.8087125296527927,
+      "grad_norm": 0.1659379005432129,
+      "learning_rate": 0.00012561612290158463,
+      "loss": 2.2775,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8087125296527927,
+      "eval_loss": 2.1113932132720947,
+      "eval_runtime": 207.6287,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8091438430019409,
+      "grad_norm": 0.17246565222740173,
+      "learning_rate": 0.00012559093375190161,
+      "loss": 2.1342,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8095751563510891,
+      "grad_norm": 0.163095623254776,
+      "learning_rate": 0.0001255657341268125,
+      "loss": 2.19,
+      "step": 1877
+    },
+    {
+      "epoch": 0.8100064697002373,
+      "grad_norm": 0.16611529886722565,
+      "learning_rate": 0.00012554052403153517,
+      "loss": 2.2269,
+      "step": 1878
+    },
+    {
+      "epoch": 0.8104377830493854,
+      "grad_norm": 0.1658940315246582,
+      "learning_rate": 0.00012551530347128964,
+      "loss": 2.1327,
+      "step": 1879
+    },
+    {
+      "epoch": 0.8108690963985336,
+      "grad_norm": 0.15693336725234985,
+      "learning_rate": 0.0001254900724512981,
+      "loss": 2.2625,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8113004097476817,
+      "grad_norm": 0.19481512904167175,
+      "learning_rate": 0.0001254648309767849,
+      "loss": 1.8088,
+      "step": 1881
+    },
+    {
+      "epoch": 0.8117317230968298,
+      "grad_norm": 0.1702975630760193,
+      "learning_rate": 0.00012543957905297658,
+      "loss": 2.2031,
+      "step": 1882
+    },
+    {
+      "epoch": 0.812163036445978,
+      "grad_norm": 0.17250478267669678,
+      "learning_rate": 0.0001254143166851018,
+      "loss": 2.2121,
+      "step": 1883
+    },
+    {
+      "epoch": 0.8125943497951261,
+      "grad_norm": 0.17849695682525635,
+      "learning_rate": 0.00012538904387839145,
+      "loss": 2.1903,
+      "step": 1884
+    },
+    {
+      "epoch": 0.8130256631442743,
+      "grad_norm": 0.1863728165626526,
+      "learning_rate": 0.00012536376063807854,
+      "loss": 2.2404,
+      "step": 1885
+    },
+    {
+      "epoch": 0.8134569764934225,
+      "grad_norm": 168.6988067626953,
+      "learning_rate": 0.00012533846696939817,
+      "loss": 2.2611,
+      "step": 1886
+    },
+    {
+      "epoch": 0.8138882898425707,
+      "grad_norm": 0.17046821117401123,
+      "learning_rate": 0.00012531316287758777,
+      "loss": 2.319,
+      "step": 1887
+    },
+    {
+      "epoch": 0.8143196031917188,
+      "grad_norm": 0.1797439604997635,
+      "learning_rate": 0.00012528784836788676,
+      "loss": 2.3515,
+      "step": 1888
+    },
+    {
+      "epoch": 0.814750916540867,
+      "grad_norm": 0.18330460786819458,
+      "learning_rate": 0.00012526252344553681,
+      "loss": 2.1221,
+      "step": 1889
+    },
+    {
+      "epoch": 0.8151822298900151,
+      "grad_norm": 0.18982136249542236,
+      "learning_rate": 0.0001252371881157817,
+      "loss": 2.3772,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8156135432391632,
+      "grad_norm": 0.17775148153305054,
+      "learning_rate": 0.00012521184238386743,
+      "loss": 2.1501,
+      "step": 1891
+    },
+    {
+      "epoch": 0.8160448565883114,
+      "grad_norm": 0.16533853113651276,
+      "learning_rate": 0.00012518648625504207,
+      "loss": 2.0584,
+      "step": 1892
+    },
+    {
+      "epoch": 0.8164761699374595,
+      "grad_norm": 0.17575526237487793,
+      "learning_rate": 0.0001251611197345559,
+      "loss": 2.1697,
+      "step": 1893
+    },
+    {
+      "epoch": 0.8169074832866077,
+      "grad_norm": 0.17966797947883606,
+      "learning_rate": 0.00012513574282766138,
+      "loss": 2.2837,
+      "step": 1894
+    },
+    {
+      "epoch": 0.8173387966357559,
+      "grad_norm": 0.19567516446113586,
+      "learning_rate": 0.000125110355539613,
+      "loss": 2.1884,
+      "step": 1895
+    },
+    {
+      "epoch": 0.8177701099849041,
+      "grad_norm": 0.20565064251422882,
+      "learning_rate": 0.00012508495787566748,
+      "loss": 2.2495,
+      "step": 1896
+    },
+    {
+      "epoch": 0.8182014233340522,
+      "grad_norm": 0.15445753931999207,
+      "learning_rate": 0.00012505954984108375,
+      "loss": 2.1746,
+      "step": 1897
+    },
+    {
+      "epoch": 0.8186327366832004,
+      "grad_norm": 0.17870637774467468,
+      "learning_rate": 0.00012503413144112277,
+      "loss": 2.3098,
+      "step": 1898
+    },
+    {
+      "epoch": 0.8190640500323485,
+      "grad_norm": 0.1896657943725586,
+      "learning_rate": 0.0001250087026810477,
+      "loss": 2.1974,
+      "step": 1899
+    },
+    {
+      "epoch": 0.8194953633814966,
+      "grad_norm": 0.17491237819194794,
+      "learning_rate": 0.00012498326356612385,
+      "loss": 2.2314,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8194953633814966,
+      "eval_loss": 2.111462354660034,
+      "eval_runtime": 199.5209,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 1900
+    },
+    {
+      "epoch": 0.8199266767306448,
+      "grad_norm": 0.1932830959558487,
+      "learning_rate": 0.00012495781410161872,
+      "loss": 2.3258,
+      "step": 1901
+    },
+    {
+      "epoch": 0.8203579900797929,
+      "grad_norm": 0.1864049881696701,
+      "learning_rate": 0.0001249323542928018,
+      "loss": 2.2642,
+      "step": 1902
+    },
+    {
+      "epoch": 0.8207893034289411,
+      "grad_norm": 0.1658972203731537,
+      "learning_rate": 0.0001249068841449449,
+      "loss": 2.0475,
+      "step": 1903
+    },
+    {
+      "epoch": 0.8212206167780893,
+      "grad_norm": 0.16217941045761108,
+      "learning_rate": 0.00012488140366332184,
+      "loss": 2.294,
+      "step": 1904
+    },
+    {
+      "epoch": 0.8216519301272375,
+      "grad_norm": 0.1949874609708786,
+      "learning_rate": 0.00012485591285320868,
+      "loss": 2.2252,
+      "step": 1905
+    },
+    {
+      "epoch": 0.8220832434763856,
+      "grad_norm": 0.2481028139591217,
+      "learning_rate": 0.00012483041171988354,
+      "loss": 2.136,
+      "step": 1906
+    },
+    {
+      "epoch": 0.8225145568255338,
+      "grad_norm": 0.15578563511371613,
+      "learning_rate": 0.00012480490026862672,
+      "loss": 2.0641,
+      "step": 1907
+    },
+    {
+      "epoch": 0.8229458701746819,
+      "grad_norm": 0.17702731490135193,
+      "learning_rate": 0.00012477937850472067,
+      "loss": 2.2588,
+      "step": 1908
+    },
+    {
+      "epoch": 0.82337718352383,
+      "grad_norm": 0.27015769481658936,
+      "learning_rate": 0.00012475384643344988,
+      "loss": 1.9521,
+      "step": 1909
+    },
+    {
+      "epoch": 0.8238084968729782,
+      "grad_norm": 0.18160024285316467,
+      "learning_rate": 0.0001247283040601011,
+      "loss": 2.3358,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8242398102221263,
+      "grad_norm": 0.18464002013206482,
+      "learning_rate": 0.00012470275138996313,
+      "loss": 2.2393,
+      "step": 1911
+    },
+    {
+      "epoch": 0.8246711235712745,
+      "grad_norm": 0.1753547638654709,
+      "learning_rate": 0.00012467718842832696,
+      "loss": 2.2028,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8251024369204227,
+      "grad_norm": 0.9519688487052917,
+      "learning_rate": 0.00012465161518048566,
+      "loss": 2.1816,
+      "step": 1913
+    },
+    {
+      "epoch": 0.8255337502695709,
+      "grad_norm": 0.1543843299150467,
+      "learning_rate": 0.0001246260316517345,
+      "loss": 2.0178,
+      "step": 1914
+    },
+    {
+      "epoch": 0.825965063618719,
+      "grad_norm": 0.1509864628314972,
+      "learning_rate": 0.0001246004378473707,
+      "loss": 2.2426,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8263963769678672,
+      "grad_norm": 0.15210863947868347,
+      "learning_rate": 0.00012457483377269392,
+      "loss": 2.2984,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8268276903170153,
+      "grad_norm": 0.18767836689949036,
+      "learning_rate": 0.0001245492194330056,
+      "loss": 2.2096,
+      "step": 1917
+    },
+    {
+      "epoch": 0.8272590036661635,
+      "grad_norm": 0.1672392636537552,
+      "learning_rate": 0.00012452359483360956,
+      "loss": 2.1577,
+      "step": 1918
+    },
+    {
+      "epoch": 0.8276903170153116,
+      "grad_norm": 0.1950211524963379,
+      "learning_rate": 0.00012449795997981162,
+      "loss": 2.1059,
+      "step": 1919
+    },
+    {
+      "epoch": 0.8281216303644597,
+      "grad_norm": 0.15297383069992065,
+      "learning_rate": 0.00012447231487691974,
+      "loss": 2.1318,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8285529437136079,
+      "grad_norm": 0.16627302765846252,
+      "learning_rate": 0.0001244466595302441,
+      "loss": 2.2125,
+      "step": 1921
+    },
+    {
+      "epoch": 0.8289842570627561,
+      "grad_norm": 0.1622379869222641,
+      "learning_rate": 0.00012442099394509683,
+      "loss": 2.1177,
+      "step": 1922
+    },
+    {
+      "epoch": 0.8294155704119043,
+      "grad_norm": 0.1775934398174286,
+      "learning_rate": 0.0001243953181267923,
+      "loss": 2.1536,
+      "step": 1923
+    },
+    {
+      "epoch": 0.8298468837610524,
+      "grad_norm": 0.20350416004657745,
+      "learning_rate": 0.00012436963208064698,
+      "loss": 2.0901,
+      "step": 1924
+    },
+    {
+      "epoch": 0.8302781971102006,
+      "grad_norm": 0.16868053376674652,
+      "learning_rate": 0.00012434393581197943,
+      "loss": 2.0385,
+      "step": 1925
+    },
+    {
+      "epoch": 0.8302781971102006,
+      "eval_loss": 2.1118338108062744,
+      "eval_runtime": 199.066,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1925
+    },
+    {
+      "epoch": 0.8307095104593487,
+      "grad_norm": 0.1899476796388626,
+      "learning_rate": 0.0001243182293261104,
+      "loss": 2.0417,
+      "step": 1926
+    },
+    {
+      "epoch": 0.8311408238084969,
+      "grad_norm": 0.1900355964899063,
+      "learning_rate": 0.0001242925126283626,
+      "loss": 2.2387,
+      "step": 1927
+    },
+    {
+      "epoch": 0.831572137157645,
+      "grad_norm": 0.16706283390522003,
+      "learning_rate": 0.00012426678572406106,
+      "loss": 2.2561,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8320034505067931,
+      "grad_norm": 0.1549728959798813,
+      "learning_rate": 0.0001242410486185327,
+      "loss": 2.0268,
+      "step": 1929
+    },
+    {
+      "epoch": 0.8324347638559414,
+      "grad_norm": 0.19593092799186707,
+      "learning_rate": 0.00012421530131710677,
+      "loss": 2.1991,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8328660772050895,
+      "grad_norm": 1.2008687257766724,
+      "learning_rate": 0.0001241895438251145,
+      "loss": 2.0635,
+      "step": 1931
+    },
+    {
+      "epoch": 0.8332973905542377,
+      "grad_norm": 0.19961075484752655,
+      "learning_rate": 0.00012416377614788923,
+      "loss": 2.2329,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8337287039033858,
+      "grad_norm": 0.17825818061828613,
+      "learning_rate": 0.00012413799829076648,
+      "loss": 2.2581,
+      "step": 1933
+    },
+    {
+      "epoch": 0.834160017252534,
+      "grad_norm": 0.17735421657562256,
+      "learning_rate": 0.0001241122102590838,
+      "loss": 2.2634,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8345913306016821,
+      "grad_norm": 0.15418070554733276,
+      "learning_rate": 0.00012408641205818093,
+      "loss": 2.2577,
+      "step": 1935
+    },
+    {
+      "epoch": 0.8350226439508303,
+      "grad_norm": 0.17420464754104614,
+      "learning_rate": 0.0001240606036933996,
+      "loss": 2.2465,
+      "step": 1936
+    },
+    {
+      "epoch": 0.8354539572999784,
+      "grad_norm": 0.17787966132164001,
+      "learning_rate": 0.00012403478517008382,
+      "loss": 2.2066,
+      "step": 1937
+    },
+    {
+      "epoch": 0.8358852706491265,
+      "grad_norm": 0.19807268679141998,
+      "learning_rate": 0.00012400895649357949,
+      "loss": 2.1749,
+      "step": 1938
+    },
+    {
+      "epoch": 0.8363165839982748,
+      "grad_norm": 0.17994818091392517,
+      "learning_rate": 0.00012398311766923477,
+      "loss": 2.1722,
+      "step": 1939
+    },
+    {
+      "epoch": 0.8367478973474229,
+      "grad_norm": 0.17680570483207703,
+      "learning_rate": 0.00012395726870239988,
+      "loss": 2.2587,
+      "step": 1940
+    },
+    {
+      "epoch": 0.8371792106965711,
+      "grad_norm": 0.19101424515247345,
+      "learning_rate": 0.00012393140959842712,
+      "loss": 2.1778,
+      "step": 1941
+    },
+    {
+      "epoch": 0.8376105240457192,
+      "grad_norm": 0.17281894385814667,
+      "learning_rate": 0.00012390554036267087,
+      "loss": 2.1408,
+      "step": 1942
+    },
+    {
+      "epoch": 0.8380418373948674,
+      "grad_norm": 0.18398477137088776,
+      "learning_rate": 0.00012387966100048771,
+      "loss": 2.3198,
+      "step": 1943
+    },
+    {
+      "epoch": 0.8384731507440155,
+      "grad_norm": 0.16956676542758942,
+      "learning_rate": 0.00012385377151723616,
+      "loss": 2.3611,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8389044640931637,
+      "grad_norm": 0.17998960614204407,
+      "learning_rate": 0.00012382787191827696,
+      "loss": 2.1598,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8393357774423118,
+      "grad_norm": 0.1674444079399109,
+      "learning_rate": 0.00012380196220897294,
+      "loss": 2.1449,
+      "step": 1946
+    },
+    {
+      "epoch": 0.83976709079146,
+      "grad_norm": 0.1496683955192566,
+      "learning_rate": 0.00012377604239468894,
+      "loss": 2.1855,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8401984041406082,
+      "grad_norm": 0.17284353077411652,
+      "learning_rate": 0.00012375011248079196,
+      "loss": 2.1736,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8406297174897563,
+      "grad_norm": 0.18436162173748016,
+      "learning_rate": 0.00012372417247265107,
+      "loss": 2.3424,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8410610308389045,
+      "grad_norm": 0.19764438271522522,
+      "learning_rate": 0.0001236982223756374,
+      "loss": 2.2667,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8410610308389045,
+      "eval_loss": 2.1127099990844727,
+      "eval_runtime": 199.1108,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8414923441880526,
+      "grad_norm": 0.19147905707359314,
+      "learning_rate": 0.00012367226219512427,
+      "loss": 2.1682,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8419236575372008,
+      "grad_norm": 0.1907227337360382,
+      "learning_rate": 0.000123646291936487,
+      "loss": 2.2614,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8423549708863489,
+      "grad_norm": 0.19571653008460999,
+      "learning_rate": 0.00012362031160510295,
+      "loss": 2.2148,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8427862842354971,
+      "grad_norm": 0.17038114368915558,
+      "learning_rate": 0.0001235943212063517,
+      "loss": 2.1184,
+      "step": 1954
+    },
+    {
+      "epoch": 0.8432175975846452,
+      "grad_norm": 0.1803564429283142,
+      "learning_rate": 0.0001235683207456148,
+      "loss": 2.25,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8436489109337934,
+      "grad_norm": 0.16071362793445587,
+      "learning_rate": 0.000123542310228276,
+      "loss": 2.2377,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8440802242829416,
+      "grad_norm": 0.16572268307209015,
+      "learning_rate": 0.000123516289659721,
+      "loss": 2.4341,
+      "step": 1957
+    },
+    {
+      "epoch": 0.8445115376320897,
+      "grad_norm": 0.19113953411579132,
+      "learning_rate": 0.00012349025904533767,
+      "loss": 2.1922,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8449428509812379,
+      "grad_norm": 0.22091901302337646,
+      "learning_rate": 0.00012346421839051587,
+      "loss": 2.4147,
+      "step": 1959
+    },
+    {
+      "epoch": 0.845374164330386,
+      "grad_norm": 0.18341030180454254,
+      "learning_rate": 0.00012343816770064771,
+      "loss": 2.3162,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8458054776795342,
+      "grad_norm": 0.1765439510345459,
+      "learning_rate": 0.0001234121069811272,
+      "loss": 2.279,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8462367910286823,
+      "grad_norm": 0.29950448870658875,
+      "learning_rate": 0.0001233860362373505,
+      "loss": 2.2917,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8466681043778305,
+      "grad_norm": 0.18259882926940918,
+      "learning_rate": 0.0001233599554747159,
+      "loss": 2.1786,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8470994177269786,
+      "grad_norm": 0.1746501475572586,
+      "learning_rate": 0.00012333386469862362,
+      "loss": 2.308,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8475307310761268,
+      "grad_norm": 0.17510280013084412,
+      "learning_rate": 0.00012330776391447613,
+      "loss": 2.5226,
+      "step": 1965
+    },
+    {
+      "epoch": 0.847962044425275,
+      "grad_norm": 0.16180452704429626,
+      "learning_rate": 0.00012328165312767777,
+      "loss": 2.0962,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8483933577744232,
+      "grad_norm": 0.166132852435112,
+      "learning_rate": 0.0001232555323436352,
+      "loss": 2.2975,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8488246711235713,
+      "grad_norm": 0.17394210398197174,
+      "learning_rate": 0.0001232294015677569,
+      "loss": 2.2925,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8492559844727194,
+      "grad_norm": 0.168202206492424,
+      "learning_rate": 0.0001232032608054536,
+      "loss": 2.0749,
+      "step": 1969
+    },
+    {
+      "epoch": 0.8496872978218676,
+      "grad_norm": 0.17130744457244873,
+      "learning_rate": 0.000123177110062138,
+      "loss": 2.2784,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8501186111710157,
+      "grad_norm": 0.16785354912281036,
+      "learning_rate": 0.0001231509493432249,
+      "loss": 2.2474,
+      "step": 1971
+    },
+    {
+      "epoch": 0.8505499245201639,
+      "grad_norm": 0.18395166099071503,
+      "learning_rate": 0.00012312477865413117,
+      "loss": 2.1545,
+      "step": 1972
+    },
+    {
+      "epoch": 0.850981237869312,
+      "grad_norm": 0.16726191341876984,
+      "learning_rate": 0.00012309859800027574,
+      "loss": 2.1333,
+      "step": 1973
+    },
+    {
+      "epoch": 0.8514125512184602,
+      "grad_norm": 0.17289185523986816,
+      "learning_rate": 0.0001230724073870796,
+      "loss": 2.1889,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8518438645676084,
+      "grad_norm": 0.16984088718891144,
+      "learning_rate": 0.00012304620681996578,
+      "loss": 2.2826,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8518438645676084,
+      "eval_loss": 2.1102747917175293,
+      "eval_runtime": 198.868,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8522751779167566,
+      "grad_norm": 0.17141634225845337,
+      "learning_rate": 0.00012301999630435944,
+      "loss": 2.1729,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8527064912659047,
+      "grad_norm": 0.17726927995681763,
+      "learning_rate": 0.00012299377584568772,
+      "loss": 2.1873,
+      "step": 1977
+    },
+    {
+      "epoch": 0.8531378046150528,
+      "grad_norm": 0.16618146002292633,
+      "learning_rate": 0.00012296754544937984,
+      "loss": 2.1897,
+      "step": 1978
+    },
+    {
+      "epoch": 0.853569117964201,
+      "grad_norm": 0.17402474582195282,
+      "learning_rate": 0.00012294130512086716,
+      "loss": 2.2492,
+      "step": 1979
+    },
+    {
+      "epoch": 0.8540004313133491,
+      "grad_norm": 0.16003389656543732,
+      "learning_rate": 0.00012291505486558291,
+      "loss": 2.232,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8544317446624973,
+      "grad_norm": 0.17695653438568115,
+      "learning_rate": 0.00012288879468896262,
+      "loss": 2.1035,
+      "step": 1981
+    },
+    {
+      "epoch": 0.8548630580116454,
+      "grad_norm": 0.17198093235492706,
+      "learning_rate": 0.00012286252459644365,
+      "loss": 2.1935,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8552943713607937,
+      "grad_norm": 0.1726846843957901,
+      "learning_rate": 0.00012283624459346557,
+      "loss": 2.1505,
+      "step": 1983
+    },
+    {
+      "epoch": 0.8557256847099418,
+      "grad_norm": 0.19706712663173676,
+      "learning_rate": 0.0001228099546854699,
+      "loss": 2.3153,
+      "step": 1984
+    },
+    {
+      "epoch": 0.85615699805909,
+      "grad_norm": 0.16633062064647675,
+      "learning_rate": 0.00012278365487790026,
+      "loss": 2.3088,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8565883114082381,
+      "grad_norm": 0.1573697030544281,
+      "learning_rate": 0.00012275734517620235,
+      "loss": 2.1636,
+      "step": 1986
+    },
+    {
+      "epoch": 0.8570196247573862,
+      "grad_norm": 0.23323068022727966,
+      "learning_rate": 0.00012273102558582385,
+      "loss": 2.0344,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8574509381065344,
+      "grad_norm": 0.22128650546073914,
+      "learning_rate": 0.00012270469611221454,
+      "loss": 2.297,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8578822514556825,
+      "grad_norm": 0.16853579878807068,
+      "learning_rate": 0.00012267835676082616,
+      "loss": 2.2457,
+      "step": 1989
+    },
+    {
+      "epoch": 0.8583135648048307,
+      "grad_norm": 0.16581346094608307,
+      "learning_rate": 0.00012265200753711268,
+      "loss": 2.3109,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8587448781539788,
+      "grad_norm": 0.19087955355644226,
+      "learning_rate": 0.00012262564844652988,
+      "loss": 2.332,
+      "step": 1991
+    },
+    {
+      "epoch": 0.8591761915031271,
+      "grad_norm": 0.19937820732593536,
+      "learning_rate": 0.00012259927949453576,
+      "loss": 2.2953,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8596075048522752,
+      "grad_norm": 0.1905757039785385,
+      "learning_rate": 0.00012257290068659027,
+      "loss": 2.2003,
+      "step": 1993
+    },
+    {
+      "epoch": 0.8600388182014234,
+      "grad_norm": 0.15634717047214508,
+      "learning_rate": 0.00012254651202815543,
+      "loss": 2.1735,
+      "step": 1994
+    },
+    {
+      "epoch": 0.8604701315505715,
+      "grad_norm": 0.16686885058879852,
+      "learning_rate": 0.00012252011352469535,
+      "loss": 2.3553,
+      "step": 1995
+    },
+    {
+      "epoch": 0.8609014448997196,
+      "grad_norm": 0.17841187119483948,
+      "learning_rate": 0.00012249370518167607,
+      "loss": 2.1238,
+      "step": 1996
+    },
+    {
+      "epoch": 0.8613327582488678,
+      "grad_norm": 0.49504896998405457,
+      "learning_rate": 0.00012246728700456578,
+      "loss": 1.7593,
+      "step": 1997
+    },
+    {
+      "epoch": 0.8617640715980159,
+      "grad_norm": 0.17972888052463531,
+      "learning_rate": 0.0001224408589988346,
+      "loss": 2.1113,
+      "step": 1998
+    },
+    {
+      "epoch": 0.8621953849471641,
+      "grad_norm": 0.18801777064800262,
+      "learning_rate": 0.00012241442116995476,
+      "loss": 2.2476,
+      "step": 1999
+    },
+    {
+      "epoch": 0.8626266982963122,
+      "grad_norm": 0.18260492384433746,
+      "learning_rate": 0.00012238797352340052,
+      "loss": 2.34,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8626266982963122,
+      "eval_loss": 2.109143018722534,
+      "eval_runtime": 200.0437,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8630580116454605,
+      "grad_norm": 0.16812659800052643,
+      "learning_rate": 0.0001223615160646481,
+      "loss": 2.2668,
+      "step": 2001
+    },
+    {
+      "epoch": 0.8634893249946086,
+      "grad_norm": 0.17180365324020386,
+      "learning_rate": 0.00012233504879917588,
+      "loss": 2.1786,
+      "step": 2002
+    },
+    {
+      "epoch": 0.8639206383437568,
+      "grad_norm": 0.16643165051937103,
+      "learning_rate": 0.00012230857173246412,
+      "loss": 2.1149,
+      "step": 2003
+    },
+    {
+      "epoch": 0.8643519516929049,
+      "grad_norm": 0.18042191863059998,
+      "learning_rate": 0.00012228208486999526,
+      "loss": 1.8244,
+      "step": 2004
+    },
+    {
+      "epoch": 0.864783265042053,
+      "grad_norm": 0.18224523961544037,
+      "learning_rate": 0.00012225558821725362,
+      "loss": 2.2373,
+      "step": 2005
+    },
+    {
+      "epoch": 0.8652145783912012,
+      "grad_norm": 0.17160971462726593,
+      "learning_rate": 0.0001222290817797257,
+      "loss": 2.1969,
+      "step": 2006
+    },
+    {
+      "epoch": 0.8656458917403493,
+      "grad_norm": 0.15949644148349762,
+      "learning_rate": 0.0001222025655628999,
+      "loss": 2.0821,
+      "step": 2007
+    },
+    {
+      "epoch": 0.8660772050894975,
+      "grad_norm": 0.20599953830242157,
+      "learning_rate": 0.00012217603957226668,
+      "loss": 2.4422,
+      "step": 2008
+    },
+    {
+      "epoch": 0.8665085184386456,
+      "grad_norm": 0.17190849781036377,
+      "learning_rate": 0.0001221495038133186,
+      "loss": 2.2824,
+      "step": 2009
+    },
+    {
+      "epoch": 0.8669398317877939,
+      "grad_norm": 0.17235378921031952,
+      "learning_rate": 0.0001221229582915501,
+      "loss": 2.1513,
+      "step": 2010
+    },
+    {
+      "epoch": 0.867371145136942,
+      "grad_norm": 0.18004624545574188,
+      "learning_rate": 0.00012209640301245775,
+      "loss": 2.3513,
+      "step": 2011
+    },
+    {
+      "epoch": 0.8678024584860902,
+      "grad_norm": 0.16808605194091797,
+      "learning_rate": 0.00012206983798154016,
+      "loss": 2.2083,
+      "step": 2012
+    },
+    {
+      "epoch": 0.8682337718352383,
+      "grad_norm": 0.1684333235025406,
+      "learning_rate": 0.00012204326320429784,
+      "loss": 2.0846,
+      "step": 2013
+    },
+    {
+      "epoch": 0.8686650851843865,
+      "grad_norm": 0.17363044619560242,
+      "learning_rate": 0.00012201667868623343,
+      "loss": 2.2877,
+      "step": 2014
+    },
+    {
+      "epoch": 0.8690963985335346,
+      "grad_norm": 0.15960228443145752,
+      "learning_rate": 0.00012199008443285151,
+      "loss": 2.0895,
+      "step": 2015
+    },
+    {
+      "epoch": 0.8695277118826827,
+      "grad_norm": 0.16592063009738922,
+      "learning_rate": 0.00012196348044965876,
+      "loss": 2.164,
+      "step": 2016
+    },
+    {
+      "epoch": 0.8699590252318309,
+      "grad_norm": 0.18730241060256958,
+      "learning_rate": 0.00012193686674216378,
+      "loss": 2.2974,
+      "step": 2017
+    },
+    {
+      "epoch": 0.870390338580979,
+      "grad_norm": 0.17575466632843018,
+      "learning_rate": 0.00012191024331587727,
+      "loss": 2.0559,
+      "step": 2018
+    },
+    {
+      "epoch": 0.8708216519301273,
+      "grad_norm": 0.1630849540233612,
+      "learning_rate": 0.00012188361017631185,
+      "loss": 2.2679,
+      "step": 2019
+    },
+    {
+      "epoch": 0.8712529652792754,
+      "grad_norm": 0.16414892673492432,
+      "learning_rate": 0.00012185696732898226,
+      "loss": 2.1962,
+      "step": 2020
+    },
+    {
+      "epoch": 0.8716842786284236,
+      "grad_norm": 0.1663442999124527,
+      "learning_rate": 0.00012183031477940516,
+      "loss": 2.2859,
+      "step": 2021
+    },
+    {
+      "epoch": 0.8721155919775717,
+      "grad_norm": 0.16488784551620483,
+      "learning_rate": 0.00012180365253309924,
+      "loss": 2.2866,
+      "step": 2022
+    },
+    {
+      "epoch": 0.8725469053267199,
+      "grad_norm": 0.19378119707107544,
+      "learning_rate": 0.00012177698059558524,
+      "loss": 2.211,
+      "step": 2023
+    },
+    {
+      "epoch": 0.872978218675868,
+      "grad_norm": 0.1780066192150116,
+      "learning_rate": 0.00012175029897238585,
+      "loss": 2.3226,
+      "step": 2024
+    },
+    {
+      "epoch": 0.8734095320250161,
+      "grad_norm": 0.1780560314655304,
+      "learning_rate": 0.00012172360766902583,
+      "loss": 2.2509,
+      "step": 2025
+    },
+    {
+      "epoch": 0.8734095320250161,
+      "eval_loss": 2.1096911430358887,
+      "eval_runtime": 200.3092,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2025
+    },
+    {
+      "epoch": 0.8738408453741643,
+      "grad_norm": 0.17343544960021973,
+      "learning_rate": 0.00012169690669103187,
+      "loss": 2.303,
+      "step": 2026
+    },
+    {
+      "epoch": 0.8742721587233124,
+      "grad_norm": 0.18043680489063263,
+      "learning_rate": 0.00012167019604393272,
+      "loss": 2.2922,
+      "step": 2027
+    },
+    {
+      "epoch": 0.8747034720724607,
+      "grad_norm": 0.17327257990837097,
+      "learning_rate": 0.00012164347573325909,
+      "loss": 2.1745,
+      "step": 2028
+    },
+    {
+      "epoch": 0.8751347854216088,
+      "grad_norm": 0.19838827848434448,
+      "learning_rate": 0.00012161674576454372,
+      "loss": 2.1947,
+      "step": 2029
+    },
+    {
+      "epoch": 0.875566098770757,
+      "grad_norm": 0.1931244134902954,
+      "learning_rate": 0.00012159000614332135,
+      "loss": 2.2768,
+      "step": 2030
+    },
+    {
+      "epoch": 0.8759974121199051,
+      "grad_norm": 0.20393235981464386,
+      "learning_rate": 0.00012156325687512871,
+      "loss": 2.233,
+      "step": 2031
+    },
+    {
+      "epoch": 0.8764287254690533,
+      "grad_norm": 0.18255698680877686,
+      "learning_rate": 0.00012153649796550453,
+      "loss": 2.0384,
+      "step": 2032
+    },
+    {
+      "epoch": 0.8768600388182014,
+      "grad_norm": 0.17171481251716614,
+      "learning_rate": 0.00012150972941998953,
+      "loss": 2.1993,
+      "step": 2033
+    },
+    {
+      "epoch": 0.8772913521673495,
+      "grad_norm": 0.1640217900276184,
+      "learning_rate": 0.00012148295124412645,
+      "loss": 2.0965,
+      "step": 2034
+    },
+    {
+      "epoch": 0.8777226655164977,
+      "grad_norm": 0.17884527146816254,
+      "learning_rate": 0.00012145616344345996,
+      "loss": 2.365,
+      "step": 2035
+    },
+    {
+      "epoch": 0.878153978865646,
+      "grad_norm": 0.15875819325447083,
+      "learning_rate": 0.00012142936602353679,
+      "loss": 2.0765,
+      "step": 2036
+    },
+    {
+      "epoch": 0.8785852922147941,
+      "grad_norm": 0.17063796520233154,
+      "learning_rate": 0.00012140255898990565,
+      "loss": 2.2196,
+      "step": 2037
+    },
+    {
+      "epoch": 0.8790166055639422,
+      "grad_norm": 0.1654708832502365,
+      "learning_rate": 0.00012137574234811722,
+      "loss": 2.1843,
+      "step": 2038
+    },
+    {
+      "epoch": 0.8794479189130904,
+      "grad_norm": 0.15875574946403503,
+      "learning_rate": 0.0001213489161037242,
+      "loss": 2.2028,
+      "step": 2039
+    },
+    {
+      "epoch": 0.8798792322622385,
+      "grad_norm": 0.16864518821239471,
+      "learning_rate": 0.00012132208026228122,
+      "loss": 2.1658,
+      "step": 2040
+    },
+    {
+      "epoch": 0.8803105456113867,
+      "grad_norm": 0.18385015428066254,
+      "learning_rate": 0.00012129523482934496,
+      "loss": 2.2669,
+      "step": 2041
+    },
+    {
+      "epoch": 0.8807418589605348,
+      "grad_norm": 0.17301145195960999,
+      "learning_rate": 0.00012126837981047405,
+      "loss": 2.1845,
+      "step": 2042
+    },
+    {
+      "epoch": 0.881173172309683,
+      "grad_norm": 0.17012512683868408,
+      "learning_rate": 0.00012124151521122911,
+      "loss": 2.2641,
+      "step": 2043
+    },
+    {
+      "epoch": 0.8816044856588311,
+      "grad_norm": 0.1657698154449463,
+      "learning_rate": 0.00012121464103717276,
+      "loss": 2.2242,
+      "step": 2044
+    },
+    {
+      "epoch": 0.8820357990079793,
+      "grad_norm": 0.19457809627056122,
+      "learning_rate": 0.0001211877572938696,
+      "loss": 1.8951,
+      "step": 2045
+    },
+    {
+      "epoch": 0.8824671123571275,
+      "grad_norm": 0.1696557104587555,
+      "learning_rate": 0.00012116086398688617,
+      "loss": 2.1189,
+      "step": 2046
+    },
+    {
+      "epoch": 0.8828984257062756,
+      "grad_norm": 0.19344045221805573,
+      "learning_rate": 0.00012113396112179105,
+      "loss": 2.0996,
+      "step": 2047
+    },
+    {
+      "epoch": 0.8833297390554238,
+      "grad_norm": 0.1660347431898117,
+      "learning_rate": 0.0001211070487041548,
+      "loss": 2.2841,
+      "step": 2048
+    },
+    {
+      "epoch": 0.8837610524045719,
+      "grad_norm": 0.18354149162769318,
+      "learning_rate": 0.00012108012673954987,
+      "loss": 2.1996,
+      "step": 2049
+    },
+    {
+      "epoch": 0.8841923657537201,
+      "grad_norm": 0.19460587203502655,
+      "learning_rate": 0.00012105319523355076,
+      "loss": 2.3914,
+      "step": 2050
+    },
+    {
+      "epoch": 0.8841923657537201,
+      "eval_loss": 2.1087186336517334,
+      "eval_runtime": 199.8968,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2050
+    },
+    {
+      "epoch": 0.8846236791028682,
+      "grad_norm": 0.1793028861284256,
+      "learning_rate": 0.00012102625419173396,
+      "loss": 2.1765,
+      "step": 2051
+    },
+    {
+      "epoch": 0.8850549924520164,
+      "grad_norm": 0.17265355587005615,
+      "learning_rate": 0.00012099930361967791,
+      "loss": 2.2571,
+      "step": 2052
+    },
+    {
+      "epoch": 0.8854863058011645,
+      "grad_norm": 0.1725883036851883,
+      "learning_rate": 0.00012097234352296297,
+      "loss": 2.324,
+      "step": 2053
+    },
+    {
+      "epoch": 0.8859176191503128,
+      "grad_norm": 0.19926157593727112,
+      "learning_rate": 0.00012094537390717161,
+      "loss": 2.14,
+      "step": 2054
+    },
+    {
+      "epoch": 0.8863489324994609,
+      "grad_norm": 0.17119097709655762,
+      "learning_rate": 0.0001209183947778881,
+      "loss": 2.2903,
+      "step": 2055
+    },
+    {
+      "epoch": 0.886780245848609,
+      "grad_norm": 0.1708480715751648,
+      "learning_rate": 0.00012089140614069881,
+      "loss": 2.2251,
+      "step": 2056
+    },
+    {
+      "epoch": 0.8872115591977572,
+      "grad_norm": 0.17137430608272552,
+      "learning_rate": 0.00012086440800119201,
+      "loss": 2.2214,
+      "step": 2057
+    },
+    {
+      "epoch": 0.8876428725469053,
+      "grad_norm": 0.16498853266239166,
+      "learning_rate": 0.00012083740036495796,
+      "loss": 1.9813,
+      "step": 2058
+    },
+    {
+      "epoch": 0.8880741858960535,
+      "grad_norm": 0.15586769580841064,
+      "learning_rate": 0.0001208103832375889,
+      "loss": 2.2625,
+      "step": 2059
+    },
+    {
+      "epoch": 0.8885054992452016,
+      "grad_norm": 0.19154292345046997,
+      "learning_rate": 0.00012078335662467903,
+      "loss": 2.235,
+      "step": 2060
+    },
+    {
+      "epoch": 0.8889368125943498,
+      "grad_norm": 0.1716182976961136,
+      "learning_rate": 0.00012075632053182447,
+      "loss": 2.1366,
+      "step": 2061
+    },
+    {
+      "epoch": 0.8893681259434979,
+      "grad_norm": 0.16451750695705414,
+      "learning_rate": 0.00012072927496462338,
+      "loss": 2.3751,
+      "step": 2062
+    },
+    {
+      "epoch": 0.8897994392926462,
+      "grad_norm": 0.16332577168941498,
+      "learning_rate": 0.00012070221992867581,
+      "loss": 2.1399,
+      "step": 2063
+    },
+    {
+      "epoch": 0.8902307526417943,
+      "grad_norm": 0.18729786574840546,
+      "learning_rate": 0.00012067515542958383,
+      "loss": 2.3277,
+      "step": 2064
+    },
+    {
+      "epoch": 0.8906620659909424,
+      "grad_norm": 0.15769799053668976,
+      "learning_rate": 0.00012064808147295142,
+      "loss": 2.2074,
+      "step": 2065
+    },
+    {
+      "epoch": 0.8910933793400906,
+      "grad_norm": 0.17490021884441376,
+      "learning_rate": 0.00012062099806438454,
+      "loss": 2.0525,
+      "step": 2066
+    },
+    {
+      "epoch": 0.8915246926892387,
+      "grad_norm": 0.16716617345809937,
+      "learning_rate": 0.00012059390520949108,
+      "loss": 2.2348,
+      "step": 2067
+    },
+    {
+      "epoch": 0.8919560060383869,
+      "grad_norm": 0.15750347077846527,
+      "learning_rate": 0.00012056680291388097,
+      "loss": 2.3384,
+      "step": 2068
+    },
+    {
+      "epoch": 0.892387319387535,
+      "grad_norm": 0.16338370740413666,
+      "learning_rate": 0.00012053969118316599,
+      "loss": 2.3072,
+      "step": 2069
+    },
+    {
+      "epoch": 0.8928186327366832,
+      "grad_norm": 0.17863647639751434,
+      "learning_rate": 0.00012051257002295996,
+      "loss": 2.2572,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8932499460858313,
+      "grad_norm": 0.17093229293823242,
+      "learning_rate": 0.00012048543943887858,
+      "loss": 2.0799,
+      "step": 2071
+    },
+    {
+      "epoch": 0.8936812594349796,
+      "grad_norm": 0.16067644953727722,
+      "learning_rate": 0.00012045829943653953,
+      "loss": 2.1233,
+      "step": 2072
+    },
+    {
+      "epoch": 0.8941125727841277,
+      "grad_norm": 0.1740393340587616,
+      "learning_rate": 0.00012043115002156247,
+      "loss": 1.9874,
+      "step": 2073
+    },
+    {
+      "epoch": 0.8945438861332758,
+      "grad_norm": 0.223533496260643,
+      "learning_rate": 0.00012040399119956896,
+      "loss": 2.2838,
+      "step": 2074
+    },
+    {
+      "epoch": 0.894975199482424,
+      "grad_norm": 0.17270897328853607,
+      "learning_rate": 0.00012037682297618256,
+      "loss": 2.1209,
+      "step": 2075
+    },
+    {
+      "epoch": 0.894975199482424,
+      "eval_loss": 2.1082229614257812,
+      "eval_runtime": 205.4679,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 2075
+    },
+    {
+      "epoch": 0.8954065128315721,
+      "grad_norm": 0.177972212433815,
+      "learning_rate": 0.00012034964535702873,
+      "loss": 2.1077,
+      "step": 2076
+    },
+    {
+      "epoch": 0.8958378261807203,
+      "grad_norm": 0.16358588635921478,
+      "learning_rate": 0.00012032245834773488,
+      "loss": 2.1419,
+      "step": 2077
+    },
+    {
+      "epoch": 0.8962691395298684,
+      "grad_norm": 0.17186710238456726,
+      "learning_rate": 0.00012029526195393039,
+      "loss": 2.3137,
+      "step": 2078
+    },
+    {
+      "epoch": 0.8967004528790166,
+      "grad_norm": 0.19588297605514526,
+      "learning_rate": 0.00012026805618124658,
+      "loss": 2.3633,
+      "step": 2079
+    },
+    {
+      "epoch": 0.8971317662281647,
+      "grad_norm": 0.18950992822647095,
+      "learning_rate": 0.0001202408410353167,
+      "loss": 2.2519,
+      "step": 2080
+    },
+    {
+      "epoch": 0.897563079577313,
+      "grad_norm": 0.14333821833133698,
+      "learning_rate": 0.00012021361652177593,
+      "loss": 2.2314,
+      "step": 2081
+    },
+    {
+      "epoch": 0.8979943929264611,
+      "grad_norm": 0.16006465256214142,
+      "learning_rate": 0.00012018638264626143,
+      "loss": 2.309,
+      "step": 2082
+    },
+    {
+      "epoch": 0.8984257062756092,
+      "grad_norm": 0.1798335760831833,
+      "learning_rate": 0.00012015913941441222,
+      "loss": 2.0984,
+      "step": 2083
+    },
+    {
+      "epoch": 0.8988570196247574,
+      "grad_norm": 0.17910893261432648,
+      "learning_rate": 0.00012013188683186934,
+      "loss": 2.1577,
+      "step": 2084
+    },
+    {
+      "epoch": 0.8992883329739055,
+      "grad_norm": 0.16627615690231323,
+      "learning_rate": 0.00012010462490427577,
+      "loss": 2.1262,
+      "step": 2085
+    },
+    {
+      "epoch": 0.8997196463230537,
+      "grad_norm": 0.18889717757701874,
+      "learning_rate": 0.00012007735363727633,
+      "loss": 2.3386,
+      "step": 2086
+    },
+    {
+      "epoch": 0.9001509596722018,
+      "grad_norm": 0.15616723895072937,
+      "learning_rate": 0.00012005007303651789,
+      "loss": 2.2993,
+      "step": 2087
+    },
+    {
+      "epoch": 0.90058227302135,
+      "grad_norm": 0.1719571053981781,
+      "learning_rate": 0.00012002278310764913,
+      "loss": 2.5332,
+      "step": 2088
+    },
+    {
+      "epoch": 0.9010135863704982,
+      "grad_norm": 0.18990328907966614,
+      "learning_rate": 0.0001199954838563208,
+      "loss": 2.4486,
+      "step": 2089
+    },
+    {
+      "epoch": 0.9014448997196464,
+      "grad_norm": 0.16839830577373505,
+      "learning_rate": 0.00011996817528818547,
+      "loss": 2.0224,
+      "step": 2090
+    },
+    {
+      "epoch": 0.9018762130687945,
+      "grad_norm": 0.15824031829833984,
+      "learning_rate": 0.00011994085740889769,
+      "loss": 2.2087,
+      "step": 2091
+    },
+    {
+      "epoch": 0.9023075264179427,
+      "grad_norm": 0.169529989361763,
+      "learning_rate": 0.00011991353022411388,
+      "loss": 2.1402,
+      "step": 2092
+    },
+    {
+      "epoch": 0.9027388397670908,
+      "grad_norm": 0.16205726563930511,
+      "learning_rate": 0.0001198861937394925,
+      "loss": 2.3329,
+      "step": 2093
+    },
+    {
+      "epoch": 0.9031701531162389,
+      "grad_norm": 0.21548905968666077,
+      "learning_rate": 0.00011985884796069387,
+      "loss": 2.3336,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9036014664653871,
+      "grad_norm": 0.1582777351140976,
+      "learning_rate": 0.00011983149289338015,
+      "loss": 2.2334,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9040327798145352,
+      "grad_norm": 0.17056939005851746,
+      "learning_rate": 0.00011980412854321559,
+      "loss": 2.2711,
+      "step": 2096
+    },
+    {
+      "epoch": 0.9044640931636834,
+      "grad_norm": 0.17518271505832672,
+      "learning_rate": 0.00011977675491586621,
+      "loss": 2.0697,
+      "step": 2097
+    },
+    {
+      "epoch": 0.9048954065128316,
+      "grad_norm": 0.16619044542312622,
+      "learning_rate": 0.00011974937201700009,
+      "loss": 2.3312,
+      "step": 2098
+    },
+    {
+      "epoch": 0.9053267198619798,
+      "grad_norm": 0.18111206591129303,
+      "learning_rate": 0.0001197219798522871,
+      "loss": 2.1856,
+      "step": 2099
+    },
+    {
+      "epoch": 0.9057580332111279,
+      "grad_norm": 0.1699441820383072,
+      "learning_rate": 0.00011969457842739911,
+      "loss": 2.2673,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9057580332111279,
+      "eval_loss": 2.1081807613372803,
+      "eval_runtime": 201.1293,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2100
+    },
+    {
+      "epoch": 0.906189346560276,
+      "grad_norm": 0.1888381838798523,
+      "learning_rate": 0.00011966716774800989,
+      "loss": 2.422,
+      "step": 2101
+    },
+    {
+      "epoch": 0.9066206599094242,
+      "grad_norm": 0.186203271150589,
+      "learning_rate": 0.00011963974781979511,
+      "loss": 2.0778,
+      "step": 2102
+    },
+    {
+      "epoch": 0.9070519732585723,
+      "grad_norm": 0.17752569913864136,
+      "learning_rate": 0.00011961231864843241,
+      "loss": 2.3665,
+      "step": 2103
+    },
+    {
+      "epoch": 0.9074832866077205,
+      "grad_norm": 0.1868693083524704,
+      "learning_rate": 0.00011958488023960123,
+      "loss": 2.0341,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9079145999568686,
+      "grad_norm": 0.17515556514263153,
+      "learning_rate": 0.00011955743259898302,
+      "loss": 2.1787,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9083459133060168,
+      "grad_norm": 0.15530408918857574,
+      "learning_rate": 0.00011952997573226113,
+      "loss": 2.1034,
+      "step": 2106
+    },
+    {
+      "epoch": 0.908777226655165,
+      "grad_norm": 0.1727735847234726,
+      "learning_rate": 0.0001195025096451208,
+      "loss": 2.2956,
+      "step": 2107
+    },
+    {
+      "epoch": 0.9092085400043132,
+      "grad_norm": 0.1671399176120758,
+      "learning_rate": 0.00011947503434324917,
+      "loss": 2.3519,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9096398533534613,
+      "grad_norm": 0.16980357468128204,
+      "learning_rate": 0.0001194475498323353,
+      "loss": 2.1494,
+      "step": 2109
+    },
+    {
+      "epoch": 0.9100711667026095,
+      "grad_norm": 0.16269958019256592,
+      "learning_rate": 0.00011942005611807017,
+      "loss": 2.1425,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9105024800517576,
+      "grad_norm": 0.20038245618343353,
+      "learning_rate": 0.00011939255320614663,
+      "loss": 2.3136,
+      "step": 2111
+    },
+    {
+      "epoch": 0.9109337934009057,
+      "grad_norm": 0.1741410195827484,
+      "learning_rate": 0.00011936504110225953,
+      "loss": 2.2522,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9113651067500539,
+      "grad_norm": 0.17310325801372528,
+      "learning_rate": 0.00011933751981210548,
+      "loss": 2.1328,
+      "step": 2113
+    },
+    {
+      "epoch": 0.911796420099202,
+      "grad_norm": 0.19136282801628113,
+      "learning_rate": 0.00011930998934138306,
+      "loss": 2.2303,
+      "step": 2114
+    },
+    {
+      "epoch": 0.9122277334483502,
+      "grad_norm": 0.18183456361293793,
+      "learning_rate": 0.00011928244969579281,
+      "loss": 2.3831,
+      "step": 2115
+    },
+    {
+      "epoch": 0.9126590467974984,
+      "grad_norm": 0.1687455177307129,
+      "learning_rate": 0.00011925490088103708,
+      "loss": 2.2036,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9130903601466466,
+      "grad_norm": 0.16309991478919983,
+      "learning_rate": 0.00011922734290282017,
+      "loss": 2.1591,
+      "step": 2117
+    },
+    {
+      "epoch": 0.9135216734957947,
+      "grad_norm": 0.16328085958957672,
+      "learning_rate": 0.00011919977576684822,
+      "loss": 2.2273,
+      "step": 2118
+    },
+    {
+      "epoch": 0.9139529868449429,
+      "grad_norm": 0.21833649277687073,
+      "learning_rate": 0.00011917219947882936,
+      "loss": 2.3208,
+      "step": 2119
+    },
+    {
+      "epoch": 0.914384300194091,
+      "grad_norm": 0.19529038667678833,
+      "learning_rate": 0.00011914461404447352,
+      "loss": 2.2636,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9148156135432391,
+      "grad_norm": 0.16871854662895203,
+      "learning_rate": 0.00011911701946949259,
+      "loss": 2.2352,
+      "step": 2121
+    },
+    {
+      "epoch": 0.9152469268923873,
+      "grad_norm": 0.19284936785697937,
+      "learning_rate": 0.00011908941575960034,
+      "loss": 2.3068,
+      "step": 2122
+    },
+    {
+      "epoch": 0.9156782402415354,
+      "grad_norm": 0.1705864518880844,
+      "learning_rate": 0.0001190618029205124,
+      "loss": 2.1109,
+      "step": 2123
+    },
+    {
+      "epoch": 0.9161095535906836,
+      "grad_norm": 0.19411548972129822,
+      "learning_rate": 0.00011903418095794631,
+      "loss": 2.3403,
+      "step": 2124
+    },
+    {
+      "epoch": 0.9165408669398318,
+      "grad_norm": 0.16737037897109985,
+      "learning_rate": 0.00011900654987762148,
+      "loss": 2.3417,
+      "step": 2125
+    },
+    {
+      "epoch": 0.9165408669398318,
+      "eval_loss": 2.1069459915161133,
+      "eval_runtime": 201.2256,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2125
+    },
+    {
+      "epoch": 0.91697218028898,
+      "grad_norm": 0.1877131313085556,
+      "learning_rate": 0.00011897890968525929,
+      "loss": 2.2791,
+      "step": 2126
+    },
+    {
+      "epoch": 0.9174034936381281,
+      "grad_norm": 0.17505024373531342,
+      "learning_rate": 0.00011895126038658288,
+      "loss": 2.2123,
+      "step": 2127
+    },
+    {
+      "epoch": 0.9178348069872763,
+      "grad_norm": 0.1770326942205429,
+      "learning_rate": 0.00011892360198731738,
+      "loss": 2.3355,
+      "step": 2128
+    },
+    {
+      "epoch": 0.9182661203364244,
+      "grad_norm": 0.1650460958480835,
+      "learning_rate": 0.00011889593449318973,
+      "loss": 2.2143,
+      "step": 2129
+    },
+    {
+      "epoch": 0.9186974336855726,
+      "grad_norm": 0.18563257157802582,
+      "learning_rate": 0.00011886825790992882,
+      "loss": 2.1454,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9191287470347207,
+      "grad_norm": 0.16880924999713898,
+      "learning_rate": 0.00011884057224326537,
+      "loss": 2.1986,
+      "step": 2131
+    },
+    {
+      "epoch": 0.9195600603838688,
+      "grad_norm": 0.15574918687343597,
+      "learning_rate": 0.000118812877498932,
+      "loss": 2.3588,
+      "step": 2132
+    },
+    {
+      "epoch": 0.919991373733017,
+      "grad_norm": 0.18395072221755981,
+      "learning_rate": 0.00011878517368266322,
+      "loss": 2.1824,
+      "step": 2133
+    },
+    {
+      "epoch": 0.9204226870821652,
+      "grad_norm": 0.17193692922592163,
+      "learning_rate": 0.00011875746080019538,
+      "loss": 2.2512,
+      "step": 2134
+    },
+    {
+      "epoch": 0.9208540004313134,
+      "grad_norm": 0.16654035449028015,
+      "learning_rate": 0.0001187297388572668,
+      "loss": 2.1752,
+      "step": 2135
+    },
+    {
+      "epoch": 0.9212853137804615,
+      "grad_norm": 0.14945533871650696,
+      "learning_rate": 0.00011870200785961754,
+      "loss": 2.1153,
+      "step": 2136
+    },
+    {
+      "epoch": 0.9217166271296097,
+      "grad_norm": 0.174434095621109,
+      "learning_rate": 0.00011867426781298962,
+      "loss": 2.1908,
+      "step": 2137
+    },
+    {
+      "epoch": 0.9221479404787578,
+      "grad_norm": 0.1762343943119049,
+      "learning_rate": 0.00011864651872312697,
+      "loss": 2.482,
+      "step": 2138
+    },
+    {
+      "epoch": 0.922579253827906,
+      "grad_norm": 0.2071942687034607,
+      "learning_rate": 0.0001186187605957753,
+      "loss": 2.175,
+      "step": 2139
+    },
+    {
+      "epoch": 0.9230105671770541,
+      "grad_norm": 0.16858017444610596,
+      "learning_rate": 0.00011859099343668223,
+      "loss": 2.136,
+      "step": 2140
+    },
+    {
+      "epoch": 0.9234418805262022,
+      "grad_norm": 0.17030109465122223,
+      "learning_rate": 0.00011856321725159727,
+      "loss": 2.2699,
+      "step": 2141
+    },
+    {
+      "epoch": 0.9238731938753505,
+      "grad_norm": 0.183243066072464,
+      "learning_rate": 0.00011853543204627178,
+      "loss": 2.135,
+      "step": 2142
+    },
+    {
+      "epoch": 0.9243045072244986,
+      "grad_norm": 0.18901191651821136,
+      "learning_rate": 0.00011850763782645899,
+      "loss": 2.1073,
+      "step": 2143
+    },
+    {
+      "epoch": 0.9247358205736468,
+      "grad_norm": 0.2096487283706665,
+      "learning_rate": 0.000118479834597914,
+      "loss": 2.363,
+      "step": 2144
+    },
+    {
+      "epoch": 0.9251671339227949,
+      "grad_norm": 0.16720110177993774,
+      "learning_rate": 0.00011845202236639378,
+      "loss": 2.0968,
+      "step": 2145
+    },
+    {
+      "epoch": 0.9255984472719431,
+      "grad_norm": 0.1674879938364029,
+      "learning_rate": 0.00011842420113765712,
+      "loss": 2.2427,
+      "step": 2146
+    },
+    {
+      "epoch": 0.9260297606210912,
+      "grad_norm": 0.18135453760623932,
+      "learning_rate": 0.00011839637091746476,
+      "loss": 2.2825,
+      "step": 2147
+    },
+    {
+      "epoch": 0.9264610739702394,
+      "grad_norm": 0.21567225456237793,
+      "learning_rate": 0.0001183685317115792,
+      "loss": 2.2642,
+      "step": 2148
+    },
+    {
+      "epoch": 0.9268923873193875,
+      "grad_norm": 0.18386003375053406,
+      "learning_rate": 0.0001183406835257649,
+      "loss": 2.1699,
+      "step": 2149
+    },
+    {
+      "epoch": 0.9273237006685356,
+      "grad_norm": 0.17996805906295776,
+      "learning_rate": 0.0001183128263657881,
+      "loss": 2.1956,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9273237006685356,
+      "eval_loss": 2.106332302093506,
+      "eval_runtime": 201.7299,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9277550140176839,
+      "grad_norm": 0.1627648025751114,
+      "learning_rate": 0.00011828496023741696,
+      "loss": 2.0021,
+      "step": 2151
+    },
+    {
+      "epoch": 0.928186327366832,
+      "grad_norm": 0.17289264500141144,
+      "learning_rate": 0.00011825708514642141,
+      "loss": 2.0195,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9286176407159802,
+      "grad_norm": 0.1960097998380661,
+      "learning_rate": 0.00011822920109857335,
+      "loss": 2.1659,
+      "step": 2153
+    },
+    {
+      "epoch": 0.9290489540651283,
+      "grad_norm": 0.16027681529521942,
+      "learning_rate": 0.00011820130809964646,
+      "loss": 2.2256,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9294802674142765,
+      "grad_norm": 0.18063746392726898,
+      "learning_rate": 0.00011817340615541629,
+      "loss": 2.1708,
+      "step": 2155
+    },
+    {
+      "epoch": 0.9299115807634246,
+      "grad_norm": 0.1574932336807251,
+      "learning_rate": 0.00011814549527166022,
+      "loss": 1.9852,
+      "step": 2156
+    },
+    {
+      "epoch": 0.9303428941125728,
+      "grad_norm": 0.16979452967643738,
+      "learning_rate": 0.00011811757545415755,
+      "loss": 2.1747,
+      "step": 2157
+    },
+    {
+      "epoch": 0.9307742074617209,
+      "grad_norm": 0.16161035001277924,
+      "learning_rate": 0.00011808964670868934,
+      "loss": 2.2419,
+      "step": 2158
+    },
+    {
+      "epoch": 0.931205520810869,
+      "grad_norm": 0.17380760610103607,
+      "learning_rate": 0.00011806170904103855,
+      "loss": 2.0742,
+      "step": 2159
+    },
+    {
+      "epoch": 0.9316368341600173,
+      "grad_norm": 0.1760004758834839,
+      "learning_rate": 0.00011803376245699,
+      "loss": 2.0483,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9320681475091654,
+      "grad_norm": 0.1811494380235672,
+      "learning_rate": 0.00011800580696233032,
+      "loss": 1.969,
+      "step": 2161
+    },
+    {
+      "epoch": 0.9324994608583136,
+      "grad_norm": 0.1914118230342865,
+      "learning_rate": 0.000117977842562848,
+      "loss": 2.1789,
+      "step": 2162
+    },
+    {
+      "epoch": 0.9329307742074617,
+      "grad_norm": 0.17660760879516602,
+      "learning_rate": 0.0001179498692643334,
+      "loss": 1.9431,
+      "step": 2163
+    },
+    {
+      "epoch": 0.9333620875566099,
+      "grad_norm": 0.20114588737487793,
+      "learning_rate": 0.00011792188707257865,
+      "loss": 2.3242,
+      "step": 2164
+    },
+    {
+      "epoch": 0.933793400905758,
+      "grad_norm": 0.18631604313850403,
+      "learning_rate": 0.0001178938959933778,
+      "loss": 2.2464,
+      "step": 2165
+    },
+    {
+      "epoch": 0.9342247142549062,
+      "grad_norm": 0.18342387676239014,
+      "learning_rate": 0.00011786589603252672,
+      "loss": 2.148,
+      "step": 2166
+    },
+    {
+      "epoch": 0.9346560276040543,
+      "grad_norm": 0.18505950272083282,
+      "learning_rate": 0.00011783788719582309,
+      "loss": 2.1848,
+      "step": 2167
+    },
+    {
+      "epoch": 0.9350873409532025,
+      "grad_norm": 0.16741490364074707,
+      "learning_rate": 0.00011780986948906645,
+      "loss": 2.3077,
+      "step": 2168
+    },
+    {
+      "epoch": 0.9355186543023507,
+      "grad_norm": 0.19790899753570557,
+      "learning_rate": 0.00011778184291805816,
+      "loss": 2.2925,
+      "step": 2169
+    },
+    {
+      "epoch": 0.9359499676514988,
+      "grad_norm": 0.18894536793231964,
+      "learning_rate": 0.00011775380748860145,
+      "loss": 2.3668,
+      "step": 2170
+    },
+    {
+      "epoch": 0.936381281000647,
+      "grad_norm": 0.19968891143798828,
+      "learning_rate": 0.00011772576320650135,
+      "loss": 2.2531,
+      "step": 2171
+    },
+    {
+      "epoch": 0.9368125943497951,
+      "grad_norm": 0.18062502145767212,
+      "learning_rate": 0.00011769771007756472,
+      "loss": 2.1539,
+      "step": 2172
+    },
+    {
+      "epoch": 0.9372439076989433,
+      "grad_norm": 0.1887185275554657,
+      "learning_rate": 0.0001176696481076003,
+      "loss": 2.3336,
+      "step": 2173
+    },
+    {
+      "epoch": 0.9376752210480914,
+      "grad_norm": 0.1643417924642563,
+      "learning_rate": 0.00011764157730241863,
+      "loss": 2.2023,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9381065343972396,
+      "grad_norm": 0.17743735015392303,
+      "learning_rate": 0.00011761349766783202,
+      "loss": 2.3152,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9381065343972396,
+      "eval_loss": 2.1060848236083984,
+      "eval_runtime": 202.8585,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9385378477463877,
+      "grad_norm": 0.15262340009212494,
+      "learning_rate": 0.00011758540920965472,
+      "loss": 2.1931,
+      "step": 2176
+    },
+    {
+      "epoch": 0.9389691610955359,
+      "grad_norm": 0.17489151656627655,
+      "learning_rate": 0.00011755731193370274,
+      "loss": 2.1653,
+      "step": 2177
+    },
+    {
+      "epoch": 0.9394004744446841,
+      "grad_norm": 0.16846604645252228,
+      "learning_rate": 0.00011752920584579393,
+      "loss": 2.3961,
+      "step": 2178
+    },
+    {
+      "epoch": 0.9398317877938323,
+      "grad_norm": 0.16123254597187042,
+      "learning_rate": 0.00011750109095174798,
+      "loss": 2.0953,
+      "step": 2179
+    },
+    {
+      "epoch": 0.9402631011429804,
+      "grad_norm": 0.16449014842510223,
+      "learning_rate": 0.0001174729672573863,
+      "loss": 2.1383,
+      "step": 2180
+    },
+    {
+      "epoch": 0.9406944144921285,
+      "grad_norm": 0.23591160774230957,
+      "learning_rate": 0.00011744483476853234,
+      "loss": 2.3589,
+      "step": 2181
+    },
+    {
+      "epoch": 0.9411257278412767,
+      "grad_norm": 0.16046296060085297,
+      "learning_rate": 0.00011741669349101113,
+      "loss": 2.1155,
+      "step": 2182
+    },
+    {
+      "epoch": 0.9415570411904248,
+      "grad_norm": 0.15319305658340454,
+      "learning_rate": 0.00011738854343064969,
+      "loss": 2.2793,
+      "step": 2183
+    },
+    {
+      "epoch": 0.941988354539573,
+      "grad_norm": 0.17600052058696747,
+      "learning_rate": 0.00011736038459327678,
+      "loss": 2.1299,
+      "step": 2184
+    },
+    {
+      "epoch": 0.9424196678887211,
+      "grad_norm": 0.17410261929035187,
+      "learning_rate": 0.00011733221698472299,
+      "loss": 2.2609,
+      "step": 2185
+    },
+    {
+      "epoch": 0.9428509812378693,
+      "grad_norm": 0.16323557496070862,
+      "learning_rate": 0.00011730404061082072,
+      "loss": 2.0666,
+      "step": 2186
+    },
+    {
+      "epoch": 0.9432822945870175,
+      "grad_norm": 0.16431361436843872,
+      "learning_rate": 0.00011727585547740425,
+      "loss": 2.1692,
+      "step": 2187
+    },
+    {
+      "epoch": 0.9437136079361657,
+      "grad_norm": 0.1787741631269455,
+      "learning_rate": 0.00011724766159030955,
+      "loss": 2.1846,
+      "step": 2188
+    },
+    {
+      "epoch": 0.9441449212853138,
+      "grad_norm": 0.17085330188274384,
+      "learning_rate": 0.00011721945895537452,
+      "loss": 2.3124,
+      "step": 2189
+    },
+    {
+      "epoch": 0.9445762346344619,
+      "grad_norm": 0.170313760638237,
+      "learning_rate": 0.0001171912475784388,
+      "loss": 2.3532,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9450075479836101,
+      "grad_norm": 0.15803568065166473,
+      "learning_rate": 0.0001171630274653439,
+      "loss": 2.1966,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9454388613327582,
+      "grad_norm": 0.17301326990127563,
+      "learning_rate": 0.00011713479862193304,
+      "loss": 2.1362,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9458701746819064,
+      "grad_norm": 0.1860528141260147,
+      "learning_rate": 0.00011710656105405136,
+      "loss": 2.365,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9463014880310545,
+      "grad_norm": 0.16941936314105988,
+      "learning_rate": 0.00011707831476754573,
+      "loss": 2.1095,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9467328013802028,
+      "grad_norm": 0.1830940544605255,
+      "learning_rate": 0.0001170500597682649,
+      "loss": 2.2742,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9471641147293509,
+      "grad_norm": 0.16178883612155914,
+      "learning_rate": 0.00011702179606205932,
+      "loss": 2.1781,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9475954280784991,
+      "grad_norm": 0.18592123687267303,
+      "learning_rate": 0.00011699352365478133,
+      "loss": 2.1072,
+      "step": 2197
+    },
+    {
+      "epoch": 0.9480267414276472,
+      "grad_norm": 0.17916928231716156,
+      "learning_rate": 0.00011696524255228504,
+      "loss": 2.1772,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9484580547767953,
+      "grad_norm": 0.1986900418996811,
+      "learning_rate": 0.00011693695276042633,
+      "loss": 2.2692,
+      "step": 2199
+    },
+    {
+      "epoch": 0.9488893681259435,
+      "grad_norm": 0.18639887869358063,
+      "learning_rate": 0.00011690865428506296,
+      "loss": 2.4306,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9488893681259435,
+      "eval_loss": 2.1058454513549805,
+      "eval_runtime": 201.9377,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9493206814750916,
+      "grad_norm": 0.1623578667640686,
+      "learning_rate": 0.0001168803471320544,
+      "loss": 2.2295,
+      "step": 2201
+    },
+    {
+      "epoch": 0.9497519948242398,
+      "grad_norm": 0.16612379252910614,
+      "learning_rate": 0.00011685203130726201,
+      "loss": 2.4427,
+      "step": 2202
+    },
+    {
+      "epoch": 0.9501833081733879,
+      "grad_norm": 0.16849513351917267,
+      "learning_rate": 0.00011682370681654879,
+      "loss": 2.2249,
+      "step": 2203
+    },
+    {
+      "epoch": 0.9506146215225362,
+      "grad_norm": 0.18068940937519073,
+      "learning_rate": 0.00011679537366577975,
+      "loss": 2.329,
+      "step": 2204
+    },
+    {
+      "epoch": 0.9510459348716843,
+      "grad_norm": 0.15872099995613098,
+      "learning_rate": 0.00011676703186082149,
+      "loss": 2.256,
+      "step": 2205
+    },
+    {
+      "epoch": 0.9514772482208325,
+      "grad_norm": 0.202578604221344,
+      "learning_rate": 0.00011673868140754254,
+      "loss": 2.2509,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9519085615699806,
+      "grad_norm": 0.15152516961097717,
+      "learning_rate": 0.00011671032231181317,
+      "loss": 2.2613,
+      "step": 2207
+    },
+    {
+      "epoch": 0.9523398749191287,
+      "grad_norm": 0.16518807411193848,
+      "learning_rate": 0.00011668195457950541,
+      "loss": 2.2611,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9527711882682769,
+      "grad_norm": 0.16880130767822266,
+      "learning_rate": 0.00011665357821649314,
+      "loss": 2.1909,
+      "step": 2209
+    },
+    {
+      "epoch": 0.953202501617425,
+      "grad_norm": 0.18211935460567474,
+      "learning_rate": 0.00011662519322865198,
+      "loss": 2.2239,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9536338149665732,
+      "grad_norm": 0.17535841464996338,
+      "learning_rate": 0.00011659679962185937,
+      "loss": 2.2193,
+      "step": 2211
+    },
+    {
+      "epoch": 0.9540651283157213,
+      "grad_norm": 0.16001532971858978,
+      "learning_rate": 0.00011656839740199448,
+      "loss": 1.8185,
+      "step": 2212
+    },
+    {
+      "epoch": 0.9544964416648696,
+      "grad_norm": 0.16617374122142792,
+      "learning_rate": 0.00011653998657493834,
+      "loss": 2.1581,
+      "step": 2213
+    },
+    {
+      "epoch": 0.9549277550140177,
+      "grad_norm": 0.18469476699829102,
+      "learning_rate": 0.00011651156714657369,
+      "loss": 2.2726,
+      "step": 2214
+    },
+    {
+      "epoch": 0.9553590683631659,
+      "grad_norm": 0.16976144909858704,
+      "learning_rate": 0.00011648313912278511,
+      "loss": 2.3153,
+      "step": 2215
+    },
+    {
+      "epoch": 0.955790381712314,
+      "grad_norm": 0.17235201597213745,
+      "learning_rate": 0.00011645470250945892,
+      "loss": 2.0512,
+      "step": 2216
+    },
+    {
+      "epoch": 0.9562216950614622,
+      "grad_norm": 17.926816940307617,
+      "learning_rate": 0.00011642625731248325,
+      "loss": 2.3241,
+      "step": 2217
+    },
+    {
+      "epoch": 0.9566530084106103,
+      "grad_norm": 0.19533559679985046,
+      "learning_rate": 0.00011639780353774797,
+      "loss": 2.2302,
+      "step": 2218
+    },
+    {
+      "epoch": 0.9570843217597584,
+      "grad_norm": 0.1967902034521103,
+      "learning_rate": 0.00011636934119114472,
+      "loss": 2.3553,
+      "step": 2219
+    },
+    {
+      "epoch": 0.9575156351089066,
+      "grad_norm": 0.17256464064121246,
+      "learning_rate": 0.00011634087027856697,
+      "loss": 2.3912,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9579469484580547,
+      "grad_norm": 0.17740106582641602,
+      "learning_rate": 0.00011631239080590998,
+      "loss": 2.2265,
+      "step": 2221
+    },
+    {
+      "epoch": 0.958378261807203,
+      "grad_norm": 0.1855895072221756,
+      "learning_rate": 0.00011628390277907065,
+      "loss": 2.1366,
+      "step": 2222
+    },
+    {
+      "epoch": 0.9588095751563511,
+      "grad_norm": 0.21072974801063538,
+      "learning_rate": 0.00011625540620394779,
+      "loss": 2.0472,
+      "step": 2223
+    },
+    {
+      "epoch": 0.9592408885054993,
+      "grad_norm": 0.3419710695743561,
+      "learning_rate": 0.00011622690108644192,
+      "loss": 2.3737,
+      "step": 2224
+    },
+    {
+      "epoch": 0.9596722018546474,
+      "grad_norm": 0.17139427363872528,
+      "learning_rate": 0.00011619838743245531,
+      "loss": 2.1824,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9596722018546474,
+      "eval_loss": 2.106443405151367,
+      "eval_runtime": 206.0581,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9601035152037956,
+      "grad_norm": 0.1631966233253479,
+      "learning_rate": 0.00011616986524789204,
+      "loss": 2.1654,
+      "step": 2226
+    },
+    {
+      "epoch": 0.9605348285529437,
+      "grad_norm": 0.16243740916252136,
+      "learning_rate": 0.00011614133453865798,
+      "loss": 2.3102,
+      "step": 2227
+    },
+    {
+      "epoch": 0.9609661419020918,
+      "grad_norm": 0.18797215819358826,
+      "learning_rate": 0.00011611279531066066,
+      "loss": 2.2347,
+      "step": 2228
+    },
+    {
+      "epoch": 0.96139745525124,
+      "grad_norm": 0.1728636771440506,
+      "learning_rate": 0.00011608424756980947,
+      "loss": 2.2274,
+      "step": 2229
+    },
+    {
+      "epoch": 0.9618287686003881,
+      "grad_norm": 0.20025914907455444,
+      "learning_rate": 0.00011605569132201557,
+      "loss": 2.1984,
+      "step": 2230
+    },
+    {
+      "epoch": 0.9622600819495364,
+      "grad_norm": 0.16293105483055115,
+      "learning_rate": 0.00011602712657319174,
+      "loss": 2.2789,
+      "step": 2231
+    },
+    {
+      "epoch": 0.9626913952986845,
+      "grad_norm": 0.17379815876483917,
+      "learning_rate": 0.00011599855332925275,
+      "loss": 2.3766,
+      "step": 2232
+    },
+    {
+      "epoch": 0.9631227086478327,
+      "grad_norm": 10.470320701599121,
+      "learning_rate": 0.0001159699715961149,
+      "loss": 1.8858,
+      "step": 2233
+    },
+    {
+      "epoch": 0.9635540219969808,
+      "grad_norm": 0.19438959658145905,
+      "learning_rate": 0.00011594138137969641,
+      "loss": 2.368,
+      "step": 2234
+    },
+    {
+      "epoch": 0.963985335346129,
+      "grad_norm": 0.19522325694561005,
+      "learning_rate": 0.00011591278268591717,
+      "loss": 2.2328,
+      "step": 2235
+    },
+    {
+      "epoch": 0.9644166486952771,
+      "grad_norm": 0.2502066493034363,
+      "learning_rate": 0.00011588417552069885,
+      "loss": 2.4032,
+      "step": 2236
+    },
+    {
+      "epoch": 0.9648479620444252,
+      "grad_norm": 0.2284168004989624,
+      "learning_rate": 0.00011585555988996491,
+      "loss": 2.3211,
+      "step": 2237
+    },
+    {
+      "epoch": 0.9652792753935734,
+      "grad_norm": 0.2279599905014038,
+      "learning_rate": 0.00011582693579964048,
+      "loss": 2.3949,
+      "step": 2238
+    },
+    {
+      "epoch": 0.9657105887427215,
+      "grad_norm": 94.24360656738281,
+      "learning_rate": 0.00011579830325565254,
+      "loss": 2.2135,
+      "step": 2239
+    },
+    {
+      "epoch": 0.9661419020918698,
+      "grad_norm": 0.17028680443763733,
+      "learning_rate": 0.00011576966226392974,
+      "loss": 2.2827,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9665732154410179,
+      "grad_norm": 0.8349494338035583,
+      "learning_rate": 0.00011574101283040251,
+      "loss": 2.224,
+      "step": 2241
+    },
+    {
+      "epoch": 0.9670045287901661,
+      "grad_norm": 0.18213754892349243,
+      "learning_rate": 0.00011571235496100303,
+      "loss": 2.2549,
+      "step": 2242
+    },
+    {
+      "epoch": 0.9674358421393142,
+      "grad_norm": 0.21162238717079163,
+      "learning_rate": 0.00011568368866166525,
+      "loss": 2.3134,
+      "step": 2243
+    },
+    {
+      "epoch": 0.9678671554884624,
+      "grad_norm": 0.18910057842731476,
+      "learning_rate": 0.00011565501393832484,
+      "loss": 2.3069,
+      "step": 2244
+    },
+    {
+      "epoch": 0.9682984688376105,
+      "grad_norm": 0.2546372413635254,
+      "learning_rate": 0.00011562633079691917,
+      "loss": 2.1317,
+      "step": 2245
+    },
+    {
+      "epoch": 0.9687297821867586,
+      "grad_norm": 0.17593315243721008,
+      "learning_rate": 0.00011559763924338746,
+      "loss": 2.0852,
+      "step": 2246
+    },
+    {
+      "epoch": 0.9691610955359068,
+      "grad_norm": 0.17111103236675262,
+      "learning_rate": 0.00011556893928367052,
+      "loss": 2.3776,
+      "step": 2247
+    },
+    {
+      "epoch": 0.969592408885055,
+      "grad_norm": 0.1876457780599594,
+      "learning_rate": 0.0001155402309237111,
+      "loss": 2.2422,
+      "step": 2248
+    },
+    {
+      "epoch": 0.9700237222342032,
+      "grad_norm": 0.16125169396400452,
+      "learning_rate": 0.0001155115141694535,
+      "loss": 2.1283,
+      "step": 2249
+    },
+    {
+      "epoch": 0.9704550355833513,
+      "grad_norm": 0.16092783212661743,
+      "learning_rate": 0.00011548278902684386,
+      "loss": 2.2404,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9704550355833513,
+      "eval_loss": 2.107475757598877,
+      "eval_runtime": 201.252,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9708863489324995,
+      "grad_norm": 0.17660339176654816,
+      "learning_rate": 0.00011545405550183006,
+      "loss": 2.3106,
+      "step": 2251
+    },
+    {
+      "epoch": 0.9713176622816476,
+      "grad_norm": 0.21650633215904236,
+      "learning_rate": 0.00011542531360036165,
+      "loss": 2.2065,
+      "step": 2252
+    },
+    {
+      "epoch": 0.9717489756307958,
+      "grad_norm": 0.18377810716629028,
+      "learning_rate": 0.00011539656332838998,
+      "loss": 2.1903,
+      "step": 2253
+    },
+    {
+      "epoch": 0.9721802889799439,
+      "grad_norm": 0.164555162191391,
+      "learning_rate": 0.00011536780469186812,
+      "loss": 2.1216,
+      "step": 2254
+    },
+    {
+      "epoch": 0.972611602329092,
+      "grad_norm": 0.16842810809612274,
+      "learning_rate": 0.00011533903769675082,
+      "loss": 2.0297,
+      "step": 2255
+    },
+    {
+      "epoch": 0.9730429156782402,
+      "grad_norm": 0.17207801342010498,
+      "learning_rate": 0.00011531026234899464,
+      "loss": 2.1705,
+      "step": 2256
+    },
+    {
+      "epoch": 0.9734742290273884,
+      "grad_norm": 0.17921404540538788,
+      "learning_rate": 0.00011528147865455781,
+      "loss": 2.247,
+      "step": 2257
+    },
+    {
+      "epoch": 0.9739055423765366,
+      "grad_norm": 0.17994186282157898,
+      "learning_rate": 0.00011525268661940033,
+      "loss": 2.0532,
+      "step": 2258
+    },
+    {
+      "epoch": 0.9743368557256847,
+      "grad_norm": 0.19421452283859253,
+      "learning_rate": 0.00011522388624948388,
+      "loss": 2.0998,
+      "step": 2259
+    },
+    {
+      "epoch": 0.9747681690748329,
+      "grad_norm": 0.1837734878063202,
+      "learning_rate": 0.00011519507755077189,
+      "loss": 2.3626,
+      "step": 2260
+    },
+    {
+      "epoch": 0.975199482423981,
+      "grad_norm": 0.18781611323356628,
+      "learning_rate": 0.00011516626052922953,
+      "loss": 2.3344,
+      "step": 2261
+    },
+    {
+      "epoch": 0.9756307957731292,
+      "grad_norm": 0.167991504073143,
+      "learning_rate": 0.00011513743519082369,
+      "loss": 2.0942,
+      "step": 2262
+    },
+    {
+      "epoch": 0.9760621091222773,
+      "grad_norm": 0.1679423600435257,
+      "learning_rate": 0.00011510860154152294,
+      "loss": 2.1936,
+      "step": 2263
+    },
+    {
+      "epoch": 0.9764934224714255,
+      "grad_norm": 0.16400964558124542,
+      "learning_rate": 0.00011507975958729763,
+      "loss": 2.0745,
+      "step": 2264
+    },
+    {
+      "epoch": 0.9769247358205736,
+      "grad_norm": 0.18759959936141968,
+      "learning_rate": 0.0001150509093341198,
+      "loss": 2.3091,
+      "step": 2265
+    },
+    {
+      "epoch": 0.9773560491697219,
+      "grad_norm": 0.16005435585975647,
+      "learning_rate": 0.00011502205078796317,
+      "loss": 2.2465,
+      "step": 2266
+    },
+    {
+      "epoch": 0.97778736251887,
+      "grad_norm": 0.16512760519981384,
+      "learning_rate": 0.0001149931839548033,
+      "loss": 2.225,
+      "step": 2267
+    },
+    {
+      "epoch": 0.9782186758680181,
+      "grad_norm": 0.1739596426486969,
+      "learning_rate": 0.0001149643088406173,
+      "loss": 2.3453,
+      "step": 2268
+    },
+    {
+      "epoch": 0.9786499892171663,
+      "grad_norm": 0.17262277007102966,
+      "learning_rate": 0.00011493542545138411,
+      "loss": 2.0907,
+      "step": 2269
+    },
+    {
+      "epoch": 0.9790813025663144,
+      "grad_norm": 0.1671365201473236,
+      "learning_rate": 0.00011490653379308439,
+      "loss": 2.3431,
+      "step": 2270
+    },
+    {
+      "epoch": 0.9795126159154626,
+      "grad_norm": 0.16103728115558624,
+      "learning_rate": 0.0001148776338717004,
+      "loss": 2.2239,
+      "step": 2271
+    },
+    {
+      "epoch": 0.9799439292646107,
+      "grad_norm": 0.16833123564720154,
+      "learning_rate": 0.00011484872569321624,
+      "loss": 2.0763,
+      "step": 2272
+    },
+    {
+      "epoch": 0.9803752426137589,
+      "grad_norm": 0.1860317438840866,
+      "learning_rate": 0.00011481980926361765,
+      "loss": 2.0637,
+      "step": 2273
+    },
+    {
+      "epoch": 0.980806555962907,
+      "grad_norm": 0.1666116863489151,
+      "learning_rate": 0.00011479088458889208,
+      "loss": 2.149,
+      "step": 2274
+    },
+    {
+      "epoch": 0.9812378693120553,
+      "grad_norm": 0.22403891384601593,
+      "learning_rate": 0.0001147619516750287,
+      "loss": 2.3037,
+      "step": 2275
+    },
+    {
+      "epoch": 0.9812378693120553,
+      "eval_loss": 2.106496572494507,
+      "eval_runtime": 198.6172,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 2275
+    },
+    {
+      "epoch": 0.9816691826612034,
+      "grad_norm": 0.17930032312870026,
+      "learning_rate": 0.00011473301052801843,
+      "loss": 2.218,
+      "step": 2276
+    },
+    {
+      "epoch": 0.9821004960103515,
+      "grad_norm": 0.2308432161808014,
+      "learning_rate": 0.00011470406115385378,
+      "loss": 1.9874,
+      "step": 2277
+    },
+    {
+      "epoch": 0.9825318093594997,
+      "grad_norm": 0.18067146837711334,
+      "learning_rate": 0.0001146751035585291,
+      "loss": 2.104,
+      "step": 2278
+    },
+    {
+      "epoch": 0.9829631227086478,
+      "grad_norm": 0.17085899412631989,
+      "learning_rate": 0.00011464613774804034,
+      "loss": 2.1093,
+      "step": 2279
+    },
+    {
+      "epoch": 0.983394436057796,
+      "grad_norm": 0.19405338168144226,
+      "learning_rate": 0.00011461716372838518,
+      "loss": 2.1491,
+      "step": 2280
+    },
+    {
+      "epoch": 0.9838257494069441,
+      "grad_norm": 0.17247186601161957,
+      "learning_rate": 0.00011458818150556304,
+      "loss": 2.1054,
+      "step": 2281
+    },
+    {
+      "epoch": 0.9842570627560923,
+      "grad_norm": 0.15420669317245483,
+      "learning_rate": 0.00011455919108557499,
+      "loss": 2.0272,
+      "step": 2282
+    },
+    {
+      "epoch": 0.9846883761052404,
+      "grad_norm": 0.18149659037590027,
+      "learning_rate": 0.0001145301924744238,
+      "loss": 2.2578,
+      "step": 2283
+    },
+    {
+      "epoch": 0.9851196894543887,
+      "grad_norm": 0.1740005910396576,
+      "learning_rate": 0.00011450118567811398,
+      "loss": 2.2191,
+      "step": 2284
+    },
+    {
+      "epoch": 0.9855510028035368,
+      "grad_norm": 0.16811902821063995,
+      "learning_rate": 0.00011447217070265168,
+      "loss": 2.0406,
+      "step": 2285
+    },
+    {
+      "epoch": 0.985982316152685,
+      "grad_norm": 0.17773064970970154,
+      "learning_rate": 0.00011444314755404478,
+      "loss": 2.4005,
+      "step": 2286
+    },
+    {
+      "epoch": 0.9864136295018331,
+      "grad_norm": 0.18015626072883606,
+      "learning_rate": 0.00011441411623830284,
+      "loss": 2.3781,
+      "step": 2287
+    },
+    {
+      "epoch": 0.9868449428509812,
+      "grad_norm": 0.1616867482662201,
+      "learning_rate": 0.0001143850767614371,
+      "loss": 2.1957,
+      "step": 2288
+    },
+    {
+      "epoch": 0.9872762562001294,
+      "grad_norm": 0.17316801846027374,
+      "learning_rate": 0.00011435602912946051,
+      "loss": 2.225,
+      "step": 2289
+    },
+    {
+      "epoch": 0.9877075695492775,
+      "grad_norm": 0.16571733355522156,
+      "learning_rate": 0.00011432697334838769,
+      "loss": 2.2129,
+      "step": 2290
+    },
+    {
+      "epoch": 0.9881388828984257,
+      "grad_norm": 0.19395996630191803,
+      "learning_rate": 0.00011429790942423497,
+      "loss": 2.198,
+      "step": 2291
+    },
+    {
+      "epoch": 0.9885701962475738,
+      "grad_norm": 0.18082626163959503,
+      "learning_rate": 0.00011426883736302036,
+      "loss": 2.2387,
+      "step": 2292
+    },
+    {
+      "epoch": 0.9890015095967221,
+      "grad_norm": 0.17537705600261688,
+      "learning_rate": 0.00011423975717076354,
+      "loss": 2.1682,
+      "step": 2293
+    },
+    {
+      "epoch": 0.9894328229458702,
+      "grad_norm": 0.15734168887138367,
+      "learning_rate": 0.00011421066885348587,
+      "loss": 2.2326,
+      "step": 2294
+    },
+    {
+      "epoch": 0.9898641362950183,
+      "grad_norm": 0.1727844476699829,
+      "learning_rate": 0.00011418157241721042,
+      "loss": 2.2387,
+      "step": 2295
+    },
+    {
+      "epoch": 0.9902954496441665,
+      "grad_norm": 0.24999341368675232,
+      "learning_rate": 0.00011415246786796189,
+      "loss": 1.9656,
+      "step": 2296
+    },
+    {
+      "epoch": 0.9907267629933146,
+      "grad_norm": 0.36071884632110596,
+      "learning_rate": 0.00011412335521176677,
+      "loss": 2.1987,
+      "step": 2297
+    },
+    {
+      "epoch": 0.9911580763424628,
+      "grad_norm": 0.16273179650306702,
+      "learning_rate": 0.00011409423445465306,
+      "loss": 2.1133,
+      "step": 2298
+    },
+    {
+      "epoch": 0.9915893896916109,
+      "grad_norm": 0.17782722413539886,
+      "learning_rate": 0.0001140651056026506,
+      "loss": 2.2404,
+      "step": 2299
+    },
+    {
+      "epoch": 0.9920207030407591,
+      "grad_norm": 0.17763769626617432,
+      "learning_rate": 0.00011403596866179084,
+      "loss": 2.3585,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9920207030407591,
+      "eval_loss": 2.1060075759887695,
+      "eval_runtime": 199.8504,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9924520163899073,
+      "grad_norm": 0.17662031948566437,
+      "learning_rate": 0.00011400682363810685,
+      "loss": 1.9006,
+      "step": 2301
+    },
+    {
+      "epoch": 0.9928833297390555,
+      "grad_norm": 0.17285652458667755,
+      "learning_rate": 0.00011397767053763346,
+      "loss": 1.9154,
+      "step": 2302
+    },
+    {
+      "epoch": 0.9933146430882036,
+      "grad_norm": 0.15654821693897247,
+      "learning_rate": 0.00011394850936640714,
+      "loss": 2.1394,
+      "step": 2303
+    },
+    {
+      "epoch": 0.9937459564373518,
+      "grad_norm": 0.15930558741092682,
+      "learning_rate": 0.000113919340130466,
+      "loss": 2.3887,
+      "step": 2304
+    },
+    {
+      "epoch": 0.9941772697864999,
+      "grad_norm": 0.23084445297718048,
+      "learning_rate": 0.00011389016283584987,
+      "loss": 2.2288,
+      "step": 2305
+    },
+    {
+      "epoch": 0.994608583135648,
+      "grad_norm": 0.16395895183086395,
+      "learning_rate": 0.00011386097748860025,
+      "loss": 2.3934,
+      "step": 2306
+    },
+    {
+      "epoch": 0.9950398964847962,
+      "grad_norm": 0.19965040683746338,
+      "learning_rate": 0.00011383178409476027,
+      "loss": 2.1694,
+      "step": 2307
+    },
+    {
+      "epoch": 0.9954712098339443,
+      "grad_norm": 0.17996490001678467,
+      "learning_rate": 0.00011380258266037472,
+      "loss": 2.0064,
+      "step": 2308
+    },
+    {
+      "epoch": 0.9959025231830925,
+      "grad_norm": 0.1590234488248825,
+      "learning_rate": 0.00011377337319149012,
+      "loss": 2.2363,
+      "step": 2309
+    },
+    {
+      "epoch": 0.9963338365322407,
+      "grad_norm": 0.17119617760181427,
+      "learning_rate": 0.00011374415569415455,
+      "loss": 2.3706,
+      "step": 2310
+    },
+    {
+      "epoch": 0.9967651498813889,
+      "grad_norm": 0.15185703337192535,
+      "learning_rate": 0.00011371493017441785,
+      "loss": 2.343,
+      "step": 2311
+    },
+    {
+      "epoch": 0.997196463230537,
+      "grad_norm": 0.18372780084609985,
+      "learning_rate": 0.00011368569663833149,
+      "loss": 2.2689,
+      "step": 2312
+    },
+    {
+      "epoch": 0.9976277765796852,
+      "grad_norm": 0.17568238079547882,
+      "learning_rate": 0.00011365645509194858,
+      "loss": 2.2873,
+      "step": 2313
+    },
+    {
+      "epoch": 0.9980590899288333,
+      "grad_norm": 0.17242491245269775,
+      "learning_rate": 0.00011362720554132387,
+      "loss": 2.1604,
+      "step": 2314
+    },
+    {
+      "epoch": 0.9984904032779814,
+      "grad_norm": 0.15822440385818481,
+      "learning_rate": 0.00011359794799251387,
+      "loss": 2.1958,
+      "step": 2315
+    },
+    {
+      "epoch": 0.9989217166271296,
+      "grad_norm": 0.17576847970485687,
+      "learning_rate": 0.00011356868245157665,
+      "loss": 2.0845,
+      "step": 2316
+    },
+    {
+      "epoch": 0.9993530299762777,
+      "grad_norm": 0.16132158041000366,
+      "learning_rate": 0.00011353940892457191,
+      "loss": 2.1926,
+      "step": 2317
+    },
+    {
+      "epoch": 0.9997843433254259,
+      "grad_norm": 0.16950838267803192,
+      "learning_rate": 0.00011351012741756111,
+      "loss": 2.3559,
+      "step": 2318
+    },
+    {
+      "epoch": 1.0004313133491483,
+      "grad_norm": 0.29467716813087463,
+      "learning_rate": 0.0001134808379366073,
+      "loss": 4.4302,
+      "step": 2319
+    },
+    {
+      "epoch": 1.0008626266982963,
+      "grad_norm": 0.15429019927978516,
+      "learning_rate": 0.00011345154048777516,
+      "loss": 2.2957,
+      "step": 2320
+    },
+    {
+      "epoch": 1.0012939400474445,
+      "grad_norm": 0.16172082722187042,
+      "learning_rate": 0.00011342223507713106,
+      "loss": 2.1032,
+      "step": 2321
+    },
+    {
+      "epoch": 1.0017252533965926,
+      "grad_norm": 0.16826793551445007,
+      "learning_rate": 0.00011339292171074302,
+      "loss": 2.2723,
+      "step": 2322
+    },
+    {
+      "epoch": 1.0021565667457408,
+      "grad_norm": 0.1786830723285675,
+      "learning_rate": 0.00011336360039468069,
+      "loss": 2.1119,
+      "step": 2323
+    },
+    {
+      "epoch": 1.0025878800948889,
+      "grad_norm": 0.15725389122962952,
+      "learning_rate": 0.00011333427113501533,
+      "loss": 1.8254,
+      "step": 2324
+    },
+    {
+      "epoch": 1.0030191934440371,
+      "grad_norm": 0.1785026639699936,
+      "learning_rate": 0.00011330493393781994,
+      "loss": 2.2904,
+      "step": 2325
+    },
+    {
+      "epoch": 1.0030191934440371,
+      "eval_loss": 2.1062262058258057,
+      "eval_runtime": 195.6116,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2325
+    },
+    {
+      "epoch": 1.0034505067931851,
+      "grad_norm": 0.18080540001392365,
+      "learning_rate": 0.0001132755888091691,
+      "loss": 2.2291,
+      "step": 2326
+    },
+    {
+      "epoch": 1.0038818201423334,
+      "grad_norm": 0.21925707161426544,
+      "learning_rate": 0.000113246235755139,
+      "loss": 2.1652,
+      "step": 2327
+    },
+    {
+      "epoch": 1.0043131334914817,
+      "grad_norm": 0.1603548377752304,
+      "learning_rate": 0.0001132168747818075,
+      "loss": 2.1039,
+      "step": 2328
+    },
+    {
+      "epoch": 1.0047444468406297,
+      "grad_norm": 0.15417563915252686,
+      "learning_rate": 0.00011318750589525418,
+      "loss": 2.0317,
+      "step": 2329
+    },
+    {
+      "epoch": 1.005175760189778,
+      "grad_norm": 0.18038475513458252,
+      "learning_rate": 0.00011315812910156015,
+      "loss": 2.1561,
+      "step": 2330
+    },
+    {
+      "epoch": 1.005607073538926,
+      "grad_norm": 0.19016197323799133,
+      "learning_rate": 0.0001131287444068082,
+      "loss": 2.2455,
+      "step": 2331
+    },
+    {
+      "epoch": 1.0060383868880742,
+      "grad_norm": 0.17915791273117065,
+      "learning_rate": 0.00011309935181708275,
+      "loss": 2.1837,
+      "step": 2332
+    },
+    {
+      "epoch": 1.0064697002372223,
+      "grad_norm": 0.1723218560218811,
+      "learning_rate": 0.00011306995133846986,
+      "loss": 2.1696,
+      "step": 2333
+    },
+    {
+      "epoch": 1.0069010135863705,
+      "grad_norm": 0.17618072032928467,
+      "learning_rate": 0.00011304054297705721,
+      "loss": 2.1869,
+      "step": 2334
+    },
+    {
+      "epoch": 1.0073323269355186,
+      "grad_norm": 0.1933479905128479,
+      "learning_rate": 0.00011301112673893413,
+      "loss": 2.1879,
+      "step": 2335
+    },
+    {
+      "epoch": 1.0077636402846668,
+      "grad_norm": 0.17005442082881927,
+      "learning_rate": 0.0001129817026301916,
+      "loss": 2.1525,
+      "step": 2336
+    },
+    {
+      "epoch": 1.008194953633815,
+      "grad_norm": 0.17357569932937622,
+      "learning_rate": 0.00011295227065692217,
+      "loss": 2.2398,
+      "step": 2337
+    },
+    {
+      "epoch": 1.008626266982963,
+      "grad_norm": 0.16071292757987976,
+      "learning_rate": 0.00011292283082522005,
+      "loss": 2.1212,
+      "step": 2338
+    },
+    {
+      "epoch": 1.0090575803321113,
+      "grad_norm": 0.18172769248485565,
+      "learning_rate": 0.0001128933831411811,
+      "loss": 2.1368,
+      "step": 2339
+    },
+    {
+      "epoch": 1.0094888936812594,
+      "grad_norm": 0.18058297038078308,
+      "learning_rate": 0.00011286392761090278,
+      "loss": 2.1471,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0099202070304076,
+      "grad_norm": 0.17868855595588684,
+      "learning_rate": 0.00011283446424048417,
+      "loss": 2.2774,
+      "step": 2341
+    },
+    {
+      "epoch": 1.0103515203795557,
+      "grad_norm": 0.18230393528938293,
+      "learning_rate": 0.000112804993036026,
+      "loss": 2.366,
+      "step": 2342
+    },
+    {
+      "epoch": 1.010782833728704,
+      "grad_norm": 0.1880486011505127,
+      "learning_rate": 0.00011277551400363057,
+      "loss": 2.1481,
+      "step": 2343
+    },
+    {
+      "epoch": 1.011214147077852,
+      "grad_norm": 0.16729071736335754,
+      "learning_rate": 0.00011274602714940186,
+      "loss": 2.2148,
+      "step": 2344
+    },
+    {
+      "epoch": 1.0116454604270002,
+      "grad_norm": 0.16809020936489105,
+      "learning_rate": 0.00011271653247944545,
+      "loss": 2.0783,
+      "step": 2345
+    },
+    {
+      "epoch": 1.0120767737761485,
+      "grad_norm": 0.2248852401971817,
+      "learning_rate": 0.00011268702999986854,
+      "loss": 2.2623,
+      "step": 2346
+    },
+    {
+      "epoch": 1.0125080871252965,
+      "grad_norm": 0.18641109764575958,
+      "learning_rate": 0.00011265751971677994,
+      "loss": 2.25,
+      "step": 2347
+    },
+    {
+      "epoch": 1.0129394004744448,
+      "grad_norm": 0.17987805604934692,
+      "learning_rate": 0.00011262800163629003,
+      "loss": 2.1126,
+      "step": 2348
+    },
+    {
+      "epoch": 1.0133707138235928,
+      "grad_norm": 0.18353773653507233,
+      "learning_rate": 0.00011259847576451093,
+      "loss": 2.3219,
+      "step": 2349
+    },
+    {
+      "epoch": 1.013802027172741,
+      "grad_norm": 0.1718786507844925,
+      "learning_rate": 0.00011256894210755622,
+      "loss": 2.2882,
+      "step": 2350
+    },
+    {
+      "epoch": 1.013802027172741,
+      "eval_loss": 2.1062445640563965,
+      "eval_runtime": 200.3454,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2350
+    },
+    {
+      "epoch": 1.014233340521889,
+      "grad_norm": 0.19025760889053345,
+      "learning_rate": 0.0001125394006715412,
+      "loss": 2.1701,
+      "step": 2351
+    },
+    {
+      "epoch": 1.0146646538710373,
+      "grad_norm": 0.1575254201889038,
+      "learning_rate": 0.00011250985146258276,
+      "loss": 2.4227,
+      "step": 2352
+    },
+    {
+      "epoch": 1.0150959672201854,
+      "grad_norm": 0.18695998191833496,
+      "learning_rate": 0.00011248029448679937,
+      "loss": 2.2756,
+      "step": 2353
+    },
+    {
+      "epoch": 1.0155272805693336,
+      "grad_norm": 0.1914224773645401,
+      "learning_rate": 0.00011245072975031115,
+      "loss": 2.1814,
+      "step": 2354
+    },
+    {
+      "epoch": 1.0159585939184819,
+      "grad_norm": 0.20052509009838104,
+      "learning_rate": 0.00011242115725923976,
+      "loss": 2.2445,
+      "step": 2355
+    },
+    {
+      "epoch": 1.01638990726763,
+      "grad_norm": 0.15901760756969452,
+      "learning_rate": 0.00011239157701970852,
+      "loss": 2.1222,
+      "step": 2356
+    },
+    {
+      "epoch": 1.0168212206167782,
+      "grad_norm": 0.1845453679561615,
+      "learning_rate": 0.00011236198903784236,
+      "loss": 2.1006,
+      "step": 2357
+    },
+    {
+      "epoch": 1.0172525339659262,
+      "grad_norm": 0.1753232330083847,
+      "learning_rate": 0.00011233239331976778,
+      "loss": 2.2954,
+      "step": 2358
+    },
+    {
+      "epoch": 1.0176838473150744,
+      "grad_norm": 0.15701685845851898,
+      "learning_rate": 0.00011230278987161294,
+      "loss": 2.2849,
+      "step": 2359
+    },
+    {
+      "epoch": 1.0181151606642225,
+      "grad_norm": 0.16856098175048828,
+      "learning_rate": 0.00011227317869950748,
+      "loss": 2.1087,
+      "step": 2360
+    },
+    {
+      "epoch": 1.0185464740133707,
+      "grad_norm": 0.19807349145412445,
+      "learning_rate": 0.00011224355980958277,
+      "loss": 2.2045,
+      "step": 2361
+    },
+    {
+      "epoch": 1.0189777873625188,
+      "grad_norm": 0.17893430590629578,
+      "learning_rate": 0.0001122139332079717,
+      "loss": 2.2461,
+      "step": 2362
+    },
+    {
+      "epoch": 1.019409100711667,
+      "grad_norm": 0.2081046998500824,
+      "learning_rate": 0.00011218429890080882,
+      "loss": 2.2196,
+      "step": 2363
+    },
+    {
+      "epoch": 1.0198404140608153,
+      "grad_norm": 0.16133612394332886,
+      "learning_rate": 0.0001121546568942302,
+      "loss": 2.153,
+      "step": 2364
+    },
+    {
+      "epoch": 1.0202717274099633,
+      "grad_norm": 0.16959582269191742,
+      "learning_rate": 0.00011212500719437354,
+      "loss": 2.1781,
+      "step": 2365
+    },
+    {
+      "epoch": 1.0207030407591116,
+      "grad_norm": 0.17506343126296997,
+      "learning_rate": 0.00011209534980737816,
+      "loss": 2.0979,
+      "step": 2366
+    },
+    {
+      "epoch": 1.0211343541082596,
+      "grad_norm": 0.1758505403995514,
+      "learning_rate": 0.00011206568473938494,
+      "loss": 2.3522,
+      "step": 2367
+    },
+    {
+      "epoch": 1.0215656674574078,
+      "grad_norm": 0.17813731729984283,
+      "learning_rate": 0.00011203601199653634,
+      "loss": 2.1947,
+      "step": 2368
+    },
+    {
+      "epoch": 1.0219969808065559,
+      "grad_norm": 0.17696397006511688,
+      "learning_rate": 0.00011200633158497645,
+      "loss": 2.0171,
+      "step": 2369
+    },
+    {
+      "epoch": 1.0224282941557041,
+      "grad_norm": 0.2053089290857315,
+      "learning_rate": 0.00011197664351085092,
+      "loss": 2.1748,
+      "step": 2370
+    },
+    {
+      "epoch": 1.0228596075048522,
+      "grad_norm": 0.1562858372926712,
+      "learning_rate": 0.00011194694778030696,
+      "loss": 2.049,
+      "step": 2371
+    },
+    {
+      "epoch": 1.0232909208540004,
+      "grad_norm": 0.18027403950691223,
+      "learning_rate": 0.0001119172443994934,
+      "loss": 2.2517,
+      "step": 2372
+    },
+    {
+      "epoch": 1.0237222342031487,
+      "grad_norm": 0.16071996092796326,
+      "learning_rate": 0.0001118875333745607,
+      "loss": 2.3241,
+      "step": 2373
+    },
+    {
+      "epoch": 1.0241535475522967,
+      "grad_norm": 0.18705935776233673,
+      "learning_rate": 0.00011185781471166082,
+      "loss": 2.1256,
+      "step": 2374
+    },
+    {
+      "epoch": 1.024584860901445,
+      "grad_norm": 0.1858339160680771,
+      "learning_rate": 0.00011182808841694732,
+      "loss": 2.0979,
+      "step": 2375
+    },
+    {
+      "epoch": 1.024584860901445,
+      "eval_loss": 2.1061363220214844,
+      "eval_runtime": 199.9588,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2375
+    },
+    {
+      "epoch": 1.025016174250593,
+      "grad_norm": 0.1685781031847,
+      "learning_rate": 0.00011179835449657537,
+      "loss": 2.1224,
+      "step": 2376
+    },
+    {
+      "epoch": 1.0254474875997412,
+      "grad_norm": 0.15608049929141998,
+      "learning_rate": 0.00011176861295670169,
+      "loss": 2.1423,
+      "step": 2377
+    },
+    {
+      "epoch": 1.0258788009488893,
+      "grad_norm": 0.18338270485401154,
+      "learning_rate": 0.00011173886380348461,
+      "loss": 2.1042,
+      "step": 2378
+    },
+    {
+      "epoch": 1.0263101142980375,
+      "grad_norm": 0.18867897987365723,
+      "learning_rate": 0.00011170910704308403,
+      "loss": 2.009,
+      "step": 2379
+    },
+    {
+      "epoch": 1.0267414276471858,
+      "grad_norm": 0.16375763714313507,
+      "learning_rate": 0.0001116793426816614,
+      "loss": 2.1258,
+      "step": 2380
+    },
+    {
+      "epoch": 1.0271727409963338,
+      "grad_norm": 0.17954808473587036,
+      "learning_rate": 0.0001116495707253797,
+      "loss": 2.2215,
+      "step": 2381
+    },
+    {
+      "epoch": 1.027604054345482,
+      "grad_norm": 0.1598980873823166,
+      "learning_rate": 0.00011161979118040363,
+      "loss": 2.0067,
+      "step": 2382
+    },
+    {
+      "epoch": 1.02803536769463,
+      "grad_norm": 0.17175449430942535,
+      "learning_rate": 0.00011159000405289931,
+      "loss": 2.2334,
+      "step": 2383
+    },
+    {
+      "epoch": 1.0284666810437784,
+      "grad_norm": 0.15543760359287262,
+      "learning_rate": 0.0001115602093490345,
+      "loss": 2.1728,
+      "step": 2384
+    },
+    {
+      "epoch": 1.0288979943929264,
+      "grad_norm": 0.19172172248363495,
+      "learning_rate": 0.00011153040707497855,
+      "loss": 2.2146,
+      "step": 2385
+    },
+    {
+      "epoch": 1.0293293077420747,
+      "grad_norm": 0.19131535291671753,
+      "learning_rate": 0.00011150059723690232,
+      "loss": 2.279,
+      "step": 2386
+    },
+    {
+      "epoch": 1.0297606210912227,
+      "grad_norm": 0.16584806144237518,
+      "learning_rate": 0.00011147077984097827,
+      "loss": 2.1,
+      "step": 2387
+    },
+    {
+      "epoch": 1.030191934440371,
+      "grad_norm": 0.18306796252727509,
+      "learning_rate": 0.0001114409548933804,
+      "loss": 2.1369,
+      "step": 2388
+    },
+    {
+      "epoch": 1.0306232477895192,
+      "grad_norm": 0.16278353333473206,
+      "learning_rate": 0.00011141112240028432,
+      "loss": 2.2203,
+      "step": 2389
+    },
+    {
+      "epoch": 1.0310545611386672,
+      "grad_norm": 0.17837879061698914,
+      "learning_rate": 0.00011138128236786716,
+      "loss": 2.1123,
+      "step": 2390
+    },
+    {
+      "epoch": 1.0314858744878155,
+      "grad_norm": 0.17795895040035248,
+      "learning_rate": 0.00011135143480230763,
+      "loss": 2.0468,
+      "step": 2391
+    },
+    {
+      "epoch": 1.0319171878369635,
+      "grad_norm": 0.17439667880535126,
+      "learning_rate": 0.00011132157970978598,
+      "loss": 2.202,
+      "step": 2392
+    },
+    {
+      "epoch": 1.0323485011861118,
+      "grad_norm": 0.17241206765174866,
+      "learning_rate": 0.00011129171709648407,
+      "loss": 2.2526,
+      "step": 2393
+    },
+    {
+      "epoch": 1.0327798145352598,
+      "grad_norm": 0.17300650477409363,
+      "learning_rate": 0.00011126184696858524,
+      "loss": 2.1694,
+      "step": 2394
+    },
+    {
+      "epoch": 1.033211127884408,
+      "grad_norm": 0.17314332723617554,
+      "learning_rate": 0.00011123196933227445,
+      "loss": 2.1224,
+      "step": 2395
+    },
+    {
+      "epoch": 1.033642441233556,
+      "grad_norm": 0.17457662522792816,
+      "learning_rate": 0.00011120208419373817,
+      "loss": 2.2439,
+      "step": 2396
+    },
+    {
+      "epoch": 1.0340737545827043,
+      "grad_norm": 0.1702902466058731,
+      "learning_rate": 0.00011117219155916444,
+      "loss": 2.1204,
+      "step": 2397
+    },
+    {
+      "epoch": 1.0345050679318526,
+      "grad_norm": 0.1870027780532837,
+      "learning_rate": 0.00011114229143474293,
+      "loss": 2.2655,
+      "step": 2398
+    },
+    {
+      "epoch": 1.0349363812810006,
+      "grad_norm": 0.18020227551460266,
+      "learning_rate": 0.0001111123838266647,
+      "loss": 2.3726,
+      "step": 2399
+    },
+    {
+      "epoch": 1.0353676946301489,
+      "grad_norm": 0.17877480387687683,
+      "learning_rate": 0.00011108246874112247,
+      "loss": 2.0178,
+      "step": 2400
+    },
+    {
+      "epoch": 1.0353676946301489,
+      "eval_loss": 2.1056740283966064,
+      "eval_runtime": 200.7673,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2400
+    },
+    {
+      "epoch": 1.035799007979297,
+      "grad_norm": 0.1728811115026474,
+      "learning_rate": 0.00011105254618431053,
+      "loss": 2.1762,
+      "step": 2401
+    },
+    {
+      "epoch": 1.0362303213284452,
+      "grad_norm": 0.17834728956222534,
+      "learning_rate": 0.00011102261616242461,
+      "loss": 2.3268,
+      "step": 2402
+    },
+    {
+      "epoch": 1.0366616346775932,
+      "grad_norm": 0.17231158912181854,
+      "learning_rate": 0.0001109926786816621,
+      "loss": 2.3475,
+      "step": 2403
+    },
+    {
+      "epoch": 1.0370929480267415,
+      "grad_norm": 0.17313973605632782,
+      "learning_rate": 0.00011096273374822182,
+      "loss": 2.2619,
+      "step": 2404
+    },
+    {
+      "epoch": 1.0375242613758895,
+      "grad_norm": 0.16596461832523346,
+      "learning_rate": 0.00011093278136830428,
+      "loss": 2.2721,
+      "step": 2405
+    },
+    {
+      "epoch": 1.0379555747250377,
+      "grad_norm": 0.19695128500461578,
+      "learning_rate": 0.00011090282154811138,
+      "loss": 2.1435,
+      "step": 2406
+    },
+    {
+      "epoch": 1.038386888074186,
+      "grad_norm": 0.16759416460990906,
+      "learning_rate": 0.00011087285429384665,
+      "loss": 2.1911,
+      "step": 2407
+    },
+    {
+      "epoch": 1.038818201423334,
+      "grad_norm": 0.1748712956905365,
+      "learning_rate": 0.00011084287961171514,
+      "loss": 2.1256,
+      "step": 2408
+    },
+    {
+      "epoch": 1.0392495147724823,
+      "grad_norm": 0.1840677261352539,
+      "learning_rate": 0.00011081289750792343,
+      "loss": 2.3436,
+      "step": 2409
+    },
+    {
+      "epoch": 1.0396808281216303,
+      "grad_norm": 0.19527612626552582,
+      "learning_rate": 0.00011078290798867967,
+      "loss": 2.1762,
+      "step": 2410
+    },
+    {
+      "epoch": 1.0401121414707786,
+      "grad_norm": 0.1802646368741989,
+      "learning_rate": 0.00011075291106019349,
+      "loss": 2.4244,
+      "step": 2411
+    },
+    {
+      "epoch": 1.0405434548199266,
+      "grad_norm": 0.17762580513954163,
+      "learning_rate": 0.00011072290672867607,
+      "loss": 2.2971,
+      "step": 2412
+    },
+    {
+      "epoch": 1.0409747681690749,
+      "grad_norm": 0.1760055422782898,
+      "learning_rate": 0.00011069289500034016,
+      "loss": 2.2714,
+      "step": 2413
+    },
+    {
+      "epoch": 1.041406081518223,
+      "grad_norm": 0.21784080564975739,
+      "learning_rate": 0.00011066287588140002,
+      "loss": 2.2483,
+      "step": 2414
+    },
+    {
+      "epoch": 1.0418373948673711,
+      "grad_norm": 0.17253689467906952,
+      "learning_rate": 0.00011063284937807142,
+      "loss": 2.0145,
+      "step": 2415
+    },
+    {
+      "epoch": 1.0422687082165194,
+      "grad_norm": 0.1615447849035263,
+      "learning_rate": 0.00011060281549657169,
+      "loss": 1.9895,
+      "step": 2416
+    },
+    {
+      "epoch": 1.0427000215656674,
+      "grad_norm": 0.16580533981323242,
+      "learning_rate": 0.00011057277424311965,
+      "loss": 2.2238,
+      "step": 2417
+    },
+    {
+      "epoch": 1.0431313349148157,
+      "grad_norm": 0.16984115540981293,
+      "learning_rate": 0.0001105427256239357,
+      "loss": 2.0539,
+      "step": 2418
+    },
+    {
+      "epoch": 1.0435626482639637,
+      "grad_norm": 0.17555032670497894,
+      "learning_rate": 0.00011051266964524171,
+      "loss": 2.2437,
+      "step": 2419
+    },
+    {
+      "epoch": 1.043993961613112,
+      "grad_norm": 0.17240315675735474,
+      "learning_rate": 0.0001104826063132611,
+      "loss": 2.2337,
+      "step": 2420
+    },
+    {
+      "epoch": 1.04442527496226,
+      "grad_norm": 0.18791107833385468,
+      "learning_rate": 0.00011045253563421886,
+      "loss": 2.1752,
+      "step": 2421
+    },
+    {
+      "epoch": 1.0448565883114083,
+      "grad_norm": 0.17155669629573822,
+      "learning_rate": 0.00011042245761434138,
+      "loss": 1.9039,
+      "step": 2422
+    },
+    {
+      "epoch": 1.0452879016605563,
+      "grad_norm": 0.1768328845500946,
+      "learning_rate": 0.00011039237225985667,
+      "loss": 2.2537,
+      "step": 2423
+    },
+    {
+      "epoch": 1.0457192150097046,
+      "grad_norm": 0.19201205670833588,
+      "learning_rate": 0.00011036227957699428,
+      "loss": 2.295,
+      "step": 2424
+    },
+    {
+      "epoch": 1.0461505283588528,
+      "grad_norm": 0.1698884516954422,
+      "learning_rate": 0.00011033217957198516,
+      "loss": 2.3538,
+      "step": 2425
+    },
+    {
+      "epoch": 1.0461505283588528,
+      "eval_loss": 2.1056013107299805,
+      "eval_runtime": 203.8965,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 2425
+    },
+    {
+      "epoch": 1.0465818417080008,
+      "grad_norm": 0.1858002245426178,
+      "learning_rate": 0.00011030207225106188,
+      "loss": 1.9397,
+      "step": 2426
+    },
+    {
+      "epoch": 1.047013155057149,
+      "grad_norm": 0.17983387410640717,
+      "learning_rate": 0.00011027195762045848,
+      "loss": 2.3015,
+      "step": 2427
+    },
+    {
+      "epoch": 1.0474444684062971,
+      "grad_norm": 0.1639018952846527,
+      "learning_rate": 0.00011024183568641054,
+      "loss": 2.1261,
+      "step": 2428
+    },
+    {
+      "epoch": 1.0478757817554454,
+      "grad_norm": 0.16954176127910614,
+      "learning_rate": 0.0001102117064551551,
+      "loss": 2.312,
+      "step": 2429
+    },
+    {
+      "epoch": 1.0483070951045934,
+      "grad_norm": 0.1672070026397705,
+      "learning_rate": 0.00011018156993293077,
+      "loss": 2.3034,
+      "step": 2430
+    },
+    {
+      "epoch": 1.0487384084537417,
+      "grad_norm": 0.18573078513145447,
+      "learning_rate": 0.00011015142612597765,
+      "loss": 2.2772,
+      "step": 2431
+    },
+    {
+      "epoch": 1.0491697218028897,
+      "grad_norm": 0.1702294945716858,
+      "learning_rate": 0.00011012127504053737,
+      "loss": 2.1996,
+      "step": 2432
+    },
+    {
+      "epoch": 1.049601035152038,
+      "grad_norm": 0.1983357071876526,
+      "learning_rate": 0.000110091116682853,
+      "loss": 2.2025,
+      "step": 2433
+    },
+    {
+      "epoch": 1.0500323485011862,
+      "grad_norm": 0.16793139278888702,
+      "learning_rate": 0.00011006095105916915,
+      "loss": 2.3513,
+      "step": 2434
+    },
+    {
+      "epoch": 1.0504636618503342,
+      "grad_norm": 0.1658577173948288,
+      "learning_rate": 0.00011003077817573198,
+      "loss": 2.0776,
+      "step": 2435
+    },
+    {
+      "epoch": 1.0508949751994825,
+      "grad_norm": 0.1889333873987198,
+      "learning_rate": 0.0001100005980387891,
+      "loss": 2.2737,
+      "step": 2436
+    },
+    {
+      "epoch": 1.0513262885486305,
+      "grad_norm": 0.17015792429447174,
+      "learning_rate": 0.00010997041065458963,
+      "loss": 2.1531,
+      "step": 2437
+    },
+    {
+      "epoch": 1.0517576018977788,
+      "grad_norm": 0.1794874370098114,
+      "learning_rate": 0.00010994021602938421,
+      "loss": 2.1614,
+      "step": 2438
+    },
+    {
+      "epoch": 1.0521889152469268,
+      "grad_norm": 0.17228356003761292,
+      "learning_rate": 0.00010991001416942495,
+      "loss": 2.3999,
+      "step": 2439
+    },
+    {
+      "epoch": 1.052620228596075,
+      "grad_norm": 0.18232309818267822,
+      "learning_rate": 0.0001098798050809655,
+      "loss": 2.0694,
+      "step": 2440
+    },
+    {
+      "epoch": 1.053051541945223,
+      "grad_norm": 0.16556480526924133,
+      "learning_rate": 0.00010984958877026095,
+      "loss": 2.186,
+      "step": 2441
+    },
+    {
+      "epoch": 1.0534828552943714,
+      "grad_norm": 0.16921713948249817,
+      "learning_rate": 0.00010981936524356795,
+      "loss": 1.9789,
+      "step": 2442
+    },
+    {
+      "epoch": 1.0539141686435196,
+      "grad_norm": 0.18567490577697754,
+      "learning_rate": 0.00010978913450714459,
+      "loss": 2.1184,
+      "step": 2443
+    },
+    {
+      "epoch": 1.0543454819926676,
+      "grad_norm": 0.16858486831188202,
+      "learning_rate": 0.00010975889656725051,
+      "loss": 2.1877,
+      "step": 2444
+    },
+    {
+      "epoch": 1.054776795341816,
+      "grad_norm": 0.19666415452957153,
+      "learning_rate": 0.00010972865143014674,
+      "loss": 2.329,
+      "step": 2445
+    },
+    {
+      "epoch": 1.055208108690964,
+      "grad_norm": 0.16206754744052887,
+      "learning_rate": 0.00010969839910209593,
+      "loss": 2.1894,
+      "step": 2446
+    },
+    {
+      "epoch": 1.0556394220401122,
+      "grad_norm": 0.17717215418815613,
+      "learning_rate": 0.00010966813958936213,
+      "loss": 1.9643,
+      "step": 2447
+    },
+    {
+      "epoch": 1.0560707353892602,
+      "grad_norm": 0.18330061435699463,
+      "learning_rate": 0.0001096378728982109,
+      "loss": 2.1257,
+      "step": 2448
+    },
+    {
+      "epoch": 1.0565020487384085,
+      "grad_norm": 0.18426138162612915,
+      "learning_rate": 0.00010960759903490928,
+      "loss": 2.2492,
+      "step": 2449
+    },
+    {
+      "epoch": 1.0569333620875565,
+      "grad_norm": 0.16594518721103668,
+      "learning_rate": 0.00010957731800572581,
+      "loss": 2.3716,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0569333620875565,
+      "eval_loss": 2.1055049896240234,
+      "eval_runtime": 200.9772,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0573646754367048,
+      "grad_norm": 0.1678372025489807,
+      "learning_rate": 0.00010954702981693055,
+      "loss": 2.1933,
+      "step": 2451
+    },
+    {
+      "epoch": 1.057795988785853,
+      "grad_norm": 0.17151105403900146,
+      "learning_rate": 0.00010951673447479495,
+      "loss": 2.1858,
+      "step": 2452
+    },
+    {
+      "epoch": 1.058227302135001,
+      "grad_norm": 0.17021241784095764,
+      "learning_rate": 0.00010948643198559201,
+      "loss": 2.3021,
+      "step": 2453
+    },
+    {
+      "epoch": 1.0586586154841493,
+      "grad_norm": 0.19027872383594513,
+      "learning_rate": 0.00010945612235559619,
+      "loss": 2.1876,
+      "step": 2454
+    },
+    {
+      "epoch": 1.0590899288332973,
+      "grad_norm": 0.1739296168088913,
+      "learning_rate": 0.00010942580559108344,
+      "loss": 2.2537,
+      "step": 2455
+    },
+    {
+      "epoch": 1.0595212421824456,
+      "grad_norm": 0.17704711854457855,
+      "learning_rate": 0.00010939548169833116,
+      "loss": 2.376,
+      "step": 2456
+    },
+    {
+      "epoch": 1.0599525555315936,
+      "grad_norm": 0.174795001745224,
+      "learning_rate": 0.00010936515068361822,
+      "loss": 2.1534,
+      "step": 2457
+    },
+    {
+      "epoch": 1.0603838688807419,
+      "grad_norm": 0.16426482796669006,
+      "learning_rate": 0.00010933481255322507,
+      "loss": 2.0385,
+      "step": 2458
+    },
+    {
+      "epoch": 1.06081518222989,
+      "grad_norm": 0.1784992516040802,
+      "learning_rate": 0.00010930446731343347,
+      "loss": 2.2563,
+      "step": 2459
+    },
+    {
+      "epoch": 1.0612464955790382,
+      "grad_norm": 0.17638806998729706,
+      "learning_rate": 0.00010927411497052679,
+      "loss": 2.145,
+      "step": 2460
+    },
+    {
+      "epoch": 1.0616778089281864,
+      "grad_norm": 0.17074482142925262,
+      "learning_rate": 0.00010924375553078977,
+      "loss": 2.2673,
+      "step": 2461
+    },
+    {
+      "epoch": 1.0621091222773344,
+      "grad_norm": 0.16658513247966766,
+      "learning_rate": 0.00010921338900050868,
+      "loss": 2.0698,
+      "step": 2462
+    },
+    {
+      "epoch": 1.0625404356264827,
+      "grad_norm": 0.16458797454833984,
+      "learning_rate": 0.00010918301538597127,
+      "loss": 2.1205,
+      "step": 2463
+    },
+    {
+      "epoch": 1.0629717489756307,
+      "grad_norm": 0.1653198003768921,
+      "learning_rate": 0.00010915263469346669,
+      "loss": 2.2109,
+      "step": 2464
+    },
+    {
+      "epoch": 1.063403062324779,
+      "grad_norm": 0.17059695720672607,
+      "learning_rate": 0.00010912224692928562,
+      "loss": 2.3215,
+      "step": 2465
+    },
+    {
+      "epoch": 1.063834375673927,
+      "grad_norm": 0.17046011984348297,
+      "learning_rate": 0.00010909185209972015,
+      "loss": 1.9422,
+      "step": 2466
+    },
+    {
+      "epoch": 1.0642656890230753,
+      "grad_norm": 0.16733916103839874,
+      "learning_rate": 0.00010906145021106391,
+      "loss": 2.3387,
+      "step": 2467
+    },
+    {
+      "epoch": 1.0646970023722235,
+      "grad_norm": 0.19165734946727753,
+      "learning_rate": 0.00010903104126961188,
+      "loss": 2.4835,
+      "step": 2468
+    },
+    {
+      "epoch": 1.0651283157213716,
+      "grad_norm": 0.17968103289604187,
+      "learning_rate": 0.0001090006252816606,
+      "loss": 2.2243,
+      "step": 2469
+    },
+    {
+      "epoch": 1.0655596290705198,
+      "grad_norm": 0.17020921409130096,
+      "learning_rate": 0.00010897020225350806,
+      "loss": 2.2879,
+      "step": 2470
+    },
+    {
+      "epoch": 1.0659909424196679,
+      "grad_norm": 0.17306166887283325,
+      "learning_rate": 0.00010893977219145362,
+      "loss": 2.3269,
+      "step": 2471
+    },
+    {
+      "epoch": 1.066422255768816,
+      "grad_norm": 0.18344846367835999,
+      "learning_rate": 0.00010890933510179822,
+      "loss": 2.2255,
+      "step": 2472
+    },
+    {
+      "epoch": 1.0668535691179641,
+      "grad_norm": 0.1960878223180771,
+      "learning_rate": 0.00010887889099084413,
+      "loss": 2.3315,
+      "step": 2473
+    },
+    {
+      "epoch": 1.0672848824671124,
+      "grad_norm": 0.16954079270362854,
+      "learning_rate": 0.00010884843986489518,
+      "loss": 2.1443,
+      "step": 2474
+    },
+    {
+      "epoch": 1.0677161958162604,
+      "grad_norm": 0.16392044723033905,
+      "learning_rate": 0.0001088179817302566,
+      "loss": 2.136,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0677161958162604,
+      "eval_loss": 2.1050617694854736,
+      "eval_runtime": 200.7502,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0681475091654087,
+      "grad_norm": 0.15941034257411957,
+      "learning_rate": 0.00010878751659323507,
+      "loss": 2.2209,
+      "step": 2476
+    },
+    {
+      "epoch": 1.068578822514557,
+      "grad_norm": 0.1872348040342331,
+      "learning_rate": 0.00010875704446013875,
+      "loss": 2.2573,
+      "step": 2477
+    },
+    {
+      "epoch": 1.069010135863705,
+      "grad_norm": 0.18529799580574036,
+      "learning_rate": 0.00010872656533727721,
+      "loss": 2.2827,
+      "step": 2478
+    },
+    {
+      "epoch": 1.0694414492128532,
+      "grad_norm": 0.1721513569355011,
+      "learning_rate": 0.00010869607923096148,
+      "loss": 2.0848,
+      "step": 2479
+    },
+    {
+      "epoch": 1.0698727625620013,
+      "grad_norm": 0.16441528499126434,
+      "learning_rate": 0.00010866558614750403,
+      "loss": 2.2571,
+      "step": 2480
+    },
+    {
+      "epoch": 1.0703040759111495,
+      "grad_norm": 0.1793617308139801,
+      "learning_rate": 0.00010863508609321885,
+      "loss": 2.1694,
+      "step": 2481
+    },
+    {
+      "epoch": 1.0707353892602975,
+      "grad_norm": 0.16509655117988586,
+      "learning_rate": 0.00010860457907442123,
+      "loss": 2.0505,
+      "step": 2482
+    },
+    {
+      "epoch": 1.0711667026094458,
+      "grad_norm": 0.19956058263778687,
+      "learning_rate": 0.00010857406509742807,
+      "loss": 2.2348,
+      "step": 2483
+    },
+    {
+      "epoch": 1.0715980159585938,
+      "grad_norm": 0.15869128704071045,
+      "learning_rate": 0.00010854354416855751,
+      "loss": 2.2292,
+      "step": 2484
+    },
+    {
+      "epoch": 1.072029329307742,
+      "grad_norm": 0.16309665143489838,
+      "learning_rate": 0.00010851301629412933,
+      "loss": 2.0799,
+      "step": 2485
+    },
+    {
+      "epoch": 1.0724606426568903,
+      "grad_norm": 0.1662570834159851,
+      "learning_rate": 0.00010848248148046464,
+      "loss": 2.0386,
+      "step": 2486
+    },
+    {
+      "epoch": 1.0728919560060384,
+      "grad_norm": 0.1775583028793335,
+      "learning_rate": 0.000108451939733886,
+      "loss": 2.3047,
+      "step": 2487
+    },
+    {
+      "epoch": 1.0733232693551866,
+      "grad_norm": 0.1622440665960312,
+      "learning_rate": 0.00010842139106071742,
+      "loss": 2.2266,
+      "step": 2488
+    },
+    {
+      "epoch": 1.0737545827043347,
+      "grad_norm": 0.18213559687137604,
+      "learning_rate": 0.00010839083546728431,
+      "loss": 2.0705,
+      "step": 2489
+    },
+    {
+      "epoch": 1.074185896053483,
+      "grad_norm": 0.179693803191185,
+      "learning_rate": 0.00010836027295991358,
+      "loss": 2.1326,
+      "step": 2490
+    },
+    {
+      "epoch": 1.074617209402631,
+      "grad_norm": 0.1911303848028183,
+      "learning_rate": 0.0001083297035449335,
+      "loss": 2.3379,
+      "step": 2491
+    },
+    {
+      "epoch": 1.0750485227517792,
+      "grad_norm": 0.16474899649620056,
+      "learning_rate": 0.00010829912722867383,
+      "loss": 1.9679,
+      "step": 2492
+    },
+    {
+      "epoch": 1.0754798361009272,
+      "grad_norm": 0.15132348239421844,
+      "learning_rate": 0.0001082685440174657,
+      "loss": 2.0117,
+      "step": 2493
+    },
+    {
+      "epoch": 1.0759111494500755,
+      "grad_norm": 0.21260079741477966,
+      "learning_rate": 0.00010823795391764172,
+      "loss": 2.2502,
+      "step": 2494
+    },
+    {
+      "epoch": 1.0763424627992237,
+      "grad_norm": 0.1629190742969513,
+      "learning_rate": 0.00010820735693553589,
+      "loss": 2.3575,
+      "step": 2495
+    },
+    {
+      "epoch": 1.0767737761483718,
+      "grad_norm": 0.21407712996006012,
+      "learning_rate": 0.00010817675307748365,
+      "loss": 2.1997,
+      "step": 2496
+    },
+    {
+      "epoch": 1.07720508949752,
+      "grad_norm": 0.18867595493793488,
+      "learning_rate": 0.00010814614234982189,
+      "loss": 2.1327,
+      "step": 2497
+    },
+    {
+      "epoch": 1.077636402846668,
+      "grad_norm": 0.1845402717590332,
+      "learning_rate": 0.00010811552475888885,
+      "loss": 2.0834,
+      "step": 2498
+    },
+    {
+      "epoch": 1.0780677161958163,
+      "grad_norm": 0.16826646029949188,
+      "learning_rate": 0.0001080849003110243,
+      "loss": 2.1984,
+      "step": 2499
+    },
+    {
+      "epoch": 1.0784990295449643,
+      "grad_norm": 0.15099520981311798,
+      "learning_rate": 0.00010805426901256929,
+      "loss": 1.981,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0784990295449643,
+      "eval_loss": 2.1044869422912598,
+      "eval_runtime": 200.3953,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0789303428941126,
+      "grad_norm": 0.16792334616184235,
+      "learning_rate": 0.00010802363086986643,
+      "loss": 2.3156,
+      "step": 2501
+    },
+    {
+      "epoch": 1.0793616562432606,
+      "grad_norm": 0.1842220574617386,
+      "learning_rate": 0.00010799298588925967,
+      "loss": 2.2968,
+      "step": 2502
+    },
+    {
+      "epoch": 1.079792969592409,
+      "grad_norm": 0.43980124592781067,
+      "learning_rate": 0.00010796233407709434,
+      "loss": 2.232,
+      "step": 2503
+    },
+    {
+      "epoch": 1.0802242829415571,
+      "grad_norm": 0.20864452421665192,
+      "learning_rate": 0.00010793167543971731,
+      "loss": 1.8391,
+      "step": 2504
+    },
+    {
+      "epoch": 1.0806555962907052,
+      "grad_norm": 0.16943299770355225,
+      "learning_rate": 0.00010790100998347671,
+      "loss": 2.2827,
+      "step": 2505
+    },
+    {
+      "epoch": 1.0810869096398534,
+      "grad_norm": 0.19129234552383423,
+      "learning_rate": 0.00010787033771472222,
+      "loss": 2.3839,
+      "step": 2506
+    },
+    {
+      "epoch": 1.0815182229890015,
+      "grad_norm": 0.1737593412399292,
+      "learning_rate": 0.00010783965863980483,
+      "loss": 2.0071,
+      "step": 2507
+    },
+    {
+      "epoch": 1.0819495363381497,
+      "grad_norm": 0.16698093712329865,
+      "learning_rate": 0.00010780897276507698,
+      "loss": 2.1422,
+      "step": 2508
+    },
+    {
+      "epoch": 1.0823808496872978,
+      "grad_norm": 0.16467595100402832,
+      "learning_rate": 0.00010777828009689255,
+      "loss": 2.1051,
+      "step": 2509
+    },
+    {
+      "epoch": 1.082812163036446,
+      "grad_norm": 0.18091897666454315,
+      "learning_rate": 0.00010774758064160676,
+      "loss": 2.3142,
+      "step": 2510
+    },
+    {
+      "epoch": 1.083243476385594,
+      "grad_norm": 0.18090017139911652,
+      "learning_rate": 0.00010771687440557628,
+      "loss": 2.2106,
+      "step": 2511
+    },
+    {
+      "epoch": 1.0836747897347423,
+      "grad_norm": 0.16148926317691803,
+      "learning_rate": 0.00010768616139515913,
+      "loss": 2.058,
+      "step": 2512
+    },
+    {
+      "epoch": 1.0841061030838905,
+      "grad_norm": 0.17057783901691437,
+      "learning_rate": 0.00010765544161671486,
+      "loss": 2.2992,
+      "step": 2513
+    },
+    {
+      "epoch": 1.0845374164330386,
+      "grad_norm": 0.1756206899881363,
+      "learning_rate": 0.00010762471507660426,
+      "loss": 2.1542,
+      "step": 2514
+    },
+    {
+      "epoch": 1.0849687297821868,
+      "grad_norm": 0.17915412783622742,
+      "learning_rate": 0.00010759398178118964,
+      "loss": 2.0949,
+      "step": 2515
+    },
+    {
+      "epoch": 1.0854000431313349,
+      "grad_norm": 0.1853836476802826,
+      "learning_rate": 0.00010756324173683465,
+      "loss": 2.1546,
+      "step": 2516
+    },
+    {
+      "epoch": 1.0858313564804831,
+      "grad_norm": 0.19392795860767365,
+      "learning_rate": 0.00010753249494990436,
+      "loss": 2.4471,
+      "step": 2517
+    },
+    {
+      "epoch": 1.0862626698296312,
+      "grad_norm": 0.1694418042898178,
+      "learning_rate": 0.00010750174142676521,
+      "loss": 2.137,
+      "step": 2518
+    },
+    {
+      "epoch": 1.0866939831787794,
+      "grad_norm": 0.17123928666114807,
+      "learning_rate": 0.00010747098117378507,
+      "loss": 2.26,
+      "step": 2519
+    },
+    {
+      "epoch": 1.0871252965279274,
+      "grad_norm": 0.18081289529800415,
+      "learning_rate": 0.00010744021419733318,
+      "loss": 2.1713,
+      "step": 2520
+    },
+    {
+      "epoch": 1.0875566098770757,
+      "grad_norm": 0.16639788448810577,
+      "learning_rate": 0.00010740944050378018,
+      "loss": 2.233,
+      "step": 2521
+    },
+    {
+      "epoch": 1.087987923226224,
+      "grad_norm": 0.1688801646232605,
+      "learning_rate": 0.00010737866009949811,
+      "loss": 2.1852,
+      "step": 2522
+    },
+    {
+      "epoch": 1.088419236575372,
+      "grad_norm": 0.1901995688676834,
+      "learning_rate": 0.00010734787299086036,
+      "loss": 2.1676,
+      "step": 2523
+    },
+    {
+      "epoch": 1.0888505499245202,
+      "grad_norm": 0.19242532551288605,
+      "learning_rate": 0.00010731707918424177,
+      "loss": 2.3821,
+      "step": 2524
+    },
+    {
+      "epoch": 1.0892818632736683,
+      "grad_norm": 0.17025871574878693,
+      "learning_rate": 0.00010728627868601852,
+      "loss": 2.0999,
+      "step": 2525
+    },
+    {
+      "epoch": 1.0892818632736683,
+      "eval_loss": 2.1046197414398193,
+      "eval_runtime": 200.6002,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 2525
+    },
+    {
+      "epoch": 1.0897131766228165,
+      "grad_norm": 0.17762216925621033,
+      "learning_rate": 0.00010725547150256818,
+      "loss": 2.4327,
+      "step": 2526
+    },
+    {
+      "epoch": 1.0901444899719646,
+      "grad_norm": 0.16256682574748993,
+      "learning_rate": 0.00010722465764026974,
+      "loss": 2.1386,
+      "step": 2527
+    },
+    {
+      "epoch": 1.0905758033211128,
+      "grad_norm": 0.16706998646259308,
+      "learning_rate": 0.00010719383710550351,
+      "loss": 2.1637,
+      "step": 2528
+    },
+    {
+      "epoch": 1.0910071166702608,
+      "grad_norm": 0.2084912806749344,
+      "learning_rate": 0.00010716300990465128,
+      "loss": 2.2047,
+      "step": 2529
+    },
+    {
+      "epoch": 1.091438430019409,
+      "grad_norm": 0.17731738090515137,
+      "learning_rate": 0.00010713217604409609,
+      "loss": 2.0935,
+      "step": 2530
+    },
+    {
+      "epoch": 1.0918697433685574,
+      "grad_norm": 0.16511079668998718,
+      "learning_rate": 0.00010710133553022247,
+      "loss": 2.2436,
+      "step": 2531
+    },
+    {
+      "epoch": 1.0923010567177054,
+      "grad_norm": 0.17809303104877472,
+      "learning_rate": 0.00010707048836941627,
+      "loss": 2.3086,
+      "step": 2532
+    },
+    {
+      "epoch": 1.0927323700668536,
+      "grad_norm": 35.525386810302734,
+      "learning_rate": 0.00010703963456806475,
+      "loss": 1.9948,
+      "step": 2533
+    },
+    {
+      "epoch": 1.0931636834160017,
+      "grad_norm": 0.197676882147789,
+      "learning_rate": 0.0001070087741325565,
+      "loss": 2.2399,
+      "step": 2534
+    },
+    {
+      "epoch": 1.09359499676515,
+      "grad_norm": 0.19233070313930511,
+      "learning_rate": 0.0001069779070692815,
+      "loss": 2.1493,
+      "step": 2535
+    },
+    {
+      "epoch": 1.094026310114298,
+      "grad_norm": 0.17287050187587738,
+      "learning_rate": 0.00010694703338463118,
+      "loss": 2.1806,
+      "step": 2536
+    },
+    {
+      "epoch": 1.0944576234634462,
+      "grad_norm": 0.1617845892906189,
+      "learning_rate": 0.00010691615308499821,
+      "loss": 2.0614,
+      "step": 2537
+    },
+    {
+      "epoch": 1.0948889368125942,
+      "grad_norm": 0.16862034797668457,
+      "learning_rate": 0.00010688526617677671,
+      "loss": 2.0655,
+      "step": 2538
+    },
+    {
+      "epoch": 1.0953202501617425,
+      "grad_norm": 0.16461607813835144,
+      "learning_rate": 0.00010685437266636217,
+      "loss": 2.1918,
+      "step": 2539
+    },
+    {
+      "epoch": 1.0957515635108908,
+      "grad_norm": 0.16629716753959656,
+      "learning_rate": 0.00010682347256015141,
+      "loss": 2.1462,
+      "step": 2540
+    },
+    {
+      "epoch": 1.0961828768600388,
+      "grad_norm": 0.17277218401432037,
+      "learning_rate": 0.00010679256586454265,
+      "loss": 2.2636,
+      "step": 2541
+    },
+    {
+      "epoch": 1.096614190209187,
+      "grad_norm": 0.18575602769851685,
+      "learning_rate": 0.00010676165258593542,
+      "loss": 2.169,
+      "step": 2542
+    },
+    {
+      "epoch": 1.097045503558335,
+      "grad_norm": 0.17312225699424744,
+      "learning_rate": 0.00010673073273073072,
+      "loss": 2.1672,
+      "step": 2543
+    },
+    {
+      "epoch": 1.0974768169074833,
+      "grad_norm": 0.17106004059314728,
+      "learning_rate": 0.00010669980630533077,
+      "loss": 2.268,
+      "step": 2544
+    },
+    {
+      "epoch": 1.0979081302566314,
+      "grad_norm": 0.17834289371967316,
+      "learning_rate": 0.00010666887331613931,
+      "loss": 2.1809,
+      "step": 2545
+    },
+    {
+      "epoch": 1.0983394436057796,
+      "grad_norm": 0.17766167223453522,
+      "learning_rate": 0.00010663793376956126,
+      "loss": 2.3147,
+      "step": 2546
+    },
+    {
+      "epoch": 1.0987707569549277,
+      "grad_norm": 0.16469496488571167,
+      "learning_rate": 0.00010660698767200306,
+      "loss": 2.1159,
+      "step": 2547
+    },
+    {
+      "epoch": 1.099202070304076,
+      "grad_norm": 0.20470421016216278,
+      "learning_rate": 0.0001065760350298724,
+      "loss": 2.386,
+      "step": 2548
+    },
+    {
+      "epoch": 1.0996333836532242,
+      "grad_norm": 0.187876895070076,
+      "learning_rate": 0.00010654507584957837,
+      "loss": 2.3584,
+      "step": 2549
+    },
+    {
+      "epoch": 1.1000646970023722,
+      "grad_norm": 0.1793707013130188,
+      "learning_rate": 0.00010651411013753145,
+      "loss": 2.2252,
+      "step": 2550
+    },
+    {
+      "epoch": 1.1000646970023722,
+      "eval_loss": 2.103973865509033,
+      "eval_runtime": 196.7781,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 2550
+    },
+    {
+      "epoch": 1.1004960103515204,
+      "grad_norm": 10.771757125854492,
+      "learning_rate": 0.00010648313790014336,
+      "loss": 2.1051,
+      "step": 2551
+    },
+    {
+      "epoch": 1.1009273237006685,
+      "grad_norm": 0.20772556960582733,
+      "learning_rate": 0.0001064521591438273,
+      "loss": 2.2376,
+      "step": 2552
+    },
+    {
+      "epoch": 1.1013586370498167,
+      "grad_norm": 0.2005288153886795,
+      "learning_rate": 0.0001064211738749977,
+      "loss": 1.88,
+      "step": 2553
+    },
+    {
+      "epoch": 1.1017899503989648,
+      "grad_norm": 0.17820362746715546,
+      "learning_rate": 0.00010639018210007047,
+      "loss": 2.3887,
+      "step": 2554
+    },
+    {
+      "epoch": 1.102221263748113,
+      "grad_norm": 0.17765012383460999,
+      "learning_rate": 0.00010635918382546275,
+      "loss": 2.3142,
+      "step": 2555
+    },
+    {
+      "epoch": 1.102652577097261,
+      "grad_norm": 0.2070050984621048,
+      "learning_rate": 0.00010632817905759306,
+      "loss": 2.0402,
+      "step": 2556
+    },
+    {
+      "epoch": 1.1030838904464093,
+      "grad_norm": 0.1713756024837494,
+      "learning_rate": 0.0001062971678028813,
+      "loss": 2.2322,
+      "step": 2557
+    },
+    {
+      "epoch": 1.1035152037955576,
+      "grad_norm": 0.1667434275150299,
+      "learning_rate": 0.00010626615006774869,
+      "loss": 2.2085,
+      "step": 2558
+    },
+    {
+      "epoch": 1.1039465171447056,
+      "grad_norm": 0.18005375564098358,
+      "learning_rate": 0.00010623512585861783,
+      "loss": 2.2999,
+      "step": 2559
+    },
+    {
+      "epoch": 1.1043778304938539,
+      "grad_norm": 0.18545310199260712,
+      "learning_rate": 0.00010620409518191252,
+      "loss": 2.183,
+      "step": 2560
+    },
+    {
+      "epoch": 1.1048091438430019,
+      "grad_norm": 0.1672646850347519,
+      "learning_rate": 0.0001061730580440581,
+      "loss": 2.1464,
+      "step": 2561
+    },
+    {
+      "epoch": 1.1052404571921501,
+      "grad_norm": 0.17547766864299774,
+      "learning_rate": 0.00010614201445148108,
+      "loss": 2.4801,
+      "step": 2562
+    },
+    {
+      "epoch": 1.1056717705412982,
+      "grad_norm": 0.18772274255752563,
+      "learning_rate": 0.00010611096441060942,
+      "loss": 2.2,
+      "step": 2563
+    },
+    {
+      "epoch": 1.1061030838904464,
+      "grad_norm": 0.19832868874073029,
+      "learning_rate": 0.00010607990792787236,
+      "loss": 2.1522,
+      "step": 2564
+    },
+    {
+      "epoch": 1.1065343972395945,
+      "grad_norm": 0.18317139148712158,
+      "learning_rate": 0.00010604884500970045,
+      "loss": 2.3309,
+      "step": 2565
+    },
+    {
+      "epoch": 1.1069657105887427,
+      "grad_norm": 0.1834000051021576,
+      "learning_rate": 0.00010601777566252568,
+      "loss": 2.2633,
+      "step": 2566
+    },
+    {
+      "epoch": 1.107397023937891,
+      "grad_norm": 0.19017237424850464,
+      "learning_rate": 0.00010598669989278121,
+      "loss": 2.2537,
+      "step": 2567
+    },
+    {
+      "epoch": 1.107828337287039,
+      "grad_norm": 0.18654869496822357,
+      "learning_rate": 0.00010595561770690171,
+      "loss": 2.1133,
+      "step": 2568
+    },
+    {
+      "epoch": 1.1082596506361873,
+      "grad_norm": 0.1633051633834839,
+      "learning_rate": 0.000105924529111323,
+      "loss": 2.2248,
+      "step": 2569
+    },
+    {
+      "epoch": 1.1086909639853353,
+      "grad_norm": 0.19110850989818573,
+      "learning_rate": 0.00010589343411248237,
+      "loss": 2.243,
+      "step": 2570
+    },
+    {
+      "epoch": 1.1091222773344835,
+      "grad_norm": 0.19738002121448517,
+      "learning_rate": 0.00010586233271681837,
+      "loss": 1.9526,
+      "step": 2571
+    },
+    {
+      "epoch": 1.1095535906836316,
+      "grad_norm": 0.1846143901348114,
+      "learning_rate": 0.00010583122493077084,
+      "loss": 2.107,
+      "step": 2572
+    },
+    {
+      "epoch": 1.1099849040327798,
+      "grad_norm": 0.1838165521621704,
+      "learning_rate": 0.00010580011076078108,
+      "loss": 2.1531,
+      "step": 2573
+    },
+    {
+      "epoch": 1.1104162173819279,
+      "grad_norm": 0.17969593405723572,
+      "learning_rate": 0.00010576899021329151,
+      "loss": 2.2307,
+      "step": 2574
+    },
+    {
+      "epoch": 1.1108475307310761,
+      "grad_norm": 0.49395978450775146,
+      "learning_rate": 0.00010573786329474607,
+      "loss": 2.063,
+      "step": 2575
+    },
+    {
+      "epoch": 1.1108475307310761,
+      "eval_loss": 2.1042685508728027,
+      "eval_runtime": 195.0821,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2575
+    },
+    {
+      "epoch": 1.1112788440802244,
+      "grad_norm": 0.17557857930660248,
+      "learning_rate": 0.00010570673001158985,
+      "loss": 1.9164,
+      "step": 2576
+    },
+    {
+      "epoch": 1.1117101574293724,
+      "grad_norm": 0.188553586602211,
+      "learning_rate": 0.00010567559037026942,
+      "loss": 2.3004,
+      "step": 2577
+    },
+    {
+      "epoch": 1.1121414707785207,
+      "grad_norm": 0.19646279513835907,
+      "learning_rate": 0.00010564444437723251,
+      "loss": 2.1013,
+      "step": 2578
+    },
+    {
+      "epoch": 1.1125727841276687,
+      "grad_norm": 0.1875109225511551,
+      "learning_rate": 0.0001056132920389283,
+      "loss": 2.4232,
+      "step": 2579
+    },
+    {
+      "epoch": 1.113004097476817,
+      "grad_norm": 0.21454770863056183,
+      "learning_rate": 0.00010558213336180716,
+      "loss": 2.2497,
+      "step": 2580
+    },
+    {
+      "epoch": 1.113435410825965,
+      "grad_norm": 0.1954226791858673,
+      "learning_rate": 0.00010555096835232089,
+      "loss": 2.1853,
+      "step": 2581
+    },
+    {
+      "epoch": 1.1138667241751132,
+      "grad_norm": 0.15640024840831757,
+      "learning_rate": 0.00010551979701692252,
+      "loss": 2.0693,
+      "step": 2582
+    },
+    {
+      "epoch": 1.1142980375242613,
+      "grad_norm": 0.17734797298908234,
+      "learning_rate": 0.0001054886193620664,
+      "loss": 2.2395,
+      "step": 2583
+    },
+    {
+      "epoch": 1.1147293508734095,
+      "grad_norm": 0.19922398030757904,
+      "learning_rate": 0.00010545743539420827,
+      "loss": 2.2398,
+      "step": 2584
+    },
+    {
+      "epoch": 1.1151606642225578,
+      "grad_norm": 0.18507863581180573,
+      "learning_rate": 0.00010542624511980499,
+      "loss": 2.2141,
+      "step": 2585
+    },
+    {
+      "epoch": 1.1155919775717058,
+      "grad_norm": 0.17060115933418274,
+      "learning_rate": 0.00010539504854531497,
+      "loss": 2.195,
+      "step": 2586
+    },
+    {
+      "epoch": 1.116023290920854,
+      "grad_norm": 0.19707359373569489,
+      "learning_rate": 0.00010536384567719777,
+      "loss": 2.1652,
+      "step": 2587
+    },
+    {
+      "epoch": 1.116454604270002,
+      "grad_norm": 0.15684722363948822,
+      "learning_rate": 0.00010533263652191422,
+      "loss": 2.0567,
+      "step": 2588
+    },
+    {
+      "epoch": 1.1168859176191503,
+      "grad_norm": 0.17799372971057892,
+      "learning_rate": 0.00010530142108592661,
+      "loss": 2.2559,
+      "step": 2589
+    },
+    {
+      "epoch": 1.1173172309682984,
+      "grad_norm": 0.16498693823814392,
+      "learning_rate": 0.00010527019937569838,
+      "loss": 2.231,
+      "step": 2590
+    },
+    {
+      "epoch": 1.1177485443174466,
+      "grad_norm": 0.1831119954586029,
+      "learning_rate": 0.00010523897139769435,
+      "loss": 2.1352,
+      "step": 2591
+    },
+    {
+      "epoch": 1.1181798576665947,
+      "grad_norm": 0.20394060015678406,
+      "learning_rate": 0.00010520773715838057,
+      "loss": 2.2722,
+      "step": 2592
+    },
+    {
+      "epoch": 1.118611171015743,
+      "grad_norm": 0.17225882411003113,
+      "learning_rate": 0.00010517649666422449,
+      "loss": 2.1723,
+      "step": 2593
+    },
+    {
+      "epoch": 1.1190424843648912,
+      "grad_norm": 0.20260009169578552,
+      "learning_rate": 0.00010514524992169477,
+      "loss": 2.3081,
+      "step": 2594
+    },
+    {
+      "epoch": 1.1194737977140392,
+      "grad_norm": 0.17399710416793823,
+      "learning_rate": 0.00010511399693726136,
+      "loss": 2.3719,
+      "step": 2595
+    },
+    {
+      "epoch": 1.1199051110631875,
+      "grad_norm": 0.18588174879550934,
+      "learning_rate": 0.00010508273771739563,
+      "loss": 2.2116,
+      "step": 2596
+    },
+    {
+      "epoch": 1.1203364244123355,
+      "grad_norm": 0.20267854630947113,
+      "learning_rate": 0.00010505147226857002,
+      "loss": 2.131,
+      "step": 2597
+    },
+    {
+      "epoch": 1.1207677377614838,
+      "grad_norm": 0.16706599295139313,
+      "learning_rate": 0.00010502020059725847,
+      "loss": 2.2558,
+      "step": 2598
+    },
+    {
+      "epoch": 1.1211990511106318,
+      "grad_norm": 0.19617630541324615,
+      "learning_rate": 0.00010498892270993606,
+      "loss": 2.1837,
+      "step": 2599
+    },
+    {
+      "epoch": 1.12163036445978,
+      "grad_norm": 0.20756325125694275,
+      "learning_rate": 0.00010495763861307929,
+      "loss": 2.1954,
+      "step": 2600
+    },
+    {
+      "epoch": 1.12163036445978,
+      "eval_loss": 2.1039798259735107,
+      "eval_runtime": 195.4933,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2600
+    },
+    {
+      "epoch": 1.122061677808928,
+      "grad_norm": 0.20020046830177307,
+      "learning_rate": 0.0001049263483131658,
+      "loss": 2.2906,
+      "step": 2601
+    },
+    {
+      "epoch": 1.1224929911580763,
+      "grad_norm": 0.2032448649406433,
+      "learning_rate": 0.00010489505181667465,
+      "loss": 2.2178,
+      "step": 2602
+    },
+    {
+      "epoch": 1.1229243045072246,
+      "grad_norm": 0.18019424378871918,
+      "learning_rate": 0.00010486374913008606,
+      "loss": 2.177,
+      "step": 2603
+    },
+    {
+      "epoch": 1.1233556178563726,
+      "grad_norm": 0.16137458384037018,
+      "learning_rate": 0.00010483244025988164,
+      "loss": 2.2681,
+      "step": 2604
+    },
+    {
+      "epoch": 1.1237869312055209,
+      "grad_norm": 0.18528112769126892,
+      "learning_rate": 0.00010480112521254425,
+      "loss": 2.3586,
+      "step": 2605
+    },
+    {
+      "epoch": 1.124218244554669,
+      "grad_norm": 0.1861640065908432,
+      "learning_rate": 0.00010476980399455795,
+      "loss": 2.0436,
+      "step": 2606
+    },
+    {
+      "epoch": 1.1246495579038172,
+      "grad_norm": 0.1724652796983719,
+      "learning_rate": 0.0001047384766124082,
+      "loss": 2.2836,
+      "step": 2607
+    },
+    {
+      "epoch": 1.1250808712529652,
+      "grad_norm": 0.16126412153244019,
+      "learning_rate": 0.00010470714307258164,
+      "loss": 2.2284,
+      "step": 2608
+    },
+    {
+      "epoch": 1.1255121846021134,
+      "grad_norm": 0.17227625846862793,
+      "learning_rate": 0.00010467580338156624,
+      "loss": 2.1648,
+      "step": 2609
+    },
+    {
+      "epoch": 1.1259434979512615,
+      "grad_norm": 0.17861312627792358,
+      "learning_rate": 0.00010464445754585122,
+      "loss": 2.1608,
+      "step": 2610
+    },
+    {
+      "epoch": 1.1263748113004097,
+      "grad_norm": 0.1870555728673935,
+      "learning_rate": 0.00010461310557192706,
+      "loss": 2.1918,
+      "step": 2611
+    },
+    {
+      "epoch": 1.126806124649558,
+      "grad_norm": 0.16503466665744781,
+      "learning_rate": 0.00010458174746628558,
+      "loss": 2.024,
+      "step": 2612
+    },
+    {
+      "epoch": 1.127237437998706,
+      "grad_norm": 0.1710997223854065,
+      "learning_rate": 0.00010455038323541976,
+      "loss": 2.112,
+      "step": 2613
+    },
+    {
+      "epoch": 1.1276687513478543,
+      "grad_norm": 0.1710243970155716,
+      "learning_rate": 0.00010451901288582397,
+      "loss": 2.3523,
+      "step": 2614
+    },
+    {
+      "epoch": 1.1281000646970023,
+      "grad_norm": 0.17720620334148407,
+      "learning_rate": 0.00010448763642399371,
+      "loss": 2.1211,
+      "step": 2615
+    },
+    {
+      "epoch": 1.1285313780461506,
+      "grad_norm": 0.17885632812976837,
+      "learning_rate": 0.00010445625385642586,
+      "loss": 2.2591,
+      "step": 2616
+    },
+    {
+      "epoch": 1.1289626913952986,
+      "grad_norm": 0.17788010835647583,
+      "learning_rate": 0.00010442486518961855,
+      "loss": 2.0312,
+      "step": 2617
+    },
+    {
+      "epoch": 1.1293940047444468,
+      "grad_norm": 0.15931367874145508,
+      "learning_rate": 0.0001043934704300711,
+      "loss": 2.4051,
+      "step": 2618
+    },
+    {
+      "epoch": 1.1298253180935949,
+      "grad_norm": 0.20527784526348114,
+      "learning_rate": 0.00010436206958428415,
+      "loss": 2.183,
+      "step": 2619
+    },
+    {
+      "epoch": 1.1302566314427431,
+      "grad_norm": 0.19679832458496094,
+      "learning_rate": 0.00010433066265875958,
+      "loss": 2.0836,
+      "step": 2620
+    },
+    {
+      "epoch": 1.1306879447918914,
+      "grad_norm": 0.1598224639892578,
+      "learning_rate": 0.0001042992496600006,
+      "loss": 2.1045,
+      "step": 2621
+    },
+    {
+      "epoch": 1.1311192581410394,
+      "grad_norm": 0.16047848761081696,
+      "learning_rate": 0.00010426783059451151,
+      "loss": 2.2245,
+      "step": 2622
+    },
+    {
+      "epoch": 1.1315505714901877,
+      "grad_norm": 0.1957550346851349,
+      "learning_rate": 0.00010423640546879807,
+      "loss": 2.3832,
+      "step": 2623
+    },
+    {
+      "epoch": 1.1319818848393357,
+      "grad_norm": 0.1612965166568756,
+      "learning_rate": 0.00010420497428936713,
+      "loss": 2.1053,
+      "step": 2624
+    },
+    {
+      "epoch": 1.132413198188484,
+      "grad_norm": 0.16824392974376678,
+      "learning_rate": 0.0001041735370627269,
+      "loss": 2.3027,
+      "step": 2625
+    },
+    {
+      "epoch": 1.132413198188484,
+      "eval_loss": 2.1031532287597656,
+      "eval_runtime": 195.1556,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2625
+    },
+    {
+      "epoch": 1.132844511537632,
+      "grad_norm": 0.17558202147483826,
+      "learning_rate": 0.00010414209379538679,
+      "loss": 2.1126,
+      "step": 2626
+    },
+    {
+      "epoch": 1.1332758248867802,
+      "grad_norm": 0.17710520327091217,
+      "learning_rate": 0.00010411064449385743,
+      "loss": 2.2288,
+      "step": 2627
+    },
+    {
+      "epoch": 1.1337071382359283,
+      "grad_norm": 0.16382630169391632,
+      "learning_rate": 0.00010407918916465083,
+      "loss": 2.2314,
+      "step": 2628
+    },
+    {
+      "epoch": 1.1341384515850765,
+      "grad_norm": 0.1876552700996399,
+      "learning_rate": 0.00010404772781428006,
+      "loss": 2.3205,
+      "step": 2629
+    },
+    {
+      "epoch": 1.1345697649342248,
+      "grad_norm": 0.1882568895816803,
+      "learning_rate": 0.00010401626044925963,
+      "loss": 2.2529,
+      "step": 2630
+    },
+    {
+      "epoch": 1.1350010782833728,
+      "grad_norm": 0.18390049040317535,
+      "learning_rate": 0.00010398478707610511,
+      "loss": 2.0675,
+      "step": 2631
+    },
+    {
+      "epoch": 1.135432391632521,
+      "grad_norm": 0.17786632478237152,
+      "learning_rate": 0.00010395330770133349,
+      "loss": 2.1582,
+      "step": 2632
+    },
+    {
+      "epoch": 1.135863704981669,
+      "grad_norm": 0.5382328033447266,
+      "learning_rate": 0.00010392182233146286,
+      "loss": 1.9208,
+      "step": 2633
+    },
+    {
+      "epoch": 1.1362950183308174,
+      "grad_norm": 0.18467487394809723,
+      "learning_rate": 0.00010389033097301264,
+      "loss": 2.1614,
+      "step": 2634
+    },
+    {
+      "epoch": 1.1367263316799656,
+      "grad_norm": 0.1702251136302948,
+      "learning_rate": 0.00010385883363250348,
+      "loss": 2.2288,
+      "step": 2635
+    },
+    {
+      "epoch": 1.1371576450291137,
+      "grad_norm": 0.17580457031726837,
+      "learning_rate": 0.00010382733031645718,
+      "loss": 2.0447,
+      "step": 2636
+    },
+    {
+      "epoch": 1.1375889583782617,
+      "grad_norm": 0.1831868290901184,
+      "learning_rate": 0.00010379582103139691,
+      "loss": 2.4478,
+      "step": 2637
+    },
+    {
+      "epoch": 1.13802027172741,
+      "grad_norm": 0.1700078547000885,
+      "learning_rate": 0.00010376430578384699,
+      "loss": 2.3375,
+      "step": 2638
+    },
+    {
+      "epoch": 1.1384515850765582,
+      "grad_norm": 0.17241442203521729,
+      "learning_rate": 0.000103732784580333,
+      "loss": 2.3493,
+      "step": 2639
+    },
+    {
+      "epoch": 1.1388828984257062,
+      "grad_norm": 0.17868413031101227,
+      "learning_rate": 0.00010370125742738173,
+      "loss": 2.3758,
+      "step": 2640
+    },
+    {
+      "epoch": 1.1393142117748545,
+      "grad_norm": 0.1892899125814438,
+      "learning_rate": 0.00010366972433152125,
+      "loss": 2.286,
+      "step": 2641
+    },
+    {
+      "epoch": 1.1397455251240025,
+      "grad_norm": 0.18568749725818634,
+      "learning_rate": 0.0001036381852992808,
+      "loss": 2.2917,
+      "step": 2642
+    },
+    {
+      "epoch": 1.1401768384731508,
+      "grad_norm": 0.1770264059305191,
+      "learning_rate": 0.00010360664033719092,
+      "loss": 2.1133,
+      "step": 2643
+    },
+    {
+      "epoch": 1.140608151822299,
+      "grad_norm": 0.17140696942806244,
+      "learning_rate": 0.00010357508945178334,
+      "loss": 2.1425,
+      "step": 2644
+    },
+    {
+      "epoch": 1.141039465171447,
+      "grad_norm": 0.179879292845726,
+      "learning_rate": 0.00010354353264959096,
+      "loss": 2.0747,
+      "step": 2645
+    },
+    {
+      "epoch": 1.1414707785205953,
+      "grad_norm": 0.18175861239433289,
+      "learning_rate": 0.00010351196993714802,
+      "loss": 2.2599,
+      "step": 2646
+    },
+    {
+      "epoch": 1.1419020918697433,
+      "grad_norm": 0.2078048437833786,
+      "learning_rate": 0.00010348040132098989,
+      "loss": 2.0717,
+      "step": 2647
+    },
+    {
+      "epoch": 1.1423334052188916,
+      "grad_norm": 0.18393027782440186,
+      "learning_rate": 0.00010344882680765322,
+      "loss": 2.179,
+      "step": 2648
+    },
+    {
+      "epoch": 1.1427647185680396,
+      "grad_norm": 0.16221146285533905,
+      "learning_rate": 0.00010341724640367585,
+      "loss": 2.234,
+      "step": 2649
+    },
+    {
+      "epoch": 1.1431960319171879,
+      "grad_norm": 0.17851437628269196,
+      "learning_rate": 0.00010338566011559683,
+      "loss": 1.992,
+      "step": 2650
+    },
+    {
+      "epoch": 1.1431960319171879,
+      "eval_loss": 2.1033437252044678,
+      "eval_runtime": 195.2141,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2650
+    },
+    {
+      "epoch": 1.143627345266336,
+      "grad_norm": 0.200714111328125,
+      "learning_rate": 0.0001033540679499565,
+      "loss": 2.3895,
+      "step": 2651
+    },
+    {
+      "epoch": 1.1440586586154842,
+      "grad_norm": 0.15753240883350372,
+      "learning_rate": 0.00010332246991329627,
+      "loss": 2.2215,
+      "step": 2652
+    },
+    {
+      "epoch": 1.1444899719646324,
+      "grad_norm": 0.17338408529758453,
+      "learning_rate": 0.00010329086601215898,
+      "loss": 2.2074,
+      "step": 2653
+    },
+    {
+      "epoch": 1.1449212853137805,
+      "grad_norm": 0.1783522218465805,
+      "learning_rate": 0.00010325925625308845,
+      "loss": 2.3323,
+      "step": 2654
+    },
+    {
+      "epoch": 1.1453525986629287,
+      "grad_norm": 0.1770525872707367,
+      "learning_rate": 0.0001032276406426299,
+      "loss": 2.0545,
+      "step": 2655
+    },
+    {
+      "epoch": 1.1457839120120767,
+      "grad_norm": 0.17786875367164612,
+      "learning_rate": 0.00010319601918732964,
+      "loss": 2.1926,
+      "step": 2656
+    },
+    {
+      "epoch": 1.146215225361225,
+      "grad_norm": 0.21019989252090454,
+      "learning_rate": 0.00010316439189373525,
+      "loss": 2.3016,
+      "step": 2657
+    },
+    {
+      "epoch": 1.146646538710373,
+      "grad_norm": 0.17054423689842224,
+      "learning_rate": 0.00010313275876839557,
+      "loss": 2.1123,
+      "step": 2658
+    },
+    {
+      "epoch": 1.1470778520595213,
+      "grad_norm": 0.16215083003044128,
+      "learning_rate": 0.00010310111981786048,
+      "loss": 2.1672,
+      "step": 2659
+    },
+    {
+      "epoch": 1.1475091654086693,
+      "grad_norm": 0.18057376146316528,
+      "learning_rate": 0.00010306947504868127,
+      "loss": 2.1375,
+      "step": 2660
+    },
+    {
+      "epoch": 1.1479404787578176,
+      "grad_norm": 0.18483436107635498,
+      "learning_rate": 0.00010303782446741026,
+      "loss": 2.1457,
+      "step": 2661
+    },
+    {
+      "epoch": 1.1483717921069658,
+      "grad_norm": 0.1700708121061325,
+      "learning_rate": 0.00010300616808060108,
+      "loss": 2.284,
+      "step": 2662
+    },
+    {
+      "epoch": 1.1488031054561139,
+      "grad_norm": 0.17855966091156006,
+      "learning_rate": 0.00010297450589480853,
+      "loss": 2.1609,
+      "step": 2663
+    },
+    {
+      "epoch": 1.1492344188052621,
+      "grad_norm": 0.16909952461719513,
+      "learning_rate": 0.0001029428379165886,
+      "loss": 2.2052,
+      "step": 2664
+    },
+    {
+      "epoch": 1.1496657321544101,
+      "grad_norm": 0.16772057116031647,
+      "learning_rate": 0.00010291116415249854,
+      "loss": 1.9577,
+      "step": 2665
+    },
+    {
+      "epoch": 1.1500970455035584,
+      "grad_norm": 0.17496979236602783,
+      "learning_rate": 0.00010287948460909667,
+      "loss": 2.1855,
+      "step": 2666
+    },
+    {
+      "epoch": 1.1505283588527064,
+      "grad_norm": 0.17731782793998718,
+      "learning_rate": 0.00010284779929294265,
+      "loss": 2.1574,
+      "step": 2667
+    },
+    {
+      "epoch": 1.1509596722018547,
+      "grad_norm": 0.1847413033246994,
+      "learning_rate": 0.00010281610821059722,
+      "loss": 2.4498,
+      "step": 2668
+    },
+    {
+      "epoch": 1.1513909855510027,
+      "grad_norm": 0.16173487901687622,
+      "learning_rate": 0.00010278441136862242,
+      "loss": 2.0799,
+      "step": 2669
+    },
+    {
+      "epoch": 1.151822298900151,
+      "grad_norm": 0.1768932342529297,
+      "learning_rate": 0.00010275270877358139,
+      "loss": 2.0765,
+      "step": 2670
+    },
+    {
+      "epoch": 1.1522536122492992,
+      "grad_norm": 0.20227771997451782,
+      "learning_rate": 0.00010272100043203851,
+      "loss": 2.1289,
+      "step": 2671
+    },
+    {
+      "epoch": 1.1526849255984473,
+      "grad_norm": 0.1585785299539566,
+      "learning_rate": 0.00010268928635055935,
+      "loss": 2.0325,
+      "step": 2672
+    },
+    {
+      "epoch": 1.1531162389475955,
+      "grad_norm": 0.17804929614067078,
+      "learning_rate": 0.00010265756653571063,
+      "loss": 2.1087,
+      "step": 2673
+    },
+    {
+      "epoch": 1.1535475522967436,
+      "grad_norm": 0.1796695739030838,
+      "learning_rate": 0.00010262584099406032,
+      "loss": 2.158,
+      "step": 2674
+    },
+    {
+      "epoch": 1.1539788656458918,
+      "grad_norm": 0.1786671131849289,
+      "learning_rate": 0.00010259410973217749,
+      "loss": 2.3043,
+      "step": 2675
+    },
+    {
+      "epoch": 1.1539788656458918,
+      "eval_loss": 2.102726697921753,
+      "eval_runtime": 195.1649,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2675
+    },
+    {
+      "epoch": 1.1544101789950398,
+      "grad_norm": 0.1836918145418167,
+      "learning_rate": 0.00010256237275663253,
+      "loss": 2.3597,
+      "step": 2676
+    },
+    {
+      "epoch": 1.154841492344188,
+      "grad_norm": 0.20627878606319427,
+      "learning_rate": 0.00010253063007399681,
+      "loss": 2.2212,
+      "step": 2677
+    },
+    {
+      "epoch": 1.1552728056933361,
+      "grad_norm": 0.16205772757530212,
+      "learning_rate": 0.0001024988816908431,
+      "loss": 2.2527,
+      "step": 2678
+    },
+    {
+      "epoch": 1.1557041190424844,
+      "grad_norm": 0.17430098354816437,
+      "learning_rate": 0.00010246712761374522,
+      "loss": 2.1506,
+      "step": 2679
+    },
+    {
+      "epoch": 1.1561354323916326,
+      "grad_norm": 0.2250378578901291,
+      "learning_rate": 0.00010243536784927817,
+      "loss": 2.1987,
+      "step": 2680
+    },
+    {
+      "epoch": 1.1565667457407807,
+      "grad_norm": 1.8225398063659668,
+      "learning_rate": 0.00010240360240401819,
+      "loss": 2.1556,
+      "step": 2681
+    },
+    {
+      "epoch": 1.156998059089929,
+      "grad_norm": 0.16364197432994843,
+      "learning_rate": 0.00010237183128454264,
+      "loss": 2.1345,
+      "step": 2682
+    },
+    {
+      "epoch": 1.157429372439077,
+      "grad_norm": 0.17830027639865875,
+      "learning_rate": 0.00010234005449743011,
+      "loss": 2.1186,
+      "step": 2683
+    },
+    {
+      "epoch": 1.1578606857882252,
+      "grad_norm": 0.17624956369400024,
+      "learning_rate": 0.00010230827204926029,
+      "loss": 2.3228,
+      "step": 2684
+    },
+    {
+      "epoch": 1.1582919991373732,
+      "grad_norm": 0.16102442145347595,
+      "learning_rate": 0.00010227648394661413,
+      "loss": 2.145,
+      "step": 2685
+    },
+    {
+      "epoch": 1.1587233124865215,
+      "grad_norm": 0.1778397262096405,
+      "learning_rate": 0.00010224469019607364,
+      "loss": 2.1796,
+      "step": 2686
+    },
+    {
+      "epoch": 1.1591546258356695,
+      "grad_norm": 0.17264363169670105,
+      "learning_rate": 0.00010221289080422212,
+      "loss": 2.0441,
+      "step": 2687
+    },
+    {
+      "epoch": 1.1595859391848178,
+      "grad_norm": 0.18140749633312225,
+      "learning_rate": 0.00010218108577764397,
+      "loss": 2.15,
+      "step": 2688
+    },
+    {
+      "epoch": 1.160017252533966,
+      "grad_norm": 0.1673058271408081,
+      "learning_rate": 0.00010214927512292473,
+      "loss": 2.1175,
+      "step": 2689
+    },
+    {
+      "epoch": 1.160448565883114,
+      "grad_norm": 0.17649729549884796,
+      "learning_rate": 0.00010211745884665124,
+      "loss": 2.3064,
+      "step": 2690
+    },
+    {
+      "epoch": 1.1608798792322623,
+      "grad_norm": 0.17438872158527374,
+      "learning_rate": 0.00010208563695541129,
+      "loss": 2.0611,
+      "step": 2691
+    },
+    {
+      "epoch": 1.1613111925814104,
+      "grad_norm": 0.17215564846992493,
+      "learning_rate": 0.00010205380945579404,
+      "loss": 2.1821,
+      "step": 2692
+    },
+    {
+      "epoch": 1.1617425059305586,
+      "grad_norm": 0.14636805653572083,
+      "learning_rate": 0.00010202197635438966,
+      "loss": 2.0414,
+      "step": 2693
+    },
+    {
+      "epoch": 1.1621738192797066,
+      "grad_norm": 0.18306148052215576,
+      "learning_rate": 0.00010199013765778958,
+      "loss": 2.1533,
+      "step": 2694
+    },
+    {
+      "epoch": 1.162605132628855,
+      "grad_norm": 0.17002162337303162,
+      "learning_rate": 0.00010195829337258634,
+      "loss": 2.0679,
+      "step": 2695
+    },
+    {
+      "epoch": 1.163036445978003,
+      "grad_norm": 0.17379575967788696,
+      "learning_rate": 0.00010192644350537362,
+      "loss": 2.2311,
+      "step": 2696
+    },
+    {
+      "epoch": 1.1634677593271512,
+      "grad_norm": 0.22427932918071747,
+      "learning_rate": 0.00010189458806274636,
+      "loss": 2.2645,
+      "step": 2697
+    },
+    {
+      "epoch": 1.1638990726762994,
+      "grad_norm": 0.18266619741916656,
+      "learning_rate": 0.00010186272705130049,
+      "loss": 2.2333,
+      "step": 2698
+    },
+    {
+      "epoch": 1.1643303860254475,
+      "grad_norm": 0.17629382014274597,
+      "learning_rate": 0.00010183086047763324,
+      "loss": 2.3974,
+      "step": 2699
+    },
+    {
+      "epoch": 1.1647616993745957,
+      "grad_norm": 0.16827429831027985,
+      "learning_rate": 0.0001017989883483429,
+      "loss": 2.1682,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1647616993745957,
+      "eval_loss": 2.1029937267303467,
+      "eval_runtime": 195.1429,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1651930127237438,
+      "grad_norm": 0.17276984453201294,
+      "learning_rate": 0.00010176711067002896,
+      "loss": 2.3793,
+      "step": 2701
+    },
+    {
+      "epoch": 1.165624326072892,
+      "grad_norm": 0.1839294582605362,
+      "learning_rate": 0.00010173522744929203,
+      "loss": 2.0978,
+      "step": 2702
+    },
+    {
+      "epoch": 1.16605563942204,
+      "grad_norm": 0.19586913287639618,
+      "learning_rate": 0.00010170333869273388,
+      "loss": 2.1481,
+      "step": 2703
+    },
+    {
+      "epoch": 1.1664869527711883,
+      "grad_norm": 0.1714116483926773,
+      "learning_rate": 0.00010167144440695744,
+      "loss": 2.1371,
+      "step": 2704
+    },
+    {
+      "epoch": 1.1669182661203363,
+      "grad_norm": 0.17770139873027802,
+      "learning_rate": 0.00010163954459856674,
+      "loss": 2.2928,
+      "step": 2705
+    },
+    {
+      "epoch": 1.1673495794694846,
+      "grad_norm": 0.16617412865161896,
+      "learning_rate": 0.00010160763927416704,
+      "loss": 2.2584,
+      "step": 2706
+    },
+    {
+      "epoch": 1.1677808928186328,
+      "grad_norm": 0.16294576227664948,
+      "learning_rate": 0.00010157572844036462,
+      "loss": 2.1495,
+      "step": 2707
+    },
+    {
+      "epoch": 1.1682122061677809,
+      "grad_norm": 0.1830517053604126,
+      "learning_rate": 0.00010154381210376703,
+      "loss": 1.9509,
+      "step": 2708
+    },
+    {
+      "epoch": 1.1686435195169291,
+      "grad_norm": 0.20103468000888824,
+      "learning_rate": 0.00010151189027098284,
+      "loss": 2.1433,
+      "step": 2709
+    },
+    {
+      "epoch": 1.1690748328660772,
+      "grad_norm": 0.18614795804023743,
+      "learning_rate": 0.00010147996294862185,
+      "loss": 2.2301,
+      "step": 2710
+    },
+    {
+      "epoch": 1.1695061462152254,
+      "grad_norm": 0.1958162486553192,
+      "learning_rate": 0.00010144803014329496,
+      "loss": 1.9723,
+      "step": 2711
+    },
+    {
+      "epoch": 1.1699374595643734,
+      "grad_norm": 0.1831211894750595,
+      "learning_rate": 0.00010141609186161417,
+      "loss": 2.4056,
+      "step": 2712
+    },
+    {
+      "epoch": 1.1703687729135217,
+      "grad_norm": 0.19260232150554657,
+      "learning_rate": 0.00010138414811019272,
+      "loss": 2.2492,
+      "step": 2713
+    },
+    {
+      "epoch": 1.1708000862626697,
+      "grad_norm": 0.17832225561141968,
+      "learning_rate": 0.00010135219889564484,
+      "loss": 2.3349,
+      "step": 2714
+    },
+    {
+      "epoch": 1.171231399611818,
+      "grad_norm": 0.19132240116596222,
+      "learning_rate": 0.00010132024422458602,
+      "loss": 2.2057,
+      "step": 2715
+    },
+    {
+      "epoch": 1.1716627129609662,
+      "grad_norm": 0.18465915322303772,
+      "learning_rate": 0.00010128828410363276,
+      "loss": 2.0808,
+      "step": 2716
+    },
+    {
+      "epoch": 1.1720940263101143,
+      "grad_norm": 0.15922652184963226,
+      "learning_rate": 0.00010125631853940283,
+      "loss": 2.1415,
+      "step": 2717
+    },
+    {
+      "epoch": 1.1725253396592625,
+      "grad_norm": 0.18575268983840942,
+      "learning_rate": 0.000101224347538515,
+      "loss": 2.287,
+      "step": 2718
+    },
+    {
+      "epoch": 1.1729566530084106,
+      "grad_norm": 0.17416925728321075,
+      "learning_rate": 0.00010119237110758921,
+      "loss": 2.0774,
+      "step": 2719
+    },
+    {
+      "epoch": 1.1733879663575588,
+      "grad_norm": 0.1689963936805725,
+      "learning_rate": 0.00010116038925324659,
+      "loss": 2.2118,
+      "step": 2720
+    },
+    {
+      "epoch": 1.1738192797067069,
+      "grad_norm": 0.18122926354408264,
+      "learning_rate": 0.00010112840198210925,
+      "loss": 2.2947,
+      "step": 2721
+    },
+    {
+      "epoch": 1.174250593055855,
+      "grad_norm": 0.17530053853988647,
+      "learning_rate": 0.00010109640930080058,
+      "loss": 2.1621,
+      "step": 2722
+    },
+    {
+      "epoch": 1.1746819064050031,
+      "grad_norm": 0.17392095923423767,
+      "learning_rate": 0.00010106441121594494,
+      "loss": 2.1327,
+      "step": 2723
+    },
+    {
+      "epoch": 1.1751132197541514,
+      "grad_norm": 0.18160422146320343,
+      "learning_rate": 0.00010103240773416795,
+      "loss": 2.2259,
+      "step": 2724
+    },
+    {
+      "epoch": 1.1755445331032996,
+      "grad_norm": 0.18598490953445435,
+      "learning_rate": 0.00010100039886209624,
+      "loss": 2.2135,
+      "step": 2725
+    },
+    {
+      "epoch": 1.1755445331032996,
+      "eval_loss": 2.1023378372192383,
+      "eval_runtime": 195.1753,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2725
+    },
+    {
+      "epoch": 1.1759758464524477,
+      "grad_norm": 0.16121797263622284,
+      "learning_rate": 0.00010096838460635761,
+      "loss": 2.1434,
+      "step": 2726
+    },
+    {
+      "epoch": 1.176407159801596,
+      "grad_norm": 0.1671607792377472,
+      "learning_rate": 0.00010093636497358096,
+      "loss": 2.0611,
+      "step": 2727
+    },
+    {
+      "epoch": 1.176838473150744,
+      "grad_norm": 0.17500080168247223,
+      "learning_rate": 0.00010090433997039631,
+      "loss": 1.9025,
+      "step": 2728
+    },
+    {
+      "epoch": 1.1772697864998922,
+      "grad_norm": 0.18734723329544067,
+      "learning_rate": 0.00010087230960343481,
+      "loss": 2.2704,
+      "step": 2729
+    },
+    {
+      "epoch": 1.1777010998490403,
+      "grad_norm": 0.17753678560256958,
+      "learning_rate": 0.00010084027387932863,
+      "loss": 2.1605,
+      "step": 2730
+    },
+    {
+      "epoch": 1.1781324131981885,
+      "grad_norm": 0.16886518895626068,
+      "learning_rate": 0.00010080823280471119,
+      "loss": 2.3079,
+      "step": 2731
+    },
+    {
+      "epoch": 1.1785637265473365,
+      "grad_norm": 0.17611104249954224,
+      "learning_rate": 0.00010077618638621688,
+      "loss": 2.2272,
+      "step": 2732
+    },
+    {
+      "epoch": 1.1789950398964848,
+      "grad_norm": 0.182244673371315,
+      "learning_rate": 0.00010074413463048132,
+      "loss": 2.2693,
+      "step": 2733
+    },
+    {
+      "epoch": 1.179426353245633,
+      "grad_norm": 0.17081747949123383,
+      "learning_rate": 0.00010071207754414116,
+      "loss": 2.1961,
+      "step": 2734
+    },
+    {
+      "epoch": 1.179857666594781,
+      "grad_norm": 0.1862158477306366,
+      "learning_rate": 0.00010068001513383413,
+      "loss": 2.3049,
+      "step": 2735
+    },
+    {
+      "epoch": 1.1802889799439293,
+      "grad_norm": 0.35091134905815125,
+      "learning_rate": 0.00010064794740619917,
+      "loss": 2.201,
+      "step": 2736
+    },
+    {
+      "epoch": 1.1807202932930774,
+      "grad_norm": 0.1901334524154663,
+      "learning_rate": 0.00010061587436787621,
+      "loss": 2.2042,
+      "step": 2737
+    },
+    {
+      "epoch": 1.1811516066422256,
+      "grad_norm": 0.1805984228849411,
+      "learning_rate": 0.00010058379602550632,
+      "loss": 2.4089,
+      "step": 2738
+    },
+    {
+      "epoch": 1.1815829199913737,
+      "grad_norm": 0.19295565783977509,
+      "learning_rate": 0.00010055171238573168,
+      "loss": 2.1412,
+      "step": 2739
+    },
+    {
+      "epoch": 1.182014233340522,
+      "grad_norm": 0.18122532963752747,
+      "learning_rate": 0.00010051962345519557,
+      "loss": 2.0987,
+      "step": 2740
+    },
+    {
+      "epoch": 1.18244554668967,
+      "grad_norm": 0.16439107060432434,
+      "learning_rate": 0.00010048752924054236,
+      "loss": 2.0193,
+      "step": 2741
+    },
+    {
+      "epoch": 1.1828768600388182,
+      "grad_norm": 0.18041017651557922,
+      "learning_rate": 0.00010045542974841747,
+      "loss": 2.3687,
+      "step": 2742
+    },
+    {
+      "epoch": 1.1833081733879665,
+      "grad_norm": 0.17560729384422302,
+      "learning_rate": 0.00010042332498546748,
+      "loss": 2.1835,
+      "step": 2743
+    },
+    {
+      "epoch": 1.1837394867371145,
+      "grad_norm": 0.1784038096666336,
+      "learning_rate": 0.00010039121495834002,
+      "loss": 2.2537,
+      "step": 2744
+    },
+    {
+      "epoch": 1.1841708000862627,
+      "grad_norm": 0.16383419930934906,
+      "learning_rate": 0.00010035909967368386,
+      "loss": 2.1551,
+      "step": 2745
+    },
+    {
+      "epoch": 1.1846021134354108,
+      "grad_norm": 0.17294475436210632,
+      "learning_rate": 0.00010032697913814875,
+      "loss": 1.996,
+      "step": 2746
+    },
+    {
+      "epoch": 1.185033426784559,
+      "grad_norm": 0.2089913785457611,
+      "learning_rate": 0.00010029485335838565,
+      "loss": 2.2704,
+      "step": 2747
+    },
+    {
+      "epoch": 1.185464740133707,
+      "grad_norm": 0.1756969392299652,
+      "learning_rate": 0.00010026272234104657,
+      "loss": 2.1276,
+      "step": 2748
+    },
+    {
+      "epoch": 1.1858960534828553,
+      "grad_norm": 0.17700275778770447,
+      "learning_rate": 0.00010023058609278453,
+      "loss": 2.1297,
+      "step": 2749
+    },
+    {
+      "epoch": 1.1863273668320033,
+      "grad_norm": 0.1597864329814911,
+      "learning_rate": 0.00010019844462025375,
+      "loss": 2.1652,
+      "step": 2750
+    },
+    {
+      "epoch": 1.1863273668320033,
+      "eval_loss": 2.1022770404815674,
+      "eval_runtime": 194.8789,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 2750
+    },
+    {
+      "epoch": 1.1867586801811516,
+      "grad_norm": 0.18585540354251862,
+      "learning_rate": 0.0001001662979301094,
+      "loss": 2.2248,
+      "step": 2751
+    },
+    {
+      "epoch": 1.1871899935302999,
+      "grad_norm": 0.1518988460302353,
+      "learning_rate": 0.0001001341460290079,
+      "loss": 2.133,
+      "step": 2752
+    },
+    {
+      "epoch": 1.187621306879448,
+      "grad_norm": 0.20666302740573883,
+      "learning_rate": 0.00010010198892360657,
+      "loss": 2.1065,
+      "step": 2753
+    },
+    {
+      "epoch": 1.1880526202285961,
+      "grad_norm": 0.17307338118553162,
+      "learning_rate": 0.00010006982662056396,
+      "loss": 2.2438,
+      "step": 2754
+    },
+    {
+      "epoch": 1.1884839335777442,
+      "grad_norm": 1.0718460083007812,
+      "learning_rate": 0.00010003765912653954,
+      "loss": 2.2488,
+      "step": 2755
+    },
+    {
+      "epoch": 1.1889152469268924,
+      "grad_norm": 0.17475785315036774,
+      "learning_rate": 0.00010000548644819402,
+      "loss": 2.0289,
+      "step": 2756
+    },
+    {
+      "epoch": 1.1893465602760405,
+      "grad_norm": 0.1769273281097412,
+      "learning_rate": 9.997330859218906e-05,
+      "loss": 2.1434,
+      "step": 2757
+    },
+    {
+      "epoch": 1.1897778736251887,
+      "grad_norm": 0.16352666914463043,
+      "learning_rate": 9.994112556518746e-05,
+      "loss": 2.2072,
+      "step": 2758
+    },
+    {
+      "epoch": 1.1902091869743368,
+      "grad_norm": 0.16758856177330017,
+      "learning_rate": 9.990893737385306e-05,
+      "loss": 2.095,
+      "step": 2759
+    },
+    {
+      "epoch": 1.190640500323485,
+      "grad_norm": 0.19053834676742554,
+      "learning_rate": 9.987674402485078e-05,
+      "loss": 1.9014,
+      "step": 2760
+    },
+    {
+      "epoch": 1.1910718136726333,
+      "grad_norm": 0.19716694951057434,
+      "learning_rate": 9.984454552484662e-05,
+      "loss": 2.2309,
+      "step": 2761
+    },
+    {
+      "epoch": 1.1915031270217813,
+      "grad_norm": 0.17813928425312042,
+      "learning_rate": 9.981234188050757e-05,
+      "loss": 2.2592,
+      "step": 2762
+    },
+    {
+      "epoch": 1.1919344403709295,
+      "grad_norm": 0.18199707567691803,
+      "learning_rate": 9.978013309850183e-05,
+      "loss": 2.0669,
+      "step": 2763
+    },
+    {
+      "epoch": 1.1923657537200776,
+      "grad_norm": 0.17666861414909363,
+      "learning_rate": 9.974791918549853e-05,
+      "loss": 2.0754,
+      "step": 2764
+    },
+    {
+      "epoch": 1.1927970670692258,
+      "grad_norm": 0.15702718496322632,
+      "learning_rate": 9.971570014816792e-05,
+      "loss": 2.0688,
+      "step": 2765
+    },
+    {
+      "epoch": 1.1932283804183739,
+      "grad_norm": 0.2100336253643036,
+      "learning_rate": 9.968347599318131e-05,
+      "loss": 2.0483,
+      "step": 2766
+    },
+    {
+      "epoch": 1.1936596937675221,
+      "grad_norm": 0.18148039281368256,
+      "learning_rate": 9.965124672721105e-05,
+      "loss": 2.2175,
+      "step": 2767
+    },
+    {
+      "epoch": 1.1940910071166702,
+      "grad_norm": 0.19504286348819733,
+      "learning_rate": 9.96190123569306e-05,
+      "loss": 2.2361,
+      "step": 2768
+    },
+    {
+      "epoch": 1.1945223204658184,
+      "grad_norm": 0.1750347912311554,
+      "learning_rate": 9.958677288901438e-05,
+      "loss": 2.2762,
+      "step": 2769
+    },
+    {
+      "epoch": 1.1949536338149667,
+      "grad_norm": 0.25596529245376587,
+      "learning_rate": 9.955452833013798e-05,
+      "loss": 2.1682,
+      "step": 2770
+    },
+    {
+      "epoch": 1.1953849471641147,
+      "grad_norm": 0.18025095760822296,
+      "learning_rate": 9.952227868697797e-05,
+      "loss": 2.4384,
+      "step": 2771
+    },
+    {
+      "epoch": 1.195816260513263,
+      "grad_norm": 0.2068176120519638,
+      "learning_rate": 9.949002396621197e-05,
+      "loss": 2.2778,
+      "step": 2772
+    },
+    {
+      "epoch": 1.196247573862411,
+      "grad_norm": 0.17091801762580872,
+      "learning_rate": 9.945776417451869e-05,
+      "loss": 2.1037,
+      "step": 2773
+    },
+    {
+      "epoch": 1.1966788872115592,
+      "grad_norm": 0.1508743315935135,
+      "learning_rate": 9.942549931857786e-05,
+      "loss": 1.7748,
+      "step": 2774
+    },
+    {
+      "epoch": 1.1971102005607073,
+      "grad_norm": 0.15512053668498993,
+      "learning_rate": 9.939322940507033e-05,
+      "loss": 2.2604,
+      "step": 2775
+    },
+    {
+      "epoch": 1.1971102005607073,
+      "eval_loss": 2.1031556129455566,
+      "eval_runtime": 213.7409,
+      "eval_samples_per_second": 0.15,
+      "eval_steps_per_second": 0.15,
+      "step": 2775
+    },
+    {
+      "epoch": 1.1975415139098555,
+      "grad_norm": 0.16651402413845062,
+      "learning_rate": 9.936095444067783e-05,
+      "loss": 2.2371,
+      "step": 2776
+    },
+    {
+      "epoch": 1.1979728272590036,
+      "grad_norm": 0.18020476400852203,
+      "learning_rate": 9.932867443208338e-05,
+      "loss": 2.1046,
+      "step": 2777
+    },
+    {
+      "epoch": 1.1984041406081518,
+      "grad_norm": 0.16545698046684265,
+      "learning_rate": 9.929638938597077e-05,
+      "loss": 2.2312,
+      "step": 2778
+    },
+    {
+      "epoch": 1.1988354539573,
+      "grad_norm": 0.18139177560806274,
+      "learning_rate": 9.926409930902506e-05,
+      "loss": 2.319,
+      "step": 2779
+    },
+    {
+      "epoch": 1.199266767306448,
+      "grad_norm": 0.1793418824672699,
+      "learning_rate": 9.923180420793225e-05,
+      "loss": 2.1874,
+      "step": 2780
+    },
+    {
+      "epoch": 1.1996980806555964,
+      "grad_norm": 0.16347436606884003,
+      "learning_rate": 9.919950408937935e-05,
+      "loss": 1.9059,
+      "step": 2781
+    },
+    {
+      "epoch": 1.2001293940047444,
+      "grad_norm": 0.165296271443367,
+      "learning_rate": 9.916719896005453e-05,
+      "loss": 2.327,
+      "step": 2782
+    },
+    {
+      "epoch": 1.2005607073538926,
+      "grad_norm": 0.19593589007854462,
+      "learning_rate": 9.913488882664684e-05,
+      "loss": 2.2942,
+      "step": 2783
+    },
+    {
+      "epoch": 1.2009920207030407,
+      "grad_norm": 0.16539724171161652,
+      "learning_rate": 9.910257369584653e-05,
+      "loss": 2.2308,
+      "step": 2784
+    },
+    {
+      "epoch": 1.201423334052189,
+      "grad_norm": 0.17547844350337982,
+      "learning_rate": 9.907025357434471e-05,
+      "loss": 2.2597,
+      "step": 2785
+    },
+    {
+      "epoch": 1.201854647401337,
+      "grad_norm": 0.17845278978347778,
+      "learning_rate": 9.903792846883367e-05,
+      "loss": 2.0745,
+      "step": 2786
+    },
+    {
+      "epoch": 1.2022859607504852,
+      "grad_norm": 0.16618625819683075,
+      "learning_rate": 9.900559838600668e-05,
+      "loss": 2.2378,
+      "step": 2787
+    },
+    {
+      "epoch": 1.2027172740996335,
+      "grad_norm": 0.1826469749212265,
+      "learning_rate": 9.897326333255803e-05,
+      "loss": 2.0667,
+      "step": 2788
+    },
+    {
+      "epoch": 1.2031485874487815,
+      "grad_norm": 0.17142803966999054,
+      "learning_rate": 9.894092331518303e-05,
+      "loss": 2.1294,
+      "step": 2789
+    },
+    {
+      "epoch": 1.2035799007979298,
+      "grad_norm": 0.15849988162517548,
+      "learning_rate": 9.890857834057802e-05,
+      "loss": 2.079,
+      "step": 2790
+    },
+    {
+      "epoch": 1.2040112141470778,
+      "grad_norm": 0.17188440263271332,
+      "learning_rate": 9.887622841544045e-05,
+      "loss": 2.0135,
+      "step": 2791
+    },
+    {
+      "epoch": 1.204442527496226,
+      "grad_norm": 0.17471234500408173,
+      "learning_rate": 9.884387354646864e-05,
+      "loss": 2.092,
+      "step": 2792
+    },
+    {
+      "epoch": 1.204873840845374,
+      "grad_norm": 0.20161357522010803,
+      "learning_rate": 9.881151374036207e-05,
+      "loss": 2.168,
+      "step": 2793
+    },
+    {
+      "epoch": 1.2053051541945223,
+      "grad_norm": 0.16633059084415436,
+      "learning_rate": 9.87791490038212e-05,
+      "loss": 2.1851,
+      "step": 2794
+    },
+    {
+      "epoch": 1.2057364675436704,
+      "grad_norm": 0.1972508728504181,
+      "learning_rate": 9.874677934354748e-05,
+      "loss": 1.6806,
+      "step": 2795
+    },
+    {
+      "epoch": 1.2061677808928186,
+      "grad_norm": 0.18068282306194305,
+      "learning_rate": 9.87144047662434e-05,
+      "loss": 2.1382,
+      "step": 2796
+    },
+    {
+      "epoch": 1.2065990942419669,
+      "grad_norm": 0.18279866874217987,
+      "learning_rate": 9.868202527861244e-05,
+      "loss": 2.0858,
+      "step": 2797
+    },
+    {
+      "epoch": 1.207030407591115,
+      "grad_norm": 0.23289109766483307,
+      "learning_rate": 9.864964088735922e-05,
+      "loss": 1.9732,
+      "step": 2798
+    },
+    {
+      "epoch": 1.2074617209402632,
+      "grad_norm": 0.17867575585842133,
+      "learning_rate": 9.86172515991892e-05,
+      "loss": 2.2441,
+      "step": 2799
+    },
+    {
+      "epoch": 1.2078930342894112,
+      "grad_norm": 0.18418186902999878,
+      "learning_rate": 9.858485742080898e-05,
+      "loss": 2.2331,
+      "step": 2800
+    },
+    {
+      "epoch": 1.2078930342894112,
+      "eval_loss": 2.101106643676758,
+      "eval_runtime": 208.9884,
+      "eval_samples_per_second": 0.153,
+      "eval_steps_per_second": 0.153,
+      "step": 2800
+    },
+    {
+      "epoch": 1.2083243476385594,
+      "grad_norm": 0.16849951446056366,
+      "learning_rate": 9.855245835892609e-05,
+      "loss": 2.1511,
+      "step": 2801
+    },
+    {
+      "epoch": 1.2087556609877075,
+      "grad_norm": 0.18521761894226074,
+      "learning_rate": 9.852005442024914e-05,
+      "loss": 1.9106,
+      "step": 2802
+    },
+    {
+      "epoch": 1.2091869743368557,
+      "grad_norm": 0.18433886766433716,
+      "learning_rate": 9.848764561148773e-05,
+      "loss": 2.1457,
+      "step": 2803
+    },
+    {
+      "epoch": 1.2096182876860038,
+      "grad_norm": 0.19022980332374573,
+      "learning_rate": 9.845523193935241e-05,
+      "loss": 2.1276,
+      "step": 2804
+    },
+    {
+      "epoch": 1.210049601035152,
+      "grad_norm": 0.19845323264598846,
+      "learning_rate": 9.842281341055484e-05,
+      "loss": 2.1331,
+      "step": 2805
+    },
+    {
+      "epoch": 1.2104809143843003,
+      "grad_norm": 0.17411771416664124,
+      "learning_rate": 9.839039003180756e-05,
+      "loss": 2.3304,
+      "step": 2806
+    },
+    {
+      "epoch": 1.2109122277334483,
+      "grad_norm": 0.16950243711471558,
+      "learning_rate": 9.835796180982428e-05,
+      "loss": 1.9149,
+      "step": 2807
+    },
+    {
+      "epoch": 1.2113435410825966,
+      "grad_norm": 0.1674221009016037,
+      "learning_rate": 9.832552875131955e-05,
+      "loss": 2.0784,
+      "step": 2808
+    },
+    {
+      "epoch": 1.2117748544317446,
+      "grad_norm": 0.16383004188537598,
+      "learning_rate": 9.829309086300901e-05,
+      "loss": 2.1236,
+      "step": 2809
+    },
+    {
+      "epoch": 1.2122061677808929,
+      "grad_norm": 0.19841207563877106,
+      "learning_rate": 9.826064815160927e-05,
+      "loss": 2.2611,
+      "step": 2810
+    },
+    {
+      "epoch": 1.2126374811300409,
+      "grad_norm": 0.17123110592365265,
+      "learning_rate": 9.822820062383797e-05,
+      "loss": 2.1434,
+      "step": 2811
+    },
+    {
+      "epoch": 1.2130687944791891,
+      "grad_norm": 0.18227308988571167,
+      "learning_rate": 9.819574828641371e-05,
+      "loss": 2.2398,
+      "step": 2812
+    },
+    {
+      "epoch": 1.2135001078283372,
+      "grad_norm": 0.16920961439609528,
+      "learning_rate": 9.816329114605609e-05,
+      "loss": 2.165,
+      "step": 2813
+    },
+    {
+      "epoch": 1.2139314211774854,
+      "grad_norm": 0.1832374781370163,
+      "learning_rate": 9.813082920948575e-05,
+      "loss": 1.884,
+      "step": 2814
+    },
+    {
+      "epoch": 1.2143627345266337,
+      "grad_norm": 0.16949114203453064,
+      "learning_rate": 9.809836248342424e-05,
+      "loss": 2.308,
+      "step": 2815
+    },
+    {
+      "epoch": 1.2147940478757817,
+      "grad_norm": 0.17201551795005798,
+      "learning_rate": 9.80658909745942e-05,
+      "loss": 2.2275,
+      "step": 2816
+    },
+    {
+      "epoch": 1.21522536122493,
+      "grad_norm": 0.2006232738494873,
+      "learning_rate": 9.803341468971918e-05,
+      "loss": 2.1329,
+      "step": 2817
+    },
+    {
+      "epoch": 1.215656674574078,
+      "grad_norm": 0.16852867603302002,
+      "learning_rate": 9.800093363552379e-05,
+      "loss": 2.1475,
+      "step": 2818
+    },
+    {
+      "epoch": 1.2160879879232263,
+      "grad_norm": 0.16948576271533966,
+      "learning_rate": 9.79684478187335e-05,
+      "loss": 2.1423,
+      "step": 2819
+    },
+    {
+      "epoch": 1.2165193012723743,
+      "grad_norm": 0.17116281390190125,
+      "learning_rate": 9.793595724607493e-05,
+      "loss": 1.9302,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2169506146215225,
+      "grad_norm": 0.17152751982212067,
+      "learning_rate": 9.790346192427563e-05,
+      "loss": 2.09,
+      "step": 2821
+    },
+    {
+      "epoch": 1.2173819279706706,
+      "grad_norm": 0.18134663999080658,
+      "learning_rate": 9.787096186006403e-05,
+      "loss": 2.1829,
+      "step": 2822
+    },
+    {
+      "epoch": 1.2178132413198188,
+      "grad_norm": 0.1795966625213623,
+      "learning_rate": 9.78384570601697e-05,
+      "loss": 2.343,
+      "step": 2823
+    },
+    {
+      "epoch": 1.218244554668967,
+      "grad_norm": 0.17307963967323303,
+      "learning_rate": 9.780594753132303e-05,
+      "loss": 2.1866,
+      "step": 2824
+    },
+    {
+      "epoch": 1.2186758680181151,
+      "grad_norm": 0.18871726095676422,
+      "learning_rate": 9.777343328025555e-05,
+      "loss": 2.3792,
+      "step": 2825
+    },
+    {
+      "epoch": 1.2186758680181151,
+      "eval_loss": 2.101025104522705,
+      "eval_runtime": 208.0421,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 2825
+    },
+    {
+      "epoch": 1.2191071813672634,
+      "grad_norm": 0.19461756944656372,
+      "learning_rate": 9.774091431369966e-05,
+      "loss": 2.2572,
+      "step": 2826
+    },
+    {
+      "epoch": 1.2195384947164114,
+      "grad_norm": 0.16747793555259705,
+      "learning_rate": 9.770839063838876e-05,
+      "loss": 2.2002,
+      "step": 2827
+    },
+    {
+      "epoch": 1.2199698080655597,
+      "grad_norm": 0.19756019115447998,
+      "learning_rate": 9.767586226105725e-05,
+      "loss": 2.1019,
+      "step": 2828
+    },
+    {
+      "epoch": 1.2204011214147077,
+      "grad_norm": 0.1785685271024704,
+      "learning_rate": 9.764332918844048e-05,
+      "loss": 2.2182,
+      "step": 2829
+    },
+    {
+      "epoch": 1.220832434763856,
+      "grad_norm": 0.17922717332839966,
+      "learning_rate": 9.761079142727479e-05,
+      "loss": 2.2225,
+      "step": 2830
+    },
+    {
+      "epoch": 1.221263748113004,
+      "grad_norm": 0.2177387773990631,
+      "learning_rate": 9.757824898429743e-05,
+      "loss": 2.2951,
+      "step": 2831
+    },
+    {
+      "epoch": 1.2216950614621522,
+      "grad_norm": 0.17728032171726227,
+      "learning_rate": 9.754570186624675e-05,
+      "loss": 2.3344,
+      "step": 2832
+    },
+    {
+      "epoch": 1.2221263748113005,
+      "grad_norm": 0.16096818447113037,
+      "learning_rate": 9.751315007986194e-05,
+      "loss": 1.9994,
+      "step": 2833
+    },
+    {
+      "epoch": 1.2225576881604485,
+      "grad_norm": 0.21062934398651123,
+      "learning_rate": 9.748059363188317e-05,
+      "loss": 2.2896,
+      "step": 2834
+    },
+    {
+      "epoch": 1.2229890015095968,
+      "grad_norm": 0.1778430938720703,
+      "learning_rate": 9.744803252905168e-05,
+      "loss": 2.3748,
+      "step": 2835
+    },
+    {
+      "epoch": 1.2234203148587448,
+      "grad_norm": 0.17728528380393982,
+      "learning_rate": 9.741546677810955e-05,
+      "loss": 2.2469,
+      "step": 2836
+    },
+    {
+      "epoch": 1.223851628207893,
+      "grad_norm": 0.1769222915172577,
+      "learning_rate": 9.738289638579991e-05,
+      "loss": 2.2029,
+      "step": 2837
+    },
+    {
+      "epoch": 1.224282941557041,
+      "grad_norm": 0.15365712344646454,
+      "learning_rate": 9.735032135886678e-05,
+      "loss": 2.1346,
+      "step": 2838
+    },
+    {
+      "epoch": 1.2247142549061893,
+      "grad_norm": 0.22072498500347137,
+      "learning_rate": 9.731774170405518e-05,
+      "loss": 1.7912,
+      "step": 2839
+    },
+    {
+      "epoch": 1.2251455682553374,
+      "grad_norm": 0.16984528303146362,
+      "learning_rate": 9.728515742811113e-05,
+      "loss": 2.2203,
+      "step": 2840
+    },
+    {
+      "epoch": 1.2255768816044856,
+      "grad_norm": 0.20495721697807312,
+      "learning_rate": 9.725256853778152e-05,
+      "loss": 2.1872,
+      "step": 2841
+    },
+    {
+      "epoch": 1.226008194953634,
+      "grad_norm": 0.16966770589351654,
+      "learning_rate": 9.721997503981423e-05,
+      "loss": 2.0506,
+      "step": 2842
+    },
+    {
+      "epoch": 1.226439508302782,
+      "grad_norm": 0.18237991631031036,
+      "learning_rate": 9.718737694095812e-05,
+      "loss": 2.1769,
+      "step": 2843
+    },
+    {
+      "epoch": 1.2268708216519302,
+      "grad_norm": 0.1720055341720581,
+      "learning_rate": 9.7154774247963e-05,
+      "loss": 2.2018,
+      "step": 2844
+    },
+    {
+      "epoch": 1.2273021350010782,
+      "grad_norm": 0.18814405798912048,
+      "learning_rate": 9.712216696757956e-05,
+      "loss": 2.2869,
+      "step": 2845
+    },
+    {
+      "epoch": 1.2277334483502265,
+      "grad_norm": 0.17777247726917267,
+      "learning_rate": 9.708955510655955e-05,
+      "loss": 2.2798,
+      "step": 2846
+    },
+    {
+      "epoch": 1.2281647616993747,
+      "grad_norm": 0.1642644703388214,
+      "learning_rate": 9.705693867165557e-05,
+      "loss": 2.2026,
+      "step": 2847
+    },
+    {
+      "epoch": 1.2285960750485228,
+      "grad_norm": 0.16237300634384155,
+      "learning_rate": 9.702431766962124e-05,
+      "loss": 2.2696,
+      "step": 2848
+    },
+    {
+      "epoch": 1.2290273883976708,
+      "grad_norm": 0.19161173701286316,
+      "learning_rate": 9.699169210721107e-05,
+      "loss": 2.3309,
+      "step": 2849
+    },
+    {
+      "epoch": 1.229458701746819,
+      "grad_norm": 0.17122620344161987,
+      "learning_rate": 9.695906199118056e-05,
+      "loss": 2.2177,
+      "step": 2850
+    },
+    {
+      "epoch": 1.229458701746819,
+      "eval_loss": 2.100764036178589,
+      "eval_runtime": 208.0607,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 2850
+    },
+    {
+      "epoch": 1.2298900150959673,
+      "grad_norm": 0.18159905076026917,
+      "learning_rate": 9.692642732828612e-05,
+      "loss": 2.3079,
+      "step": 2851
+    },
+    {
+      "epoch": 1.2303213284451153,
+      "grad_norm": 0.1928827464580536,
+      "learning_rate": 9.68937881252851e-05,
+      "loss": 2.2603,
+      "step": 2852
+    },
+    {
+      "epoch": 1.2307526417942636,
+      "grad_norm": 0.20083266496658325,
+      "learning_rate": 9.686114438893586e-05,
+      "loss": 2.3624,
+      "step": 2853
+    },
+    {
+      "epoch": 1.2311839551434116,
+      "grad_norm": 0.16932237148284912,
+      "learning_rate": 9.682849612599759e-05,
+      "loss": 2.1182,
+      "step": 2854
+    },
+    {
+      "epoch": 1.2316152684925599,
+      "grad_norm": 0.17478060722351074,
+      "learning_rate": 9.679584334323047e-05,
+      "loss": 2.2539,
+      "step": 2855
+    },
+    {
+      "epoch": 1.2320465818417081,
+      "grad_norm": 0.18880511820316315,
+      "learning_rate": 9.676318604739565e-05,
+      "loss": 2.1343,
+      "step": 2856
+    },
+    {
+      "epoch": 1.2324778951908562,
+      "grad_norm": 0.1596124768257141,
+      "learning_rate": 9.673052424525515e-05,
+      "loss": 2.1123,
+      "step": 2857
+    },
+    {
+      "epoch": 1.2329092085400044,
+      "grad_norm": 0.17469385266304016,
+      "learning_rate": 9.669785794357198e-05,
+      "loss": 2.152,
+      "step": 2858
+    },
+    {
+      "epoch": 1.2333405218891524,
+      "grad_norm": 0.20866814255714417,
+      "learning_rate": 9.666518714911001e-05,
+      "loss": 2.0822,
+      "step": 2859
+    },
+    {
+      "epoch": 1.2337718352383007,
+      "grad_norm": 0.19376201927661896,
+      "learning_rate": 9.663251186863415e-05,
+      "loss": 2.1318,
+      "step": 2860
+    },
+    {
+      "epoch": 1.2342031485874487,
+      "grad_norm": 0.1695946455001831,
+      "learning_rate": 9.659983210891011e-05,
+      "loss": 2.1996,
+      "step": 2861
+    },
+    {
+      "epoch": 1.234634461936597,
+      "grad_norm": 0.16476131975650787,
+      "learning_rate": 9.656714787670463e-05,
+      "loss": 2.1671,
+      "step": 2862
+    },
+    {
+      "epoch": 1.235065775285745,
+      "grad_norm": 0.17561404407024384,
+      "learning_rate": 9.653445917878532e-05,
+      "loss": 2.3061,
+      "step": 2863
+    },
+    {
+      "epoch": 1.2354970886348933,
+      "grad_norm": 0.18235552310943604,
+      "learning_rate": 9.650176602192076e-05,
+      "loss": 2.2073,
+      "step": 2864
+    },
+    {
+      "epoch": 1.2359284019840415,
+      "grad_norm": 0.16104130446910858,
+      "learning_rate": 9.646906841288038e-05,
+      "loss": 2.0979,
+      "step": 2865
+    },
+    {
+      "epoch": 1.2363597153331896,
+      "grad_norm": 0.344385027885437,
+      "learning_rate": 9.643636635843462e-05,
+      "loss": 2.0115,
+      "step": 2866
+    },
+    {
+      "epoch": 1.2367910286823378,
+      "grad_norm": 0.1773248314857483,
+      "learning_rate": 9.64036598653548e-05,
+      "loss": 2.1553,
+      "step": 2867
+    },
+    {
+      "epoch": 1.2372223420314858,
+      "grad_norm": 0.1927177459001541,
+      "learning_rate": 9.637094894041308e-05,
+      "loss": 2.3593,
+      "step": 2868
+    },
+    {
+      "epoch": 1.237653655380634,
+      "grad_norm": 0.17691570520401,
+      "learning_rate": 9.633823359038273e-05,
+      "loss": 2.2298,
+      "step": 2869
+    },
+    {
+      "epoch": 1.2380849687297821,
+      "grad_norm": 0.18220536410808563,
+      "learning_rate": 9.630551382203773e-05,
+      "loss": 2.1005,
+      "step": 2870
+    },
+    {
+      "epoch": 1.2385162820789304,
+      "grad_norm": 0.17187197506427765,
+      "learning_rate": 9.627278964215313e-05,
+      "loss": 2.2041,
+      "step": 2871
+    },
+    {
+      "epoch": 1.2389475954280784,
+      "grad_norm": 0.18310484290122986,
+      "learning_rate": 9.624006105750477e-05,
+      "loss": 2.3803,
+      "step": 2872
+    },
+    {
+      "epoch": 1.2393789087772267,
+      "grad_norm": 0.18291732668876648,
+      "learning_rate": 9.62073280748695e-05,
+      "loss": 2.2872,
+      "step": 2873
+    },
+    {
+      "epoch": 1.239810222126375,
+      "grad_norm": 0.18414902687072754,
+      "learning_rate": 9.617459070102504e-05,
+      "loss": 2.3016,
+      "step": 2874
+    },
+    {
+      "epoch": 1.240241535475523,
+      "grad_norm": 0.17588050663471222,
+      "learning_rate": 9.614184894275e-05,
+      "loss": 2.2289,
+      "step": 2875
+    },
+    {
+      "epoch": 1.240241535475523,
+      "eval_loss": 2.1006932258605957,
+      "eval_runtime": 208.1806,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 2875
+    },
+    {
+      "epoch": 1.2406728488246712,
+      "grad_norm": 0.1735721081495285,
+      "learning_rate": 9.610910280682398e-05,
+      "loss": 2.1857,
+      "step": 2876
+    },
+    {
+      "epoch": 1.2411041621738192,
+      "grad_norm": 0.22414951026439667,
+      "learning_rate": 9.607635230002733e-05,
+      "loss": 2.3306,
+      "step": 2877
+    },
+    {
+      "epoch": 1.2415354755229675,
+      "grad_norm": 0.16612106561660767,
+      "learning_rate": 9.604359742914145e-05,
+      "loss": 2.1791,
+      "step": 2878
+    },
+    {
+      "epoch": 1.2419667888721155,
+      "grad_norm": 0.17590609192848206,
+      "learning_rate": 9.601083820094862e-05,
+      "loss": 2.1513,
+      "step": 2879
+    },
+    {
+      "epoch": 1.2423981022212638,
+      "grad_norm": 0.18331202864646912,
+      "learning_rate": 9.597807462223195e-05,
+      "loss": 2.1928,
+      "step": 2880
+    },
+    {
+      "epoch": 1.2428294155704118,
+      "grad_norm": 0.18726730346679688,
+      "learning_rate": 9.594530669977552e-05,
+      "loss": 2.2287,
+      "step": 2881
+    },
+    {
+      "epoch": 1.24326072891956,
+      "grad_norm": 0.2933848798274994,
+      "learning_rate": 9.591253444036426e-05,
+      "loss": 2.2461,
+      "step": 2882
+    },
+    {
+      "epoch": 1.2436920422687083,
+      "grad_norm": 0.2122485637664795,
+      "learning_rate": 9.587975785078409e-05,
+      "loss": 2.3764,
+      "step": 2883
+    },
+    {
+      "epoch": 1.2441233556178564,
+      "grad_norm": 0.18504616618156433,
+      "learning_rate": 9.584697693782171e-05,
+      "loss": 2.2626,
+      "step": 2884
+    },
+    {
+      "epoch": 1.2445546689670046,
+      "grad_norm": 0.16713541746139526,
+      "learning_rate": 9.581419170826476e-05,
+      "loss": 2.056,
+      "step": 2885
+    },
+    {
+      "epoch": 1.2449859823161527,
+      "grad_norm": 0.20109900832176208,
+      "learning_rate": 9.57814021689018e-05,
+      "loss": 2.0518,
+      "step": 2886
+    },
+    {
+      "epoch": 1.245417295665301,
+      "grad_norm": 0.19723549485206604,
+      "learning_rate": 9.574860832652229e-05,
+      "loss": 2.1423,
+      "step": 2887
+    },
+    {
+      "epoch": 1.245848609014449,
+      "grad_norm": 0.19441846013069153,
+      "learning_rate": 9.57158101879165e-05,
+      "loss": 2.1949,
+      "step": 2888
+    },
+    {
+      "epoch": 1.2462799223635972,
+      "grad_norm": 0.18298396468162537,
+      "learning_rate": 9.568300775987565e-05,
+      "loss": 2.2847,
+      "step": 2889
+    },
+    {
+      "epoch": 1.2467112357127452,
+      "grad_norm": 0.18244996666908264,
+      "learning_rate": 9.565020104919188e-05,
+      "loss": 2.1066,
+      "step": 2890
+    },
+    {
+      "epoch": 1.2471425490618935,
+      "grad_norm": 0.20596863329410553,
+      "learning_rate": 9.561739006265813e-05,
+      "loss": 2.1475,
+      "step": 2891
+    },
+    {
+      "epoch": 1.2475738624110417,
+      "grad_norm": 0.2012377828359604,
+      "learning_rate": 9.558457480706834e-05,
+      "loss": 2.3096,
+      "step": 2892
+    },
+    {
+      "epoch": 1.2480051757601898,
+      "grad_norm": 0.17559053003787994,
+      "learning_rate": 9.55517552892172e-05,
+      "loss": 2.2434,
+      "step": 2893
+    },
+    {
+      "epoch": 1.248436489109338,
+      "grad_norm": 0.1771698296070099,
+      "learning_rate": 9.551893151590039e-05,
+      "loss": 2.1215,
+      "step": 2894
+    },
+    {
+      "epoch": 1.248867802458486,
+      "grad_norm": 0.16106769442558289,
+      "learning_rate": 9.548610349391441e-05,
+      "loss": 2.1251,
+      "step": 2895
+    },
+    {
+      "epoch": 1.2492991158076343,
+      "grad_norm": 0.2000650316476822,
+      "learning_rate": 9.545327123005667e-05,
+      "loss": 2.0942,
+      "step": 2896
+    },
+    {
+      "epoch": 1.2497304291567823,
+      "grad_norm": 0.17449326813220978,
+      "learning_rate": 9.542043473112544e-05,
+      "loss": 2.1445,
+      "step": 2897
+    },
+    {
+      "epoch": 1.2501617425059306,
+      "grad_norm": 0.2125837802886963,
+      "learning_rate": 9.538759400391989e-05,
+      "loss": 2.1557,
+      "step": 2898
+    },
+    {
+      "epoch": 1.2505930558550786,
+      "grad_norm": 0.2138071060180664,
+      "learning_rate": 9.535474905524007e-05,
+      "loss": 1.7936,
+      "step": 2899
+    },
+    {
+      "epoch": 1.2510243692042269,
+      "grad_norm": 0.17955918610095978,
+      "learning_rate": 9.532189989188683e-05,
+      "loss": 2.1389,
+      "step": 2900
+    },
+    {
+      "epoch": 1.2510243692042269,
+      "eval_loss": 2.099691867828369,
+      "eval_runtime": 208.0343,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 2900
+    },
+    {
+      "epoch": 1.2514556825533751,
+      "grad_norm": 0.16655541956424713,
+      "learning_rate": 9.528904652066199e-05,
+      "loss": 2.0131,
+      "step": 2901
+    },
+    {
+      "epoch": 1.2518869959025232,
+      "grad_norm": 0.16214993596076965,
+      "learning_rate": 9.52561889483682e-05,
+      "loss": 2.2244,
+      "step": 2902
+    },
+    {
+      "epoch": 1.2523183092516712,
+      "grad_norm": 0.17968380451202393,
+      "learning_rate": 9.522332718180896e-05,
+      "loss": 2.2194,
+      "step": 2903
+    },
+    {
+      "epoch": 1.2527496226008195,
+      "grad_norm": 0.1735077202320099,
+      "learning_rate": 9.519046122778868e-05,
+      "loss": 2.0717,
+      "step": 2904
+    },
+    {
+      "epoch": 1.2531809359499677,
+      "grad_norm": 0.1795034110546112,
+      "learning_rate": 9.515759109311258e-05,
+      "loss": 2.4281,
+      "step": 2905
+    },
+    {
+      "epoch": 1.2536122492991157,
+      "grad_norm": 0.17288808524608612,
+      "learning_rate": 9.512471678458684e-05,
+      "loss": 2.1008,
+      "step": 2906
+    },
+    {
+      "epoch": 1.254043562648264,
+      "grad_norm": 0.17616263031959534,
+      "learning_rate": 9.509183830901838e-05,
+      "loss": 2.1797,
+      "step": 2907
+    },
+    {
+      "epoch": 1.254474875997412,
+      "grad_norm": 0.16509383916854858,
+      "learning_rate": 9.50589556732151e-05,
+      "loss": 2.1597,
+      "step": 2908
+    },
+    {
+      "epoch": 1.2549061893465603,
+      "grad_norm": 0.18447110056877136,
+      "learning_rate": 9.502606888398567e-05,
+      "loss": 2.1389,
+      "step": 2909
+    },
+    {
+      "epoch": 1.2553375026957085,
+      "grad_norm": 0.1685207039117813,
+      "learning_rate": 9.499317794813968e-05,
+      "loss": 2.0595,
+      "step": 2910
+    },
+    {
+      "epoch": 1.2557688160448566,
+      "grad_norm": 0.16968344151973724,
+      "learning_rate": 9.496028287248756e-05,
+      "loss": 1.8777,
+      "step": 2911
+    },
+    {
+      "epoch": 1.2562001293940048,
+      "grad_norm": 0.19373373687267303,
+      "learning_rate": 9.492738366384059e-05,
+      "loss": 2.2779,
+      "step": 2912
+    },
+    {
+      "epoch": 1.2566314427431529,
+      "grad_norm": 0.19972391426563263,
+      "learning_rate": 9.489448032901089e-05,
+      "loss": 2.1135,
+      "step": 2913
+    },
+    {
+      "epoch": 1.2570627560923011,
+      "grad_norm": 0.17108936607837677,
+      "learning_rate": 9.486157287481148e-05,
+      "loss": 2.1786,
+      "step": 2914
+    },
+    {
+      "epoch": 1.2574940694414491,
+      "grad_norm": 0.18781358003616333,
+      "learning_rate": 9.482866130805623e-05,
+      "loss": 2.2741,
+      "step": 2915
+    },
+    {
+      "epoch": 1.2579253827905974,
+      "grad_norm": 0.17684100568294525,
+      "learning_rate": 9.47957456355598e-05,
+      "loss": 2.0681,
+      "step": 2916
+    },
+    {
+      "epoch": 1.2583566961397454,
+      "grad_norm": 0.17751702666282654,
+      "learning_rate": 9.476282586413774e-05,
+      "loss": 2.1477,
+      "step": 2917
+    },
+    {
+      "epoch": 1.2587880094888937,
+      "grad_norm": 0.16708078980445862,
+      "learning_rate": 9.472990200060648e-05,
+      "loss": 2.1511,
+      "step": 2918
+    },
+    {
+      "epoch": 1.259219322838042,
+      "grad_norm": 0.1751113384962082,
+      "learning_rate": 9.469697405178324e-05,
+      "loss": 2.0367,
+      "step": 2919
+    },
+    {
+      "epoch": 1.25965063618719,
+      "grad_norm": 0.1651589274406433,
+      "learning_rate": 9.466404202448613e-05,
+      "loss": 2.0795,
+      "step": 2920
+    },
+    {
+      "epoch": 1.2600819495363382,
+      "grad_norm": 0.16996340453624725,
+      "learning_rate": 9.463110592553405e-05,
+      "loss": 2.1443,
+      "step": 2921
+    },
+    {
+      "epoch": 1.2605132628854863,
+      "grad_norm": 0.20765995979309082,
+      "learning_rate": 9.459816576174685e-05,
+      "loss": 2.1737,
+      "step": 2922
+    },
+    {
+      "epoch": 1.2609445762346345,
+      "grad_norm": 0.21429865062236786,
+      "learning_rate": 9.456522153994505e-05,
+      "loss": 2.3622,
+      "step": 2923
+    },
+    {
+      "epoch": 1.2613758895837826,
+      "grad_norm": 0.19938156008720398,
+      "learning_rate": 9.453227326695021e-05,
+      "loss": 2.1501,
+      "step": 2924
+    },
+    {
+      "epoch": 1.2618072029329308,
+      "grad_norm": 0.17281770706176758,
+      "learning_rate": 9.449932094958458e-05,
+      "loss": 2.2994,
+      "step": 2925
+    },
+    {
+      "epoch": 1.2618072029329308,
+      "eval_loss": 2.097632646560669,
+      "eval_runtime": 208.7814,
+      "eval_samples_per_second": 0.153,
+      "eval_steps_per_second": 0.153,
+      "step": 2925
+    },
+    {
+      "epoch": 1.2622385162820788,
+      "grad_norm": 0.19390475749969482,
+      "learning_rate": 9.446636459467131e-05,
+      "loss": 2.1672,
+      "step": 2926
+    },
+    {
+      "epoch": 1.262669829631227,
+      "grad_norm": 0.18728719651699066,
+      "learning_rate": 9.443340420903438e-05,
+      "loss": 2.2925,
+      "step": 2927
+    },
+    {
+      "epoch": 1.2631011429803753,
+      "grad_norm": 0.17728185653686523,
+      "learning_rate": 9.440043979949857e-05,
+      "loss": 2.2727,
+      "step": 2928
+    },
+    {
+      "epoch": 1.2635324563295234,
+      "grad_norm": 0.17605188488960266,
+      "learning_rate": 9.436747137288955e-05,
+      "loss": 2.2959,
+      "step": 2929
+    },
+    {
+      "epoch": 1.2639637696786716,
+      "grad_norm": 0.18089932203292847,
+      "learning_rate": 9.433449893603376e-05,
+      "loss": 2.1482,
+      "step": 2930
+    },
+    {
+      "epoch": 1.2643950830278197,
+      "grad_norm": 0.17335166037082672,
+      "learning_rate": 9.430152249575855e-05,
+      "loss": 2.3316,
+      "step": 2931
+    },
+    {
+      "epoch": 1.264826396376968,
+      "grad_norm": 0.1741197407245636,
+      "learning_rate": 9.426854205889202e-05,
+      "loss": 2.2759,
+      "step": 2932
+    },
+    {
+      "epoch": 1.265257709726116,
+      "grad_norm": 0.1794310063123703,
+      "learning_rate": 9.423555763226314e-05,
+      "loss": 2.2444,
+      "step": 2933
+    },
+    {
+      "epoch": 1.2656890230752642,
+      "grad_norm": 0.18344342708587646,
+      "learning_rate": 9.420256922270169e-05,
+      "loss": 2.4161,
+      "step": 2934
+    },
+    {
+      "epoch": 1.2661203364244122,
+      "grad_norm": 0.19544778764247894,
+      "learning_rate": 9.416957683703831e-05,
+      "loss": 2.1683,
+      "step": 2935
+    },
+    {
+      "epoch": 1.2665516497735605,
+      "grad_norm": 0.16360361874103546,
+      "learning_rate": 9.413658048210441e-05,
+      "loss": 2.0059,
+      "step": 2936
+    },
+    {
+      "epoch": 1.2669829631227087,
+      "grad_norm": 0.1777682900428772,
+      "learning_rate": 9.410358016473222e-05,
+      "loss": 2.1029,
+      "step": 2937
+    },
+    {
+      "epoch": 1.2674142764718568,
+      "grad_norm": 0.18484915792942047,
+      "learning_rate": 9.407057589175484e-05,
+      "loss": 2.3951,
+      "step": 2938
+    },
+    {
+      "epoch": 1.267845589821005,
+      "grad_norm": 0.19149358570575714,
+      "learning_rate": 9.40375676700062e-05,
+      "loss": 2.0752,
+      "step": 2939
+    },
+    {
+      "epoch": 1.268276903170153,
+      "grad_norm": 0.1761498898267746,
+      "learning_rate": 9.400455550632097e-05,
+      "loss": 2.2717,
+      "step": 2940
+    },
+    {
+      "epoch": 1.2687082165193013,
+      "grad_norm": 0.1742338240146637,
+      "learning_rate": 9.397153940753468e-05,
+      "loss": 2.3397,
+      "step": 2941
+    },
+    {
+      "epoch": 1.2691395298684494,
+      "grad_norm": 0.18861375749111176,
+      "learning_rate": 9.393851938048371e-05,
+      "loss": 2.1523,
+      "step": 2942
+    },
+    {
+      "epoch": 1.2695708432175976,
+      "grad_norm": 0.16949686408042908,
+      "learning_rate": 9.390549543200518e-05,
+      "loss": 2.2293,
+      "step": 2943
+    },
+    {
+      "epoch": 1.2700021565667456,
+      "grad_norm": 0.17178656160831451,
+      "learning_rate": 9.387246756893706e-05,
+      "loss": 2.1625,
+      "step": 2944
+    },
+    {
+      "epoch": 1.270433469915894,
+      "grad_norm": 0.1995915025472641,
+      "learning_rate": 9.383943579811817e-05,
+      "loss": 2.0885,
+      "step": 2945
+    },
+    {
+      "epoch": 1.2708647832650422,
+      "grad_norm": 0.20199699699878693,
+      "learning_rate": 9.380640012638805e-05,
+      "loss": 2.2046,
+      "step": 2946
+    },
+    {
+      "epoch": 1.2712960966141902,
+      "grad_norm": 0.1766977161169052,
+      "learning_rate": 9.377336056058714e-05,
+      "loss": 2.1567,
+      "step": 2947
+    },
+    {
+      "epoch": 1.2717274099633384,
+      "grad_norm": 0.16584517061710358,
+      "learning_rate": 9.374031710755662e-05,
+      "loss": 2.2226,
+      "step": 2948
+    },
+    {
+      "epoch": 1.2721587233124865,
+      "grad_norm": 0.1809447705745697,
+      "learning_rate": 9.370726977413851e-05,
+      "loss": 2.128,
+      "step": 2949
+    },
+    {
+      "epoch": 1.2725900366616347,
+      "grad_norm": 0.19714750349521637,
+      "learning_rate": 9.36742185671756e-05,
+      "loss": 2.2823,
+      "step": 2950
+    },
+    {
+      "epoch": 1.2725900366616347,
+      "eval_loss": 2.097557544708252,
+      "eval_runtime": 203.2645,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 2950
+    },
+    {
+      "epoch": 1.2730213500107828,
+      "grad_norm": 0.18128915131092072,
+      "learning_rate": 9.364116349351153e-05,
+      "loss": 2.2105,
+      "step": 2951
+    },
+    {
+      "epoch": 1.273452663359931,
+      "grad_norm": 0.19274786114692688,
+      "learning_rate": 9.360810455999067e-05,
+      "loss": 2.1284,
+      "step": 2952
+    },
+    {
+      "epoch": 1.273883976709079,
+      "grad_norm": 0.18371281027793884,
+      "learning_rate": 9.357504177345829e-05,
+      "loss": 2.2204,
+      "step": 2953
+    },
+    {
+      "epoch": 1.2743152900582273,
+      "grad_norm": 0.1810397356748581,
+      "learning_rate": 9.354197514076038e-05,
+      "loss": 2.2558,
+      "step": 2954
+    },
+    {
+      "epoch": 1.2747466034073756,
+      "grad_norm": 0.1709802746772766,
+      "learning_rate": 9.350890466874374e-05,
+      "loss": 2.0882,
+      "step": 2955
+    },
+    {
+      "epoch": 1.2751779167565236,
+      "grad_norm": 0.1786419302225113,
+      "learning_rate": 9.347583036425597e-05,
+      "loss": 2.1403,
+      "step": 2956
+    },
+    {
+      "epoch": 1.2756092301056718,
+      "grad_norm": 0.1712750643491745,
+      "learning_rate": 9.34427522341455e-05,
+      "loss": 2.2502,
+      "step": 2957
+    },
+    {
+      "epoch": 1.2760405434548199,
+      "grad_norm": 0.1820487231016159,
+      "learning_rate": 9.340967028526148e-05,
+      "loss": 2.138,
+      "step": 2958
+    },
+    {
+      "epoch": 1.2764718568039681,
+      "grad_norm": 0.16608822345733643,
+      "learning_rate": 9.337658452445391e-05,
+      "loss": 2.2128,
+      "step": 2959
+    },
+    {
+      "epoch": 1.2769031701531162,
+      "grad_norm": 0.20153141021728516,
+      "learning_rate": 9.334349495857353e-05,
+      "loss": 2.1863,
+      "step": 2960
+    },
+    {
+      "epoch": 1.2773344835022644,
+      "grad_norm": 0.18792590498924255,
+      "learning_rate": 9.331040159447194e-05,
+      "loss": 2.1333,
+      "step": 2961
+    },
+    {
+      "epoch": 1.2777657968514124,
+      "grad_norm": 0.1876603364944458,
+      "learning_rate": 9.327730443900146e-05,
+      "loss": 2.2312,
+      "step": 2962
+    },
+    {
+      "epoch": 1.2781971102005607,
+      "grad_norm": 0.17630432546138763,
+      "learning_rate": 9.32442034990152e-05,
+      "loss": 2.2698,
+      "step": 2963
+    },
+    {
+      "epoch": 1.278628423549709,
+      "grad_norm": 0.17087142169475555,
+      "learning_rate": 9.321109878136709e-05,
+      "loss": 2.0537,
+      "step": 2964
+    },
+    {
+      "epoch": 1.279059736898857,
+      "grad_norm": 0.20455554127693176,
+      "learning_rate": 9.317799029291185e-05,
+      "loss": 2.2909,
+      "step": 2965
+    },
+    {
+      "epoch": 1.2794910502480052,
+      "grad_norm": 0.1775055080652237,
+      "learning_rate": 9.31448780405049e-05,
+      "loss": 2.3168,
+      "step": 2966
+    },
+    {
+      "epoch": 1.2799223635971533,
+      "grad_norm": 0.18382766842842102,
+      "learning_rate": 9.311176203100253e-05,
+      "loss": 2.3394,
+      "step": 2967
+    },
+    {
+      "epoch": 1.2803536769463015,
+      "grad_norm": 0.38945379853248596,
+      "learning_rate": 9.307864227126179e-05,
+      "loss": 2.3609,
+      "step": 2968
+    },
+    {
+      "epoch": 1.2807849902954496,
+      "grad_norm": 0.1864595115184784,
+      "learning_rate": 9.304551876814043e-05,
+      "loss": 2.0455,
+      "step": 2969
+    },
+    {
+      "epoch": 1.2812163036445978,
+      "grad_norm": 0.19058941304683685,
+      "learning_rate": 9.301239152849708e-05,
+      "loss": 1.9847,
+      "step": 2970
+    },
+    {
+      "epoch": 1.2816476169937459,
+      "grad_norm": 0.17958027124404907,
+      "learning_rate": 9.297926055919109e-05,
+      "loss": 2.5394,
+      "step": 2971
+    },
+    {
+      "epoch": 1.282078930342894,
+      "grad_norm": 0.18691030144691467,
+      "learning_rate": 9.294612586708257e-05,
+      "loss": 2.3476,
+      "step": 2972
+    },
+    {
+      "epoch": 1.2825102436920424,
+      "grad_norm": 0.17819638550281525,
+      "learning_rate": 9.291298745903245e-05,
+      "loss": 2.4438,
+      "step": 2973
+    },
+    {
+      "epoch": 1.2829415570411904,
+      "grad_norm": 0.18015502393245697,
+      "learning_rate": 9.287984534190237e-05,
+      "loss": 2.1254,
+      "step": 2974
+    },
+    {
+      "epoch": 1.2833728703903386,
+      "grad_norm": 0.21266379952430725,
+      "learning_rate": 9.28466995225548e-05,
+      "loss": 1.987,
+      "step": 2975
+    },
+    {
+      "epoch": 1.2833728703903386,
+      "eval_loss": 2.0974252223968506,
+      "eval_runtime": 202.1486,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 2975
+    },
+    {
+      "epoch": 1.2838041837394867,
+      "grad_norm": 0.18659232556819916,
+      "learning_rate": 9.28135500078529e-05,
+      "loss": 2.0334,
+      "step": 2976
+    },
+    {
+      "epoch": 1.284235497088635,
+      "grad_norm": 0.1787107288837433,
+      "learning_rate": 9.278039680466068e-05,
+      "loss": 2.1156,
+      "step": 2977
+    },
+    {
+      "epoch": 1.284666810437783,
+      "grad_norm": 0.17784705758094788,
+      "learning_rate": 9.274723991984285e-05,
+      "loss": 2.2095,
+      "step": 2978
+    },
+    {
+      "epoch": 1.2850981237869312,
+      "grad_norm": 0.18107645213603973,
+      "learning_rate": 9.271407936026492e-05,
+      "loss": 2.2255,
+      "step": 2979
+    },
+    {
+      "epoch": 1.2855294371360793,
+      "grad_norm": 0.18629951775074005,
+      "learning_rate": 9.268091513279314e-05,
+      "loss": 2.3375,
+      "step": 2980
+    },
+    {
+      "epoch": 1.2859607504852275,
+      "grad_norm": 0.1789446324110031,
+      "learning_rate": 9.264774724429454e-05,
+      "loss": 2.2594,
+      "step": 2981
+    },
+    {
+      "epoch": 1.2863920638343758,
+      "grad_norm": 0.17903368175029755,
+      "learning_rate": 9.261457570163685e-05,
+      "loss": 2.2187,
+      "step": 2982
+    },
+    {
+      "epoch": 1.2868233771835238,
+      "grad_norm": 0.18009735643863678,
+      "learning_rate": 9.258140051168865e-05,
+      "loss": 2.2597,
+      "step": 2983
+    },
+    {
+      "epoch": 1.287254690532672,
+      "grad_norm": 0.17903639376163483,
+      "learning_rate": 9.25482216813192e-05,
+      "loss": 2.3297,
+      "step": 2984
+    },
+    {
+      "epoch": 1.28768600388182,
+      "grad_norm": 0.169985830783844,
+      "learning_rate": 9.251503921739855e-05,
+      "loss": 2.207,
+      "step": 2985
+    },
+    {
+      "epoch": 1.2881173172309683,
+      "grad_norm": 0.17052622139453888,
+      "learning_rate": 9.248185312679749e-05,
+      "loss": 2.292,
+      "step": 2986
+    },
+    {
+      "epoch": 1.2885486305801164,
+      "grad_norm": 0.17440353333950043,
+      "learning_rate": 9.244866341638759e-05,
+      "loss": 2.2154,
+      "step": 2987
+    },
+    {
+      "epoch": 1.2889799439292646,
+      "grad_norm": 0.16670095920562744,
+      "learning_rate": 9.241547009304106e-05,
+      "loss": 1.9959,
+      "step": 2988
+    },
+    {
+      "epoch": 1.2894112572784127,
+      "grad_norm": 0.17001476883888245,
+      "learning_rate": 9.238227316363104e-05,
+      "loss": 2.269,
+      "step": 2989
+    },
+    {
+      "epoch": 1.289842570627561,
+      "grad_norm": 0.162721186876297,
+      "learning_rate": 9.234907263503124e-05,
+      "loss": 2.0803,
+      "step": 2990
+    },
+    {
+      "epoch": 1.2902738839767092,
+      "grad_norm": 0.17348825931549072,
+      "learning_rate": 9.231586851411626e-05,
+      "loss": 2.0746,
+      "step": 2991
+    },
+    {
+      "epoch": 1.2907051973258572,
+      "grad_norm": 0.17610464990139008,
+      "learning_rate": 9.228266080776129e-05,
+      "loss": 2.127,
+      "step": 2992
+    },
+    {
+      "epoch": 1.2911365106750055,
+      "grad_norm": 0.18986164033412933,
+      "learning_rate": 9.224944952284243e-05,
+      "loss": 2.3315,
+      "step": 2993
+    },
+    {
+      "epoch": 1.2915678240241535,
+      "grad_norm": 0.18086329102516174,
+      "learning_rate": 9.221623466623642e-05,
+      "loss": 2.1828,
+      "step": 2994
+    },
+    {
+      "epoch": 1.2919991373733017,
+      "grad_norm": 0.1708398312330246,
+      "learning_rate": 9.218301624482074e-05,
+      "loss": 2.2815,
+      "step": 2995
+    },
+    {
+      "epoch": 1.29243045072245,
+      "grad_norm": 0.18381886184215546,
+      "learning_rate": 9.214979426547364e-05,
+      "loss": 2.3363,
+      "step": 2996
+    },
+    {
+      "epoch": 1.292861764071598,
+      "grad_norm": 0.1695384830236435,
+      "learning_rate": 9.211656873507407e-05,
+      "loss": 2.1781,
+      "step": 2997
+    },
+    {
+      "epoch": 1.293293077420746,
+      "grad_norm": 0.18817633390426636,
+      "learning_rate": 9.208333966050178e-05,
+      "loss": 2.3369,
+      "step": 2998
+    },
+    {
+      "epoch": 1.2937243907698943,
+      "grad_norm": 0.20488505065441132,
+      "learning_rate": 9.205010704863718e-05,
+      "loss": 2.0694,
+      "step": 2999
+    },
+    {
+      "epoch": 1.2941557041190426,
+      "grad_norm": 0.17080359160900116,
+      "learning_rate": 9.201687090636147e-05,
+      "loss": 2.1132,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2941557041190426,
+      "eval_loss": 2.0969784259796143,
+      "eval_runtime": 208.7919,
+      "eval_samples_per_second": 0.153,
+      "eval_steps_per_second": 0.153,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2945870174681906,
+      "grad_norm": 3.370145082473755,
+      "learning_rate": 9.198363124055655e-05,
+      "loss": 2.1544,
+      "step": 3001
+    },
+    {
+      "epoch": 1.2950183308173389,
+      "grad_norm": 0.18138986825942993,
+      "learning_rate": 9.195038805810504e-05,
+      "loss": 2.0651,
+      "step": 3002
+    },
+    {
+      "epoch": 1.295449644166487,
+      "grad_norm": 0.16984714567661285,
+      "learning_rate": 9.191714136589032e-05,
+      "loss": 2.3752,
+      "step": 3003
+    },
+    {
+      "epoch": 1.2958809575156351,
+      "grad_norm": 0.19066904485225677,
+      "learning_rate": 9.188389117079647e-05,
+      "loss": 2.3016,
+      "step": 3004
+    },
+    {
+      "epoch": 1.2963122708647834,
+      "grad_norm": 0.38692083954811096,
+      "learning_rate": 9.185063747970833e-05,
+      "loss": 2.1112,
+      "step": 3005
+    },
+    {
+      "epoch": 1.2967435842139314,
+      "grad_norm": 0.16445212066173553,
+      "learning_rate": 9.18173802995114e-05,
+      "loss": 2.2301,
+      "step": 3006
+    },
+    {
+      "epoch": 1.2971748975630795,
+      "grad_norm": 0.17288804054260254,
+      "learning_rate": 9.178411963709198e-05,
+      "loss": 2.274,
+      "step": 3007
+    },
+    {
+      "epoch": 1.2976062109122277,
+      "grad_norm": 0.16802461445331573,
+      "learning_rate": 9.175085549933706e-05,
+      "loss": 2.0457,
+      "step": 3008
+    },
+    {
+      "epoch": 1.298037524261376,
+      "grad_norm": 0.17567449808120728,
+      "learning_rate": 9.171758789313431e-05,
+      "loss": 2.1486,
+      "step": 3009
+    },
+    {
+      "epoch": 1.298468837610524,
+      "grad_norm": 0.17225013673305511,
+      "learning_rate": 9.168431682537216e-05,
+      "loss": 2.0606,
+      "step": 3010
+    },
+    {
+      "epoch": 1.2989001509596723,
+      "grad_norm": 0.17275701463222504,
+      "learning_rate": 9.165104230293976e-05,
+      "loss": 2.1349,
+      "step": 3011
+    },
+    {
+      "epoch": 1.2993314643088203,
+      "grad_norm": 0.18286320567131042,
+      "learning_rate": 9.161776433272695e-05,
+      "loss": 2.2249,
+      "step": 3012
+    },
+    {
+      "epoch": 1.2997627776579685,
+      "grad_norm": 0.18570812046527863,
+      "learning_rate": 9.158448292162433e-05,
+      "loss": 2.2915,
+      "step": 3013
+    },
+    {
+      "epoch": 1.3001940910071168,
+      "grad_norm": 0.1986859291791916,
+      "learning_rate": 9.155119807652314e-05,
+      "loss": 2.1963,
+      "step": 3014
+    },
+    {
+      "epoch": 1.3006254043562648,
+      "grad_norm": 0.17509645223617554,
+      "learning_rate": 9.151790980431537e-05,
+      "loss": 2.259,
+      "step": 3015
+    },
+    {
+      "epoch": 1.3010567177054129,
+      "grad_norm": 0.1850888580083847,
+      "learning_rate": 9.148461811189375e-05,
+      "loss": 2.196,
+      "step": 3016
+    },
+    {
+      "epoch": 1.3014880310545611,
+      "grad_norm": 0.18214569985866547,
+      "learning_rate": 9.14513230061517e-05,
+      "loss": 1.9803,
+      "step": 3017
+    },
+    {
+      "epoch": 1.3019193444037094,
+      "grad_norm": 0.19960394501686096,
+      "learning_rate": 9.14180244939833e-05,
+      "loss": 2.2102,
+      "step": 3018
+    },
+    {
+      "epoch": 1.3023506577528574,
+      "grad_norm": 0.18149137496948242,
+      "learning_rate": 9.138472258228337e-05,
+      "loss": 2.2894,
+      "step": 3019
+    },
+    {
+      "epoch": 1.3027819711020057,
+      "grad_norm": 0.17491424083709717,
+      "learning_rate": 9.135141727794745e-05,
+      "loss": 2.0448,
+      "step": 3020
+    },
+    {
+      "epoch": 1.3032132844511537,
+      "grad_norm": 0.16022880375385284,
+      "learning_rate": 9.131810858787179e-05,
+      "loss": 1.9723,
+      "step": 3021
+    },
+    {
+      "epoch": 1.303644597800302,
+      "grad_norm": 0.17898283898830414,
+      "learning_rate": 9.128479651895327e-05,
+      "loss": 2.097,
+      "step": 3022
+    },
+    {
+      "epoch": 1.3040759111494502,
+      "grad_norm": 0.2150205373764038,
+      "learning_rate": 9.125148107808955e-05,
+      "loss": 2.0051,
+      "step": 3023
+    },
+    {
+      "epoch": 1.3045072244985982,
+      "grad_norm": 0.20482948422431946,
+      "learning_rate": 9.121816227217895e-05,
+      "loss": 2.1562,
+      "step": 3024
+    },
+    {
+      "epoch": 1.3049385378477463,
+      "grad_norm": 0.16900508105754852,
+      "learning_rate": 9.118484010812051e-05,
+      "loss": 2.2336,
+      "step": 3025
+    },
+    {
+      "epoch": 1.3049385378477463,
+      "eval_loss": 2.0978634357452393,
+      "eval_runtime": 204.1551,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 3025
+    },
+    {
+      "epoch": 1.3053698511968945,
+      "grad_norm": 0.18790045380592346,
+      "learning_rate": 9.115151459281391e-05,
+      "loss": 2.2516,
+      "step": 3026
+    },
+    {
+      "epoch": 1.3058011645460428,
+      "grad_norm": 0.1752679944038391,
+      "learning_rate": 9.111818573315961e-05,
+      "loss": 2.111,
+      "step": 3027
+    },
+    {
+      "epoch": 1.3062324778951908,
+      "grad_norm": 0.25007501244544983,
+      "learning_rate": 9.108485353605866e-05,
+      "loss": 2.2764,
+      "step": 3028
+    },
+    {
+      "epoch": 1.306663791244339,
+      "grad_norm": 0.19131995737552643,
+      "learning_rate": 9.105151800841288e-05,
+      "loss": 2.2796,
+      "step": 3029
+    },
+    {
+      "epoch": 1.307095104593487,
+      "grad_norm": 0.18395444750785828,
+      "learning_rate": 9.101817915712478e-05,
+      "loss": 2.27,
+      "step": 3030
+    },
+    {
+      "epoch": 1.3075264179426354,
+      "grad_norm": 0.17417460680007935,
+      "learning_rate": 9.09848369890975e-05,
+      "loss": 2.0593,
+      "step": 3031
+    },
+    {
+      "epoch": 1.3079577312917836,
+      "grad_norm": 0.18455031514167786,
+      "learning_rate": 9.095149151123491e-05,
+      "loss": 2.1211,
+      "step": 3032
+    },
+    {
+      "epoch": 1.3083890446409316,
+      "grad_norm": 0.16537529230117798,
+      "learning_rate": 9.091814273044159e-05,
+      "loss": 2.2592,
+      "step": 3033
+    },
+    {
+      "epoch": 1.3088203579900797,
+      "grad_norm": 0.19210714101791382,
+      "learning_rate": 9.088479065362273e-05,
+      "loss": 2.1661,
+      "step": 3034
+    },
+    {
+      "epoch": 1.309251671339228,
+      "grad_norm": 0.2028828263282776,
+      "learning_rate": 9.085143528768424e-05,
+      "loss": 2.3139,
+      "step": 3035
+    },
+    {
+      "epoch": 1.3096829846883762,
+      "grad_norm": 0.19314557313919067,
+      "learning_rate": 9.081807663953272e-05,
+      "loss": 2.2221,
+      "step": 3036
+    },
+    {
+      "epoch": 1.3101142980375242,
+      "grad_norm": 0.19766665995121002,
+      "learning_rate": 9.078471471607547e-05,
+      "loss": 2.2387,
+      "step": 3037
+    },
+    {
+      "epoch": 1.3105456113866725,
+      "grad_norm": 0.21651652455329895,
+      "learning_rate": 9.07513495242204e-05,
+      "loss": 1.8804,
+      "step": 3038
+    },
+    {
+      "epoch": 1.3109769247358205,
+      "grad_norm": 0.20336247980594635,
+      "learning_rate": 9.071798107087614e-05,
+      "loss": 2.1254,
+      "step": 3039
+    },
+    {
+      "epoch": 1.3114082380849688,
+      "grad_norm": 0.18429327011108398,
+      "learning_rate": 9.068460936295205e-05,
+      "loss": 1.9877,
+      "step": 3040
+    },
+    {
+      "epoch": 1.311839551434117,
+      "grad_norm": 0.1765599399805069,
+      "learning_rate": 9.065123440735805e-05,
+      "loss": 2.0811,
+      "step": 3041
+    },
+    {
+      "epoch": 1.312270864783265,
+      "grad_norm": 0.16899888217449188,
+      "learning_rate": 9.061785621100481e-05,
+      "loss": 2.2005,
+      "step": 3042
+    },
+    {
+      "epoch": 1.312702178132413,
+      "grad_norm": 0.18764810264110565,
+      "learning_rate": 9.058447478080366e-05,
+      "loss": 2.2184,
+      "step": 3043
+    },
+    {
+      "epoch": 1.3131334914815613,
+      "grad_norm": 0.18549354374408722,
+      "learning_rate": 9.055109012366655e-05,
+      "loss": 2.1128,
+      "step": 3044
+    },
+    {
+      "epoch": 1.3135648048307096,
+      "grad_norm": 0.19809360802173615,
+      "learning_rate": 9.051770224650617e-05,
+      "loss": 2.2453,
+      "step": 3045
+    },
+    {
+      "epoch": 1.3139961181798576,
+      "grad_norm": 0.16052544116973877,
+      "learning_rate": 9.048431115623585e-05,
+      "loss": 2.136,
+      "step": 3046
+    },
+    {
+      "epoch": 1.3144274315290059,
+      "grad_norm": 0.1702003926038742,
+      "learning_rate": 9.045091685976957e-05,
+      "loss": 2.0531,
+      "step": 3047
+    },
+    {
+      "epoch": 1.314858744878154,
+      "grad_norm": 0.1690002977848053,
+      "learning_rate": 9.041751936402199e-05,
+      "loss": 2.0023,
+      "step": 3048
+    },
+    {
+      "epoch": 1.3152900582273022,
+      "grad_norm": 0.1664373129606247,
+      "learning_rate": 9.038411867590842e-05,
+      "loss": 2.1576,
+      "step": 3049
+    },
+    {
+      "epoch": 1.3157213715764504,
+      "grad_norm": 0.18579117953777313,
+      "learning_rate": 9.035071480234485e-05,
+      "loss": 2.159,
+      "step": 3050
+    },
+    {
+      "epoch": 1.3157213715764504,
+      "eval_loss": 2.097856044769287,
+      "eval_runtime": 203.1222,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3050
+    },
+    {
+      "epoch": 1.3161526849255984,
+      "grad_norm": 0.16586829721927643,
+      "learning_rate": 9.03173077502479e-05,
+      "loss": 1.9904,
+      "step": 3051
+    },
+    {
+      "epoch": 1.3165839982747465,
+      "grad_norm": 0.1913750171661377,
+      "learning_rate": 9.028389752653487e-05,
+      "loss": 2.1929,
+      "step": 3052
+    },
+    {
+      "epoch": 1.3170153116238947,
+      "grad_norm": 0.17231319844722748,
+      "learning_rate": 9.02504841381237e-05,
+      "loss": 2.132,
+      "step": 3053
+    },
+    {
+      "epoch": 1.317446624973043,
+      "grad_norm": 0.21133653819561005,
+      "learning_rate": 9.021706759193305e-05,
+      "loss": 2.2717,
+      "step": 3054
+    },
+    {
+      "epoch": 1.317877938322191,
+      "grad_norm": 0.17870862782001495,
+      "learning_rate": 9.018364789488211e-05,
+      "loss": 2.3588,
+      "step": 3055
+    },
+    {
+      "epoch": 1.3183092516713393,
+      "grad_norm": 0.18890124559402466,
+      "learning_rate": 9.015022505389084e-05,
+      "loss": 2.2494,
+      "step": 3056
+    },
+    {
+      "epoch": 1.3187405650204873,
+      "grad_norm": 0.1804681122303009,
+      "learning_rate": 9.01167990758798e-05,
+      "loss": 2.1849,
+      "step": 3057
+    },
+    {
+      "epoch": 1.3191718783696356,
+      "grad_norm": 0.17999668419361115,
+      "learning_rate": 9.00833699677702e-05,
+      "loss": 2.0232,
+      "step": 3058
+    },
+    {
+      "epoch": 1.3196031917187838,
+      "grad_norm": 0.16746574640274048,
+      "learning_rate": 9.004993773648389e-05,
+      "loss": 1.9494,
+      "step": 3059
+    },
+    {
+      "epoch": 1.3200345050679319,
+      "grad_norm": 0.209141343832016,
+      "learning_rate": 9.00165023889434e-05,
+      "loss": 2.3657,
+      "step": 3060
+    },
+    {
+      "epoch": 1.3204658184170799,
+      "grad_norm": 0.170698344707489,
+      "learning_rate": 8.998306393207183e-05,
+      "loss": 2.1525,
+      "step": 3061
+    },
+    {
+      "epoch": 1.3208971317662281,
+      "grad_norm": 0.18231551349163055,
+      "learning_rate": 8.994962237279306e-05,
+      "loss": 2.3213,
+      "step": 3062
+    },
+    {
+      "epoch": 1.3213284451153764,
+      "grad_norm": 0.19542478024959564,
+      "learning_rate": 8.991617771803148e-05,
+      "loss": 2.1471,
+      "step": 3063
+    },
+    {
+      "epoch": 1.3217597584645244,
+      "grad_norm": 0.19738048315048218,
+      "learning_rate": 8.988272997471219e-05,
+      "loss": 2.1084,
+      "step": 3064
+    },
+    {
+      "epoch": 1.3221910718136727,
+      "grad_norm": 0.24825488030910492,
+      "learning_rate": 8.984927914976091e-05,
+      "loss": 2.1697,
+      "step": 3065
+    },
+    {
+      "epoch": 1.3226223851628207,
+      "grad_norm": 0.1763385534286499,
+      "learning_rate": 8.981582525010398e-05,
+      "loss": 1.9507,
+      "step": 3066
+    },
+    {
+      "epoch": 1.323053698511969,
+      "grad_norm": 0.16423176229000092,
+      "learning_rate": 8.97823682826684e-05,
+      "loss": 2.2201,
+      "step": 3067
+    },
+    {
+      "epoch": 1.3234850118611172,
+      "grad_norm": 0.17444059252738953,
+      "learning_rate": 8.974890825438183e-05,
+      "loss": 2.2739,
+      "step": 3068
+    },
+    {
+      "epoch": 1.3239163252102653,
+      "grad_norm": 0.204995796084404,
+      "learning_rate": 8.97154451721725e-05,
+      "loss": 2.2925,
+      "step": 3069
+    },
+    {
+      "epoch": 1.3243476385594133,
+      "grad_norm": 0.19501759111881256,
+      "learning_rate": 8.968197904296933e-05,
+      "loss": 2.4235,
+      "step": 3070
+    },
+    {
+      "epoch": 1.3247789519085615,
+      "grad_norm": 0.17075209319591522,
+      "learning_rate": 8.964850987370182e-05,
+      "loss": 2.266,
+      "step": 3071
+    },
+    {
+      "epoch": 1.3252102652577098,
+      "grad_norm": 0.17689673602581024,
+      "learning_rate": 8.961503767130018e-05,
+      "loss": 2.1537,
+      "step": 3072
+    },
+    {
+      "epoch": 1.3256415786068578,
+      "grad_norm": 0.1606774777173996,
+      "learning_rate": 8.958156244269514e-05,
+      "loss": 2.1582,
+      "step": 3073
+    },
+    {
+      "epoch": 1.326072891956006,
+      "grad_norm": 0.17741374671459198,
+      "learning_rate": 8.954808419481815e-05,
+      "loss": 2.3015,
+      "step": 3074
+    },
+    {
+      "epoch": 1.3265042053051541,
+      "grad_norm": 0.19335578382015228,
+      "learning_rate": 8.951460293460124e-05,
+      "loss": 2.2417,
+      "step": 3075
+    },
+    {
+      "epoch": 1.3265042053051541,
+      "eval_loss": 2.097837209701538,
+      "eval_runtime": 202.0834,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3075
+    },
+    {
+      "epoch": 1.3269355186543024,
+      "grad_norm": 0.17709966003894806,
+      "learning_rate": 8.948111866897701e-05,
+      "loss": 2.171,
+      "step": 3076
+    },
+    {
+      "epoch": 1.3273668320034506,
+      "grad_norm": 0.17744885385036469,
+      "learning_rate": 8.944763140487885e-05,
+      "loss": 2.234,
+      "step": 3077
+    },
+    {
+      "epoch": 1.3277981453525987,
+      "grad_norm": 0.1813969761133194,
+      "learning_rate": 8.941414114924059e-05,
+      "loss": 2.2137,
+      "step": 3078
+    },
+    {
+      "epoch": 1.3282294587017467,
+      "grad_norm": 0.17423033714294434,
+      "learning_rate": 8.93806479089968e-05,
+      "loss": 1.9412,
+      "step": 3079
+    },
+    {
+      "epoch": 1.328660772050895,
+      "grad_norm": 0.17788293957710266,
+      "learning_rate": 8.934715169108257e-05,
+      "loss": 2.3469,
+      "step": 3080
+    },
+    {
+      "epoch": 1.3290920854000432,
+      "grad_norm": 0.1929638832807541,
+      "learning_rate": 8.931365250243368e-05,
+      "loss": 2.1577,
+      "step": 3081
+    },
+    {
+      "epoch": 1.3295233987491912,
+      "grad_norm": 0.18380503356456757,
+      "learning_rate": 8.928015034998653e-05,
+      "loss": 2.2365,
+      "step": 3082
+    },
+    {
+      "epoch": 1.3299547120983395,
+      "grad_norm": 0.1732213944196701,
+      "learning_rate": 8.924664524067807e-05,
+      "loss": 2.3558,
+      "step": 3083
+    },
+    {
+      "epoch": 1.3303860254474875,
+      "grad_norm": 0.16939488053321838,
+      "learning_rate": 8.921313718144592e-05,
+      "loss": 2.079,
+      "step": 3084
+    },
+    {
+      "epoch": 1.3308173387966358,
+      "grad_norm": 0.17569899559020996,
+      "learning_rate": 8.917962617922827e-05,
+      "loss": 2.0108,
+      "step": 3085
+    },
+    {
+      "epoch": 1.331248652145784,
+      "grad_norm": 0.16894367337226868,
+      "learning_rate": 8.914611224096398e-05,
+      "loss": 2.0602,
+      "step": 3086
+    },
+    {
+      "epoch": 1.331679965494932,
+      "grad_norm": 0.17975656688213348,
+      "learning_rate": 8.911259537359244e-05,
+      "loss": 2.2368,
+      "step": 3087
+    },
+    {
+      "epoch": 1.33211127884408,
+      "grad_norm": 0.17068015038967133,
+      "learning_rate": 8.90790755840537e-05,
+      "loss": 2.2709,
+      "step": 3088
+    },
+    {
+      "epoch": 1.3325425921932283,
+      "grad_norm": 0.17485836148262024,
+      "learning_rate": 8.904555287928838e-05,
+      "loss": 2.1873,
+      "step": 3089
+    },
+    {
+      "epoch": 1.3329739055423766,
+      "grad_norm": 0.1645931601524353,
+      "learning_rate": 8.901202726623775e-05,
+      "loss": 2.1562,
+      "step": 3090
+    },
+    {
+      "epoch": 1.3334052188915246,
+      "grad_norm": 0.16713939607143402,
+      "learning_rate": 8.897849875184363e-05,
+      "loss": 2.0063,
+      "step": 3091
+    },
+    {
+      "epoch": 1.333836532240673,
+      "grad_norm": 0.16838465631008148,
+      "learning_rate": 8.894496734304849e-05,
+      "loss": 2.1059,
+      "step": 3092
+    },
+    {
+      "epoch": 1.334267845589821,
+      "grad_norm": 0.17659622430801392,
+      "learning_rate": 8.891143304679534e-05,
+      "loss": 2.4106,
+      "step": 3093
+    },
+    {
+      "epoch": 1.3346991589389692,
+      "grad_norm": 0.19652199745178223,
+      "learning_rate": 8.887789587002787e-05,
+      "loss": 2.3738,
+      "step": 3094
+    },
+    {
+      "epoch": 1.3351304722881174,
+      "grad_norm": 0.196419358253479,
+      "learning_rate": 8.884435581969029e-05,
+      "loss": 2.0439,
+      "step": 3095
+    },
+    {
+      "epoch": 1.3355617856372655,
+      "grad_norm": 0.1740923374891281,
+      "learning_rate": 8.881081290272742e-05,
+      "loss": 2.1699,
+      "step": 3096
+    },
+    {
+      "epoch": 1.3359930989864135,
+      "grad_norm": 0.18634657561779022,
+      "learning_rate": 8.877726712608473e-05,
+      "loss": 2.2341,
+      "step": 3097
+    },
+    {
+      "epoch": 1.3364244123355618,
+      "grad_norm": 0.18123063445091248,
+      "learning_rate": 8.874371849670819e-05,
+      "loss": 2.2009,
+      "step": 3098
+    },
+    {
+      "epoch": 1.33685572568471,
+      "grad_norm": 0.16630500555038452,
+      "learning_rate": 8.871016702154445e-05,
+      "loss": 2.2254,
+      "step": 3099
+    },
+    {
+      "epoch": 1.337287039033858,
+      "grad_norm": 0.2106820046901703,
+      "learning_rate": 8.86766127075407e-05,
+      "loss": 2.1707,
+      "step": 3100
+    },
+    {
+      "epoch": 1.337287039033858,
+      "eval_loss": 2.0972704887390137,
+      "eval_runtime": 203.8969,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 3100
+    },
+    {
+      "epoch": 1.3377183523830063,
+      "grad_norm": 0.18919944763183594,
+      "learning_rate": 8.864305556164473e-05,
+      "loss": 2.1498,
+      "step": 3101
+    },
+    {
+      "epoch": 1.3381496657321543,
+      "grad_norm": 0.16261659562587738,
+      "learning_rate": 8.860949559080488e-05,
+      "loss": 2.0927,
+      "step": 3102
+    },
+    {
+      "epoch": 1.3385809790813026,
+      "grad_norm": 0.1766456514596939,
+      "learning_rate": 8.857593280197016e-05,
+      "loss": 2.1982,
+      "step": 3103
+    },
+    {
+      "epoch": 1.3390122924304508,
+      "grad_norm": 0.18764159083366394,
+      "learning_rate": 8.85423672020901e-05,
+      "loss": 2.1275,
+      "step": 3104
+    },
+    {
+      "epoch": 1.3394436057795989,
+      "grad_norm": 0.18761979043483734,
+      "learning_rate": 8.85087987981148e-05,
+      "loss": 2.3064,
+      "step": 3105
+    },
+    {
+      "epoch": 1.339874919128747,
+      "grad_norm": 0.17793145775794983,
+      "learning_rate": 8.847522759699496e-05,
+      "loss": 2.148,
+      "step": 3106
+    },
+    {
+      "epoch": 1.3403062324778952,
+      "grad_norm": 0.17504742741584778,
+      "learning_rate": 8.84416536056819e-05,
+      "loss": 2.2292,
+      "step": 3107
+    },
+    {
+      "epoch": 1.3407375458270434,
+      "grad_norm": 0.16357029974460602,
+      "learning_rate": 8.840807683112748e-05,
+      "loss": 2.0971,
+      "step": 3108
+    },
+    {
+      "epoch": 1.3411688591761914,
+      "grad_norm": 0.40526726841926575,
+      "learning_rate": 8.83744972802841e-05,
+      "loss": 2.1609,
+      "step": 3109
+    },
+    {
+      "epoch": 1.3416001725253397,
+      "grad_norm": 0.15176741778850555,
+      "learning_rate": 8.834091496010482e-05,
+      "loss": 1.9961,
+      "step": 3110
+    },
+    {
+      "epoch": 1.3420314858744877,
+      "grad_norm": 0.19039228558540344,
+      "learning_rate": 8.830732987754319e-05,
+      "loss": 1.9417,
+      "step": 3111
+    },
+    {
+      "epoch": 1.342462799223636,
+      "grad_norm": 0.1844143122434616,
+      "learning_rate": 8.827374203955338e-05,
+      "loss": 2.1261,
+      "step": 3112
+    },
+    {
+      "epoch": 1.3428941125727842,
+      "grad_norm": 0.176142156124115,
+      "learning_rate": 8.824015145309014e-05,
+      "loss": 2.254,
+      "step": 3113
+    },
+    {
+      "epoch": 1.3433254259219323,
+      "grad_norm": 0.17060662806034088,
+      "learning_rate": 8.820655812510874e-05,
+      "loss": 2.3113,
+      "step": 3114
+    },
+    {
+      "epoch": 1.3437567392710803,
+      "grad_norm": 0.16808998584747314,
+      "learning_rate": 8.817296206256504e-05,
+      "loss": 2.1092,
+      "step": 3115
+    },
+    {
+      "epoch": 1.3441880526202286,
+      "grad_norm": 0.1780100166797638,
+      "learning_rate": 8.813936327241549e-05,
+      "loss": 2.092,
+      "step": 3116
+    },
+    {
+      "epoch": 1.3446193659693768,
+      "grad_norm": 0.18709628283977509,
+      "learning_rate": 8.81057617616171e-05,
+      "loss": 2.2783,
+      "step": 3117
+    },
+    {
+      "epoch": 1.3450506793185248,
+      "grad_norm": 0.1694032847881317,
+      "learning_rate": 8.80721575371274e-05,
+      "loss": 2.0234,
+      "step": 3118
+    },
+    {
+      "epoch": 1.345481992667673,
+      "grad_norm": 0.18214310705661774,
+      "learning_rate": 8.803855060590455e-05,
+      "loss": 2.1824,
+      "step": 3119
+    },
+    {
+      "epoch": 1.3459133060168211,
+      "grad_norm": 0.19179998338222504,
+      "learning_rate": 8.80049409749072e-05,
+      "loss": 2.4183,
+      "step": 3120
+    },
+    {
+      "epoch": 1.3463446193659694,
+      "grad_norm": 0.18185855448246002,
+      "learning_rate": 8.79713286510946e-05,
+      "loss": 2.3265,
+      "step": 3121
+    },
+    {
+      "epoch": 1.3467759327151176,
+      "grad_norm": 0.18518030643463135,
+      "learning_rate": 8.793771364142653e-05,
+      "loss": 2.2387,
+      "step": 3122
+    },
+    {
+      "epoch": 1.3472072460642657,
+      "grad_norm": 0.17802715301513672,
+      "learning_rate": 8.790409595286339e-05,
+      "loss": 2.3093,
+      "step": 3123
+    },
+    {
+      "epoch": 1.347638559413414,
+      "grad_norm": 0.1845805048942566,
+      "learning_rate": 8.787047559236606e-05,
+      "loss": 2.0402,
+      "step": 3124
+    },
+    {
+      "epoch": 1.348069872762562,
+      "grad_norm": 0.20575165748596191,
+      "learning_rate": 8.7836852566896e-05,
+      "loss": 2.1048,
+      "step": 3125
+    },
+    {
+      "epoch": 1.348069872762562,
+      "eval_loss": 2.0977463722229004,
+      "eval_runtime": 201.6906,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3125
+    },
+    {
+      "epoch": 1.3485011861117102,
+      "grad_norm": 0.17934945225715637,
+      "learning_rate": 8.780322688341525e-05,
+      "loss": 2.1051,
+      "step": 3126
+    },
+    {
+      "epoch": 1.3489324994608582,
+      "grad_norm": 0.18327248096466064,
+      "learning_rate": 8.776959854888636e-05,
+      "loss": 2.3397,
+      "step": 3127
+    },
+    {
+      "epoch": 1.3493638128100065,
+      "grad_norm": 0.18778805434703827,
+      "learning_rate": 8.773596757027244e-05,
+      "loss": 2.0872,
+      "step": 3128
+    },
+    {
+      "epoch": 1.3497951261591545,
+      "grad_norm": 0.16798192262649536,
+      "learning_rate": 8.770233395453714e-05,
+      "loss": 2.0782,
+      "step": 3129
+    },
+    {
+      "epoch": 1.3502264395083028,
+      "grad_norm": 0.18652935326099396,
+      "learning_rate": 8.766869770864471e-05,
+      "loss": 2.1258,
+      "step": 3130
+    },
+    {
+      "epoch": 1.350657752857451,
+      "grad_norm": 0.1795220524072647,
+      "learning_rate": 8.763505883955986e-05,
+      "loss": 2.4144,
+      "step": 3131
+    },
+    {
+      "epoch": 1.351089066206599,
+      "grad_norm": 0.1760927140712738,
+      "learning_rate": 8.760141735424793e-05,
+      "loss": 2.3392,
+      "step": 3132
+    },
+    {
+      "epoch": 1.3515203795557473,
+      "grad_norm": 0.17478467524051666,
+      "learning_rate": 8.75677732596747e-05,
+      "loss": 2.1875,
+      "step": 3133
+    },
+    {
+      "epoch": 1.3519516929048954,
+      "grad_norm": 0.1812528520822525,
+      "learning_rate": 8.753412656280659e-05,
+      "loss": 2.2426,
+      "step": 3134
+    },
+    {
+      "epoch": 1.3523830062540436,
+      "grad_norm": 0.18718136847019196,
+      "learning_rate": 8.75004772706105e-05,
+      "loss": 2.2056,
+      "step": 3135
+    },
+    {
+      "epoch": 1.3528143196031917,
+      "grad_norm": 0.3149339556694031,
+      "learning_rate": 8.746682539005389e-05,
+      "loss": 2.1533,
+      "step": 3136
+    },
+    {
+      "epoch": 1.35324563295234,
+      "grad_norm": 0.1598491221666336,
+      "learning_rate": 8.743317092810475e-05,
+      "loss": 2.0824,
+      "step": 3137
+    },
+    {
+      "epoch": 1.353676946301488,
+      "grad_norm": 0.16594962775707245,
+      "learning_rate": 8.739951389173159e-05,
+      "loss": 2.0977,
+      "step": 3138
+    },
+    {
+      "epoch": 1.3541082596506362,
+      "grad_norm": 0.1650681346654892,
+      "learning_rate": 8.736585428790348e-05,
+      "loss": 2.1672,
+      "step": 3139
+    },
+    {
+      "epoch": 1.3545395729997844,
+      "grad_norm": 0.17856769263744354,
+      "learning_rate": 8.733219212359004e-05,
+      "loss": 2.1519,
+      "step": 3140
+    },
+    {
+      "epoch": 1.3549708863489325,
+      "grad_norm": 0.17163975536823273,
+      "learning_rate": 8.729852740576132e-05,
+      "loss": 2.1415,
+      "step": 3141
+    },
+    {
+      "epoch": 1.3554021996980807,
+      "grad_norm": 0.1770739108324051,
+      "learning_rate": 8.726486014138803e-05,
+      "loss": 2.0446,
+      "step": 3142
+    },
+    {
+      "epoch": 1.3558335130472288,
+      "grad_norm": 0.17098559439182281,
+      "learning_rate": 8.723119033744133e-05,
+      "loss": 2.2327,
+      "step": 3143
+    },
+    {
+      "epoch": 1.356264826396377,
+      "grad_norm": 0.1768409013748169,
+      "learning_rate": 8.71975180008929e-05,
+      "loss": 2.1411,
+      "step": 3144
+    },
+    {
+      "epoch": 1.356696139745525,
+      "grad_norm": 0.18088935315608978,
+      "learning_rate": 8.716384313871495e-05,
+      "loss": 2.1351,
+      "step": 3145
+    },
+    {
+      "epoch": 1.3571274530946733,
+      "grad_norm": 0.1696568876504898,
+      "learning_rate": 8.71301657578803e-05,
+      "loss": 2.0054,
+      "step": 3146
+    },
+    {
+      "epoch": 1.3575587664438213,
+      "grad_norm": 0.19234417378902435,
+      "learning_rate": 8.709648586536214e-05,
+      "loss": 2.1927,
+      "step": 3147
+    },
+    {
+      "epoch": 1.3579900797929696,
+      "grad_norm": 0.1776120811700821,
+      "learning_rate": 8.706280346813434e-05,
+      "loss": 2.0688,
+      "step": 3148
+    },
+    {
+      "epoch": 1.3584213931421179,
+      "grad_norm": 0.7187220454216003,
+      "learning_rate": 8.702911857317112e-05,
+      "loss": 2.2263,
+      "step": 3149
+    },
+    {
+      "epoch": 1.3588527064912659,
+      "grad_norm": 0.1841685175895691,
+      "learning_rate": 8.699543118744739e-05,
+      "loss": 2.2218,
+      "step": 3150
+    },
+    {
+      "epoch": 1.3588527064912659,
+      "eval_loss": 2.0971086025238037,
+      "eval_runtime": 202.3868,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3150
+    },
+    {
+      "epoch": 1.3592840198404141,
+      "grad_norm": 0.16733163595199585,
+      "learning_rate": 8.696174131793843e-05,
+      "loss": 2.1474,
+      "step": 3151
+    },
+    {
+      "epoch": 1.3597153331895622,
+      "grad_norm": 0.17958466708660126,
+      "learning_rate": 8.692804897162011e-05,
+      "loss": 2.3287,
+      "step": 3152
+    },
+    {
+      "epoch": 1.3601466465387104,
+      "grad_norm": 0.18487928807735443,
+      "learning_rate": 8.689435415546883e-05,
+      "loss": 2.2847,
+      "step": 3153
+    },
+    {
+      "epoch": 1.3605779598878585,
+      "grad_norm": 0.1870775669813156,
+      "learning_rate": 8.686065687646145e-05,
+      "loss": 2.1921,
+      "step": 3154
+    },
+    {
+      "epoch": 1.3610092732370067,
+      "grad_norm": 0.17926892638206482,
+      "learning_rate": 8.682695714157533e-05,
+      "loss": 2.0102,
+      "step": 3155
+    },
+    {
+      "epoch": 1.3614405865861547,
+      "grad_norm": 0.1682741492986679,
+      "learning_rate": 8.679325495778842e-05,
+      "loss": 2.0879,
+      "step": 3156
+    },
+    {
+      "epoch": 1.361871899935303,
+      "grad_norm": 0.18406817317008972,
+      "learning_rate": 8.67595503320791e-05,
+      "loss": 2.1946,
+      "step": 3157
+    },
+    {
+      "epoch": 1.3623032132844513,
+      "grad_norm": 0.18852907419204712,
+      "learning_rate": 8.672584327142627e-05,
+      "loss": 2.285,
+      "step": 3158
+    },
+    {
+      "epoch": 1.3627345266335993,
+      "grad_norm": 0.1969786286354065,
+      "learning_rate": 8.669213378280938e-05,
+      "loss": 2.4023,
+      "step": 3159
+    },
+    {
+      "epoch": 1.3631658399827475,
+      "grad_norm": 0.17748579382896423,
+      "learning_rate": 8.66584218732083e-05,
+      "loss": 2.1213,
+      "step": 3160
+    },
+    {
+      "epoch": 1.3635971533318956,
+      "grad_norm": 0.16388718783855438,
+      "learning_rate": 8.662470754960349e-05,
+      "loss": 1.9148,
+      "step": 3161
+    },
+    {
+      "epoch": 1.3640284666810438,
+      "grad_norm": 0.27124521136283875,
+      "learning_rate": 8.659099081897587e-05,
+      "loss": 2.1991,
+      "step": 3162
+    },
+    {
+      "epoch": 1.3644597800301919,
+      "grad_norm": 0.18605460226535797,
+      "learning_rate": 8.655727168830681e-05,
+      "loss": 2.2165,
+      "step": 3163
+    },
+    {
+      "epoch": 1.3648910933793401,
+      "grad_norm": 0.17285169661045074,
+      "learning_rate": 8.652355016457831e-05,
+      "loss": 2.1461,
+      "step": 3164
+    },
+    {
+      "epoch": 1.3653224067284881,
+      "grad_norm": 0.18387822806835175,
+      "learning_rate": 8.648982625477268e-05,
+      "loss": 2.1026,
+      "step": 3165
+    },
+    {
+      "epoch": 1.3657537200776364,
+      "grad_norm": 0.17600493133068085,
+      "learning_rate": 8.64560999658729e-05,
+      "loss": 2.1583,
+      "step": 3166
+    },
+    {
+      "epoch": 1.3661850334267847,
+      "grad_norm": 0.1766958236694336,
+      "learning_rate": 8.642237130486234e-05,
+      "loss": 2.1425,
+      "step": 3167
+    },
+    {
+      "epoch": 1.3666163467759327,
+      "grad_norm": 1.2923835515975952,
+      "learning_rate": 8.638864027872487e-05,
+      "loss": 2.142,
+      "step": 3168
+    },
+    {
+      "epoch": 1.367047660125081,
+      "grad_norm": 0.18593095242977142,
+      "learning_rate": 8.63549068944449e-05,
+      "loss": 2.2189,
+      "step": 3169
+    },
+    {
+      "epoch": 1.367478973474229,
+      "grad_norm": 0.19588585197925568,
+      "learning_rate": 8.632117115900728e-05,
+      "loss": 2.2355,
+      "step": 3170
+    },
+    {
+      "epoch": 1.3679102868233772,
+      "grad_norm": 0.1888759285211563,
+      "learning_rate": 8.628743307939737e-05,
+      "loss": 2.1629,
+      "step": 3171
+    },
+    {
+      "epoch": 1.3683416001725253,
+      "grad_norm": 0.17903682589530945,
+      "learning_rate": 8.6253692662601e-05,
+      "loss": 2.0798,
+      "step": 3172
+    },
+    {
+      "epoch": 1.3687729135216735,
+      "grad_norm": 0.18080520629882812,
+      "learning_rate": 8.621994991560449e-05,
+      "loss": 2.2332,
+      "step": 3173
+    },
+    {
+      "epoch": 1.3692042268708216,
+      "grad_norm": 0.20339682698249817,
+      "learning_rate": 8.618620484539467e-05,
+      "loss": 2.1999,
+      "step": 3174
+    },
+    {
+      "epoch": 1.3696355402199698,
+      "grad_norm": 0.1827765554189682,
+      "learning_rate": 8.615245745895877e-05,
+      "loss": 2.0979,
+      "step": 3175
+    },
+    {
+      "epoch": 1.3696355402199698,
+      "eval_loss": 2.0981106758117676,
+      "eval_runtime": 202.6773,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3175
+    },
+    {
+      "epoch": 1.370066853569118,
+      "grad_norm": 0.19208596646785736,
+      "learning_rate": 8.611870776328463e-05,
+      "loss": 2.1504,
+      "step": 3176
+    },
+    {
+      "epoch": 1.370498166918266,
+      "grad_norm": 0.17461636662483215,
+      "learning_rate": 8.608495576536042e-05,
+      "loss": 1.9311,
+      "step": 3177
+    },
+    {
+      "epoch": 1.3709294802674143,
+      "grad_norm": 0.17411647737026215,
+      "learning_rate": 8.605120147217493e-05,
+      "loss": 2.305,
+      "step": 3178
+    },
+    {
+      "epoch": 1.3713607936165624,
+      "grad_norm": 0.1835533082485199,
+      "learning_rate": 8.601744489071731e-05,
+      "loss": 2.1118,
+      "step": 3179
+    },
+    {
+      "epoch": 1.3717921069657106,
+      "grad_norm": 0.17653480172157288,
+      "learning_rate": 8.598368602797723e-05,
+      "loss": 2.1825,
+      "step": 3180
+    },
+    {
+      "epoch": 1.3722234203148587,
+      "grad_norm": 0.17974716424942017,
+      "learning_rate": 8.594992489094486e-05,
+      "loss": 2.2029,
+      "step": 3181
+    },
+    {
+      "epoch": 1.372654733664007,
+      "grad_norm": 0.1939789205789566,
+      "learning_rate": 8.591616148661078e-05,
+      "loss": 2.2997,
+      "step": 3182
+    },
+    {
+      "epoch": 1.373086047013155,
+      "grad_norm": 0.19809889793395996,
+      "learning_rate": 8.58823958219661e-05,
+      "loss": 2.1236,
+      "step": 3183
+    },
+    {
+      "epoch": 1.3735173603623032,
+      "grad_norm": 0.19726596772670746,
+      "learning_rate": 8.584862790400232e-05,
+      "loss": 2.1344,
+      "step": 3184
+    },
+    {
+      "epoch": 1.3739486737114515,
+      "grad_norm": 0.18464700877666473,
+      "learning_rate": 8.581485773971153e-05,
+      "loss": 2.198,
+      "step": 3185
+    },
+    {
+      "epoch": 1.3743799870605995,
+      "grad_norm": 0.18575282394886017,
+      "learning_rate": 8.578108533608618e-05,
+      "loss": 2.3213,
+      "step": 3186
+    },
+    {
+      "epoch": 1.3748113004097477,
+      "grad_norm": 0.1865396797657013,
+      "learning_rate": 8.57473107001192e-05,
+      "loss": 2.3166,
+      "step": 3187
+    },
+    {
+      "epoch": 1.3752426137588958,
+      "grad_norm": 0.1792069673538208,
+      "learning_rate": 8.571353383880401e-05,
+      "loss": 2.152,
+      "step": 3188
+    },
+    {
+      "epoch": 1.375673927108044,
+      "grad_norm": 0.16752682626247406,
+      "learning_rate": 8.567975475913448e-05,
+      "loss": 2.3021,
+      "step": 3189
+    },
+    {
+      "epoch": 1.376105240457192,
+      "grad_norm": 0.1645505130290985,
+      "learning_rate": 8.564597346810492e-05,
+      "loss": 1.9309,
+      "step": 3190
+    },
+    {
+      "epoch": 1.3765365538063403,
+      "grad_norm": 0.18013405799865723,
+      "learning_rate": 8.561218997271014e-05,
+      "loss": 2.1577,
+      "step": 3191
+    },
+    {
+      "epoch": 1.3769678671554884,
+      "grad_norm": 0.1741134524345398,
+      "learning_rate": 8.557840427994536e-05,
+      "loss": 2.1166,
+      "step": 3192
+    },
+    {
+      "epoch": 1.3773991805046366,
+      "grad_norm": 0.184743270277977,
+      "learning_rate": 8.554461639680632e-05,
+      "loss": 2.2432,
+      "step": 3193
+    },
+    {
+      "epoch": 1.3778304938537849,
+      "grad_norm": 0.19026319682598114,
+      "learning_rate": 8.551082633028913e-05,
+      "loss": 2.2845,
+      "step": 3194
+    },
+    {
+      "epoch": 1.378261807202933,
+      "grad_norm": 0.1513228416442871,
+      "learning_rate": 8.547703408739042e-05,
+      "loss": 2.0565,
+      "step": 3195
+    },
+    {
+      "epoch": 1.3786931205520812,
+      "grad_norm": 0.18880301713943481,
+      "learning_rate": 8.544323967510718e-05,
+      "loss": 2.3363,
+      "step": 3196
+    },
+    {
+      "epoch": 1.3791244339012292,
+      "grad_norm": 0.18461208045482635,
+      "learning_rate": 8.540944310043702e-05,
+      "loss": 2.3256,
+      "step": 3197
+    },
+    {
+      "epoch": 1.3795557472503774,
+      "grad_norm": 0.18815840780735016,
+      "learning_rate": 8.537564437037778e-05,
+      "loss": 2.179,
+      "step": 3198
+    },
+    {
+      "epoch": 1.3799870605995255,
+      "grad_norm": 0.1892249882221222,
+      "learning_rate": 8.534184349192792e-05,
+      "loss": 2.2513,
+      "step": 3199
+    },
+    {
+      "epoch": 1.3804183739486737,
+      "grad_norm": 0.17379261553287506,
+      "learning_rate": 8.530804047208627e-05,
+      "loss": 1.7714,
+      "step": 3200
+    },
+    {
+      "epoch": 1.3804183739486737,
+      "eval_loss": 2.0976791381835938,
+      "eval_runtime": 196.5304,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 3200
+    },
+    {
+      "epoch": 1.3808496872978218,
+      "grad_norm": 0.17144569754600525,
+      "learning_rate": 8.527423531785213e-05,
+      "loss": 1.9854,
+      "step": 3201
+    },
+    {
+      "epoch": 1.38128100064697,
+      "grad_norm": 0.184304341673851,
+      "learning_rate": 8.524042803622519e-05,
+      "loss": 2.1147,
+      "step": 3202
+    },
+    {
+      "epoch": 1.3817123139961183,
+      "grad_norm": 0.18346154689788818,
+      "learning_rate": 8.520661863420565e-05,
+      "loss": 1.8588,
+      "step": 3203
+    },
+    {
+      "epoch": 1.3821436273452663,
+      "grad_norm": 0.17854824662208557,
+      "learning_rate": 8.51728071187941e-05,
+      "loss": 2.1437,
+      "step": 3204
+    },
+    {
+      "epoch": 1.3825749406944146,
+      "grad_norm": 0.23320263624191284,
+      "learning_rate": 8.513899349699158e-05,
+      "loss": 2.0817,
+      "step": 3205
+    },
+    {
+      "epoch": 1.3830062540435626,
+      "grad_norm": 0.19302494823932648,
+      "learning_rate": 8.51051777757996e-05,
+      "loss": 2.2875,
+      "step": 3206
+    },
+    {
+      "epoch": 1.3834375673927108,
+      "grad_norm": 0.1735406219959259,
+      "learning_rate": 8.507135996222002e-05,
+      "loss": 2.0925,
+      "step": 3207
+    },
+    {
+      "epoch": 1.383868880741859,
+      "grad_norm": 0.1883225440979004,
+      "learning_rate": 8.503754006325524e-05,
+      "loss": 2.1741,
+      "step": 3208
+    },
+    {
+      "epoch": 1.3843001940910071,
+      "grad_norm": 0.20691944658756256,
+      "learning_rate": 8.500371808590802e-05,
+      "loss": 2.2111,
+      "step": 3209
+    },
+    {
+      "epoch": 1.3847315074401552,
+      "grad_norm": 0.22449901700019836,
+      "learning_rate": 8.496989403718159e-05,
+      "loss": 1.9673,
+      "step": 3210
+    },
+    {
+      "epoch": 1.3851628207893034,
+      "grad_norm": 0.19177977740764618,
+      "learning_rate": 8.493606792407954e-05,
+      "loss": 2.0311,
+      "step": 3211
+    },
+    {
+      "epoch": 1.3855941341384517,
+      "grad_norm": 0.1945643126964569,
+      "learning_rate": 8.4902239753606e-05,
+      "loss": 2.1568,
+      "step": 3212
+    },
+    {
+      "epoch": 1.3860254474875997,
+      "grad_norm": 1.4558762311935425,
+      "learning_rate": 8.486840953276541e-05,
+      "loss": 2.2195,
+      "step": 3213
+    },
+    {
+      "epoch": 1.386456760836748,
+      "grad_norm": 0.21869879961013794,
+      "learning_rate": 8.483457726856268e-05,
+      "loss": 2.2545,
+      "step": 3214
+    },
+    {
+      "epoch": 1.386888074185896,
+      "grad_norm": 0.1923229843378067,
+      "learning_rate": 8.480074296800322e-05,
+      "loss": 2.0498,
+      "step": 3215
+    },
+    {
+      "epoch": 1.3873193875350442,
+      "grad_norm": 0.18915557861328125,
+      "learning_rate": 8.476690663809273e-05,
+      "loss": 2.1517,
+      "step": 3216
+    },
+    {
+      "epoch": 1.3877507008841925,
+      "grad_norm": 0.16886180639266968,
+      "learning_rate": 8.473306828583742e-05,
+      "loss": 2.1885,
+      "step": 3217
+    },
+    {
+      "epoch": 1.3881820142333405,
+      "grad_norm": 0.18701143562793732,
+      "learning_rate": 8.469922791824387e-05,
+      "loss": 2.1815,
+      "step": 3218
+    },
+    {
+      "epoch": 1.3886133275824886,
+      "grad_norm": 0.17433500289916992,
+      "learning_rate": 8.466538554231913e-05,
+      "loss": 2.1176,
+      "step": 3219
+    },
+    {
+      "epoch": 1.3890446409316368,
+      "grad_norm": 0.18689289689064026,
+      "learning_rate": 8.463154116507061e-05,
+      "loss": 1.9017,
+      "step": 3220
+    },
+    {
+      "epoch": 1.389475954280785,
+      "grad_norm": 0.18619681894779205,
+      "learning_rate": 8.459769479350614e-05,
+      "loss": 2.2011,
+      "step": 3221
+    },
+    {
+      "epoch": 1.389907267629933,
+      "grad_norm": 0.19147801399230957,
+      "learning_rate": 8.456384643463404e-05,
+      "loss": 2.2078,
+      "step": 3222
+    },
+    {
+      "epoch": 1.3903385809790814,
+      "grad_norm": 0.18116207420825958,
+      "learning_rate": 8.452999609546291e-05,
+      "loss": 2.1176,
+      "step": 3223
+    },
+    {
+      "epoch": 1.3907698943282294,
+      "grad_norm": 0.17653116583824158,
+      "learning_rate": 8.449614378300189e-05,
+      "loss": 2.2895,
+      "step": 3224
+    },
+    {
+      "epoch": 1.3912012076773776,
+      "grad_norm": 0.18579091131687164,
+      "learning_rate": 8.446228950426045e-05,
+      "loss": 2.1845,
+      "step": 3225
+    },
+    {
+      "epoch": 1.3912012076773776,
+      "eval_loss": 2.097705364227295,
+      "eval_runtime": 194.6266,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3225
+    },
+    {
+      "epoch": 1.391632521026526,
+      "grad_norm": 0.18824826180934906,
+      "learning_rate": 8.44284332662485e-05,
+      "loss": 2.1208,
+      "step": 3226
+    },
+    {
+      "epoch": 1.392063834375674,
+      "grad_norm": 0.19434987008571625,
+      "learning_rate": 8.439457507597633e-05,
+      "loss": 2.2117,
+      "step": 3227
+    },
+    {
+      "epoch": 1.392495147724822,
+      "grad_norm": 0.17461137473583221,
+      "learning_rate": 8.436071494045466e-05,
+      "loss": 2.0738,
+      "step": 3228
+    },
+    {
+      "epoch": 1.3929264610739702,
+      "grad_norm": 0.18477419018745422,
+      "learning_rate": 8.432685286669458e-05,
+      "loss": 2.208,
+      "step": 3229
+    },
+    {
+      "epoch": 1.3933577744231185,
+      "grad_norm": 0.17233210802078247,
+      "learning_rate": 8.42929888617076e-05,
+      "loss": 2.2569,
+      "step": 3230
+    },
+    {
+      "epoch": 1.3937890877722665,
+      "grad_norm": 0.20839418470859528,
+      "learning_rate": 8.425912293250567e-05,
+      "loss": 2.1946,
+      "step": 3231
+    },
+    {
+      "epoch": 1.3942204011214148,
+      "grad_norm": 0.18050678074359894,
+      "learning_rate": 8.422525508610109e-05,
+      "loss": 2.0627,
+      "step": 3232
+    },
+    {
+      "epoch": 1.3946517144705628,
+      "grad_norm": 0.19521097838878632,
+      "learning_rate": 8.419138532950655e-05,
+      "loss": 2.3172,
+      "step": 3233
+    },
+    {
+      "epoch": 1.395083027819711,
+      "grad_norm": 0.16553284227848053,
+      "learning_rate": 8.415751366973517e-05,
+      "loss": 2.004,
+      "step": 3234
+    },
+    {
+      "epoch": 1.3955143411688593,
+      "grad_norm": 0.18260253965854645,
+      "learning_rate": 8.412364011380042e-05,
+      "loss": 2.2002,
+      "step": 3235
+    },
+    {
+      "epoch": 1.3959456545180073,
+      "grad_norm": 0.17164035141468048,
+      "learning_rate": 8.408976466871623e-05,
+      "loss": 2.2397,
+      "step": 3236
+    },
+    {
+      "epoch": 1.3963769678671554,
+      "grad_norm": 0.2340926080942154,
+      "learning_rate": 8.405588734149684e-05,
+      "loss": 2.125,
+      "step": 3237
+    },
+    {
+      "epoch": 1.3968082812163036,
+      "grad_norm": 0.18080025911331177,
+      "learning_rate": 8.402200813915698e-05,
+      "loss": 2.1483,
+      "step": 3238
+    },
+    {
+      "epoch": 1.3972395945654519,
+      "grad_norm": 0.18892081081867218,
+      "learning_rate": 8.398812706871163e-05,
+      "loss": 2.2249,
+      "step": 3239
+    },
+    {
+      "epoch": 1.3976709079146,
+      "grad_norm": 0.2113853543996811,
+      "learning_rate": 8.395424413717628e-05,
+      "loss": 2.2859,
+      "step": 3240
+    },
+    {
+      "epoch": 1.3981022212637482,
+      "grad_norm": 0.18064041435718536,
+      "learning_rate": 8.39203593515668e-05,
+      "loss": 2.06,
+      "step": 3241
+    },
+    {
+      "epoch": 1.3985335346128962,
+      "grad_norm": 0.1873074322938919,
+      "learning_rate": 8.388647271889933e-05,
+      "loss": 2.2682,
+      "step": 3242
+    },
+    {
+      "epoch": 1.3989648479620445,
+      "grad_norm": 0.18234270811080933,
+      "learning_rate": 8.385258424619053e-05,
+      "loss": 2.1766,
+      "step": 3243
+    },
+    {
+      "epoch": 1.3993961613111927,
+      "grad_norm": 0.19495293498039246,
+      "learning_rate": 8.38186939404573e-05,
+      "loss": 2.4004,
+      "step": 3244
+    },
+    {
+      "epoch": 1.3998274746603407,
+      "grad_norm": 0.16893860697746277,
+      "learning_rate": 8.37848018087171e-05,
+      "loss": 2.0107,
+      "step": 3245
+    },
+    {
+      "epoch": 1.4002587880094888,
+      "grad_norm": 0.18764536082744598,
+      "learning_rate": 8.37509078579876e-05,
+      "loss": 2.2513,
+      "step": 3246
+    },
+    {
+      "epoch": 1.400690101358637,
+      "grad_norm": 0.1757303923368454,
+      "learning_rate": 8.371701209528691e-05,
+      "loss": 2.1406,
+      "step": 3247
+    },
+    {
+      "epoch": 1.4011214147077853,
+      "grad_norm": 0.1827249675989151,
+      "learning_rate": 8.368311452763358e-05,
+      "loss": 2.3251,
+      "step": 3248
+    },
+    {
+      "epoch": 1.4015527280569333,
+      "grad_norm": 0.17690004408359528,
+      "learning_rate": 8.364921516204641e-05,
+      "loss": 2.175,
+      "step": 3249
+    },
+    {
+      "epoch": 1.4019840414060816,
+      "grad_norm": 0.2701353132724762,
+      "learning_rate": 8.361531400554465e-05,
+      "loss": 2.0582,
+      "step": 3250
+    },
+    {
+      "epoch": 1.4019840414060816,
+      "eval_loss": 2.096698522567749,
+      "eval_runtime": 194.8302,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3250
+    },
+    {
+      "epoch": 1.4024153547552296,
+      "grad_norm": 0.19098925590515137,
+      "learning_rate": 8.358141106514789e-05,
+      "loss": 2.2413,
+      "step": 3251
+    },
+    {
+      "epoch": 1.4028466681043779,
+      "grad_norm": 0.19250550866127014,
+      "learning_rate": 8.354750634787618e-05,
+      "loss": 1.9567,
+      "step": 3252
+    },
+    {
+      "epoch": 1.4032779814535261,
+      "grad_norm": 0.19768285751342773,
+      "learning_rate": 8.351359986074975e-05,
+      "loss": 2.133,
+      "step": 3253
+    },
+    {
+      "epoch": 1.4037092948026741,
+      "grad_norm": 0.18588483333587646,
+      "learning_rate": 8.34796916107894e-05,
+      "loss": 2.3247,
+      "step": 3254
+    },
+    {
+      "epoch": 1.4041406081518222,
+      "grad_norm": 0.19821996986865997,
+      "learning_rate": 8.344578160501615e-05,
+      "loss": 2.224,
+      "step": 3255
+    },
+    {
+      "epoch": 1.4045719215009704,
+      "grad_norm": 0.18002888560295105,
+      "learning_rate": 8.341186985045148e-05,
+      "loss": 2.1396,
+      "step": 3256
+    },
+    {
+      "epoch": 1.4050032348501187,
+      "grad_norm": 0.18589261174201965,
+      "learning_rate": 8.337795635411715e-05,
+      "loss": 2.18,
+      "step": 3257
+    },
+    {
+      "epoch": 1.4054345481992667,
+      "grad_norm": 0.1814439445734024,
+      "learning_rate": 8.334404112303535e-05,
+      "loss": 2.1433,
+      "step": 3258
+    },
+    {
+      "epoch": 1.405865861548415,
+      "grad_norm": 0.179869145154953,
+      "learning_rate": 8.331012416422855e-05,
+      "loss": 2.1902,
+      "step": 3259
+    },
+    {
+      "epoch": 1.406297174897563,
+      "grad_norm": 0.19356581568717957,
+      "learning_rate": 8.327620548471969e-05,
+      "loss": 2.0897,
+      "step": 3260
+    },
+    {
+      "epoch": 1.4067284882467113,
+      "grad_norm": 0.19356249272823334,
+      "learning_rate": 8.324228509153199e-05,
+      "loss": 2.3182,
+      "step": 3261
+    },
+    {
+      "epoch": 1.4071598015958595,
+      "grad_norm": 0.1817573457956314,
+      "learning_rate": 8.320836299168901e-05,
+      "loss": 2.288,
+      "step": 3262
+    },
+    {
+      "epoch": 1.4075911149450075,
+      "grad_norm": 0.17940185964107513,
+      "learning_rate": 8.317443919221471e-05,
+      "loss": 2.054,
+      "step": 3263
+    },
+    {
+      "epoch": 1.4080224282941556,
+      "grad_norm": 0.1818537712097168,
+      "learning_rate": 8.314051370013338e-05,
+      "loss": 2.1608,
+      "step": 3264
+    },
+    {
+      "epoch": 1.4084537416433038,
+      "grad_norm": 0.19189421832561493,
+      "learning_rate": 8.310658652246967e-05,
+      "loss": 2.1291,
+      "step": 3265
+    },
+    {
+      "epoch": 1.408885054992452,
+      "grad_norm": 0.19296544790267944,
+      "learning_rate": 8.307265766624857e-05,
+      "loss": 2.2991,
+      "step": 3266
+    },
+    {
+      "epoch": 1.4093163683416001,
+      "grad_norm": 0.20070980489253998,
+      "learning_rate": 8.30387271384954e-05,
+      "loss": 2.2874,
+      "step": 3267
+    },
+    {
+      "epoch": 1.4097476816907484,
+      "grad_norm": 0.18332870304584503,
+      "learning_rate": 8.300479494623589e-05,
+      "loss": 2.2083,
+      "step": 3268
+    },
+    {
+      "epoch": 1.4101789950398964,
+      "grad_norm": 0.19442547857761383,
+      "learning_rate": 8.297086109649604e-05,
+      "loss": 2.1842,
+      "step": 3269
+    },
+    {
+      "epoch": 1.4106103083890447,
+      "grad_norm": 0.17771577835083008,
+      "learning_rate": 8.293692559630223e-05,
+      "loss": 2.188,
+      "step": 3270
+    },
+    {
+      "epoch": 1.411041621738193,
+      "grad_norm": 0.1854981929063797,
+      "learning_rate": 8.290298845268117e-05,
+      "loss": 2.2063,
+      "step": 3271
+    },
+    {
+      "epoch": 1.411472935087341,
+      "grad_norm": 0.1690223664045334,
+      "learning_rate": 8.286904967265994e-05,
+      "loss": 2.028,
+      "step": 3272
+    },
+    {
+      "epoch": 1.411904248436489,
+      "grad_norm": 0.18008871376514435,
+      "learning_rate": 8.283510926326592e-05,
+      "loss": 2.2517,
+      "step": 3273
+    },
+    {
+      "epoch": 1.4123355617856372,
+      "grad_norm": 0.17644241452217102,
+      "learning_rate": 8.280116723152681e-05,
+      "loss": 2.1543,
+      "step": 3274
+    },
+    {
+      "epoch": 1.4127668751347855,
+      "grad_norm": 0.18835556507110596,
+      "learning_rate": 8.276722358447078e-05,
+      "loss": 2.2019,
+      "step": 3275
+    },
+    {
+      "epoch": 1.4127668751347855,
+      "eval_loss": 2.0969247817993164,
+      "eval_runtime": 194.9691,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3275
+    },
+    {
+      "epoch": 1.4131981884839335,
+      "grad_norm": 0.21935436129570007,
+      "learning_rate": 8.273327832912611e-05,
+      "loss": 2.4498,
+      "step": 3276
+    },
+    {
+      "epoch": 1.4136295018330818,
+      "grad_norm": 0.18742847442626953,
+      "learning_rate": 8.26993314725216e-05,
+      "loss": 2.2589,
+      "step": 3277
+    },
+    {
+      "epoch": 1.4140608151822298,
+      "grad_norm": 0.19992081820964813,
+      "learning_rate": 8.266538302168632e-05,
+      "loss": 2.1686,
+      "step": 3278
+    },
+    {
+      "epoch": 1.414492128531378,
+      "grad_norm": 0.18198950588703156,
+      "learning_rate": 8.263143298364966e-05,
+      "loss": 2.1204,
+      "step": 3279
+    },
+    {
+      "epoch": 1.4149234418805263,
+      "grad_norm": 0.19453582167625427,
+      "learning_rate": 8.259748136544134e-05,
+      "loss": 2.3806,
+      "step": 3280
+    },
+    {
+      "epoch": 1.4153547552296744,
+      "grad_norm": 0.18559814989566803,
+      "learning_rate": 8.256352817409142e-05,
+      "loss": 2.2214,
+      "step": 3281
+    },
+    {
+      "epoch": 1.4157860685788224,
+      "grad_norm": 0.1632603406906128,
+      "learning_rate": 8.252957341663028e-05,
+      "loss": 2.2278,
+      "step": 3282
+    },
+    {
+      "epoch": 1.4162173819279706,
+      "grad_norm": 0.16929636895656586,
+      "learning_rate": 8.249561710008861e-05,
+      "loss": 2.2441,
+      "step": 3283
+    },
+    {
+      "epoch": 1.416648695277119,
+      "grad_norm": 0.17531128227710724,
+      "learning_rate": 8.246165923149748e-05,
+      "loss": 2.1605,
+      "step": 3284
+    },
+    {
+      "epoch": 1.417080008626267,
+      "grad_norm": 0.18277880549430847,
+      "learning_rate": 8.242769981788817e-05,
+      "loss": 2.1051,
+      "step": 3285
+    },
+    {
+      "epoch": 1.4175113219754152,
+      "grad_norm": 0.1829131543636322,
+      "learning_rate": 8.239373886629242e-05,
+      "loss": 2.048,
+      "step": 3286
+    },
+    {
+      "epoch": 1.4179426353245632,
+      "grad_norm": 0.18608812987804413,
+      "learning_rate": 8.235977638374216e-05,
+      "loss": 2.3122,
+      "step": 3287
+    },
+    {
+      "epoch": 1.4183739486737115,
+      "grad_norm": 0.17488327622413635,
+      "learning_rate": 8.232581237726971e-05,
+      "loss": 2.095,
+      "step": 3288
+    },
+    {
+      "epoch": 1.4188052620228597,
+      "grad_norm": 0.17854848504066467,
+      "learning_rate": 8.229184685390771e-05,
+      "loss": 2.2336,
+      "step": 3289
+    },
+    {
+      "epoch": 1.4192365753720078,
+      "grad_norm": 0.18223245441913605,
+      "learning_rate": 8.225787982068904e-05,
+      "loss": 2.0173,
+      "step": 3290
+    },
+    {
+      "epoch": 1.4196678887211558,
+      "grad_norm": 0.1716955453157425,
+      "learning_rate": 8.222391128464704e-05,
+      "loss": 2.0861,
+      "step": 3291
+    },
+    {
+      "epoch": 1.420099202070304,
+      "grad_norm": 0.1973477154970169,
+      "learning_rate": 8.218994125281517e-05,
+      "loss": 2.3174,
+      "step": 3292
+    },
+    {
+      "epoch": 1.4205305154194523,
+      "grad_norm": 0.18362484872341156,
+      "learning_rate": 8.215596973222734e-05,
+      "loss": 2.1593,
+      "step": 3293
+    },
+    {
+      "epoch": 1.4209618287686003,
+      "grad_norm": 0.1910424381494522,
+      "learning_rate": 8.212199672991771e-05,
+      "loss": 2.2483,
+      "step": 3294
+    },
+    {
+      "epoch": 1.4213931421177486,
+      "grad_norm": 0.18225796520709991,
+      "learning_rate": 8.208802225292081e-05,
+      "loss": 2.1716,
+      "step": 3295
+    },
+    {
+      "epoch": 1.4218244554668966,
+      "grad_norm": 0.1668190360069275,
+      "learning_rate": 8.205404630827136e-05,
+      "loss": 2.1617,
+      "step": 3296
+    },
+    {
+      "epoch": 1.4222557688160449,
+      "grad_norm": 0.1746533215045929,
+      "learning_rate": 8.202006890300447e-05,
+      "loss": 2.2301,
+      "step": 3297
+    },
+    {
+      "epoch": 1.4226870821651931,
+      "grad_norm": 0.18129593133926392,
+      "learning_rate": 8.198609004415557e-05,
+      "loss": 2.1187,
+      "step": 3298
+    },
+    {
+      "epoch": 1.4231183955143412,
+      "grad_norm": 0.18112768232822418,
+      "learning_rate": 8.195210973876032e-05,
+      "loss": 2.08,
+      "step": 3299
+    },
+    {
+      "epoch": 1.4235497088634892,
+      "grad_norm": 0.18262645602226257,
+      "learning_rate": 8.191812799385473e-05,
+      "loss": 2.2791,
+      "step": 3300
+    },
+    {
+      "epoch": 1.4235497088634892,
+      "eval_loss": 2.096240282058716,
+      "eval_runtime": 194.9073,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3300
+    },
+    {
+      "epoch": 1.4239810222126374,
+      "grad_norm": 0.1792260706424713,
+      "learning_rate": 8.188414481647506e-05,
+      "loss": 2.2229,
+      "step": 3301
+    },
+    {
+      "epoch": 1.4244123355617857,
+      "grad_norm": 0.1991865336894989,
+      "learning_rate": 8.185016021365794e-05,
+      "loss": 2.2453,
+      "step": 3302
+    },
+    {
+      "epoch": 1.4248436489109337,
+      "grad_norm": 0.21256665885448456,
+      "learning_rate": 8.181617419244023e-05,
+      "loss": 1.7668,
+      "step": 3303
+    },
+    {
+      "epoch": 1.425274962260082,
+      "grad_norm": 0.17974048852920532,
+      "learning_rate": 8.178218675985912e-05,
+      "loss": 2.0113,
+      "step": 3304
+    },
+    {
+      "epoch": 1.42570627560923,
+      "grad_norm": 0.17181038856506348,
+      "learning_rate": 8.174819792295207e-05,
+      "loss": 2.0498,
+      "step": 3305
+    },
+    {
+      "epoch": 1.4261375889583783,
+      "grad_norm": 0.1864602416753769,
+      "learning_rate": 8.171420768875683e-05,
+      "loss": 2.2682,
+      "step": 3306
+    },
+    {
+      "epoch": 1.4265689023075265,
+      "grad_norm": 0.1754862517118454,
+      "learning_rate": 8.168021606431148e-05,
+      "loss": 2.2322,
+      "step": 3307
+    },
+    {
+      "epoch": 1.4270002156566746,
+      "grad_norm": 0.17984440922737122,
+      "learning_rate": 8.164622305665431e-05,
+      "loss": 2.0406,
+      "step": 3308
+    },
+    {
+      "epoch": 1.4274315290058226,
+      "grad_norm": 0.1859109103679657,
+      "learning_rate": 8.161222867282401e-05,
+      "loss": 1.9864,
+      "step": 3309
+    },
+    {
+      "epoch": 1.4278628423549709,
+      "grad_norm": 0.1888839453458786,
+      "learning_rate": 8.157823291985944e-05,
+      "loss": 2.2825,
+      "step": 3310
+    },
+    {
+      "epoch": 1.428294155704119,
+      "grad_norm": 0.18982118368148804,
+      "learning_rate": 8.15442358047998e-05,
+      "loss": 2.1656,
+      "step": 3311
+    },
+    {
+      "epoch": 1.4287254690532671,
+      "grad_norm": 0.18888141214847565,
+      "learning_rate": 8.151023733468458e-05,
+      "loss": 2.3547,
+      "step": 3312
+    },
+    {
+      "epoch": 1.4291567824024154,
+      "grad_norm": 0.1889471858739853,
+      "learning_rate": 8.147623751655351e-05,
+      "loss": 2.0428,
+      "step": 3313
+    },
+    {
+      "epoch": 1.4295880957515634,
+      "grad_norm": 0.1735839992761612,
+      "learning_rate": 8.144223635744666e-05,
+      "loss": 2.0846,
+      "step": 3314
+    },
+    {
+      "epoch": 1.4300194091007117,
+      "grad_norm": 0.17066586017608643,
+      "learning_rate": 8.14082338644043e-05,
+      "loss": 2.1956,
+      "step": 3315
+    },
+    {
+      "epoch": 1.43045072244986,
+      "grad_norm": 0.18497024476528168,
+      "learning_rate": 8.137423004446706e-05,
+      "loss": 2.1482,
+      "step": 3316
+    },
+    {
+      "epoch": 1.430882035799008,
+      "grad_norm": 0.19920971989631653,
+      "learning_rate": 8.134022490467577e-05,
+      "loss": 2.0942,
+      "step": 3317
+    },
+    {
+      "epoch": 1.431313349148156,
+      "grad_norm": 0.19529315829277039,
+      "learning_rate": 8.130621845207159e-05,
+      "loss": 1.8287,
+      "step": 3318
+    },
+    {
+      "epoch": 1.4317446624973043,
+      "grad_norm": 0.1871774047613144,
+      "learning_rate": 8.12722106936959e-05,
+      "loss": 2.2785,
+      "step": 3319
+    },
+    {
+      "epoch": 1.4321759758464525,
+      "grad_norm": 0.17630258202552795,
+      "learning_rate": 8.123820163659038e-05,
+      "loss": 2.0975,
+      "step": 3320
+    },
+    {
+      "epoch": 1.4326072891956005,
+      "grad_norm": 0.1827884465456009,
+      "learning_rate": 8.120419128779703e-05,
+      "loss": 2.1393,
+      "step": 3321
+    },
+    {
+      "epoch": 1.4330386025447488,
+      "grad_norm": 0.2376793920993805,
+      "learning_rate": 8.1170179654358e-05,
+      "loss": 2.2405,
+      "step": 3322
+    },
+    {
+      "epoch": 1.4334699158938968,
+      "grad_norm": 0.1873629242181778,
+      "learning_rate": 8.113616674331581e-05,
+      "loss": 2.2394,
+      "step": 3323
+    },
+    {
+      "epoch": 1.433901229243045,
+      "grad_norm": 0.17424309253692627,
+      "learning_rate": 8.110215256171318e-05,
+      "loss": 2.2561,
+      "step": 3324
+    },
+    {
+      "epoch": 1.4343325425921933,
+      "grad_norm": 0.20326147973537445,
+      "learning_rate": 8.106813711659313e-05,
+      "loss": 2.2324,
+      "step": 3325
+    },
+    {
+      "epoch": 1.4343325425921933,
+      "eval_loss": 2.095902919769287,
+      "eval_runtime": 194.7412,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3325
+    },
+    {
+      "epoch": 1.4347638559413414,
+      "grad_norm": 0.18723958730697632,
+      "learning_rate": 8.103412041499895e-05,
+      "loss": 2.294,
+      "step": 3326
+    },
+    {
+      "epoch": 1.4351951692904894,
+      "grad_norm": 0.18472136557102203,
+      "learning_rate": 8.100010246397413e-05,
+      "loss": 2.2449,
+      "step": 3327
+    },
+    {
+      "epoch": 1.4356264826396377,
+      "grad_norm": 0.18690557777881622,
+      "learning_rate": 8.09660832705625e-05,
+      "loss": 2.2492,
+      "step": 3328
+    },
+    {
+      "epoch": 1.436057795988786,
+      "grad_norm": 0.18417558073997498,
+      "learning_rate": 8.093206284180804e-05,
+      "loss": 2.0997,
+      "step": 3329
+    },
+    {
+      "epoch": 1.436489109337934,
+      "grad_norm": 0.19392870366573334,
+      "learning_rate": 8.089804118475515e-05,
+      "loss": 2.1467,
+      "step": 3330
+    },
+    {
+      "epoch": 1.4369204226870822,
+      "grad_norm": 0.18073543906211853,
+      "learning_rate": 8.08640183064483e-05,
+      "loss": 2.1225,
+      "step": 3331
+    },
+    {
+      "epoch": 1.4373517360362302,
+      "grad_norm": 0.18332725763320923,
+      "learning_rate": 8.082999421393233e-05,
+      "loss": 2.1037,
+      "step": 3332
+    },
+    {
+      "epoch": 1.4377830493853785,
+      "grad_norm": 0.4294225871562958,
+      "learning_rate": 8.079596891425231e-05,
+      "loss": 2.1481,
+      "step": 3333
+    },
+    {
+      "epoch": 1.4382143627345267,
+      "grad_norm": 0.17668937146663666,
+      "learning_rate": 8.076194241445356e-05,
+      "loss": 2.2179,
+      "step": 3334
+    },
+    {
+      "epoch": 1.4386456760836748,
+      "grad_norm": 0.17035427689552307,
+      "learning_rate": 8.07279147215816e-05,
+      "loss": 2.2046,
+      "step": 3335
+    },
+    {
+      "epoch": 1.439076989432823,
+      "grad_norm": 0.1593809276819229,
+      "learning_rate": 8.069388584268226e-05,
+      "loss": 2.1098,
+      "step": 3336
+    },
+    {
+      "epoch": 1.439508302781971,
+      "grad_norm": 2.597872734069824,
+      "learning_rate": 8.06598557848016e-05,
+      "loss": 2.3244,
+      "step": 3337
+    },
+    {
+      "epoch": 1.4399396161311193,
+      "grad_norm": 0.19738531112670898,
+      "learning_rate": 8.06258245549859e-05,
+      "loss": 2.2166,
+      "step": 3338
+    },
+    {
+      "epoch": 1.4403709294802673,
+      "grad_norm": 0.19736456871032715,
+      "learning_rate": 8.059179216028168e-05,
+      "loss": 2.2696,
+      "step": 3339
+    },
+    {
+      "epoch": 1.4408022428294156,
+      "grad_norm": 0.17271456122398376,
+      "learning_rate": 8.055775860773577e-05,
+      "loss": 1.9545,
+      "step": 3340
+    },
+    {
+      "epoch": 1.4412335561785636,
+      "grad_norm": 0.20499290525913239,
+      "learning_rate": 8.052372390439517e-05,
+      "loss": 2.2657,
+      "step": 3341
+    },
+    {
+      "epoch": 1.441664869527712,
+      "grad_norm": 0.20654812455177307,
+      "learning_rate": 8.048968805730712e-05,
+      "loss": 2.1996,
+      "step": 3342
+    },
+    {
+      "epoch": 1.4420961828768601,
+      "grad_norm": 0.1679326891899109,
+      "learning_rate": 8.045565107351909e-05,
+      "loss": 2.0968,
+      "step": 3343
+    },
+    {
+      "epoch": 1.4425274962260082,
+      "grad_norm": 0.1864512413740158,
+      "learning_rate": 8.04216129600789e-05,
+      "loss": 2.2755,
+      "step": 3344
+    },
+    {
+      "epoch": 1.4429588095751564,
+      "grad_norm": 0.18227176368236542,
+      "learning_rate": 8.038757372403441e-05,
+      "loss": 2.2069,
+      "step": 3345
+    },
+    {
+      "epoch": 1.4433901229243045,
+      "grad_norm": 0.1615314930677414,
+      "learning_rate": 8.035353337243389e-05,
+      "loss": 1.9317,
+      "step": 3346
+    },
+    {
+      "epoch": 1.4438214362734527,
+      "grad_norm": 0.18779605627059937,
+      "learning_rate": 8.031949191232572e-05,
+      "loss": 2.3717,
+      "step": 3347
+    },
+    {
+      "epoch": 1.4442527496226008,
+      "grad_norm": 0.17955152690410614,
+      "learning_rate": 8.02854493507586e-05,
+      "loss": 2.1175,
+      "step": 3348
+    },
+    {
+      "epoch": 1.444684062971749,
+      "grad_norm": 0.1738167554140091,
+      "learning_rate": 8.025140569478137e-05,
+      "loss": 2.1624,
+      "step": 3349
+    },
+    {
+      "epoch": 1.445115376320897,
+      "grad_norm": 0.18269379436969757,
+      "learning_rate": 8.021736095144318e-05,
+      "loss": 2.2026,
+      "step": 3350
+    },
+    {
+      "epoch": 1.445115376320897,
+      "eval_loss": 2.096182346343994,
+      "eval_runtime": 194.6214,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3350
+    },
+    {
+      "epoch": 1.4455466896700453,
+      "grad_norm": 0.18059170246124268,
+      "learning_rate": 8.018331512779334e-05,
+      "loss": 2.3293,
+      "step": 3351
+    },
+    {
+      "epoch": 1.4459780030191935,
+      "grad_norm": 0.21296890079975128,
+      "learning_rate": 8.01492682308814e-05,
+      "loss": 2.0689,
+      "step": 3352
+    },
+    {
+      "epoch": 1.4464093163683416,
+      "grad_norm": 0.1927291601896286,
+      "learning_rate": 8.01152202677572e-05,
+      "loss": 2.1997,
+      "step": 3353
+    },
+    {
+      "epoch": 1.4468406297174898,
+      "grad_norm": 0.1816396266222,
+      "learning_rate": 8.008117124547068e-05,
+      "loss": 2.0234,
+      "step": 3354
+    },
+    {
+      "epoch": 1.4472719430666379,
+      "grad_norm": 0.19203664362430573,
+      "learning_rate": 8.004712117107209e-05,
+      "loss": 2.2696,
+      "step": 3355
+    },
+    {
+      "epoch": 1.4477032564157861,
+      "grad_norm": 0.17730873823165894,
+      "learning_rate": 8.001307005161187e-05,
+      "loss": 2.2438,
+      "step": 3356
+    },
+    {
+      "epoch": 1.4481345697649342,
+      "grad_norm": 0.18266335129737854,
+      "learning_rate": 7.997901789414068e-05,
+      "loss": 2.119,
+      "step": 3357
+    },
+    {
+      "epoch": 1.4485658831140824,
+      "grad_norm": 0.17785349488258362,
+      "learning_rate": 7.994496470570938e-05,
+      "loss": 2.3753,
+      "step": 3358
+    },
+    {
+      "epoch": 1.4489971964632304,
+      "grad_norm": 0.17622007429599762,
+      "learning_rate": 7.991091049336904e-05,
+      "loss": 2.0939,
+      "step": 3359
+    },
+    {
+      "epoch": 1.4494285098123787,
+      "grad_norm": 0.1801026314496994,
+      "learning_rate": 7.987685526417101e-05,
+      "loss": 2.2183,
+      "step": 3360
+    },
+    {
+      "epoch": 1.449859823161527,
+      "grad_norm": 0.1778012067079544,
+      "learning_rate": 7.984279902516674e-05,
+      "loss": 2.2588,
+      "step": 3361
+    },
+    {
+      "epoch": 1.450291136510675,
+      "grad_norm": 0.17804737389087677,
+      "learning_rate": 7.980874178340799e-05,
+      "loss": 2.0134,
+      "step": 3362
+    },
+    {
+      "epoch": 1.4507224498598232,
+      "grad_norm": 3.638392925262451,
+      "learning_rate": 7.977468354594665e-05,
+      "loss": 2.1824,
+      "step": 3363
+    },
+    {
+      "epoch": 1.4511537632089713,
+      "grad_norm": 0.18926359713077545,
+      "learning_rate": 7.974062431983489e-05,
+      "loss": 2.3343,
+      "step": 3364
+    },
+    {
+      "epoch": 1.4515850765581195,
+      "grad_norm": 0.17392082512378693,
+      "learning_rate": 7.9706564112125e-05,
+      "loss": 1.9891,
+      "step": 3365
+    },
+    {
+      "epoch": 1.4520163899072676,
+      "grad_norm": 0.2056262493133545,
+      "learning_rate": 7.967250292986954e-05,
+      "loss": 2.204,
+      "step": 3366
+    },
+    {
+      "epoch": 1.4524477032564158,
+      "grad_norm": 0.1681697815656662,
+      "learning_rate": 7.963844078012128e-05,
+      "loss": 2.1408,
+      "step": 3367
+    },
+    {
+      "epoch": 1.4528790166055638,
+      "grad_norm": 0.18251508474349976,
+      "learning_rate": 7.960437766993309e-05,
+      "loss": 2.1993,
+      "step": 3368
+    },
+    {
+      "epoch": 1.453310329954712,
+      "grad_norm": 0.17431260645389557,
+      "learning_rate": 7.957031360635819e-05,
+      "loss": 2.0966,
+      "step": 3369
+    },
+    {
+      "epoch": 1.4537416433038604,
+      "grad_norm": 0.17686443030834198,
+      "learning_rate": 7.953624859644984e-05,
+      "loss": 2.05,
+      "step": 3370
+    },
+    {
+      "epoch": 1.4541729566530084,
+      "grad_norm": 0.16880741715431213,
+      "learning_rate": 7.950218264726164e-05,
+      "loss": 2.0137,
+      "step": 3371
+    },
+    {
+      "epoch": 1.4546042700021566,
+      "grad_norm": 0.1719847470521927,
+      "learning_rate": 7.946811576584729e-05,
+      "loss": 2.154,
+      "step": 3372
+    },
+    {
+      "epoch": 1.4550355833513047,
+      "grad_norm": 0.16456718742847443,
+      "learning_rate": 7.943404795926068e-05,
+      "loss": 2.3106,
+      "step": 3373
+    },
+    {
+      "epoch": 1.455466896700453,
+      "grad_norm": 0.17277392745018005,
+      "learning_rate": 7.939997923455598e-05,
+      "loss": 2.1343,
+      "step": 3374
+    },
+    {
+      "epoch": 1.455898210049601,
+      "grad_norm": 0.17675180733203888,
+      "learning_rate": 7.936590959878741e-05,
+      "loss": 2.0205,
+      "step": 3375
+    },
+    {
+      "epoch": 1.455898210049601,
+      "eval_loss": 2.0960140228271484,
+      "eval_runtime": 205.7204,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 3375
+    },
+    {
+      "epoch": 1.4563295233987492,
+      "grad_norm": 0.18122228980064392,
+      "learning_rate": 7.933183905900957e-05,
+      "loss": 2.1369,
+      "step": 3376
+    },
+    {
+      "epoch": 1.4567608367478972,
+      "grad_norm": 0.18892325460910797,
+      "learning_rate": 7.929776762227706e-05,
+      "loss": 2.1501,
+      "step": 3377
+    },
+    {
+      "epoch": 1.4571921500970455,
+      "grad_norm": 0.18425828218460083,
+      "learning_rate": 7.926369529564477e-05,
+      "loss": 2.0671,
+      "step": 3378
+    },
+    {
+      "epoch": 1.4576234634461938,
+      "grad_norm": 0.1792745143175125,
+      "learning_rate": 7.922962208616773e-05,
+      "loss": 2.175,
+      "step": 3379
+    },
+    {
+      "epoch": 1.4580547767953418,
+      "grad_norm": 0.17559273540973663,
+      "learning_rate": 7.91955480009012e-05,
+      "loss": 2.1323,
+      "step": 3380
+    },
+    {
+      "epoch": 1.45848609014449,
+      "grad_norm": 0.1783163994550705,
+      "learning_rate": 7.916147304690056e-05,
+      "loss": 2.1548,
+      "step": 3381
+    },
+    {
+      "epoch": 1.458917403493638,
+      "grad_norm": 0.15873871743679047,
+      "learning_rate": 7.912739723122142e-05,
+      "loss": 2.0485,
+      "step": 3382
+    },
+    {
+      "epoch": 1.4593487168427863,
+      "grad_norm": 0.14900639653205872,
+      "learning_rate": 7.909332056091955e-05,
+      "loss": 1.9628,
+      "step": 3383
+    },
+    {
+      "epoch": 1.4597800301919344,
+      "grad_norm": 0.17377440631389618,
+      "learning_rate": 7.905924304305088e-05,
+      "loss": 2.154,
+      "step": 3384
+    },
+    {
+      "epoch": 1.4602113435410826,
+      "grad_norm": 0.2151319533586502,
+      "learning_rate": 7.902516468467158e-05,
+      "loss": 2.2311,
+      "step": 3385
+    },
+    {
+      "epoch": 1.4606426568902307,
+      "grad_norm": 0.19471900165081024,
+      "learning_rate": 7.899108549283788e-05,
+      "loss": 2.1155,
+      "step": 3386
+    },
+    {
+      "epoch": 1.461073970239379,
+      "grad_norm": 0.17895857989788055,
+      "learning_rate": 7.89570054746063e-05,
+      "loss": 1.9349,
+      "step": 3387
+    },
+    {
+      "epoch": 1.4615052835885272,
+      "grad_norm": 0.17501680552959442,
+      "learning_rate": 7.892292463703346e-05,
+      "loss": 2.1263,
+      "step": 3388
+    },
+    {
+      "epoch": 1.4619365969376752,
+      "grad_norm": 0.186353400349617,
+      "learning_rate": 7.888884298717617e-05,
+      "loss": 2.199,
+      "step": 3389
+    },
+    {
+      "epoch": 1.4623679102868234,
+      "grad_norm": 0.17816871404647827,
+      "learning_rate": 7.885476053209143e-05,
+      "loss": 2.2223,
+      "step": 3390
+    },
+    {
+      "epoch": 1.4627992236359715,
+      "grad_norm": 0.16904309391975403,
+      "learning_rate": 7.882067727883633e-05,
+      "loss": 2.1566,
+      "step": 3391
+    },
+    {
+      "epoch": 1.4632305369851197,
+      "grad_norm": 0.16086207330226898,
+      "learning_rate": 7.878659323446826e-05,
+      "loss": 2.0205,
+      "step": 3392
+    },
+    {
+      "epoch": 1.4636618503342678,
+      "grad_norm": 0.19864903390407562,
+      "learning_rate": 7.875250840604461e-05,
+      "loss": 2.3275,
+      "step": 3393
+    },
+    {
+      "epoch": 1.464093163683416,
+      "grad_norm": 0.5341060757637024,
+      "learning_rate": 7.871842280062309e-05,
+      "loss": 2.1908,
+      "step": 3394
+    },
+    {
+      "epoch": 1.464524477032564,
+      "grad_norm": 0.1994878649711609,
+      "learning_rate": 7.868433642526145e-05,
+      "loss": 2.2238,
+      "step": 3395
+    },
+    {
+      "epoch": 1.4649557903817123,
+      "grad_norm": 0.181796133518219,
+      "learning_rate": 7.865024928701767e-05,
+      "loss": 2.1112,
+      "step": 3396
+    },
+    {
+      "epoch": 1.4653871037308606,
+      "grad_norm": 0.1800110638141632,
+      "learning_rate": 7.861616139294986e-05,
+      "loss": 1.9921,
+      "step": 3397
+    },
+    {
+      "epoch": 1.4658184170800086,
+      "grad_norm": 0.18481217324733734,
+      "learning_rate": 7.858207275011626e-05,
+      "loss": 2.0483,
+      "step": 3398
+    },
+    {
+      "epoch": 1.4662497304291569,
+      "grad_norm": 0.19453781843185425,
+      "learning_rate": 7.854798336557535e-05,
+      "loss": 1.7871,
+      "step": 3399
+    },
+    {
+      "epoch": 1.4666810437783049,
+      "grad_norm": 0.1700371354818344,
+      "learning_rate": 7.851389324638567e-05,
+      "loss": 1.9453,
+      "step": 3400
+    },
+    {
+      "epoch": 1.4666810437783049,
+      "eval_loss": 2.0953893661499023,
+      "eval_runtime": 199.7551,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 3400
+    },
+    {
+      "epoch": 1.4671123571274531,
+      "grad_norm": 0.1720868945121765,
+      "learning_rate": 7.847980239960598e-05,
+      "loss": 2.1746,
+      "step": 3401
+    },
+    {
+      "epoch": 1.4675436704766012,
+      "grad_norm": 0.1633797287940979,
+      "learning_rate": 7.844571083229514e-05,
+      "loss": 1.9359,
+      "step": 3402
+    },
+    {
+      "epoch": 1.4679749838257494,
+      "grad_norm": 0.1909305900335312,
+      "learning_rate": 7.841161855151221e-05,
+      "loss": 1.981,
+      "step": 3403
+    },
+    {
+      "epoch": 1.4684062971748975,
+      "grad_norm": 14.385684967041016,
+      "learning_rate": 7.837752556431632e-05,
+      "loss": 2.1872,
+      "step": 3404
+    },
+    {
+      "epoch": 1.4688376105240457,
+      "grad_norm": 0.17449794709682465,
+      "learning_rate": 7.834343187776683e-05,
+      "loss": 2.1808,
+      "step": 3405
+    },
+    {
+      "epoch": 1.469268923873194,
+      "grad_norm": 0.18831457197666168,
+      "learning_rate": 7.830933749892323e-05,
+      "loss": 2.2185,
+      "step": 3406
+    },
+    {
+      "epoch": 1.469700237222342,
+      "grad_norm": 0.18120522797107697,
+      "learning_rate": 7.827524243484509e-05,
+      "loss": 2.0987,
+      "step": 3407
+    },
+    {
+      "epoch": 1.4701315505714903,
+      "grad_norm": 0.19174909591674805,
+      "learning_rate": 7.82411466925922e-05,
+      "loss": 2.4923,
+      "step": 3408
+    },
+    {
+      "epoch": 1.4705628639206383,
+      "grad_norm": 0.1888199895620346,
+      "learning_rate": 7.820705027922442e-05,
+      "loss": 2.2071,
+      "step": 3409
+    },
+    {
+      "epoch": 1.4709941772697865,
+      "grad_norm": 0.1881730556488037,
+      "learning_rate": 7.817295320180183e-05,
+      "loss": 1.9944,
+      "step": 3410
+    },
+    {
+      "epoch": 1.4714254906189346,
+      "grad_norm": 0.19050532579421997,
+      "learning_rate": 7.813885546738457e-05,
+      "loss": 2.2363,
+      "step": 3411
+    },
+    {
+      "epoch": 1.4718568039680828,
+      "grad_norm": 0.18670903146266937,
+      "learning_rate": 7.810475708303294e-05,
+      "loss": 2.0754,
+      "step": 3412
+    },
+    {
+      "epoch": 1.4722881173172309,
+      "grad_norm": 0.18117989599704742,
+      "learning_rate": 7.807065805580745e-05,
+      "loss": 2.1655,
+      "step": 3413
+    },
+    {
+      "epoch": 1.4727194306663791,
+      "grad_norm": 0.16975900530815125,
+      "learning_rate": 7.803655839276858e-05,
+      "loss": 2.1339,
+      "step": 3414
+    },
+    {
+      "epoch": 1.4731507440155274,
+      "grad_norm": 0.18406729400157928,
+      "learning_rate": 7.800245810097714e-05,
+      "loss": 2.1403,
+      "step": 3415
+    },
+    {
+      "epoch": 1.4735820573646754,
+      "grad_norm": 0.16810061037540436,
+      "learning_rate": 7.796835718749388e-05,
+      "loss": 1.9413,
+      "step": 3416
+    },
+    {
+      "epoch": 1.4740133707138237,
+      "grad_norm": 0.20537112653255463,
+      "learning_rate": 7.793425565937982e-05,
+      "loss": 2.0279,
+      "step": 3417
+    },
+    {
+      "epoch": 1.4744446840629717,
+      "grad_norm": 0.30291008949279785,
+      "learning_rate": 7.790015352369603e-05,
+      "loss": 2.292,
+      "step": 3418
+    },
+    {
+      "epoch": 1.47487599741212,
+      "grad_norm": 0.18326658010482788,
+      "learning_rate": 7.786605078750374e-05,
+      "loss": 2.1898,
+      "step": 3419
+    },
+    {
+      "epoch": 1.4753073107612682,
+      "grad_norm": 0.18363992869853973,
+      "learning_rate": 7.783194745786429e-05,
+      "loss": 2.2315,
+      "step": 3420
+    },
+    {
+      "epoch": 1.4757386241104162,
+      "grad_norm": 0.17302939295768738,
+      "learning_rate": 7.779784354183912e-05,
+      "loss": 2.2566,
+      "step": 3421
+    },
+    {
+      "epoch": 1.4761699374595643,
+      "grad_norm": 0.19719082117080688,
+      "learning_rate": 7.77637390464899e-05,
+      "loss": 2.0581,
+      "step": 3422
+    },
+    {
+      "epoch": 1.4766012508087125,
+      "grad_norm": 0.20065642893314362,
+      "learning_rate": 7.772963397887825e-05,
+      "loss": 2.2128,
+      "step": 3423
+    },
+    {
+      "epoch": 1.4770325641578608,
+      "grad_norm": 0.1718757152557373,
+      "learning_rate": 7.769552834606606e-05,
+      "loss": 2.0527,
+      "step": 3424
+    },
+    {
+      "epoch": 1.4774638775070088,
+      "grad_norm": 0.19068890810012817,
+      "learning_rate": 7.766142215511523e-05,
+      "loss": 2.0046,
+      "step": 3425
+    },
+    {
+      "epoch": 1.4774638775070088,
+      "eval_loss": 2.095679998397827,
+      "eval_runtime": 198.9871,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 3425
+    },
+    {
+      "epoch": 1.477895190856157,
+      "grad_norm": 0.17807812988758087,
+      "learning_rate": 7.762731541308785e-05,
+      "loss": 2.0032,
+      "step": 3426
+    },
+    {
+      "epoch": 1.478326504205305,
+      "grad_norm": 0.19017374515533447,
+      "learning_rate": 7.759320812704607e-05,
+      "loss": 2.3483,
+      "step": 3427
+    },
+    {
+      "epoch": 1.4787578175544533,
+      "grad_norm": 0.19937926530838013,
+      "learning_rate": 7.755910030405217e-05,
+      "loss": 1.9971,
+      "step": 3428
+    },
+    {
+      "epoch": 1.4791891309036016,
+      "grad_norm": 0.1809537261724472,
+      "learning_rate": 7.752499195116861e-05,
+      "loss": 2.0348,
+      "step": 3429
+    },
+    {
+      "epoch": 1.4796204442527496,
+      "grad_norm": 0.19054853916168213,
+      "learning_rate": 7.749088307545783e-05,
+      "loss": 2.1322,
+      "step": 3430
+    },
+    {
+      "epoch": 1.4800517576018977,
+      "grad_norm": 0.19482950866222382,
+      "learning_rate": 7.745677368398248e-05,
+      "loss": 2.1562,
+      "step": 3431
+    },
+    {
+      "epoch": 1.480483070951046,
+      "grad_norm": 0.18106971681118011,
+      "learning_rate": 7.742266378380524e-05,
+      "loss": 2.2104,
+      "step": 3432
+    },
+    {
+      "epoch": 1.4809143843001942,
+      "grad_norm": 0.17390477657318115,
+      "learning_rate": 7.7388553381989e-05,
+      "loss": 2.2025,
+      "step": 3433
+    },
+    {
+      "epoch": 1.4813456976493422,
+      "grad_norm": 0.1953706592321396,
+      "learning_rate": 7.735444248559666e-05,
+      "loss": 2.2455,
+      "step": 3434
+    },
+    {
+      "epoch": 1.4817770109984905,
+      "grad_norm": 0.18527862429618835,
+      "learning_rate": 7.732033110169123e-05,
+      "loss": 2.1696,
+      "step": 3435
+    },
+    {
+      "epoch": 1.4822083243476385,
+      "grad_norm": 0.19061794877052307,
+      "learning_rate": 7.728621923733592e-05,
+      "loss": 2.547,
+      "step": 3436
+    },
+    {
+      "epoch": 1.4826396376967867,
+      "grad_norm": 0.17946751415729523,
+      "learning_rate": 7.725210689959387e-05,
+      "loss": 2.134,
+      "step": 3437
+    },
+    {
+      "epoch": 1.483070951045935,
+      "grad_norm": 0.19174616038799286,
+      "learning_rate": 7.721799409552849e-05,
+      "loss": 2.0701,
+      "step": 3438
+    },
+    {
+      "epoch": 1.483502264395083,
+      "grad_norm": 0.18157075345516205,
+      "learning_rate": 7.718388083220315e-05,
+      "loss": 2.4425,
+      "step": 3439
+    },
+    {
+      "epoch": 1.483933577744231,
+      "grad_norm": 0.1765347272157669,
+      "learning_rate": 7.71497671166814e-05,
+      "loss": 2.1589,
+      "step": 3440
+    },
+    {
+      "epoch": 1.4843648910933793,
+      "grad_norm": 0.18968167901039124,
+      "learning_rate": 7.711565295602689e-05,
+      "loss": 2.2585,
+      "step": 3441
+    },
+    {
+      "epoch": 1.4847962044425276,
+      "grad_norm": 0.2010895311832428,
+      "learning_rate": 7.708153835730329e-05,
+      "loss": 2.373,
+      "step": 3442
+    },
+    {
+      "epoch": 1.4852275177916756,
+      "grad_norm": 0.17231179773807526,
+      "learning_rate": 7.70474233275744e-05,
+      "loss": 2.1558,
+      "step": 3443
+    },
+    {
+      "epoch": 1.4856588311408239,
+      "grad_norm": 0.17638081312179565,
+      "learning_rate": 7.70133078739041e-05,
+      "loss": 2.1843,
+      "step": 3444
+    },
+    {
+      "epoch": 1.486090144489972,
+      "grad_norm": 0.18057771027088165,
+      "learning_rate": 7.697919200335643e-05,
+      "loss": 2.0534,
+      "step": 3445
+    },
+    {
+      "epoch": 1.4865214578391202,
+      "grad_norm": 0.1979825794696808,
+      "learning_rate": 7.694507572299537e-05,
+      "loss": 2.3043,
+      "step": 3446
+    },
+    {
+      "epoch": 1.4869527711882684,
+      "grad_norm": 0.18936648964881897,
+      "learning_rate": 7.691095903988515e-05,
+      "loss": 1.9956,
+      "step": 3447
+    },
+    {
+      "epoch": 1.4873840845374164,
+      "grad_norm": 0.1901467740535736,
+      "learning_rate": 7.687684196108992e-05,
+      "loss": 2.2853,
+      "step": 3448
+    },
+    {
+      "epoch": 1.4878153978865645,
+      "grad_norm": 0.1820063441991806,
+      "learning_rate": 7.684272449367407e-05,
+      "loss": 2.0876,
+      "step": 3449
+    },
+    {
+      "epoch": 1.4882467112357127,
+      "grad_norm": 0.17177510261535645,
+      "learning_rate": 7.680860664470195e-05,
+      "loss": 2.0898,
+      "step": 3450
+    },
+    {
+      "epoch": 1.4882467112357127,
+      "eval_loss": 2.095259428024292,
+      "eval_runtime": 199.7432,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 3450
+    },
+    {
+      "epoch": 1.488678024584861,
+      "grad_norm": 0.18127751350402832,
+      "learning_rate": 7.677448842123805e-05,
+      "loss": 2.1625,
+      "step": 3451
+    },
+    {
+      "epoch": 1.489109337934009,
+      "grad_norm": 0.18150100111961365,
+      "learning_rate": 7.674036983034692e-05,
+      "loss": 2.0997,
+      "step": 3452
+    },
+    {
+      "epoch": 1.4895406512831573,
+      "grad_norm": 0.18290740251541138,
+      "learning_rate": 7.670625087909316e-05,
+      "loss": 2.194,
+      "step": 3453
+    },
+    {
+      "epoch": 1.4899719646323053,
+      "grad_norm": 0.17931239306926727,
+      "learning_rate": 7.667213157454153e-05,
+      "loss": 2.0456,
+      "step": 3454
+    },
+    {
+      "epoch": 1.4904032779814536,
+      "grad_norm": 0.18845637142658234,
+      "learning_rate": 7.663801192375675e-05,
+      "loss": 2.4016,
+      "step": 3455
+    },
+    {
+      "epoch": 1.4908345913306018,
+      "grad_norm": 0.1789526492357254,
+      "learning_rate": 7.66038919338037e-05,
+      "loss": 2.1449,
+      "step": 3456
+    },
+    {
+      "epoch": 1.4912659046797498,
+      "grad_norm": 0.21393506228923798,
+      "learning_rate": 7.656977161174728e-05,
+      "loss": 2.1899,
+      "step": 3457
+    },
+    {
+      "epoch": 1.4916972180288979,
+      "grad_norm": 0.1698625236749649,
+      "learning_rate": 7.653565096465247e-05,
+      "loss": 2.0487,
+      "step": 3458
+    },
+    {
+      "epoch": 1.4921285313780461,
+      "grad_norm": 0.19428661465644836,
+      "learning_rate": 7.650152999958436e-05,
+      "loss": 2.0786,
+      "step": 3459
+    },
+    {
+      "epoch": 1.4925598447271944,
+      "grad_norm": 0.16949570178985596,
+      "learning_rate": 7.646740872360802e-05,
+      "loss": 2.1475,
+      "step": 3460
+    },
+    {
+      "epoch": 1.4929911580763424,
+      "grad_norm": 0.18819738924503326,
+      "learning_rate": 7.643328714378866e-05,
+      "loss": 2.1348,
+      "step": 3461
+    },
+    {
+      "epoch": 1.4934224714254907,
+      "grad_norm": 0.17552988231182098,
+      "learning_rate": 7.639916526719153e-05,
+      "loss": 2.1711,
+      "step": 3462
+    },
+    {
+      "epoch": 1.4938537847746387,
+      "grad_norm": 0.17839846014976501,
+      "learning_rate": 7.636504310088192e-05,
+      "loss": 2.0626,
+      "step": 3463
+    },
+    {
+      "epoch": 1.494285098123787,
+      "grad_norm": 0.161084845662117,
+      "learning_rate": 7.633092065192523e-05,
+      "loss": 1.9977,
+      "step": 3464
+    },
+    {
+      "epoch": 1.4947164114729352,
+      "grad_norm": 0.22303293645381927,
+      "learning_rate": 7.629679792738686e-05,
+      "loss": 2.1712,
+      "step": 3465
+    },
+    {
+      "epoch": 1.4951477248220832,
+      "grad_norm": 0.17171968519687653,
+      "learning_rate": 7.626267493433227e-05,
+      "loss": 2.1384,
+      "step": 3466
+    },
+    {
+      "epoch": 1.4955790381712313,
+      "grad_norm": 0.1698496788740158,
+      "learning_rate": 7.622855167982703e-05,
+      "loss": 2.2687,
+      "step": 3467
+    },
+    {
+      "epoch": 1.4960103515203795,
+      "grad_norm": 0.17629973590373993,
+      "learning_rate": 7.619442817093677e-05,
+      "loss": 2.0812,
+      "step": 3468
+    },
+    {
+      "epoch": 1.4964416648695278,
+      "grad_norm": 0.17922022938728333,
+      "learning_rate": 7.616030441472706e-05,
+      "loss": 2.306,
+      "step": 3469
+    },
+    {
+      "epoch": 1.4968729782186758,
+      "grad_norm": 0.18257419764995575,
+      "learning_rate": 7.612618041826366e-05,
+      "loss": 2.1525,
+      "step": 3470
+    },
+    {
+      "epoch": 1.497304291567824,
+      "grad_norm": 0.17645780742168427,
+      "learning_rate": 7.609205618861225e-05,
+      "loss": 2.0277,
+      "step": 3471
+    },
+    {
+      "epoch": 1.497735604916972,
+      "grad_norm": 0.2862652540206909,
+      "learning_rate": 7.605793173283869e-05,
+      "loss": 1.8123,
+      "step": 3472
+    },
+    {
+      "epoch": 1.4981669182661204,
+      "grad_norm": 0.28746744990348816,
+      "learning_rate": 7.602380705800879e-05,
+      "loss": 2.2131,
+      "step": 3473
+    },
+    {
+      "epoch": 1.4985982316152686,
+      "grad_norm": 0.17240886390209198,
+      "learning_rate": 7.598968217118841e-05,
+      "loss": 2.3652,
+      "step": 3474
+    },
+    {
+      "epoch": 1.4990295449644166,
+      "grad_norm": 0.1765579879283905,
+      "learning_rate": 7.595555707944358e-05,
+      "loss": 2.2843,
+      "step": 3475
+    },
+    {
+      "epoch": 1.4990295449644166,
+      "eval_loss": 2.0947535037994385,
+      "eval_runtime": 199.7196,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 3475
+    },
+    {
+      "epoch": 1.4994608583135647,
+      "grad_norm": 0.17466917634010315,
+      "learning_rate": 7.592143178984015e-05,
+      "loss": 2.0196,
+      "step": 3476
+    },
+    {
+      "epoch": 1.499892171662713,
+      "grad_norm": 0.1990082859992981,
+      "learning_rate": 7.588730630944424e-05,
+      "loss": 2.3536,
+      "step": 3477
+    },
+    {
+      "epoch": 1.5003234850118612,
+      "grad_norm": 0.19265393912792206,
+      "learning_rate": 7.58531806453218e-05,
+      "loss": 2.0515,
+      "step": 3478
+    },
+    {
+      "epoch": 1.5007547983610092,
+      "grad_norm": 0.1856214702129364,
+      "learning_rate": 7.581905480453901e-05,
+      "loss": 2.1362,
+      "step": 3479
+    },
+    {
+      "epoch": 1.5011861117101575,
+      "grad_norm": 0.16783815622329712,
+      "learning_rate": 7.578492879416196e-05,
+      "loss": 2.2121,
+      "step": 3480
+    },
+    {
+      "epoch": 1.5016174250593055,
+      "grad_norm": 0.1762775331735611,
+      "learning_rate": 7.575080262125684e-05,
+      "loss": 1.9548,
+      "step": 3481
+    },
+    {
+      "epoch": 1.5020487384084538,
+      "grad_norm": 0.18140533566474915,
+      "learning_rate": 7.571667629288979e-05,
+      "loss": 2.0891,
+      "step": 3482
+    },
+    {
+      "epoch": 1.502480051757602,
+      "grad_norm": 0.1749914437532425,
+      "learning_rate": 7.568254981612709e-05,
+      "loss": 2.1629,
+      "step": 3483
+    },
+    {
+      "epoch": 1.50291136510675,
+      "grad_norm": 0.20092450082302094,
+      "learning_rate": 7.564842319803501e-05,
+      "loss": 2.183,
+      "step": 3484
+    },
+    {
+      "epoch": 1.503342678455898,
+      "grad_norm": 0.18456554412841797,
+      "learning_rate": 7.561429644567977e-05,
+      "loss": 2.2325,
+      "step": 3485
+    },
+    {
+      "epoch": 1.5037739918050463,
+      "grad_norm": 0.189732164144516,
+      "learning_rate": 7.558016956612779e-05,
+      "loss": 2.2595,
+      "step": 3486
+    },
+    {
+      "epoch": 1.5042053051541946,
+      "grad_norm": 0.179192915558815,
+      "learning_rate": 7.55460425664453e-05,
+      "loss": 2.1265,
+      "step": 3487
+    },
+    {
+      "epoch": 1.5046366185033426,
+      "grad_norm": 0.17854931950569153,
+      "learning_rate": 7.551191545369876e-05,
+      "loss": 2.2346,
+      "step": 3488
+    },
+    {
+      "epoch": 1.5050679318524909,
+      "grad_norm": 0.21214659512043,
+      "learning_rate": 7.547778823495452e-05,
+      "loss": 2.2265,
+      "step": 3489
+    },
+    {
+      "epoch": 1.505499245201639,
+      "grad_norm": 0.17340655624866486,
+      "learning_rate": 7.544366091727897e-05,
+      "loss": 2.3306,
+      "step": 3490
+    },
+    {
+      "epoch": 1.5059305585507872,
+      "grad_norm": 0.18704965710639954,
+      "learning_rate": 7.540953350773863e-05,
+      "loss": 2.2339,
+      "step": 3491
+    },
+    {
+      "epoch": 1.5063618718999354,
+      "grad_norm": 0.17873235046863556,
+      "learning_rate": 7.537540601339985e-05,
+      "loss": 2.1256,
+      "step": 3492
+    },
+    {
+      "epoch": 1.5067931852490835,
+      "grad_norm": 0.16924327611923218,
+      "learning_rate": 7.53412784413292e-05,
+      "loss": 2.0724,
+      "step": 3493
+    },
+    {
+      "epoch": 1.5072244985982315,
+      "grad_norm": 0.18192622065544128,
+      "learning_rate": 7.530715079859307e-05,
+      "loss": 2.1323,
+      "step": 3494
+    },
+    {
+      "epoch": 1.5076558119473797,
+      "grad_norm": 0.17550119757652283,
+      "learning_rate": 7.527302309225803e-05,
+      "loss": 1.9776,
+      "step": 3495
+    },
+    {
+      "epoch": 1.508087125296528,
+      "grad_norm": 0.186650350689888,
+      "learning_rate": 7.523889532939057e-05,
+      "loss": 2.4068,
+      "step": 3496
+    },
+    {
+      "epoch": 1.508518438645676,
+      "grad_norm": 0.1668436974287033,
+      "learning_rate": 7.520476751705721e-05,
+      "loss": 2.0524,
+      "step": 3497
+    },
+    {
+      "epoch": 1.5089497519948243,
+      "grad_norm": 0.186021625995636,
+      "learning_rate": 7.517063966232452e-05,
+      "loss": 2.1887,
+      "step": 3498
+    },
+    {
+      "epoch": 1.5093810653439723,
+      "grad_norm": 0.20452392101287842,
+      "learning_rate": 7.513651177225899e-05,
+      "loss": 2.0384,
+      "step": 3499
+    },
+    {
+      "epoch": 1.5098123786931206,
+      "grad_norm": 0.18916206061840057,
+      "learning_rate": 7.510238385392724e-05,
+      "loss": 2.2156,
+      "step": 3500
+    },
+    {
+      "epoch": 1.5098123786931206,
+      "eval_loss": 2.0948498249053955,
+      "eval_runtime": 199.8624,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 3500
+    },
+    {
+      "epoch": 1.5102436920422688,
+      "grad_norm": 0.19154448807239532,
+      "learning_rate": 7.506825591439576e-05,
+      "loss": 2.2353,
+      "step": 3501
+    },
+    {
+      "epoch": 1.5106750053914169,
+      "grad_norm": 0.1821613907814026,
+      "learning_rate": 7.503412796073116e-05,
+      "loss": 2.1699,
+      "step": 3502
+    },
+    {
+      "epoch": 1.511106318740565,
+      "grad_norm": 0.1843271255493164,
+      "learning_rate": 7.5e-05,
+      "loss": 2.1596,
+      "step": 3503
+    },
+    {
+      "epoch": 1.5115376320897131,
+      "grad_norm": 0.1836938112974167,
+      "learning_rate": 7.496587203926884e-05,
+      "loss": 1.9115,
+      "step": 3504
+    },
+    {
+      "epoch": 1.5119689454388614,
+      "grad_norm": 0.1645289808511734,
+      "learning_rate": 7.493174408560423e-05,
+      "loss": 2.0508,
+      "step": 3505
+    },
+    {
+      "epoch": 1.5124002587880097,
+      "grad_norm": 0.18941354751586914,
+      "learning_rate": 7.489761614607276e-05,
+      "loss": 2.2317,
+      "step": 3506
+    },
+    {
+      "epoch": 1.5128315721371577,
+      "grad_norm": 0.17841602861881256,
+      "learning_rate": 7.486348822774101e-05,
+      "loss": 2.2352,
+      "step": 3507
+    },
+    {
+      "epoch": 1.5132628854863057,
+      "grad_norm": 0.20502035319805145,
+      "learning_rate": 7.48293603376755e-05,
+      "loss": 2.2623,
+      "step": 3508
+    },
+    {
+      "epoch": 1.513694198835454,
+      "grad_norm": 0.18774735927581787,
+      "learning_rate": 7.479523248294277e-05,
+      "loss": 2.1597,
+      "step": 3509
+    },
+    {
+      "epoch": 1.5141255121846022,
+      "grad_norm": 0.17617200314998627,
+      "learning_rate": 7.476110467060943e-05,
+      "loss": 2.013,
+      "step": 3510
+    },
+    {
+      "epoch": 1.5145568255337503,
+      "grad_norm": 0.17860938608646393,
+      "learning_rate": 7.472697690774197e-05,
+      "loss": 2.1755,
+      "step": 3511
+    },
+    {
+      "epoch": 1.5149881388828983,
+      "grad_norm": 0.21400243043899536,
+      "learning_rate": 7.469284920140692e-05,
+      "loss": 2.1347,
+      "step": 3512
+    },
+    {
+      "epoch": 1.5154194522320465,
+      "grad_norm": 0.18597297370433807,
+      "learning_rate": 7.465872155867081e-05,
+      "loss": 2.2397,
+      "step": 3513
+    },
+    {
+      "epoch": 1.5158507655811948,
+      "grad_norm": 0.17704488337039948,
+      "learning_rate": 7.462459398660014e-05,
+      "loss": 2.127,
+      "step": 3514
+    },
+    {
+      "epoch": 1.516282078930343,
+      "grad_norm": 0.19206072390079498,
+      "learning_rate": 7.459046649226138e-05,
+      "loss": 2.3177,
+      "step": 3515
+    },
+    {
+      "epoch": 1.516713392279491,
+      "grad_norm": 0.17340447008609772,
+      "learning_rate": 7.4556339082721e-05,
+      "loss": 1.8396,
+      "step": 3516
+    },
+    {
+      "epoch": 1.5171447056286391,
+      "grad_norm": 0.19418282806873322,
+      "learning_rate": 7.452221176504548e-05,
+      "loss": 1.8297,
+      "step": 3517
+    },
+    {
+      "epoch": 1.5175760189777874,
+      "grad_norm": 0.2816286087036133,
+      "learning_rate": 7.448808454630124e-05,
+      "loss": 2.2843,
+      "step": 3518
+    },
+    {
+      "epoch": 1.5180073323269356,
+      "grad_norm": 0.1821245402097702,
+      "learning_rate": 7.445395743355468e-05,
+      "loss": 2.1009,
+      "step": 3519
+    },
+    {
+      "epoch": 1.5184386456760837,
+      "grad_norm": 0.19121207296848297,
+      "learning_rate": 7.441983043387221e-05,
+      "loss": 2.0815,
+      "step": 3520
+    },
+    {
+      "epoch": 1.5188699590252317,
+      "grad_norm": 0.18853315711021423,
+      "learning_rate": 7.438570355432022e-05,
+      "loss": 2.3358,
+      "step": 3521
+    },
+    {
+      "epoch": 1.51930127237438,
+      "grad_norm": 0.19356144964694977,
+      "learning_rate": 7.435157680196502e-05,
+      "loss": 2.3547,
+      "step": 3522
+    },
+    {
+      "epoch": 1.5197325857235282,
+      "grad_norm": 0.18097326159477234,
+      "learning_rate": 7.43174501838729e-05,
+      "loss": 2.0562,
+      "step": 3523
+    },
+    {
+      "epoch": 1.5201638990726765,
+      "grad_norm": 0.18228799104690552,
+      "learning_rate": 7.42833237071102e-05,
+      "loss": 2.1742,
+      "step": 3524
+    },
+    {
+      "epoch": 1.5205952124218245,
+      "grad_norm": 0.17227162420749664,
+      "learning_rate": 7.424919737874316e-05,
+      "loss": 2.0236,
+      "step": 3525
+    },
+    {
+      "epoch": 1.5205952124218245,
+      "eval_loss": 2.0941295623779297,
+      "eval_runtime": 201.3088,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3525
+    },
+    {
+      "epoch": 1.5210265257709725,
+      "grad_norm": 0.20134098827838898,
+      "learning_rate": 7.421507120583801e-05,
+      "loss": 2.143,
+      "step": 3526
+    },
+    {
+      "epoch": 1.5214578391201208,
+      "grad_norm": 0.18537122011184692,
+      "learning_rate": 7.418094519546099e-05,
+      "loss": 2.3952,
+      "step": 3527
+    },
+    {
+      "epoch": 1.521889152469269,
+      "grad_norm": 0.21926040947437286,
+      "learning_rate": 7.414681935467819e-05,
+      "loss": 2.2343,
+      "step": 3528
+    },
+    {
+      "epoch": 1.522320465818417,
+      "grad_norm": 0.17132803797721863,
+      "learning_rate": 7.411269369055577e-05,
+      "loss": 2.0471,
+      "step": 3529
+    },
+    {
+      "epoch": 1.522751779167565,
+      "grad_norm": 0.19180268049240112,
+      "learning_rate": 7.407856821015984e-05,
+      "loss": 2.1841,
+      "step": 3530
+    },
+    {
+      "epoch": 1.5231830925167134,
+      "grad_norm": 0.17654922604560852,
+      "learning_rate": 7.404444292055642e-05,
+      "loss": 2.1316,
+      "step": 3531
+    },
+    {
+      "epoch": 1.5236144058658616,
+      "grad_norm": 0.20351848006248474,
+      "learning_rate": 7.401031782881157e-05,
+      "loss": 2.1818,
+      "step": 3532
+    },
+    {
+      "epoch": 1.5240457192150099,
+      "grad_norm": 0.172661691904068,
+      "learning_rate": 7.397619294199123e-05,
+      "loss": 2.1867,
+      "step": 3533
+    },
+    {
+      "epoch": 1.524477032564158,
+      "grad_norm": 0.19422678649425507,
+      "learning_rate": 7.394206826716131e-05,
+      "loss": 2.024,
+      "step": 3534
+    },
+    {
+      "epoch": 1.524908345913306,
+      "grad_norm": 0.18358619511127472,
+      "learning_rate": 7.390794381138775e-05,
+      "loss": 2.2407,
+      "step": 3535
+    },
+    {
+      "epoch": 1.5253396592624542,
+      "grad_norm": 0.18279418349266052,
+      "learning_rate": 7.387381958173636e-05,
+      "loss": 2.1805,
+      "step": 3536
+    },
+    {
+      "epoch": 1.5257709726116024,
+      "grad_norm": 0.17657078802585602,
+      "learning_rate": 7.383969558527293e-05,
+      "loss": 2.171,
+      "step": 3537
+    },
+    {
+      "epoch": 1.5262022859607505,
+      "grad_norm": 0.18388250470161438,
+      "learning_rate": 7.380557182906323e-05,
+      "loss": 2.073,
+      "step": 3538
+    },
+    {
+      "epoch": 1.5266335993098985,
+      "grad_norm": 0.18958689272403717,
+      "learning_rate": 7.377144832017296e-05,
+      "loss": 2.1475,
+      "step": 3539
+    },
+    {
+      "epoch": 1.5270649126590468,
+      "grad_norm": 0.16585297882556915,
+      "learning_rate": 7.373732506566773e-05,
+      "loss": 2.0459,
+      "step": 3540
+    },
+    {
+      "epoch": 1.527496226008195,
+      "grad_norm": 0.182152658700943,
+      "learning_rate": 7.370320207261315e-05,
+      "loss": 2.0672,
+      "step": 3541
+    },
+    {
+      "epoch": 1.5279275393573433,
+      "grad_norm": 0.17660200595855713,
+      "learning_rate": 7.366907934807477e-05,
+      "loss": 2.2686,
+      "step": 3542
+    },
+    {
+      "epoch": 1.5283588527064913,
+      "grad_norm": 0.1848592460155487,
+      "learning_rate": 7.363495689911807e-05,
+      "loss": 2.2825,
+      "step": 3543
+    },
+    {
+      "epoch": 1.5287901660556393,
+      "grad_norm": 0.18739467859268188,
+      "learning_rate": 7.360083473280844e-05,
+      "loss": 2.2374,
+      "step": 3544
+    },
+    {
+      "epoch": 1.5292214794047876,
+      "grad_norm": 0.17787355184555054,
+      "learning_rate": 7.356671285621132e-05,
+      "loss": 2.1722,
+      "step": 3545
+    },
+    {
+      "epoch": 1.5296527927539358,
+      "grad_norm": 0.182230144739151,
+      "learning_rate": 7.353259127639198e-05,
+      "loss": 2.2587,
+      "step": 3546
+    },
+    {
+      "epoch": 1.5300841061030839,
+      "grad_norm": 0.1774773746728897,
+      "learning_rate": 7.349847000041566e-05,
+      "loss": 2.1517,
+      "step": 3547
+    },
+    {
+      "epoch": 1.530515419452232,
+      "grad_norm": 0.18556547164916992,
+      "learning_rate": 7.346434903534751e-05,
+      "loss": 2.2128,
+      "step": 3548
+    },
+    {
+      "epoch": 1.5309467328013802,
+      "grad_norm": 0.17297889292240143,
+      "learning_rate": 7.343022838825272e-05,
+      "loss": 2.0662,
+      "step": 3549
+    },
+    {
+      "epoch": 1.5313780461505284,
+      "grad_norm": 0.1797797530889511,
+      "learning_rate": 7.33961080661963e-05,
+      "loss": 2.256,
+      "step": 3550
+    },
+    {
+      "epoch": 1.5313780461505284,
+      "eval_loss": 2.0944180488586426,
+      "eval_runtime": 201.7022,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3550
+    },
+    {
+      "epoch": 1.5318093594996767,
+      "grad_norm": 0.18436093628406525,
+      "learning_rate": 7.336198807624322e-05,
+      "loss": 2.1237,
+      "step": 3551
+    },
+    {
+      "epoch": 1.5322406728488247,
+      "grad_norm": 0.19445446133613586,
+      "learning_rate": 7.332786842545845e-05,
+      "loss": 2.3032,
+      "step": 3552
+    },
+    {
+      "epoch": 1.5326719861979727,
+      "grad_norm": 0.16594868898391724,
+      "learning_rate": 7.329374912090683e-05,
+      "loss": 2.1842,
+      "step": 3553
+    },
+    {
+      "epoch": 1.533103299547121,
+      "grad_norm": 0.28609853982925415,
+      "learning_rate": 7.325963016965308e-05,
+      "loss": 2.1166,
+      "step": 3554
+    },
+    {
+      "epoch": 1.5335346128962692,
+      "grad_norm": 0.19114753603935242,
+      "learning_rate": 7.322551157876194e-05,
+      "loss": 1.9692,
+      "step": 3555
+    },
+    {
+      "epoch": 1.5339659262454173,
+      "grad_norm": 0.18666251003742218,
+      "learning_rate": 7.319139335529804e-05,
+      "loss": 2.1514,
+      "step": 3556
+    },
+    {
+      "epoch": 1.5343972395945653,
+      "grad_norm": 0.1783144474029541,
+      "learning_rate": 7.315727550632593e-05,
+      "loss": 2.049,
+      "step": 3557
+    },
+    {
+      "epoch": 1.5348285529437136,
+      "grad_norm": 0.18049302697181702,
+      "learning_rate": 7.312315803891005e-05,
+      "loss": 1.9799,
+      "step": 3558
+    },
+    {
+      "epoch": 1.5352598662928618,
+      "grad_norm": 0.1771216243505478,
+      "learning_rate": 7.308904096011485e-05,
+      "loss": 2.2921,
+      "step": 3559
+    },
+    {
+      "epoch": 1.53569117964201,
+      "grad_norm": 0.17605052888393402,
+      "learning_rate": 7.305492427700463e-05,
+      "loss": 2.3114,
+      "step": 3560
+    },
+    {
+      "epoch": 1.536122492991158,
+      "grad_norm": 0.19665959477424622,
+      "learning_rate": 7.30208079966436e-05,
+      "loss": 2.288,
+      "step": 3561
+    },
+    {
+      "epoch": 1.5365538063403061,
+      "grad_norm": 0.18665872514247894,
+      "learning_rate": 7.298669212609588e-05,
+      "loss": 2.1297,
+      "step": 3562
+    },
+    {
+      "epoch": 1.5369851196894544,
+      "grad_norm": 0.17894567549228668,
+      "learning_rate": 7.295257667242561e-05,
+      "loss": 2.1004,
+      "step": 3563
+    },
+    {
+      "epoch": 1.5374164330386026,
+      "grad_norm": 0.19599032402038574,
+      "learning_rate": 7.291846164269671e-05,
+      "loss": 2.1506,
+      "step": 3564
+    },
+    {
+      "epoch": 1.5378477463877507,
+      "grad_norm": 0.17925117909908295,
+      "learning_rate": 7.288434704397308e-05,
+      "loss": 2.3149,
+      "step": 3565
+    },
+    {
+      "epoch": 1.5382790597368987,
+      "grad_norm": 0.17601613700389862,
+      "learning_rate": 7.285023288331858e-05,
+      "loss": 2.0923,
+      "step": 3566
+    },
+    {
+      "epoch": 1.538710373086047,
+      "grad_norm": 0.27391019463539124,
+      "learning_rate": 7.281611916779685e-05,
+      "loss": 2.2239,
+      "step": 3567
+    },
+    {
+      "epoch": 1.5391416864351952,
+      "grad_norm": 0.18218959867954254,
+      "learning_rate": 7.278200590447153e-05,
+      "loss": 2.1551,
+      "step": 3568
+    },
+    {
+      "epoch": 1.5395729997843435,
+      "grad_norm": 0.18566249310970306,
+      "learning_rate": 7.274789310040612e-05,
+      "loss": 2.1783,
+      "step": 3569
+    },
+    {
+      "epoch": 1.5400043131334915,
+      "grad_norm": 0.19422903656959534,
+      "learning_rate": 7.271378076266408e-05,
+      "loss": 2.1221,
+      "step": 3570
+    },
+    {
+      "epoch": 1.5404356264826395,
+      "grad_norm": 0.17731690406799316,
+      "learning_rate": 7.267966889830876e-05,
+      "loss": 2.2625,
+      "step": 3571
+    },
+    {
+      "epoch": 1.5408669398317878,
+      "grad_norm": 0.17667533457279205,
+      "learning_rate": 7.264555751440332e-05,
+      "loss": 1.8725,
+      "step": 3572
+    },
+    {
+      "epoch": 1.541298253180936,
+      "grad_norm": 0.17535041272640228,
+      "learning_rate": 7.261144661801099e-05,
+      "loss": 2.143,
+      "step": 3573
+    },
+    {
+      "epoch": 1.541729566530084,
+      "grad_norm": 0.19393585622310638,
+      "learning_rate": 7.257733621619474e-05,
+      "loss": 2.2788,
+      "step": 3574
+    },
+    {
+      "epoch": 1.5421608798792321,
+      "grad_norm": 0.18830327689647675,
+      "learning_rate": 7.254322631601752e-05,
+      "loss": 2.1969,
+      "step": 3575
+    },
+    {
+      "epoch": 1.5421608798792321,
+      "eval_loss": 2.0939698219299316,
+      "eval_runtime": 206.7306,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 3575
+    },
+    {
+      "epoch": 1.5425921932283804,
+      "grad_norm": 0.16435548663139343,
+      "learning_rate": 7.250911692454216e-05,
+      "loss": 2.0484,
+      "step": 3576
+    },
+    {
+      "epoch": 1.5430235065775286,
+      "grad_norm": 0.20261076092720032,
+      "learning_rate": 7.247500804883138e-05,
+      "loss": 2.3425,
+      "step": 3577
+    },
+    {
+      "epoch": 1.5434548199266769,
+      "grad_norm": 0.22808308899402618,
+      "learning_rate": 7.244089969594782e-05,
+      "loss": 2.2497,
+      "step": 3578
+    },
+    {
+      "epoch": 1.543886133275825,
+      "grad_norm": 0.2001747488975525,
+      "learning_rate": 7.240679187295394e-05,
+      "loss": 2.3458,
+      "step": 3579
+    },
+    {
+      "epoch": 1.544317446624973,
+      "grad_norm": 0.19855140149593353,
+      "learning_rate": 7.237268458691216e-05,
+      "loss": 2.085,
+      "step": 3580
+    },
+    {
+      "epoch": 1.5447487599741212,
+      "grad_norm": 0.18147879838943481,
+      "learning_rate": 7.233857784488476e-05,
+      "loss": 2.1479,
+      "step": 3581
+    },
+    {
+      "epoch": 1.5451800733232695,
+      "grad_norm": 0.16147656738758087,
+      "learning_rate": 7.230447165393394e-05,
+      "loss": 2.3397,
+      "step": 3582
+    },
+    {
+      "epoch": 1.5456113866724175,
+      "grad_norm": 0.19551271200180054,
+      "learning_rate": 7.227036602112172e-05,
+      "loss": 2.2916,
+      "step": 3583
+    },
+    {
+      "epoch": 1.5460427000215655,
+      "grad_norm": 0.19639210402965546,
+      "learning_rate": 7.22362609535101e-05,
+      "loss": 2.1183,
+      "step": 3584
+    },
+    {
+      "epoch": 1.5464740133707138,
+      "grad_norm": 0.1812535971403122,
+      "learning_rate": 7.220215645816086e-05,
+      "loss": 2.4531,
+      "step": 3585
+    },
+    {
+      "epoch": 1.546905326719862,
+      "grad_norm": 0.19537201523780823,
+      "learning_rate": 7.216805254213573e-05,
+      "loss": 2.1848,
+      "step": 3586
+    },
+    {
+      "epoch": 1.5473366400690103,
+      "grad_norm": 0.17963844537734985,
+      "learning_rate": 7.213394921249626e-05,
+      "loss": 2.1498,
+      "step": 3587
+    },
+    {
+      "epoch": 1.5477679534181583,
+      "grad_norm": 0.17501026391983032,
+      "learning_rate": 7.209984647630397e-05,
+      "loss": 2.2232,
+      "step": 3588
+    },
+    {
+      "epoch": 1.5481992667673063,
+      "grad_norm": 0.20665574073791504,
+      "learning_rate": 7.206574434062018e-05,
+      "loss": 2.1346,
+      "step": 3589
+    },
+    {
+      "epoch": 1.5486305801164546,
+      "grad_norm": 0.18055406212806702,
+      "learning_rate": 7.20316428125061e-05,
+      "loss": 1.9602,
+      "step": 3590
+    },
+    {
+      "epoch": 1.5490618934656029,
+      "grad_norm": 0.18960954248905182,
+      "learning_rate": 7.199754189902286e-05,
+      "loss": 2.3107,
+      "step": 3591
+    },
+    {
+      "epoch": 1.549493206814751,
+      "grad_norm": 0.17563574016094208,
+      "learning_rate": 7.19634416072314e-05,
+      "loss": 1.9254,
+      "step": 3592
+    },
+    {
+      "epoch": 1.549924520163899,
+      "grad_norm": 0.17406399548053741,
+      "learning_rate": 7.192934194419255e-05,
+      "loss": 2.3134,
+      "step": 3593
+    },
+    {
+      "epoch": 1.5503558335130472,
+      "grad_norm": 0.19203811883926392,
+      "learning_rate": 7.189524291696703e-05,
+      "loss": 2.1172,
+      "step": 3594
+    },
+    {
+      "epoch": 1.5507871468621954,
+      "grad_norm": 0.19079047441482544,
+      "learning_rate": 7.186114453261542e-05,
+      "loss": 2.0095,
+      "step": 3595
+    },
+    {
+      "epoch": 1.5512184602113437,
+      "grad_norm": 0.1765333116054535,
+      "learning_rate": 7.182704679819817e-05,
+      "loss": 2.0992,
+      "step": 3596
+    },
+    {
+      "epoch": 1.5516497735604917,
+      "grad_norm": 0.18226292729377747,
+      "learning_rate": 7.179294972077555e-05,
+      "loss": 2.2139,
+      "step": 3597
+    },
+    {
+      "epoch": 1.5520810869096398,
+      "grad_norm": 0.1825229823589325,
+      "learning_rate": 7.175885330740779e-05,
+      "loss": 2.2704,
+      "step": 3598
+    },
+    {
+      "epoch": 1.552512400258788,
+      "grad_norm": 0.1848277598619461,
+      "learning_rate": 7.172475756515491e-05,
+      "loss": 2.1899,
+      "step": 3599
+    },
+    {
+      "epoch": 1.5529437136079363,
+      "grad_norm": 0.17505969107151031,
+      "learning_rate": 7.169066250107678e-05,
+      "loss": 1.9208,
+      "step": 3600
+    },
+    {
+      "epoch": 1.5529437136079363,
+      "eval_loss": 2.093771457672119,
+      "eval_runtime": 201.1603,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3600
+    },
+    {
+      "epoch": 1.5533750269570843,
+      "grad_norm": 0.19374166429042816,
+      "learning_rate": 7.165656812223316e-05,
+      "loss": 2.1802,
+      "step": 3601
+    },
+    {
+      "epoch": 1.5538063403062323,
+      "grad_norm": 0.18536177277565002,
+      "learning_rate": 7.162247443568368e-05,
+      "loss": 2.3327,
+      "step": 3602
+    },
+    {
+      "epoch": 1.5542376536553806,
+      "grad_norm": 0.19441302120685577,
+      "learning_rate": 7.15883814484878e-05,
+      "loss": 2.087,
+      "step": 3603
+    },
+    {
+      "epoch": 1.5546689670045288,
+      "grad_norm": 0.18043111264705658,
+      "learning_rate": 7.155428916770484e-05,
+      "loss": 2.2256,
+      "step": 3604
+    },
+    {
+      "epoch": 1.555100280353677,
+      "grad_norm": 0.20430094003677368,
+      "learning_rate": 7.152019760039402e-05,
+      "loss": 2.2689,
+      "step": 3605
+    },
+    {
+      "epoch": 1.5555315937028251,
+      "grad_norm": 0.20238332450389862,
+      "learning_rate": 7.148610675361433e-05,
+      "loss": 2.3419,
+      "step": 3606
+    },
+    {
+      "epoch": 1.5559629070519732,
+      "grad_norm": 0.19608813524246216,
+      "learning_rate": 7.145201663442465e-05,
+      "loss": 2.2375,
+      "step": 3607
+    },
+    {
+      "epoch": 1.5563942204011214,
+      "grad_norm": 0.1908615380525589,
+      "learning_rate": 7.141792724988373e-05,
+      "loss": 2.2167,
+      "step": 3608
+    },
+    {
+      "epoch": 1.5568255337502697,
+      "grad_norm": 0.15812508761882782,
+      "learning_rate": 7.138383860705015e-05,
+      "loss": 2.2137,
+      "step": 3609
+    },
+    {
+      "epoch": 1.5572568470994177,
+      "grad_norm": 0.16722245514392853,
+      "learning_rate": 7.134975071298235e-05,
+      "loss": 2.1217,
+      "step": 3610
+    },
+    {
+      "epoch": 1.5576881604485657,
+      "grad_norm": 0.2262086719274521,
+      "learning_rate": 7.131566357473854e-05,
+      "loss": 2.1413,
+      "step": 3611
+    },
+    {
+      "epoch": 1.558119473797714,
+      "grad_norm": 0.1825946569442749,
+      "learning_rate": 7.128157719937691e-05,
+      "loss": 2.1258,
+      "step": 3612
+    },
+    {
+      "epoch": 1.5585507871468622,
+      "grad_norm": 0.18516826629638672,
+      "learning_rate": 7.124749159395538e-05,
+      "loss": 2.3571,
+      "step": 3613
+    },
+    {
+      "epoch": 1.5589821004960105,
+      "grad_norm": 0.8397002816200256,
+      "learning_rate": 7.121340676553175e-05,
+      "loss": 2.2926,
+      "step": 3614
+    },
+    {
+      "epoch": 1.5594134138451585,
+      "grad_norm": 0.18018440902233124,
+      "learning_rate": 7.117932272116365e-05,
+      "loss": 2.2957,
+      "step": 3615
+    },
+    {
+      "epoch": 1.5598447271943066,
+      "grad_norm": 0.1772436648607254,
+      "learning_rate": 7.114523946790857e-05,
+      "loss": 2.0046,
+      "step": 3616
+    },
+    {
+      "epoch": 1.5602760405434548,
+      "grad_norm": 0.18658606708049774,
+      "learning_rate": 7.111115701282384e-05,
+      "loss": 2.1106,
+      "step": 3617
+    },
+    {
+      "epoch": 1.560707353892603,
+      "grad_norm": 0.19837208092212677,
+      "learning_rate": 7.107707536296652e-05,
+      "loss": 2.2816,
+      "step": 3618
+    },
+    {
+      "epoch": 1.561138667241751,
+      "grad_norm": 0.18111276626586914,
+      "learning_rate": 7.10429945253937e-05,
+      "loss": 2.2865,
+      "step": 3619
+    },
+    {
+      "epoch": 1.5615699805908991,
+      "grad_norm": 0.17805011570453644,
+      "learning_rate": 7.10089145071621e-05,
+      "loss": 2.0251,
+      "step": 3620
+    },
+    {
+      "epoch": 1.5620012939400474,
+      "grad_norm": 0.18128082156181335,
+      "learning_rate": 7.097483531532842e-05,
+      "loss": 2.1825,
+      "step": 3621
+    },
+    {
+      "epoch": 1.5624326072891956,
+      "grad_norm": 0.17830413579940796,
+      "learning_rate": 7.094075695694909e-05,
+      "loss": 2.236,
+      "step": 3622
+    },
+    {
+      "epoch": 1.562863920638344,
+      "grad_norm": 0.17792202532291412,
+      "learning_rate": 7.090667943908044e-05,
+      "loss": 2.2288,
+      "step": 3623
+    },
+    {
+      "epoch": 1.563295233987492,
+      "grad_norm": 0.20085816085338593,
+      "learning_rate": 7.087260276877858e-05,
+      "loss": 2.1816,
+      "step": 3624
+    },
+    {
+      "epoch": 1.56372654733664,
+      "grad_norm": 0.18358451128005981,
+      "learning_rate": 7.083852695309944e-05,
+      "loss": 2.1069,
+      "step": 3625
+    },
+    {
+      "epoch": 1.56372654733664,
+      "eval_loss": 2.093684673309326,
+      "eval_runtime": 201.6838,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3625
+    },
+    {
+      "epoch": 1.5641578606857882,
+      "grad_norm": 0.17750532925128937,
+      "learning_rate": 7.08044519990988e-05,
+      "loss": 2.1724,
+      "step": 3626
+    },
+    {
+      "epoch": 1.5645891740349365,
+      "grad_norm": 0.201032355427742,
+      "learning_rate": 7.077037791383226e-05,
+      "loss": 2.1289,
+      "step": 3627
+    },
+    {
+      "epoch": 1.5650204873840845,
+      "grad_norm": 0.18252480030059814,
+      "learning_rate": 7.073630470435522e-05,
+      "loss": 2.2806,
+      "step": 3628
+    },
+    {
+      "epoch": 1.5654518007332325,
+      "grad_norm": 0.17951056361198425,
+      "learning_rate": 7.070223237772291e-05,
+      "loss": 2.1345,
+      "step": 3629
+    },
+    {
+      "epoch": 1.5658831140823808,
+      "grad_norm": 0.18737827241420746,
+      "learning_rate": 7.066816094099042e-05,
+      "loss": 2.1661,
+      "step": 3630
+    },
+    {
+      "epoch": 1.566314427431529,
+      "grad_norm": 0.1997365951538086,
+      "learning_rate": 7.063409040121257e-05,
+      "loss": 2.1505,
+      "step": 3631
+    },
+    {
+      "epoch": 1.5667457407806773,
+      "grad_norm": 0.19382046163082123,
+      "learning_rate": 7.060002076544405e-05,
+      "loss": 2.1109,
+      "step": 3632
+    },
+    {
+      "epoch": 1.5671770541298253,
+      "grad_norm": 0.17259393632411957,
+      "learning_rate": 7.056595204073931e-05,
+      "loss": 2.1547,
+      "step": 3633
+    },
+    {
+      "epoch": 1.5676083674789734,
+      "grad_norm": 0.17684011161327362,
+      "learning_rate": 7.053188423415273e-05,
+      "loss": 2.2844,
+      "step": 3634
+    },
+    {
+      "epoch": 1.5680396808281216,
+      "grad_norm": 0.17988254129886627,
+      "learning_rate": 7.049781735273836e-05,
+      "loss": 2.0416,
+      "step": 3635
+    },
+    {
+      "epoch": 1.5684709941772699,
+      "grad_norm": 0.164667010307312,
+      "learning_rate": 7.046375140355013e-05,
+      "loss": 2.1502,
+      "step": 3636
+    },
+    {
+      "epoch": 1.568902307526418,
+      "grad_norm": 0.1714576631784439,
+      "learning_rate": 7.042968639364181e-05,
+      "loss": 2.112,
+      "step": 3637
+    },
+    {
+      "epoch": 1.569333620875566,
+      "grad_norm": 0.1917424499988556,
+      "learning_rate": 7.039562233006691e-05,
+      "loss": 2.2119,
+      "step": 3638
+    },
+    {
+      "epoch": 1.5697649342247142,
+      "grad_norm": 0.16879668831825256,
+      "learning_rate": 7.036155921987874e-05,
+      "loss": 1.9152,
+      "step": 3639
+    },
+    {
+      "epoch": 1.5701962475738624,
+      "grad_norm": 0.1793883591890335,
+      "learning_rate": 7.032749707013045e-05,
+      "loss": 2.1579,
+      "step": 3640
+    },
+    {
+      "epoch": 1.5706275609230107,
+      "grad_norm": 0.1909743845462799,
+      "learning_rate": 7.029343588787499e-05,
+      "loss": 2.1412,
+      "step": 3641
+    },
+    {
+      "epoch": 1.5710588742721587,
+      "grad_norm": 0.1718597561120987,
+      "learning_rate": 7.025937568016511e-05,
+      "loss": 2.0683,
+      "step": 3642
+    },
+    {
+      "epoch": 1.5714901876213068,
+      "grad_norm": 0.2004898637533188,
+      "learning_rate": 7.022531645405332e-05,
+      "loss": 2.1873,
+      "step": 3643
+    },
+    {
+      "epoch": 1.571921500970455,
+      "grad_norm": 0.18730789422988892,
+      "learning_rate": 7.0191258216592e-05,
+      "loss": 2.1092,
+      "step": 3644
+    },
+    {
+      "epoch": 1.5723528143196033,
+      "grad_norm": 0.17451775074005127,
+      "learning_rate": 7.015720097483326e-05,
+      "loss": 2.2293,
+      "step": 3645
+    },
+    {
+      "epoch": 1.5727841276687513,
+      "grad_norm": 0.17554758489131927,
+      "learning_rate": 7.0123144735829e-05,
+      "loss": 2.0978,
+      "step": 3646
+    },
+    {
+      "epoch": 1.5732154410178993,
+      "grad_norm": 0.18201641738414764,
+      "learning_rate": 7.008908950663095e-05,
+      "loss": 2.0228,
+      "step": 3647
+    },
+    {
+      "epoch": 1.5736467543670476,
+      "grad_norm": 0.18830415606498718,
+      "learning_rate": 7.005503529429064e-05,
+      "loss": 2.0015,
+      "step": 3648
+    },
+    {
+      "epoch": 1.5740780677161959,
+      "grad_norm": 0.1909087747335434,
+      "learning_rate": 7.002098210585932e-05,
+      "loss": 2.0628,
+      "step": 3649
+    },
+    {
+      "epoch": 1.574509381065344,
+      "grad_norm": 0.18361161649227142,
+      "learning_rate": 6.998692994838812e-05,
+      "loss": 1.994,
+      "step": 3650
+    },
+    {
+      "epoch": 1.574509381065344,
+      "eval_loss": 2.093738317489624,
+      "eval_runtime": 200.716,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3650
+    },
+    {
+      "epoch": 1.5749406944144921,
+      "grad_norm": 0.19482986629009247,
+      "learning_rate": 6.995287882892791e-05,
+      "loss": 2.1545,
+      "step": 3651
+    },
+    {
+      "epoch": 1.5753720077636402,
+      "grad_norm": 0.1884821504354477,
+      "learning_rate": 6.991882875452932e-05,
+      "loss": 2.1563,
+      "step": 3652
+    },
+    {
+      "epoch": 1.5758033211127884,
+      "grad_norm": 0.1934521347284317,
+      "learning_rate": 6.988477973224281e-05,
+      "loss": 2.2272,
+      "step": 3653
+    },
+    {
+      "epoch": 1.5762346344619367,
+      "grad_norm": 0.19240306317806244,
+      "learning_rate": 6.985073176911858e-05,
+      "loss": 2.0432,
+      "step": 3654
+    },
+    {
+      "epoch": 1.5766659478110847,
+      "grad_norm": 0.20495979487895966,
+      "learning_rate": 6.981668487220666e-05,
+      "loss": 2.3219,
+      "step": 3655
+    },
+    {
+      "epoch": 1.5770972611602327,
+      "grad_norm": 0.19982969760894775,
+      "learning_rate": 6.978263904855683e-05,
+      "loss": 2.2596,
+      "step": 3656
+    },
+    {
+      "epoch": 1.577528574509381,
+      "grad_norm": 0.1917852759361267,
+      "learning_rate": 6.97485943052186e-05,
+      "loss": 2.1474,
+      "step": 3657
+    },
+    {
+      "epoch": 1.5779598878585293,
+      "grad_norm": 0.1615539789199829,
+      "learning_rate": 6.97145506492414e-05,
+      "loss": 1.9549,
+      "step": 3658
+    },
+    {
+      "epoch": 1.5783912012076775,
+      "grad_norm": 0.19644218683242798,
+      "learning_rate": 6.968050808767427e-05,
+      "loss": 2.1014,
+      "step": 3659
+    },
+    {
+      "epoch": 1.5788225145568255,
+      "grad_norm": 0.1665312498807907,
+      "learning_rate": 6.964646662756611e-05,
+      "loss": 1.6771,
+      "step": 3660
+    },
+    {
+      "epoch": 1.5792538279059736,
+      "grad_norm": 0.18535394966602325,
+      "learning_rate": 6.961242627596557e-05,
+      "loss": 2.1066,
+      "step": 3661
+    },
+    {
+      "epoch": 1.5796851412551218,
+      "grad_norm": 0.2052871584892273,
+      "learning_rate": 6.95783870399211e-05,
+      "loss": 1.9451,
+      "step": 3662
+    },
+    {
+      "epoch": 1.58011645460427,
+      "grad_norm": 0.18887032568454742,
+      "learning_rate": 6.95443489264809e-05,
+      "loss": 1.8664,
+      "step": 3663
+    },
+    {
+      "epoch": 1.5805477679534181,
+      "grad_norm": 0.17410443723201752,
+      "learning_rate": 6.951031194269287e-05,
+      "loss": 1.9715,
+      "step": 3664
+    },
+    {
+      "epoch": 1.5809790813025664,
+      "grad_norm": 0.17642557621002197,
+      "learning_rate": 6.947627609560483e-05,
+      "loss": 2.1207,
+      "step": 3665
+    },
+    {
+      "epoch": 1.5814103946517144,
+      "grad_norm": 0.18183539807796478,
+      "learning_rate": 6.944224139226421e-05,
+      "loss": 2.0773,
+      "step": 3666
+    },
+    {
+      "epoch": 1.5818417080008627,
+      "grad_norm": 0.18124890327453613,
+      "learning_rate": 6.94082078397183e-05,
+      "loss": 2.489,
+      "step": 3667
+    },
+    {
+      "epoch": 1.582273021350011,
+      "grad_norm": 0.21090415120124817,
+      "learning_rate": 6.937417544501408e-05,
+      "loss": 2.1516,
+      "step": 3668
+    },
+    {
+      "epoch": 1.582704334699159,
+      "grad_norm": 0.21677644550800323,
+      "learning_rate": 6.934014421519839e-05,
+      "loss": 2.1173,
+      "step": 3669
+    },
+    {
+      "epoch": 1.583135648048307,
+      "grad_norm": 0.1930365264415741,
+      "learning_rate": 6.930611415731774e-05,
+      "loss": 2.0612,
+      "step": 3670
+    },
+    {
+      "epoch": 1.5835669613974552,
+      "grad_norm": 0.1924310177564621,
+      "learning_rate": 6.927208527841841e-05,
+      "loss": 2.0412,
+      "step": 3671
+    },
+    {
+      "epoch": 1.5839982747466035,
+      "grad_norm": 0.16797201335430145,
+      "learning_rate": 6.923805758554644e-05,
+      "loss": 2.1477,
+      "step": 3672
+    },
+    {
+      "epoch": 1.5844295880957515,
+      "grad_norm": 0.18496067821979523,
+      "learning_rate": 6.920403108574767e-05,
+      "loss": 2.0006,
+      "step": 3673
+    },
+    {
+      "epoch": 1.5848609014448998,
+      "grad_norm": 0.19043207168579102,
+      "learning_rate": 6.917000578606766e-05,
+      "loss": 2.2337,
+      "step": 3674
+    },
+    {
+      "epoch": 1.5852922147940478,
+      "grad_norm": 0.1813650131225586,
+      "learning_rate": 6.91359816935517e-05,
+      "loss": 2.2809,
+      "step": 3675
+    },
+    {
+      "epoch": 1.5852922147940478,
+      "eval_loss": 2.093628406524658,
+      "eval_runtime": 203.0369,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3675
+    },
+    {
+      "epoch": 1.585723528143196,
+      "grad_norm": 0.19538749754428864,
+      "learning_rate": 6.910195881524485e-05,
+      "loss": 2.1929,
+      "step": 3676
+    },
+    {
+      "epoch": 1.5861548414923443,
+      "grad_norm": 0.20857664942741394,
+      "learning_rate": 6.906793715819195e-05,
+      "loss": 2.0994,
+      "step": 3677
+    },
+    {
+      "epoch": 1.5865861548414923,
+      "grad_norm": 0.16829629242420197,
+      "learning_rate": 6.903391672943752e-05,
+      "loss": 2.1483,
+      "step": 3678
+    },
+    {
+      "epoch": 1.5870174681906404,
+      "grad_norm": 0.1961994618177414,
+      "learning_rate": 6.899989753602587e-05,
+      "loss": 2.1908,
+      "step": 3679
+    },
+    {
+      "epoch": 1.5874487815397886,
+      "grad_norm": 0.1927735060453415,
+      "learning_rate": 6.896587958500105e-05,
+      "loss": 2.154,
+      "step": 3680
+    },
+    {
+      "epoch": 1.5878800948889369,
+      "grad_norm": 0.17948473989963531,
+      "learning_rate": 6.893186288340687e-05,
+      "loss": 1.9965,
+      "step": 3681
+    },
+    {
+      "epoch": 1.588311408238085,
+      "grad_norm": 0.1843881756067276,
+      "learning_rate": 6.88978474382868e-05,
+      "loss": 2.0506,
+      "step": 3682
+    },
+    {
+      "epoch": 1.5887427215872332,
+      "grad_norm": 0.16958780586719513,
+      "learning_rate": 6.886383325668417e-05,
+      "loss": 2.0875,
+      "step": 3683
+    },
+    {
+      "epoch": 1.5891740349363812,
+      "grad_norm": 0.17877309024333954,
+      "learning_rate": 6.8829820345642e-05,
+      "loss": 2.1458,
+      "step": 3684
+    },
+    {
+      "epoch": 1.5896053482855295,
+      "grad_norm": 0.17716002464294434,
+      "learning_rate": 6.879580871220298e-05,
+      "loss": 2.167,
+      "step": 3685
+    },
+    {
+      "epoch": 1.5900366616346777,
+      "grad_norm": 0.18612174689769745,
+      "learning_rate": 6.876179836340959e-05,
+      "loss": 2.0139,
+      "step": 3686
+    },
+    {
+      "epoch": 1.5904679749838257,
+      "grad_norm": 0.20165590941905975,
+      "learning_rate": 6.87277893063041e-05,
+      "loss": 2.1366,
+      "step": 3687
+    },
+    {
+      "epoch": 1.5908992883329738,
+      "grad_norm": 0.19413705170154572,
+      "learning_rate": 6.869378154792841e-05,
+      "loss": 2.0338,
+      "step": 3688
+    },
+    {
+      "epoch": 1.591330601682122,
+      "grad_norm": 0.18128186464309692,
+      "learning_rate": 6.865977509532421e-05,
+      "loss": 2.1962,
+      "step": 3689
+    },
+    {
+      "epoch": 1.5917619150312703,
+      "grad_norm": 0.20013929903507233,
+      "learning_rate": 6.862576995553294e-05,
+      "loss": 2.0225,
+      "step": 3690
+    },
+    {
+      "epoch": 1.5921932283804183,
+      "grad_norm": 0.19913722574710846,
+      "learning_rate": 6.85917661355957e-05,
+      "loss": 2.1976,
+      "step": 3691
+    },
+    {
+      "epoch": 1.5926245417295666,
+      "grad_norm": 0.16997520625591278,
+      "learning_rate": 6.855776364255334e-05,
+      "loss": 2.1994,
+      "step": 3692
+    },
+    {
+      "epoch": 1.5930558550787146,
+      "grad_norm": 0.17715983092784882,
+      "learning_rate": 6.852376248344648e-05,
+      "loss": 2.0439,
+      "step": 3693
+    },
+    {
+      "epoch": 1.5934871684278629,
+      "grad_norm": 0.19184236228466034,
+      "learning_rate": 6.848976266531541e-05,
+      "loss": 2.2112,
+      "step": 3694
+    },
+    {
+      "epoch": 1.5939184817770111,
+      "grad_norm": 0.20436334609985352,
+      "learning_rate": 6.845576419520021e-05,
+      "loss": 1.9827,
+      "step": 3695
+    },
+    {
+      "epoch": 1.5943497951261592,
+      "grad_norm": 0.18236425518989563,
+      "learning_rate": 6.842176708014054e-05,
+      "loss": 2.1338,
+      "step": 3696
+    },
+    {
+      "epoch": 1.5947811084753072,
+      "grad_norm": 0.18234539031982422,
+      "learning_rate": 6.838777132717599e-05,
+      "loss": 2.2777,
+      "step": 3697
+    },
+    {
+      "epoch": 1.5952124218244554,
+      "grad_norm": 0.1843755543231964,
+      "learning_rate": 6.835377694334568e-05,
+      "loss": 2.1487,
+      "step": 3698
+    },
+    {
+      "epoch": 1.5956437351736037,
+      "grad_norm": 0.17670325934886932,
+      "learning_rate": 6.831978393568853e-05,
+      "loss": 2.1982,
+      "step": 3699
+    },
+    {
+      "epoch": 1.5960750485227517,
+      "grad_norm": 0.19596204161643982,
+      "learning_rate": 6.828579231124316e-05,
+      "loss": 2.1467,
+      "step": 3700
+    },
+    {
+      "epoch": 1.5960750485227517,
+      "eval_loss": 2.093451976776123,
+      "eval_runtime": 201.8392,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3700
+    },
+    {
+      "epoch": 1.5965063618719,
+      "grad_norm": 0.18570303916931152,
+      "learning_rate": 6.825180207704793e-05,
+      "loss": 1.9782,
+      "step": 3701
+    },
+    {
+      "epoch": 1.596937675221048,
+      "grad_norm": 0.2120261937379837,
+      "learning_rate": 6.82178132401409e-05,
+      "loss": 2.0996,
+      "step": 3702
+    },
+    {
+      "epoch": 1.5973689885701963,
+      "grad_norm": 0.20858530700206757,
+      "learning_rate": 6.818382580755975e-05,
+      "loss": 2.1682,
+      "step": 3703
+    },
+    {
+      "epoch": 1.5978003019193445,
+      "grad_norm": 0.1919013112783432,
+      "learning_rate": 6.814983978634206e-05,
+      "loss": 2.0944,
+      "step": 3704
+    },
+    {
+      "epoch": 1.5982316152684926,
+      "grad_norm": 0.1854456216096878,
+      "learning_rate": 6.811585518352493e-05,
+      "loss": 2.2461,
+      "step": 3705
+    },
+    {
+      "epoch": 1.5986629286176406,
+      "grad_norm": 0.1700373888015747,
+      "learning_rate": 6.808187200614528e-05,
+      "loss": 1.8029,
+      "step": 3706
+    },
+    {
+      "epoch": 1.5990942419667888,
+      "grad_norm": 0.2353130578994751,
+      "learning_rate": 6.804789026123967e-05,
+      "loss": 2.0473,
+      "step": 3707
+    },
+    {
+      "epoch": 1.599525555315937,
+      "grad_norm": 0.18630927801132202,
+      "learning_rate": 6.801390995584441e-05,
+      "loss": 2.3925,
+      "step": 3708
+    },
+    {
+      "epoch": 1.5999568686650851,
+      "grad_norm": 0.2129543125629425,
+      "learning_rate": 6.797993109699552e-05,
+      "loss": 2.2899,
+      "step": 3709
+    },
+    {
+      "epoch": 1.6003881820142334,
+      "grad_norm": 0.18666571378707886,
+      "learning_rate": 6.794595369172862e-05,
+      "loss": 1.9599,
+      "step": 3710
+    },
+    {
+      "epoch": 1.6008194953633814,
+      "grad_norm": 0.16544368863105774,
+      "learning_rate": 6.791197774707919e-05,
+      "loss": 2.0883,
+      "step": 3711
+    },
+    {
+      "epoch": 1.6012508087125297,
+      "grad_norm": 0.18772681057453156,
+      "learning_rate": 6.787800327008227e-05,
+      "loss": 2.3356,
+      "step": 3712
+    },
+    {
+      "epoch": 1.601682122061678,
+      "grad_norm": 0.19081822037696838,
+      "learning_rate": 6.784403026777266e-05,
+      "loss": 2.0163,
+      "step": 3713
+    },
+    {
+      "epoch": 1.602113435410826,
+      "grad_norm": 0.17879170179367065,
+      "learning_rate": 6.781005874718481e-05,
+      "loss": 2.2735,
+      "step": 3714
+    },
+    {
+      "epoch": 1.602544748759974,
+      "grad_norm": 0.19571247696876526,
+      "learning_rate": 6.777608871535296e-05,
+      "loss": 2.1857,
+      "step": 3715
+    },
+    {
+      "epoch": 1.6029760621091222,
+      "grad_norm": 0.19109483063220978,
+      "learning_rate": 6.774212017931095e-05,
+      "loss": 2.1034,
+      "step": 3716
+    },
+    {
+      "epoch": 1.6034073754582705,
+      "grad_norm": 0.1833193004131317,
+      "learning_rate": 6.770815314609228e-05,
+      "loss": 2.0422,
+      "step": 3717
+    },
+    {
+      "epoch": 1.6038386888074188,
+      "grad_norm": 0.1861751526594162,
+      "learning_rate": 6.767418762273028e-05,
+      "loss": 2.1788,
+      "step": 3718
+    },
+    {
+      "epoch": 1.6042700021565668,
+      "grad_norm": 0.2010440081357956,
+      "learning_rate": 6.764022361625784e-05,
+      "loss": 2.1818,
+      "step": 3719
+    },
+    {
+      "epoch": 1.6047013155057148,
+      "grad_norm": 0.20387910306453705,
+      "learning_rate": 6.76062611337076e-05,
+      "loss": 2.0423,
+      "step": 3720
+    },
+    {
+      "epoch": 1.605132628854863,
+      "grad_norm": 0.19214536249637604,
+      "learning_rate": 6.75723001821118e-05,
+      "loss": 2.2937,
+      "step": 3721
+    },
+    {
+      "epoch": 1.6055639422040113,
+      "grad_norm": 0.1803417056798935,
+      "learning_rate": 6.753834076850252e-05,
+      "loss": 2.038,
+      "step": 3722
+    },
+    {
+      "epoch": 1.6059952555531594,
+      "grad_norm": 0.19501323997974396,
+      "learning_rate": 6.750438289991139e-05,
+      "loss": 2.2286,
+      "step": 3723
+    },
+    {
+      "epoch": 1.6064265689023074,
+      "grad_norm": 0.177200585603714,
+      "learning_rate": 6.747042658336972e-05,
+      "loss": 2.261,
+      "step": 3724
+    },
+    {
+      "epoch": 1.6068578822514556,
+      "grad_norm": 0.19286774098873138,
+      "learning_rate": 6.743647182590857e-05,
+      "loss": 2.1549,
+      "step": 3725
+    },
+    {
+      "epoch": 1.6068578822514556,
+      "eval_loss": 2.0928478240966797,
+      "eval_runtime": 201.4313,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3725
+    },
+    {
+      "epoch": 1.607289195600604,
+      "grad_norm": 0.2139384150505066,
+      "learning_rate": 6.740251863455866e-05,
+      "loss": 2.2656,
+      "step": 3726
+    },
+    {
+      "epoch": 1.6077205089497522,
+      "grad_norm": 0.184177428483963,
+      "learning_rate": 6.736856701635033e-05,
+      "loss": 2.3855,
+      "step": 3727
+    },
+    {
+      "epoch": 1.6081518222989002,
+      "grad_norm": 0.18360666930675507,
+      "learning_rate": 6.733461697831366e-05,
+      "loss": 2.1589,
+      "step": 3728
+    },
+    {
+      "epoch": 1.6085831356480482,
+      "grad_norm": 0.9161972403526306,
+      "learning_rate": 6.73006685274784e-05,
+      "loss": 2.1925,
+      "step": 3729
+    },
+    {
+      "epoch": 1.6090144489971965,
+      "grad_norm": 0.19454066455364227,
+      "learning_rate": 6.72667216708739e-05,
+      "loss": 2.0233,
+      "step": 3730
+    },
+    {
+      "epoch": 1.6094457623463447,
+      "grad_norm": 0.18415388464927673,
+      "learning_rate": 6.723277641552924e-05,
+      "loss": 2.0935,
+      "step": 3731
+    },
+    {
+      "epoch": 1.6098770756954928,
+      "grad_norm": 0.19692564010620117,
+      "learning_rate": 6.719883276847316e-05,
+      "loss": 2.2279,
+      "step": 3732
+    },
+    {
+      "epoch": 1.6103083890446408,
+      "grad_norm": 0.21135157346725464,
+      "learning_rate": 6.716489073673408e-05,
+      "loss": 2.1733,
+      "step": 3733
+    },
+    {
+      "epoch": 1.610739702393789,
+      "grad_norm": 0.2029101848602295,
+      "learning_rate": 6.713095032734007e-05,
+      "loss": 2.1401,
+      "step": 3734
+    },
+    {
+      "epoch": 1.6111710157429373,
+      "grad_norm": 0.17743490636348724,
+      "learning_rate": 6.70970115473188e-05,
+      "loss": 1.9196,
+      "step": 3735
+    },
+    {
+      "epoch": 1.6116023290920856,
+      "grad_norm": 0.18146555125713348,
+      "learning_rate": 6.706307440369778e-05,
+      "loss": 2.3626,
+      "step": 3736
+    },
+    {
+      "epoch": 1.6120336424412336,
+      "grad_norm": 0.1897459179162979,
+      "learning_rate": 6.702913890350396e-05,
+      "loss": 2.4544,
+      "step": 3737
+    },
+    {
+      "epoch": 1.6124649557903816,
+      "grad_norm": 0.17637895047664642,
+      "learning_rate": 6.699520505376411e-05,
+      "loss": 2.1949,
+      "step": 3738
+    },
+    {
+      "epoch": 1.6128962691395299,
+      "grad_norm": 0.19048604369163513,
+      "learning_rate": 6.696127286150458e-05,
+      "loss": 2.1958,
+      "step": 3739
+    },
+    {
+      "epoch": 1.6133275824886781,
+      "grad_norm": 0.1806829422712326,
+      "learning_rate": 6.692734233375143e-05,
+      "loss": 1.9665,
+      "step": 3740
+    },
+    {
+      "epoch": 1.6137588958378262,
+      "grad_norm": 0.17890702188014984,
+      "learning_rate": 6.689341347753034e-05,
+      "loss": 2.1628,
+      "step": 3741
+    },
+    {
+      "epoch": 1.6141902091869742,
+      "grad_norm": 0.21568112075328827,
+      "learning_rate": 6.685948629986659e-05,
+      "loss": 2.2335,
+      "step": 3742
+    },
+    {
+      "epoch": 1.6146215225361225,
+      "grad_norm": 0.19446949660778046,
+      "learning_rate": 6.682556080778529e-05,
+      "loss": 2.008,
+      "step": 3743
+    },
+    {
+      "epoch": 1.6150528358852707,
+      "grad_norm": 0.18394096195697784,
+      "learning_rate": 6.679163700831099e-05,
+      "loss": 2.1141,
+      "step": 3744
+    },
+    {
+      "epoch": 1.615484149234419,
+      "grad_norm": 0.19576245546340942,
+      "learning_rate": 6.675771490846802e-05,
+      "loss": 2.2772,
+      "step": 3745
+    },
+    {
+      "epoch": 1.615915462583567,
+      "grad_norm": 0.18927359580993652,
+      "learning_rate": 6.672379451528029e-05,
+      "loss": 2.2442,
+      "step": 3746
+    },
+    {
+      "epoch": 1.616346775932715,
+      "grad_norm": 0.19549596309661865,
+      "learning_rate": 6.668987583577143e-05,
+      "loss": 2.1024,
+      "step": 3747
+    },
+    {
+      "epoch": 1.6167780892818633,
+      "grad_norm": 0.18620263040065765,
+      "learning_rate": 6.665595887696468e-05,
+      "loss": 2.1882,
+      "step": 3748
+    },
+    {
+      "epoch": 1.6172094026310115,
+      "grad_norm": 0.19507049024105072,
+      "learning_rate": 6.662204364588283e-05,
+      "loss": 2.1049,
+      "step": 3749
+    },
+    {
+      "epoch": 1.6176407159801596,
+      "grad_norm": 0.20146001875400543,
+      "learning_rate": 6.658813014954852e-05,
+      "loss": 2.2832,
+      "step": 3750
+    },
+    {
+      "epoch": 1.6176407159801596,
+      "eval_loss": 2.09356689453125,
+      "eval_runtime": 201.387,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 3750
+    },
+    {
+      "epoch": 1.6180720293293076,
+      "grad_norm": 0.22598698735237122,
+      "learning_rate": 6.655421839498385e-05,
+      "loss": 1.9081,
+      "step": 3751
+    },
+    {
+      "epoch": 1.6185033426784559,
+      "grad_norm": 0.1829482764005661,
+      "learning_rate": 6.65203083892106e-05,
+      "loss": 2.0644,
+      "step": 3752
+    },
+    {
+      "epoch": 1.6189346560276041,
+      "grad_norm": 0.20377013087272644,
+      "learning_rate": 6.648640013925023e-05,
+      "loss": 2.233,
+      "step": 3753
+    },
+    {
+      "epoch": 1.6193659693767524,
+      "grad_norm": 0.19465422630310059,
+      "learning_rate": 6.645249365212382e-05,
+      "loss": 2.3635,
+      "step": 3754
+    },
+    {
+      "epoch": 1.6197972827259004,
+      "grad_norm": 0.18360711634159088,
+      "learning_rate": 6.64185889348521e-05,
+      "loss": 2.2299,
+      "step": 3755
+    },
+    {
+      "epoch": 1.6202285960750484,
+      "grad_norm": 0.19367410242557526,
+      "learning_rate": 6.638468599445533e-05,
+      "loss": 2.1938,
+      "step": 3756
+    },
+    {
+      "epoch": 1.6206599094241967,
+      "grad_norm": 0.1897435337305069,
+      "learning_rate": 6.635078483795359e-05,
+      "loss": 1.9657,
+      "step": 3757
+    },
+    {
+      "epoch": 1.621091222773345,
+      "grad_norm": 0.18628539144992828,
+      "learning_rate": 6.631688547236642e-05,
+      "loss": 2.2478,
+      "step": 3758
+    },
+    {
+      "epoch": 1.621522536122493,
+      "grad_norm": 0.17115898430347443,
+      "learning_rate": 6.628298790471307e-05,
+      "loss": 2.3662,
+      "step": 3759
+    },
+    {
+      "epoch": 1.621953849471641,
+      "grad_norm": 0.17931334674358368,
+      "learning_rate": 6.624909214201239e-05,
+      "loss": 2.146,
+      "step": 3760
+    },
+    {
+      "epoch": 1.6223851628207893,
+      "grad_norm": 0.22137954831123352,
+      "learning_rate": 6.621519819128289e-05,
+      "loss": 2.087,
+      "step": 3761
+    },
+    {
+      "epoch": 1.6228164761699375,
+      "grad_norm": 0.17036446928977966,
+      "learning_rate": 6.61813060595427e-05,
+      "loss": 2.0117,
+      "step": 3762
+    },
+    {
+      "epoch": 1.6232477895190858,
+      "grad_norm": 0.17956706881523132,
+      "learning_rate": 6.614741575380946e-05,
+      "loss": 2.0999,
+      "step": 3763
+    },
+    {
+      "epoch": 1.6236791028682338,
+      "grad_norm": 0.19675466418266296,
+      "learning_rate": 6.611352728110067e-05,
+      "loss": 2.1033,
+      "step": 3764
+    },
+    {
+      "epoch": 1.6241104162173818,
+      "grad_norm": 0.20353537797927856,
+      "learning_rate": 6.60796406484332e-05,
+      "loss": 2.1655,
+      "step": 3765
+    },
+    {
+      "epoch": 1.62454172956653,
+      "grad_norm": 0.21810005605220795,
+      "learning_rate": 6.60457558628237e-05,
+      "loss": 2.2077,
+      "step": 3766
+    },
+    {
+      "epoch": 1.6249730429156783,
+      "grad_norm": 0.18552573025226593,
+      "learning_rate": 6.601187293128836e-05,
+      "loss": 2.2549,
+      "step": 3767
+    },
+    {
+      "epoch": 1.6254043562648264,
+      "grad_norm": 0.20179718732833862,
+      "learning_rate": 6.597799186084302e-05,
+      "loss": 2.1223,
+      "step": 3768
+    },
+    {
+      "epoch": 1.6258356696139744,
+      "grad_norm": 0.18860100209712982,
+      "learning_rate": 6.594411265850315e-05,
+      "loss": 2.3134,
+      "step": 3769
+    },
+    {
+      "epoch": 1.6262669829631227,
+      "grad_norm": 0.18685784935951233,
+      "learning_rate": 6.591023533128378e-05,
+      "loss": 2.1947,
+      "step": 3770
+    },
+    {
+      "epoch": 1.626698296312271,
+      "grad_norm": 0.20574668049812317,
+      "learning_rate": 6.587635988619957e-05,
+      "loss": 2.1871,
+      "step": 3771
+    },
+    {
+      "epoch": 1.6271296096614192,
+      "grad_norm": 0.1886902153491974,
+      "learning_rate": 6.584248633026483e-05,
+      "loss": 2.0815,
+      "step": 3772
+    },
+    {
+      "epoch": 1.6275609230105672,
+      "grad_norm": 0.19360224902629852,
+      "learning_rate": 6.580861467049345e-05,
+      "loss": 2.2357,
+      "step": 3773
+    },
+    {
+      "epoch": 1.6279922363597152,
+      "grad_norm": 0.17270958423614502,
+      "learning_rate": 6.57747449138989e-05,
+      "loss": 2.201,
+      "step": 3774
+    },
+    {
+      "epoch": 1.6284235497088635,
+      "grad_norm": 0.18650749325752258,
+      "learning_rate": 6.574087706749432e-05,
+      "loss": 2.0524,
+      "step": 3775
+    },
+    {
+      "epoch": 1.6284235497088635,
+      "eval_loss": 2.093459367752075,
+      "eval_runtime": 202.3645,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 3775
+    },
+    {
+      "epoch": 1.6288548630580117,
+      "grad_norm": 0.19357305765151978,
+      "learning_rate": 6.57070111382924e-05,
+      "loss": 2.1072,
+      "step": 3776
+    },
+    {
+      "epoch": 1.6292861764071598,
+      "grad_norm": 0.17448721826076508,
+      "learning_rate": 6.567314713330544e-05,
+      "loss": 1.9925,
+      "step": 3777
+    },
+    {
+      "epoch": 1.6297174897563078,
+      "grad_norm": 0.176400825381279,
+      "learning_rate": 6.563928505954534e-05,
+      "loss": 2.1545,
+      "step": 3778
+    },
+    {
+      "epoch": 1.630148803105456,
+      "grad_norm": 0.1928510069847107,
+      "learning_rate": 6.560542492402367e-05,
+      "loss": 2.1201,
+      "step": 3779
+    },
+    {
+      "epoch": 1.6305801164546043,
+      "grad_norm": 0.19245097041130066,
+      "learning_rate": 6.557156673375151e-05,
+      "loss": 2.2759,
+      "step": 3780
+    },
+    {
+      "epoch": 1.6310114298037526,
+      "grad_norm": 0.200565367937088,
+      "learning_rate": 6.553771049573952e-05,
+      "loss": 2.1643,
+      "step": 3781
+    },
+    {
+      "epoch": 1.6314427431529006,
+      "grad_norm": 0.1743149757385254,
+      "learning_rate": 6.55038562169981e-05,
+      "loss": 2.2006,
+      "step": 3782
+    },
+    {
+      "epoch": 1.6318740565020486,
+      "grad_norm": 0.18234965205192566,
+      "learning_rate": 6.547000390453708e-05,
+      "loss": 2.3149,
+      "step": 3783
+    },
+    {
+      "epoch": 1.632305369851197,
+      "grad_norm": 0.19367872178554535,
+      "learning_rate": 6.543615356536596e-05,
+      "loss": 2.2829,
+      "step": 3784
+    },
+    {
+      "epoch": 1.6327366832003452,
+      "grad_norm": 0.19692032039165497,
+      "learning_rate": 6.540230520649383e-05,
+      "loss": 2.2099,
+      "step": 3785
+    },
+    {
+      "epoch": 1.6331679965494932,
+      "grad_norm": 0.193384051322937,
+      "learning_rate": 6.536845883492939e-05,
+      "loss": 2.4338,
+      "step": 3786
+    },
+    {
+      "epoch": 1.6335993098986412,
+      "grad_norm": 0.1894853711128235,
+      "learning_rate": 6.533461445768087e-05,
+      "loss": 2.0901,
+      "step": 3787
+    },
+    {
+      "epoch": 1.6340306232477895,
+      "grad_norm": 0.1653389036655426,
+      "learning_rate": 6.530077208175609e-05,
+      "loss": 1.8634,
+      "step": 3788
+    },
+    {
+      "epoch": 1.6344619365969377,
+      "grad_norm": 0.1730147749185562,
+      "learning_rate": 6.526693171416258e-05,
+      "loss": 2.1809,
+      "step": 3789
+    },
+    {
+      "epoch": 1.634893249946086,
+      "grad_norm": 0.18234191834926605,
+      "learning_rate": 6.523309336190727e-05,
+      "loss": 2.1617,
+      "step": 3790
+    },
+    {
+      "epoch": 1.635324563295234,
+      "grad_norm": 0.19211749732494354,
+      "learning_rate": 6.519925703199678e-05,
+      "loss": 2.1767,
+      "step": 3791
+    },
+    {
+      "epoch": 1.635755876644382,
+      "grad_norm": 0.2109699696302414,
+      "learning_rate": 6.516542273143729e-05,
+      "loss": 2.3465,
+      "step": 3792
+    },
+    {
+      "epoch": 1.6361871899935303,
+      "grad_norm": 0.18605192005634308,
+      "learning_rate": 6.513159046723459e-05,
+      "loss": 2.1536,
+      "step": 3793
+    },
+    {
+      "epoch": 1.6366185033426786,
+      "grad_norm": 0.1837363839149475,
+      "learning_rate": 6.509776024639403e-05,
+      "loss": 2.1865,
+      "step": 3794
+    },
+    {
+      "epoch": 1.6370498166918266,
+      "grad_norm": 0.18559573590755463,
+      "learning_rate": 6.506393207592043e-05,
+      "loss": 2.0815,
+      "step": 3795
+    },
+    {
+      "epoch": 1.6374811300409746,
+      "grad_norm": 0.1857297122478485,
+      "learning_rate": 6.503010596281841e-05,
+      "loss": 2.2236,
+      "step": 3796
+    },
+    {
+      "epoch": 1.6379124433901229,
+      "grad_norm": 0.20153896510601044,
+      "learning_rate": 6.499628191409196e-05,
+      "loss": 2.2147,
+      "step": 3797
+    },
+    {
+      "epoch": 1.6383437567392711,
+      "grad_norm": 0.1797814816236496,
+      "learning_rate": 6.496245993674476e-05,
+      "loss": 2.1808,
+      "step": 3798
+    },
+    {
+      "epoch": 1.6387750700884194,
+      "grad_norm": 0.18175220489501953,
+      "learning_rate": 6.492864003777996e-05,
+      "loss": 2.2005,
+      "step": 3799
+    },
+    {
+      "epoch": 1.6392063834375674,
+      "grad_norm": 0.20047171413898468,
+      "learning_rate": 6.489482222420039e-05,
+      "loss": 2.1684,
+      "step": 3800
+    },
+    {
+      "epoch": 1.6392063834375674,
+      "eval_loss": 2.092963218688965,
+      "eval_runtime": 196.5623,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 3800
+    },
+    {
+      "epoch": 1.6396376967867154,
+      "grad_norm": 0.22544299066066742,
+      "learning_rate": 6.486100650300842e-05,
+      "loss": 2.0748,
+      "step": 3801
+    },
+    {
+      "epoch": 1.6400690101358637,
+      "grad_norm": 0.19890649616718292,
+      "learning_rate": 6.482719288120587e-05,
+      "loss": 2.1556,
+      "step": 3802
+    },
+    {
+      "epoch": 1.640500323485012,
+      "grad_norm": 0.18018411099910736,
+      "learning_rate": 6.479338136579435e-05,
+      "loss": 2.1624,
+      "step": 3803
+    },
+    {
+      "epoch": 1.64093163683416,
+      "grad_norm": 0.19493593275547028,
+      "learning_rate": 6.47595719637748e-05,
+      "loss": 2.2055,
+      "step": 3804
+    },
+    {
+      "epoch": 1.641362950183308,
+      "grad_norm": 0.1730249971151352,
+      "learning_rate": 6.472576468214787e-05,
+      "loss": 1.8869,
+      "step": 3805
+    },
+    {
+      "epoch": 1.6417942635324563,
+      "grad_norm": 0.18684785068035126,
+      "learning_rate": 6.46919595279137e-05,
+      "loss": 2.2683,
+      "step": 3806
+    },
+    {
+      "epoch": 1.6422255768816045,
+      "grad_norm": 0.18305426836013794,
+      "learning_rate": 6.465815650807206e-05,
+      "loss": 2.3451,
+      "step": 3807
+    },
+    {
+      "epoch": 1.6426568902307528,
+      "grad_norm": 0.23286522924900055,
+      "learning_rate": 6.462435562962222e-05,
+      "loss": 2.1248,
+      "step": 3808
+    },
+    {
+      "epoch": 1.6430882035799008,
+      "grad_norm": 0.1933186948299408,
+      "learning_rate": 6.4590556899563e-05,
+      "loss": 2.0885,
+      "step": 3809
+    },
+    {
+      "epoch": 1.6435195169290489,
+      "grad_norm": 0.16845665872097015,
+      "learning_rate": 6.45567603248928e-05,
+      "loss": 1.8661,
+      "step": 3810
+    },
+    {
+      "epoch": 1.643950830278197,
+      "grad_norm": 0.19671374559402466,
+      "learning_rate": 6.45229659126096e-05,
+      "loss": 1.7578,
+      "step": 3811
+    },
+    {
+      "epoch": 1.6443821436273454,
+      "grad_norm": 0.19340135157108307,
+      "learning_rate": 6.448917366971085e-05,
+      "loss": 2.192,
+      "step": 3812
+    },
+    {
+      "epoch": 1.6448134569764934,
+      "grad_norm": 0.18395887315273285,
+      "learning_rate": 6.445538360319366e-05,
+      "loss": 2.2982,
+      "step": 3813
+    },
+    {
+      "epoch": 1.6452447703256414,
+      "grad_norm": 0.2056943029165268,
+      "learning_rate": 6.442159572005462e-05,
+      "loss": 2.2493,
+      "step": 3814
+    },
+    {
+      "epoch": 1.6456760836747897,
+      "grad_norm": 0.20504404604434967,
+      "learning_rate": 6.438781002728985e-05,
+      "loss": 2.1617,
+      "step": 3815
+    },
+    {
+      "epoch": 1.646107397023938,
+      "grad_norm": 0.18038319051265717,
+      "learning_rate": 6.435402653189509e-05,
+      "loss": 2.1422,
+      "step": 3816
+    },
+    {
+      "epoch": 1.6465387103730862,
+      "grad_norm": 0.18079359829425812,
+      "learning_rate": 6.432024524086552e-05,
+      "loss": 2.1337,
+      "step": 3817
+    },
+    {
+      "epoch": 1.6469700237222342,
+      "grad_norm": 0.18782946467399597,
+      "learning_rate": 6.428646616119599e-05,
+      "loss": 1.9715,
+      "step": 3818
+    },
+    {
+      "epoch": 1.6474013370713823,
+      "grad_norm": 0.1954319030046463,
+      "learning_rate": 6.425268929988081e-05,
+      "loss": 2.3628,
+      "step": 3819
+    },
+    {
+      "epoch": 1.6478326504205305,
+      "grad_norm": 0.23721271753311157,
+      "learning_rate": 6.421891466391381e-05,
+      "loss": 2.0577,
+      "step": 3820
+    },
+    {
+      "epoch": 1.6482639637696788,
+      "grad_norm": 0.1829412579536438,
+      "learning_rate": 6.418514226028847e-05,
+      "loss": 2.3678,
+      "step": 3821
+    },
+    {
+      "epoch": 1.6486952771188268,
+      "grad_norm": 0.2008577436208725,
+      "learning_rate": 6.415137209599767e-05,
+      "loss": 2.1813,
+      "step": 3822
+    },
+    {
+      "epoch": 1.6491265904679748,
+      "grad_norm": 0.1968059092760086,
+      "learning_rate": 6.411760417803392e-05,
+      "loss": 2.2353,
+      "step": 3823
+    },
+    {
+      "epoch": 1.649557903817123,
+      "grad_norm": 0.18984900414943695,
+      "learning_rate": 6.408383851338922e-05,
+      "loss": 2.1367,
+      "step": 3824
+    },
+    {
+      "epoch": 1.6499892171662713,
+      "grad_norm": 0.2056960016489029,
+      "learning_rate": 6.405007510905515e-05,
+      "loss": 2.2562,
+      "step": 3825
+    },
+    {
+      "epoch": 1.6499892171662713,
+      "eval_loss": 2.092493772506714,
+      "eval_runtime": 194.8674,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3825
+    },
+    {
+      "epoch": 1.6504205305154196,
+      "grad_norm": 0.1952376812696457,
+      "learning_rate": 6.401631397202277e-05,
+      "loss": 2.2311,
+      "step": 3826
+    },
+    {
+      "epoch": 1.6508518438645676,
+      "grad_norm": 0.17466172575950623,
+      "learning_rate": 6.398255510928268e-05,
+      "loss": 2.0367,
+      "step": 3827
+    },
+    {
+      "epoch": 1.6512831572137157,
+      "grad_norm": 0.3534952998161316,
+      "learning_rate": 6.394879852782507e-05,
+      "loss": 2.0924,
+      "step": 3828
+    },
+    {
+      "epoch": 1.651714470562864,
+      "grad_norm": 0.25892743468284607,
+      "learning_rate": 6.391504423463957e-05,
+      "loss": 2.1704,
+      "step": 3829
+    },
+    {
+      "epoch": 1.6521457839120122,
+      "grad_norm": 0.1772749274969101,
+      "learning_rate": 6.388129223671538e-05,
+      "loss": 2.1152,
+      "step": 3830
+    },
+    {
+      "epoch": 1.6525770972611602,
+      "grad_norm": 0.25078848004341125,
+      "learning_rate": 6.384754254104121e-05,
+      "loss": 2.1433,
+      "step": 3831
+    },
+    {
+      "epoch": 1.6530084106103082,
+      "grad_norm": 0.17856615781784058,
+      "learning_rate": 6.381379515460533e-05,
+      "loss": 2.166,
+      "step": 3832
+    },
+    {
+      "epoch": 1.6534397239594565,
+      "grad_norm": 0.17271089553833008,
+      "learning_rate": 6.378005008439552e-05,
+      "loss": 2.1329,
+      "step": 3833
+    },
+    {
+      "epoch": 1.6538710373086047,
+      "grad_norm": 0.18587853014469147,
+      "learning_rate": 6.374630733739898e-05,
+      "loss": 2.1766,
+      "step": 3834
+    },
+    {
+      "epoch": 1.654302350657753,
+      "grad_norm": 0.19487249851226807,
+      "learning_rate": 6.371256692060263e-05,
+      "loss": 2.2771,
+      "step": 3835
+    },
+    {
+      "epoch": 1.654733664006901,
+      "grad_norm": 0.20743614435195923,
+      "learning_rate": 6.367882884099272e-05,
+      "loss": 2.1246,
+      "step": 3836
+    },
+    {
+      "epoch": 1.655164977356049,
+      "grad_norm": 0.18915899097919464,
+      "learning_rate": 6.364509310555509e-05,
+      "loss": 2.35,
+      "step": 3837
+    },
+    {
+      "epoch": 1.6555962907051973,
+      "grad_norm": 0.176448255777359,
+      "learning_rate": 6.361135972127512e-05,
+      "loss": 2.1189,
+      "step": 3838
+    },
+    {
+      "epoch": 1.6560276040543456,
+      "grad_norm": 0.18612118065357208,
+      "learning_rate": 6.357762869513766e-05,
+      "loss": 2.1382,
+      "step": 3839
+    },
+    {
+      "epoch": 1.6564589174034936,
+      "grad_norm": 0.17284585535526276,
+      "learning_rate": 6.35439000341271e-05,
+      "loss": 2.1476,
+      "step": 3840
+    },
+    {
+      "epoch": 1.6568902307526416,
+      "grad_norm": 0.17270182073116302,
+      "learning_rate": 6.351017374522729e-05,
+      "loss": 1.9631,
+      "step": 3841
+    },
+    {
+      "epoch": 1.65732154410179,
+      "grad_norm": 0.19759193062782288,
+      "learning_rate": 6.34764498354217e-05,
+      "loss": 2.0957,
+      "step": 3842
+    },
+    {
+      "epoch": 1.6577528574509381,
+      "grad_norm": 0.2020193189382553,
+      "learning_rate": 6.344272831169317e-05,
+      "loss": 2.1451,
+      "step": 3843
+    },
+    {
+      "epoch": 1.6581841708000864,
+      "grad_norm": 0.18171140551567078,
+      "learning_rate": 6.340900918102413e-05,
+      "loss": 2.0582,
+      "step": 3844
+    },
+    {
+      "epoch": 1.6586154841492344,
+      "grad_norm": 0.18665428459644318,
+      "learning_rate": 6.33752924503965e-05,
+      "loss": 2.0344,
+      "step": 3845
+    },
+    {
+      "epoch": 1.6590467974983825,
+      "grad_norm": 0.18399561941623688,
+      "learning_rate": 6.334157812679168e-05,
+      "loss": 2.1475,
+      "step": 3846
+    },
+    {
+      "epoch": 1.6594781108475307,
+      "grad_norm": 0.1734449863433838,
+      "learning_rate": 6.330786621719064e-05,
+      "loss": 2.1052,
+      "step": 3847
+    },
+    {
+      "epoch": 1.659909424196679,
+      "grad_norm": 0.22349125146865845,
+      "learning_rate": 6.327415672857371e-05,
+      "loss": 2.2059,
+      "step": 3848
+    },
+    {
+      "epoch": 1.660340737545827,
+      "grad_norm": 0.17899243533611298,
+      "learning_rate": 6.32404496679209e-05,
+      "loss": 2.1341,
+      "step": 3849
+    },
+    {
+      "epoch": 1.660772050894975,
+      "grad_norm": 0.17724715173244476,
+      "learning_rate": 6.320674504221158e-05,
+      "loss": 2.0841,
+      "step": 3850
+    },
+    {
+      "epoch": 1.660772050894975,
+      "eval_loss": 2.09236741065979,
+      "eval_runtime": 194.9891,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3850
+    },
+    {
+      "epoch": 1.6612033642441233,
+      "grad_norm": 0.19121047854423523,
+      "learning_rate": 6.317304285842465e-05,
+      "loss": 2.2765,
+      "step": 3851
+    },
+    {
+      "epoch": 1.6616346775932715,
+      "grad_norm": 0.17873340845108032,
+      "learning_rate": 6.313934312353854e-05,
+      "loss": 2.0656,
+      "step": 3852
+    },
+    {
+      "epoch": 1.6620659909424198,
+      "grad_norm": 0.18030403554439545,
+      "learning_rate": 6.310564584453118e-05,
+      "loss": 2.1507,
+      "step": 3853
+    },
+    {
+      "epoch": 1.6624973042915678,
+      "grad_norm": 0.19685925543308258,
+      "learning_rate": 6.307195102837989e-05,
+      "loss": 2.2512,
+      "step": 3854
+    },
+    {
+      "epoch": 1.6629286176407159,
+      "grad_norm": 0.18864823877811432,
+      "learning_rate": 6.303825868206156e-05,
+      "loss": 2.1788,
+      "step": 3855
+    },
+    {
+      "epoch": 1.6633599309898641,
+      "grad_norm": 0.2153341919183731,
+      "learning_rate": 6.300456881255261e-05,
+      "loss": 2.3064,
+      "step": 3856
+    },
+    {
+      "epoch": 1.6637912443390124,
+      "grad_norm": 0.17508377134799957,
+      "learning_rate": 6.297088142682887e-05,
+      "loss": 1.8921,
+      "step": 3857
+    },
+    {
+      "epoch": 1.6642225576881604,
+      "grad_norm": 0.198838472366333,
+      "learning_rate": 6.293719653186569e-05,
+      "loss": 2.0898,
+      "step": 3858
+    },
+    {
+      "epoch": 1.6646538710373084,
+      "grad_norm": 0.17895318567752838,
+      "learning_rate": 6.290351413463782e-05,
+      "loss": 2.1175,
+      "step": 3859
+    },
+    {
+      "epoch": 1.6650851843864567,
+      "grad_norm": 0.1883961409330368,
+      "learning_rate": 6.28698342421197e-05,
+      "loss": 2.4141,
+      "step": 3860
+    },
+    {
+      "epoch": 1.665516497735605,
+      "grad_norm": 0.1782238483428955,
+      "learning_rate": 6.283615686128505e-05,
+      "loss": 2.0513,
+      "step": 3861
+    },
+    {
+      "epoch": 1.6659478110847532,
+      "grad_norm": 0.16665376722812653,
+      "learning_rate": 6.28024819991071e-05,
+      "loss": 2.0893,
+      "step": 3862
+    },
+    {
+      "epoch": 1.6663791244339012,
+      "grad_norm": 0.16384795308113098,
+      "learning_rate": 6.276880966255868e-05,
+      "loss": 1.988,
+      "step": 3863
+    },
+    {
+      "epoch": 1.6668104377830493,
+      "grad_norm": 0.17633698880672455,
+      "learning_rate": 6.273513985861196e-05,
+      "loss": 2.0995,
+      "step": 3864
+    },
+    {
+      "epoch": 1.6672417511321975,
+      "grad_norm": 0.4395500123500824,
+      "learning_rate": 6.270147259423868e-05,
+      "loss": 2.1884,
+      "step": 3865
+    },
+    {
+      "epoch": 1.6676730644813458,
+      "grad_norm": 0.20222854614257812,
+      "learning_rate": 6.266780787640995e-05,
+      "loss": 2.2195,
+      "step": 3866
+    },
+    {
+      "epoch": 1.6681043778304938,
+      "grad_norm": 0.16490493714809418,
+      "learning_rate": 6.26341457120965e-05,
+      "loss": 2.0757,
+      "step": 3867
+    },
+    {
+      "epoch": 1.6685356911796418,
+      "grad_norm": 0.17670033872127533,
+      "learning_rate": 6.26004861082684e-05,
+      "loss": 2.1008,
+      "step": 3868
+    },
+    {
+      "epoch": 1.66896700452879,
+      "grad_norm": 0.22973522543907166,
+      "learning_rate": 6.256682907189525e-05,
+      "loss": 1.952,
+      "step": 3869
+    },
+    {
+      "epoch": 1.6693983178779384,
+      "grad_norm": 0.18370388448238373,
+      "learning_rate": 6.253317460994611e-05,
+      "loss": 2.169,
+      "step": 3870
+    },
+    {
+      "epoch": 1.6698296312270866,
+      "grad_norm": 0.1987435668706894,
+      "learning_rate": 6.24995227293895e-05,
+      "loss": 2.1858,
+      "step": 3871
+    },
+    {
+      "epoch": 1.6702609445762346,
+      "grad_norm": 0.18324199318885803,
+      "learning_rate": 6.246587343719343e-05,
+      "loss": 2.0375,
+      "step": 3872
+    },
+    {
+      "epoch": 1.6706922579253827,
+      "grad_norm": 0.18455226719379425,
+      "learning_rate": 6.243222674032529e-05,
+      "loss": 2.232,
+      "step": 3873
+    },
+    {
+      "epoch": 1.671123571274531,
+      "grad_norm": 0.19305765628814697,
+      "learning_rate": 6.239858264575209e-05,
+      "loss": 2.2791,
+      "step": 3874
+    },
+    {
+      "epoch": 1.6715548846236792,
+      "grad_norm": 0.21840623021125793,
+      "learning_rate": 6.236494116044012e-05,
+      "loss": 1.8552,
+      "step": 3875
+    },
+    {
+      "epoch": 1.6715548846236792,
+      "eval_loss": 2.0930511951446533,
+      "eval_runtime": 195.0966,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3875
+    },
+    {
+      "epoch": 1.6719861979728272,
+      "grad_norm": 0.20301920175552368,
+      "learning_rate": 6.23313022913553e-05,
+      "loss": 2.2956,
+      "step": 3876
+    },
+    {
+      "epoch": 1.6724175113219755,
+      "grad_norm": 0.27590247988700867,
+      "learning_rate": 6.229766604546285e-05,
+      "loss": 2.2569,
+      "step": 3877
+    },
+    {
+      "epoch": 1.6728488246711235,
+      "grad_norm": 0.19490353763103485,
+      "learning_rate": 6.226403242972755e-05,
+      "loss": 2.2655,
+      "step": 3878
+    },
+    {
+      "epoch": 1.6732801380202718,
+      "grad_norm": 0.1732746809720993,
+      "learning_rate": 6.223040145111365e-05,
+      "loss": 2.1003,
+      "step": 3879
+    },
+    {
+      "epoch": 1.67371145136942,
+      "grad_norm": 0.18505042791366577,
+      "learning_rate": 6.219677311658473e-05,
+      "loss": 1.9627,
+      "step": 3880
+    },
+    {
+      "epoch": 1.674142764718568,
+      "grad_norm": 0.20846937596797943,
+      "learning_rate": 6.216314743310398e-05,
+      "loss": 2.2654,
+      "step": 3881
+    },
+    {
+      "epoch": 1.674574078067716,
+      "grad_norm": 0.17472702264785767,
+      "learning_rate": 6.212952440763394e-05,
+      "loss": 2.0093,
+      "step": 3882
+    },
+    {
+      "epoch": 1.6750053914168643,
+      "grad_norm": 0.17258580029010773,
+      "learning_rate": 6.209590404713663e-05,
+      "loss": 1.9649,
+      "step": 3883
+    },
+    {
+      "epoch": 1.6754367047660126,
+      "grad_norm": 0.1834079474210739,
+      "learning_rate": 6.206228635857345e-05,
+      "loss": 2.0188,
+      "step": 3884
+    },
+    {
+      "epoch": 1.6758680181151606,
+      "grad_norm": 0.1837594360113144,
+      "learning_rate": 6.20286713489054e-05,
+      "loss": 2.0756,
+      "step": 3885
+    },
+    {
+      "epoch": 1.6762993314643089,
+      "grad_norm": 0.19340825080871582,
+      "learning_rate": 6.199505902509283e-05,
+      "loss": 2.1349,
+      "step": 3886
+    },
+    {
+      "epoch": 1.676730644813457,
+      "grad_norm": 0.18126502633094788,
+      "learning_rate": 6.196144939409543e-05,
+      "loss": 2.2139,
+      "step": 3887
+    },
+    {
+      "epoch": 1.6771619581626052,
+      "grad_norm": 0.18060217797756195,
+      "learning_rate": 6.19278424628726e-05,
+      "loss": 2.0478,
+      "step": 3888
+    },
+    {
+      "epoch": 1.6775932715117534,
+      "grad_norm": 0.19164370000362396,
+      "learning_rate": 6.18942382383829e-05,
+      "loss": 2.3741,
+      "step": 3889
+    },
+    {
+      "epoch": 1.6780245848609014,
+      "grad_norm": 0.7620417475700378,
+      "learning_rate": 6.186063672758451e-05,
+      "loss": 2.0286,
+      "step": 3890
+    },
+    {
+      "epoch": 1.6784558982100495,
+      "grad_norm": 0.19031478464603424,
+      "learning_rate": 6.182703793743495e-05,
+      "loss": 2.2822,
+      "step": 3891
+    },
+    {
+      "epoch": 1.6788872115591977,
+      "grad_norm": 0.19598186016082764,
+      "learning_rate": 6.179344187489126e-05,
+      "loss": 2.1966,
+      "step": 3892
+    },
+    {
+      "epoch": 1.679318524908346,
+      "grad_norm": 0.1745317429304123,
+      "learning_rate": 6.175984854690988e-05,
+      "loss": 1.9842,
+      "step": 3893
+    },
+    {
+      "epoch": 1.679749838257494,
+      "grad_norm": 0.17918013036251068,
+      "learning_rate": 6.172625796044661e-05,
+      "loss": 2.2293,
+      "step": 3894
+    },
+    {
+      "epoch": 1.6801811516066423,
+      "grad_norm": 0.18294218182563782,
+      "learning_rate": 6.169267012245681e-05,
+      "loss": 2.1592,
+      "step": 3895
+    },
+    {
+      "epoch": 1.6806124649557903,
+      "grad_norm": 0.16389308869838715,
+      "learning_rate": 6.165908503989518e-05,
+      "loss": 2.0937,
+      "step": 3896
+    },
+    {
+      "epoch": 1.6810437783049386,
+      "grad_norm": 0.2014763355255127,
+      "learning_rate": 6.162550271971589e-05,
+      "loss": 2.0043,
+      "step": 3897
+    },
+    {
+      "epoch": 1.6814750916540868,
+      "grad_norm": 0.17000114917755127,
+      "learning_rate": 6.159192316887251e-05,
+      "loss": 2.0175,
+      "step": 3898
+    },
+    {
+      "epoch": 1.6819064050032349,
+      "grad_norm": 0.20296160876750946,
+      "learning_rate": 6.155834639431809e-05,
+      "loss": 2.135,
+      "step": 3899
+    },
+    {
+      "epoch": 1.6823377183523829,
+      "grad_norm": 0.16654804348945618,
+      "learning_rate": 6.152477240300504e-05,
+      "loss": 1.8905,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6823377183523829,
+      "eval_loss": 2.0927248001098633,
+      "eval_runtime": 194.9672,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3900
+    },
+    {
+      "epoch": 1.6827690317015311,
+      "grad_norm": 0.1887587010860443,
+      "learning_rate": 6.14912012018852e-05,
+      "loss": 2.2891,
+      "step": 3901
+    },
+    {
+      "epoch": 1.6832003450506794,
+      "grad_norm": 0.174739271402359,
+      "learning_rate": 6.14576327979099e-05,
+      "loss": 2.0998,
+      "step": 3902
+    },
+    {
+      "epoch": 1.6836316583998274,
+      "grad_norm": 0.20673100650310516,
+      "learning_rate": 6.142406719802983e-05,
+      "loss": 2.1349,
+      "step": 3903
+    },
+    {
+      "epoch": 1.6840629717489757,
+      "grad_norm": 0.1663362681865692,
+      "learning_rate": 6.139050440919512e-05,
+      "loss": 2.0095,
+      "step": 3904
+    },
+    {
+      "epoch": 1.6844942850981237,
+      "grad_norm": 0.18062900006771088,
+      "learning_rate": 6.135694443835525e-05,
+      "loss": 2.2951,
+      "step": 3905
+    },
+    {
+      "epoch": 1.684925598447272,
+      "grad_norm": 0.1740148812532425,
+      "learning_rate": 6.13233872924593e-05,
+      "loss": 2.2085,
+      "step": 3906
+    },
+    {
+      "epoch": 1.6853569117964202,
+      "grad_norm": 0.19164739549160004,
+      "learning_rate": 6.128983297845554e-05,
+      "loss": 2.1054,
+      "step": 3907
+    },
+    {
+      "epoch": 1.6857882251455683,
+      "grad_norm": 0.18385522067546844,
+      "learning_rate": 6.12562815032918e-05,
+      "loss": 2.0579,
+      "step": 3908
+    },
+    {
+      "epoch": 1.6862195384947163,
+      "grad_norm": 0.17914317548274994,
+      "learning_rate": 6.122273287391527e-05,
+      "loss": 2.0341,
+      "step": 3909
+    },
+    {
+      "epoch": 1.6866508518438645,
+      "grad_norm": 0.1921696662902832,
+      "learning_rate": 6.118918709727257e-05,
+      "loss": 2.2172,
+      "step": 3910
+    },
+    {
+      "epoch": 1.6870821651930128,
+      "grad_norm": 0.20252445340156555,
+      "learning_rate": 6.115564418030973e-05,
+      "loss": 2.2343,
+      "step": 3911
+    },
+    {
+      "epoch": 1.6875134785421608,
+      "grad_norm": 0.18932776153087616,
+      "learning_rate": 6.112210412997212e-05,
+      "loss": 2.2026,
+      "step": 3912
+    },
+    {
+      "epoch": 1.687944791891309,
+      "grad_norm": 0.2137858271598816,
+      "learning_rate": 6.108856695320465e-05,
+      "loss": 2.3142,
+      "step": 3913
+    },
+    {
+      "epoch": 1.6883761052404571,
+      "grad_norm": 0.18023236095905304,
+      "learning_rate": 6.105503265695152e-05,
+      "loss": 2.1168,
+      "step": 3914
+    },
+    {
+      "epoch": 1.6888074185896054,
+      "grad_norm": 0.18548575043678284,
+      "learning_rate": 6.1021501248156375e-05,
+      "loss": 2.1705,
+      "step": 3915
+    },
+    {
+      "epoch": 1.6892387319387536,
+      "grad_norm": 0.18816471099853516,
+      "learning_rate": 6.098797273376224e-05,
+      "loss": 2.1692,
+      "step": 3916
+    },
+    {
+      "epoch": 1.6896700452879017,
+      "grad_norm": 0.1845218688249588,
+      "learning_rate": 6.095444712071161e-05,
+      "loss": 2.1266,
+      "step": 3917
+    },
+    {
+      "epoch": 1.6901013586370497,
+      "grad_norm": 0.19871877133846283,
+      "learning_rate": 6.0920924415946324e-05,
+      "loss": 2.2414,
+      "step": 3918
+    },
+    {
+      "epoch": 1.690532671986198,
+      "grad_norm": 0.17879559099674225,
+      "learning_rate": 6.088740462640754e-05,
+      "loss": 1.9735,
+      "step": 3919
+    },
+    {
+      "epoch": 1.6909639853353462,
+      "grad_norm": 0.1785094290971756,
+      "learning_rate": 6.085388775903602e-05,
+      "loss": 2.3052,
+      "step": 3920
+    },
+    {
+      "epoch": 1.6913952986844942,
+      "grad_norm": 0.19720037281513214,
+      "learning_rate": 6.082037382077171e-05,
+      "loss": 2.0464,
+      "step": 3921
+    },
+    {
+      "epoch": 1.6918266120336425,
+      "grad_norm": 0.18563510477542877,
+      "learning_rate": 6.078686281855408e-05,
+      "loss": 2.1952,
+      "step": 3922
+    },
+    {
+      "epoch": 1.6922579253827905,
+      "grad_norm": 0.19016218185424805,
+      "learning_rate": 6.0753354759321914e-05,
+      "loss": 2.162,
+      "step": 3923
+    },
+    {
+      "epoch": 1.6926892387319388,
+      "grad_norm": 0.1768539696931839,
+      "learning_rate": 6.071984965001346e-05,
+      "loss": 2.277,
+      "step": 3924
+    },
+    {
+      "epoch": 1.693120552081087,
+      "grad_norm": 0.18435247242450714,
+      "learning_rate": 6.068634749756632e-05,
+      "loss": 2.1464,
+      "step": 3925
+    },
+    {
+      "epoch": 1.693120552081087,
+      "eval_loss": 2.092581033706665,
+      "eval_runtime": 194.7985,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3925
+    },
+    {
+      "epoch": 1.693551865430235,
+      "grad_norm": 0.17748647928237915,
+      "learning_rate": 6.065284830891741e-05,
+      "loss": 2.134,
+      "step": 3926
+    },
+    {
+      "epoch": 1.693983178779383,
+      "grad_norm": 0.16803781688213348,
+      "learning_rate": 6.061935209100321e-05,
+      "loss": 1.9717,
+      "step": 3927
+    },
+    {
+      "epoch": 1.6944144921285313,
+      "grad_norm": 0.2016642689704895,
+      "learning_rate": 6.0585858850759395e-05,
+      "loss": 2.3664,
+      "step": 3928
+    },
+    {
+      "epoch": 1.6948458054776796,
+      "grad_norm": 0.17812767624855042,
+      "learning_rate": 6.055236859512114e-05,
+      "loss": 2.073,
+      "step": 3929
+    },
+    {
+      "epoch": 1.6952771188268279,
+      "grad_norm": 0.18797719478607178,
+      "learning_rate": 6.051888133102296e-05,
+      "loss": 2.1412,
+      "step": 3930
+    },
+    {
+      "epoch": 1.6957084321759759,
+      "grad_norm": 0.19550077617168427,
+      "learning_rate": 6.0485397065398764e-05,
+      "loss": 2.1034,
+      "step": 3931
+    },
+    {
+      "epoch": 1.696139745525124,
+      "grad_norm": 0.1779249608516693,
+      "learning_rate": 6.045191580518186e-05,
+      "loss": 1.975,
+      "step": 3932
+    },
+    {
+      "epoch": 1.6965710588742722,
+      "grad_norm": 0.1735992133617401,
+      "learning_rate": 6.0418437557304845e-05,
+      "loss": 2.1543,
+      "step": 3933
+    },
+    {
+      "epoch": 1.6970023722234204,
+      "grad_norm": 0.18777874112129211,
+      "learning_rate": 6.038496232869982e-05,
+      "loss": 2.3913,
+      "step": 3934
+    },
+    {
+      "epoch": 1.6974336855725685,
+      "grad_norm": 0.19871315360069275,
+      "learning_rate": 6.035149012629816e-05,
+      "loss": 2.2304,
+      "step": 3935
+    },
+    {
+      "epoch": 1.6978649989217165,
+      "grad_norm": 0.1709122359752655,
+      "learning_rate": 6.0318020957030674e-05,
+      "loss": 2.0261,
+      "step": 3936
+    },
+    {
+      "epoch": 1.6982963122708647,
+      "grad_norm": 0.18706877529621124,
+      "learning_rate": 6.0284554827827495e-05,
+      "loss": 2.1231,
+      "step": 3937
+    },
+    {
+      "epoch": 1.698727625620013,
+      "grad_norm": 0.18627259135246277,
+      "learning_rate": 6.025109174561818e-05,
+      "loss": 2.2471,
+      "step": 3938
+    },
+    {
+      "epoch": 1.6991589389691613,
+      "grad_norm": 0.19945907592773438,
+      "learning_rate": 6.0217631717331606e-05,
+      "loss": 2.0648,
+      "step": 3939
+    },
+    {
+      "epoch": 1.6995902523183093,
+      "grad_norm": 0.19371074438095093,
+      "learning_rate": 6.018417474989602e-05,
+      "loss": 2.0709,
+      "step": 3940
+    },
+    {
+      "epoch": 1.7000215656674573,
+      "grad_norm": 0.19727398455142975,
+      "learning_rate": 6.01507208502391e-05,
+      "loss": 2.2179,
+      "step": 3941
+    },
+    {
+      "epoch": 1.7004528790166056,
+      "grad_norm": 0.1947556436061859,
+      "learning_rate": 6.011727002528781e-05,
+      "loss": 2.1538,
+      "step": 3942
+    },
+    {
+      "epoch": 1.7008841923657538,
+      "grad_norm": 0.17219044268131256,
+      "learning_rate": 6.0083822281968536e-05,
+      "loss": 2.1982,
+      "step": 3943
+    },
+    {
+      "epoch": 1.7013155057149019,
+      "grad_norm": 0.18319495022296906,
+      "learning_rate": 6.005037762720692e-05,
+      "loss": 2.2547,
+      "step": 3944
+    },
+    {
+      "epoch": 1.70174681906405,
+      "grad_norm": 0.1841065138578415,
+      "learning_rate": 6.001693606792816e-05,
+      "loss": 2.2998,
+      "step": 3945
+    },
+    {
+      "epoch": 1.7021781324131982,
+      "grad_norm": 0.180063396692276,
+      "learning_rate": 5.9983497611056633e-05,
+      "loss": 2.1797,
+      "step": 3946
+    },
+    {
+      "epoch": 1.7026094457623464,
+      "grad_norm": 0.1915334016084671,
+      "learning_rate": 5.9950062263516114e-05,
+      "loss": 2.1179,
+      "step": 3947
+    },
+    {
+      "epoch": 1.7030407591114947,
+      "grad_norm": 0.19555287063121796,
+      "learning_rate": 5.9916630032229804e-05,
+      "loss": 2.2253,
+      "step": 3948
+    },
+    {
+      "epoch": 1.7034720724606427,
+      "grad_norm": 0.1670524626970291,
+      "learning_rate": 5.98832009241202e-05,
+      "loss": 1.9075,
+      "step": 3949
+    },
+    {
+      "epoch": 1.7039033858097907,
+      "grad_norm": 0.18049435317516327,
+      "learning_rate": 5.9849774946109164e-05,
+      "loss": 2.2191,
+      "step": 3950
+    },
+    {
+      "epoch": 1.7039033858097907,
+      "eval_loss": 2.092341184616089,
+      "eval_runtime": 195.676,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3950
+    },
+    {
+      "epoch": 1.704334699158939,
+      "grad_norm": 0.18992845714092255,
+      "learning_rate": 5.9816352105117874e-05,
+      "loss": 2.1745,
+      "step": 3951
+    },
+    {
+      "epoch": 1.7047660125080872,
+      "grad_norm": 0.1593848168849945,
+      "learning_rate": 5.978293240806696e-05,
+      "loss": 1.9007,
+      "step": 3952
+    },
+    {
+      "epoch": 1.7051973258572353,
+      "grad_norm": 0.1983330100774765,
+      "learning_rate": 5.974951586187629e-05,
+      "loss": 2.1191,
+      "step": 3953
+    },
+    {
+      "epoch": 1.7056286392063833,
+      "grad_norm": 0.18438118696212769,
+      "learning_rate": 5.971610247346513e-05,
+      "loss": 2.2795,
+      "step": 3954
+    },
+    {
+      "epoch": 1.7060599525555316,
+      "grad_norm": 0.18407520651817322,
+      "learning_rate": 5.968269224975209e-05,
+      "loss": 2.0768,
+      "step": 3955
+    },
+    {
+      "epoch": 1.7064912659046798,
+      "grad_norm": 0.18830512464046478,
+      "learning_rate": 5.964928519765515e-05,
+      "loss": 2.1758,
+      "step": 3956
+    },
+    {
+      "epoch": 1.706922579253828,
+      "grad_norm": 0.16916070878505707,
+      "learning_rate": 5.961588132409158e-05,
+      "loss": 2.2333,
+      "step": 3957
+    },
+    {
+      "epoch": 1.707353892602976,
+      "grad_norm": 0.19552044570446014,
+      "learning_rate": 5.9582480635977984e-05,
+      "loss": 2.2937,
+      "step": 3958
+    },
+    {
+      "epoch": 1.7077852059521241,
+      "grad_norm": 0.18797554075717926,
+      "learning_rate": 5.954908314023043e-05,
+      "loss": 2.0564,
+      "step": 3959
+    },
+    {
+      "epoch": 1.7082165193012724,
+      "grad_norm": 0.20710021257400513,
+      "learning_rate": 5.951568884376415e-05,
+      "loss": 2.0914,
+      "step": 3960
+    },
+    {
+      "epoch": 1.7086478326504206,
+      "grad_norm": 0.191316157579422,
+      "learning_rate": 5.948229775349383e-05,
+      "loss": 2.3974,
+      "step": 3961
+    },
+    {
+      "epoch": 1.7090791459995687,
+      "grad_norm": 0.1908051073551178,
+      "learning_rate": 5.944890987633344e-05,
+      "loss": 2.2372,
+      "step": 3962
+    },
+    {
+      "epoch": 1.7095104593487167,
+      "grad_norm": 0.1851019263267517,
+      "learning_rate": 5.941552521919635e-05,
+      "loss": 2.31,
+      "step": 3963
+    },
+    {
+      "epoch": 1.709941772697865,
+      "grad_norm": 1.414542317390442,
+      "learning_rate": 5.93821437889952e-05,
+      "loss": 2.346,
+      "step": 3964
+    },
+    {
+      "epoch": 1.7103730860470132,
+      "grad_norm": 0.1777050644159317,
+      "learning_rate": 5.9348765592641924e-05,
+      "loss": 2.3289,
+      "step": 3965
+    },
+    {
+      "epoch": 1.7108043993961615,
+      "grad_norm": 0.20037689805030823,
+      "learning_rate": 5.931539063704795e-05,
+      "loss": 1.9037,
+      "step": 3966
+    },
+    {
+      "epoch": 1.7112357127453095,
+      "grad_norm": 0.18143126368522644,
+      "learning_rate": 5.9282018929123845e-05,
+      "loss": 2.2531,
+      "step": 3967
+    },
+    {
+      "epoch": 1.7116670260944575,
+      "grad_norm": 0.19299474358558655,
+      "learning_rate": 5.924865047577961e-05,
+      "loss": 2.2543,
+      "step": 3968
+    },
+    {
+      "epoch": 1.7120983394436058,
+      "grad_norm": 0.20400649309158325,
+      "learning_rate": 5.921528528392453e-05,
+      "loss": 2.0489,
+      "step": 3969
+    },
+    {
+      "epoch": 1.712529652792754,
+      "grad_norm": 0.18670541048049927,
+      "learning_rate": 5.9181923360467264e-05,
+      "loss": 2.0749,
+      "step": 3970
+    },
+    {
+      "epoch": 1.712960966141902,
+      "grad_norm": 0.1974756121635437,
+      "learning_rate": 5.914856471231577e-05,
+      "loss": 2.1085,
+      "step": 3971
+    },
+    {
+      "epoch": 1.71339227949105,
+      "grad_norm": 0.1814204305410385,
+      "learning_rate": 5.911520934637727e-05,
+      "loss": 1.8131,
+      "step": 3972
+    },
+    {
+      "epoch": 1.7138235928401984,
+      "grad_norm": 0.1980465054512024,
+      "learning_rate": 5.9081857269558406e-05,
+      "loss": 2.2717,
+      "step": 3973
+    },
+    {
+      "epoch": 1.7142549061893466,
+      "grad_norm": 0.1712466925382614,
+      "learning_rate": 5.9048508488765066e-05,
+      "loss": 2.0018,
+      "step": 3974
+    },
+    {
+      "epoch": 1.7146862195384949,
+      "grad_norm": 0.17620418965816498,
+      "learning_rate": 5.901516301090249e-05,
+      "loss": 2.2338,
+      "step": 3975
+    },
+    {
+      "epoch": 1.7146862195384949,
+      "eval_loss": 2.092210054397583,
+      "eval_runtime": 195.6295,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 3975
+    },
+    {
+      "epoch": 1.715117532887643,
+      "grad_norm": 0.18851706385612488,
+      "learning_rate": 5.898182084287521e-05,
+      "loss": 2.1636,
+      "step": 3976
+    },
+    {
+      "epoch": 1.715548846236791,
+      "grad_norm": 0.16978512704372406,
+      "learning_rate": 5.894848199158712e-05,
+      "loss": 2.191,
+      "step": 3977
+    },
+    {
+      "epoch": 1.7159801595859392,
+      "grad_norm": 0.1951020061969757,
+      "learning_rate": 5.891514646394135e-05,
+      "loss": 2.2491,
+      "step": 3978
+    },
+    {
+      "epoch": 1.7164114729350874,
+      "grad_norm": 0.2247585654258728,
+      "learning_rate": 5.88818142668404e-05,
+      "loss": 2.1196,
+      "step": 3979
+    },
+    {
+      "epoch": 1.7168427862842355,
+      "grad_norm": 0.2103835642337799,
+      "learning_rate": 5.884848540718608e-05,
+      "loss": 2.0868,
+      "step": 3980
+    },
+    {
+      "epoch": 1.7172740996333835,
+      "grad_norm": 0.1812957227230072,
+      "learning_rate": 5.881515989187949e-05,
+      "loss": 2.2188,
+      "step": 3981
+    },
+    {
+      "epoch": 1.7177054129825318,
+      "grad_norm": 0.17861905694007874,
+      "learning_rate": 5.878183772782105e-05,
+      "loss": 2.0906,
+      "step": 3982
+    },
+    {
+      "epoch": 1.71813672633168,
+      "grad_norm": 0.16796641051769257,
+      "learning_rate": 5.8748518921910434e-05,
+      "loss": 1.9122,
+      "step": 3983
+    },
+    {
+      "epoch": 1.7185680396808283,
+      "grad_norm": 0.18065188825130463,
+      "learning_rate": 5.8715203481046726e-05,
+      "loss": 2.0403,
+      "step": 3984
+    },
+    {
+      "epoch": 1.7189993530299763,
+      "grad_norm": 0.18287047743797302,
+      "learning_rate": 5.8681891412128224e-05,
+      "loss": 2.2585,
+      "step": 3985
+    },
+    {
+      "epoch": 1.7194306663791243,
+      "grad_norm": 0.18636803328990936,
+      "learning_rate": 5.864858272205253e-05,
+      "loss": 2.1291,
+      "step": 3986
+    },
+    {
+      "epoch": 1.7198619797282726,
+      "grad_norm": 0.1851028949022293,
+      "learning_rate": 5.861527741771663e-05,
+      "loss": 2.2074,
+      "step": 3987
+    },
+    {
+      "epoch": 1.7202932930774208,
+      "grad_norm": 0.18895862996578217,
+      "learning_rate": 5.858197550601671e-05,
+      "loss": 2.2836,
+      "step": 3988
+    },
+    {
+      "epoch": 1.7207246064265689,
+      "grad_norm": 0.1789833903312683,
+      "learning_rate": 5.854867699384832e-05,
+      "loss": 2.0446,
+      "step": 3989
+    },
+    {
+      "epoch": 1.721155919775717,
+      "grad_norm": 0.17047496140003204,
+      "learning_rate": 5.851538188810622e-05,
+      "loss": 2.2997,
+      "step": 3990
+    },
+    {
+      "epoch": 1.7215872331248652,
+      "grad_norm": 0.19940529763698578,
+      "learning_rate": 5.848209019568463e-05,
+      "loss": 2.1693,
+      "step": 3991
+    },
+    {
+      "epoch": 1.7220185464740134,
+      "grad_norm": 0.16889867186546326,
+      "learning_rate": 5.8448801923476876e-05,
+      "loss": 2.3546,
+      "step": 3992
+    },
+    {
+      "epoch": 1.7224498598231617,
+      "grad_norm": 0.16920462250709534,
+      "learning_rate": 5.841551707837567e-05,
+      "loss": 2.225,
+      "step": 3993
+    },
+    {
+      "epoch": 1.7228811731723097,
+      "grad_norm": 0.18416011333465576,
+      "learning_rate": 5.8382235667273034e-05,
+      "loss": 2.1297,
+      "step": 3994
+    },
+    {
+      "epoch": 1.7233124865214577,
+      "grad_norm": 0.18310239911079407,
+      "learning_rate": 5.834895769706024e-05,
+      "loss": 2.3231,
+      "step": 3995
+    },
+    {
+      "epoch": 1.723743799870606,
+      "grad_norm": 0.1977590173482895,
+      "learning_rate": 5.831568317462785e-05,
+      "loss": 2.2033,
+      "step": 3996
+    },
+    {
+      "epoch": 1.7241751132197543,
+      "grad_norm": 0.1837632954120636,
+      "learning_rate": 5.828241210686568e-05,
+      "loss": 2.5726,
+      "step": 3997
+    },
+    {
+      "epoch": 1.7246064265689023,
+      "grad_norm": 0.1981734335422516,
+      "learning_rate": 5.8249144500662945e-05,
+      "loss": 2.1677,
+      "step": 3998
+    },
+    {
+      "epoch": 1.7250377399180503,
+      "grad_norm": 0.1921195387840271,
+      "learning_rate": 5.821588036290801e-05,
+      "loss": 2.1464,
+      "step": 3999
+    },
+    {
+      "epoch": 1.7254690532671986,
+      "grad_norm": 0.20350435376167297,
+      "learning_rate": 5.8182619700488576e-05,
+      "loss": 2.0763,
+      "step": 4000
+    },
+    {
+      "epoch": 1.7254690532671986,
+      "eval_loss": 2.091747283935547,
+      "eval_runtime": 207.3983,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 4000
+    },
+    {
+      "epoch": 1.7259003666163468,
+      "grad_norm": 0.18566665053367615,
+      "learning_rate": 5.814936252029167e-05,
+      "loss": 2.1219,
+      "step": 4001
+    },
+    {
+      "epoch": 1.726331679965495,
+      "grad_norm": 0.1934313178062439,
+      "learning_rate": 5.811610882920352e-05,
+      "loss": 2.3388,
+      "step": 4002
+    },
+    {
+      "epoch": 1.7267629933146431,
+      "grad_norm": 0.18000420928001404,
+      "learning_rate": 5.808285863410969e-05,
+      "loss": 2.2346,
+      "step": 4003
+    },
+    {
+      "epoch": 1.7271943066637911,
+      "grad_norm": 0.20419535040855408,
+      "learning_rate": 5.804961194189495e-05,
+      "loss": 2.2095,
+      "step": 4004
+    },
+    {
+      "epoch": 1.7276256200129394,
+      "grad_norm": 0.1848215013742447,
+      "learning_rate": 5.8016368759443454e-05,
+      "loss": 2.267,
+      "step": 4005
+    },
+    {
+      "epoch": 1.7280569333620877,
+      "grad_norm": 0.191674143075943,
+      "learning_rate": 5.7983129093638527e-05,
+      "loss": 1.9493,
+      "step": 4006
+    },
+    {
+      "epoch": 1.7284882467112357,
+      "grad_norm": 0.19241303205490112,
+      "learning_rate": 5.794989295136282e-05,
+      "loss": 2.0437,
+      "step": 4007
+    },
+    {
+      "epoch": 1.7289195600603837,
+      "grad_norm": 0.19676679372787476,
+      "learning_rate": 5.791666033949821e-05,
+      "loss": 2.0497,
+      "step": 4008
+    },
+    {
+      "epoch": 1.729350873409532,
+      "grad_norm": 0.17912554740905762,
+      "learning_rate": 5.788343126492591e-05,
+      "loss": 2.3007,
+      "step": 4009
+    },
+    {
+      "epoch": 1.7297821867586802,
+      "grad_norm": 0.19069208204746246,
+      "learning_rate": 5.7850205734526376e-05,
+      "loss": 2.2575,
+      "step": 4010
+    },
+    {
+      "epoch": 1.7302135001078285,
+      "grad_norm": 0.2015782594680786,
+      "learning_rate": 5.7816983755179237e-05,
+      "loss": 2.197,
+      "step": 4011
+    },
+    {
+      "epoch": 1.7306448134569765,
+      "grad_norm": 0.20349441468715668,
+      "learning_rate": 5.778376533376357e-05,
+      "loss": 2.1082,
+      "step": 4012
+    },
+    {
+      "epoch": 1.7310761268061245,
+      "grad_norm": 0.19610483944416046,
+      "learning_rate": 5.775055047715755e-05,
+      "loss": 2.3583,
+      "step": 4013
+    },
+    {
+      "epoch": 1.7315074401552728,
+      "grad_norm": 0.19321462512016296,
+      "learning_rate": 5.771733919223869e-05,
+      "loss": 2.2541,
+      "step": 4014
+    },
+    {
+      "epoch": 1.731938753504421,
+      "grad_norm": 0.18723595142364502,
+      "learning_rate": 5.7684131485883744e-05,
+      "loss": 2.2065,
+      "step": 4015
+    },
+    {
+      "epoch": 1.732370066853569,
+      "grad_norm": 0.18089692294597626,
+      "learning_rate": 5.765092736496874e-05,
+      "loss": 1.8975,
+      "step": 4016
+    },
+    {
+      "epoch": 1.7328013802027171,
+      "grad_norm": 0.18110182881355286,
+      "learning_rate": 5.761772683636897e-05,
+      "loss": 2.2158,
+      "step": 4017
+    },
+    {
+      "epoch": 1.7332326935518654,
+      "grad_norm": 0.19009125232696533,
+      "learning_rate": 5.758452990695892e-05,
+      "loss": 2.1776,
+      "step": 4018
+    },
+    {
+      "epoch": 1.7336640069010136,
+      "grad_norm": 0.2081516981124878,
+      "learning_rate": 5.755133658361242e-05,
+      "loss": 2.197,
+      "step": 4019
+    },
+    {
+      "epoch": 1.7340953202501619,
+      "grad_norm": 0.18723157048225403,
+      "learning_rate": 5.75181468732025e-05,
+      "loss": 2.1137,
+      "step": 4020
+    },
+    {
+      "epoch": 1.73452663359931,
+      "grad_norm": 0.18066422641277313,
+      "learning_rate": 5.748496078260144e-05,
+      "loss": 2.1427,
+      "step": 4021
+    },
+    {
+      "epoch": 1.734957946948458,
+      "grad_norm": 0.19186154007911682,
+      "learning_rate": 5.7451778318680775e-05,
+      "loss": 2.1782,
+      "step": 4022
+    },
+    {
+      "epoch": 1.7353892602976062,
+      "grad_norm": 0.1925586760044098,
+      "learning_rate": 5.741859948831135e-05,
+      "loss": 2.3438,
+      "step": 4023
+    },
+    {
+      "epoch": 1.7358205736467545,
+      "grad_norm": 0.17659415304660797,
+      "learning_rate": 5.738542429836315e-05,
+      "loss": 2.0247,
+      "step": 4024
+    },
+    {
+      "epoch": 1.7362518869959025,
+      "grad_norm": 0.18985515832901,
+      "learning_rate": 5.735225275570547e-05,
+      "loss": 2.2613,
+      "step": 4025
+    },
+    {
+      "epoch": 1.7362518869959025,
+      "eval_loss": 2.0919628143310547,
+      "eval_runtime": 201.674,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4025
+    },
+    {
+      "epoch": 1.7366832003450505,
+      "grad_norm": 0.19430814683437347,
+      "learning_rate": 5.731908486720685e-05,
+      "loss": 2.1622,
+      "step": 4026
+    },
+    {
+      "epoch": 1.7371145136941988,
+      "grad_norm": 0.18821769952774048,
+      "learning_rate": 5.728592063973508e-05,
+      "loss": 1.995,
+      "step": 4027
+    },
+    {
+      "epoch": 1.737545827043347,
+      "grad_norm": 0.2075040340423584,
+      "learning_rate": 5.725276008015716e-05,
+      "loss": 2.2193,
+      "step": 4028
+    },
+    {
+      "epoch": 1.7379771403924953,
+      "grad_norm": 0.18539956212043762,
+      "learning_rate": 5.721960319533931e-05,
+      "loss": 2.2047,
+      "step": 4029
+    },
+    {
+      "epoch": 1.7384084537416433,
+      "grad_norm": 0.22715280950069427,
+      "learning_rate": 5.71864499921471e-05,
+      "loss": 2.1533,
+      "step": 4030
+    },
+    {
+      "epoch": 1.7388397670907914,
+      "grad_norm": 0.21007855236530304,
+      "learning_rate": 5.715330047744522e-05,
+      "loss": 2.1163,
+      "step": 4031
+    },
+    {
+      "epoch": 1.7392710804399396,
+      "grad_norm": 0.1632479578256607,
+      "learning_rate": 5.712015465809762e-05,
+      "loss": 2.1047,
+      "step": 4032
+    },
+    {
+      "epoch": 1.7397023937890879,
+      "grad_norm": 0.19230559468269348,
+      "learning_rate": 5.708701254096755e-05,
+      "loss": 2.1609,
+      "step": 4033
+    },
+    {
+      "epoch": 1.740133707138236,
+      "grad_norm": 0.19357123970985413,
+      "learning_rate": 5.705387413291742e-05,
+      "loss": 2.1058,
+      "step": 4034
+    },
+    {
+      "epoch": 1.740565020487384,
+      "grad_norm": 0.17949312925338745,
+      "learning_rate": 5.702073944080893e-05,
+      "loss": 2.0622,
+      "step": 4035
+    },
+    {
+      "epoch": 1.7409963338365322,
+      "grad_norm": 0.202627494931221,
+      "learning_rate": 5.6987608471502894e-05,
+      "loss": 2.1767,
+      "step": 4036
+    },
+    {
+      "epoch": 1.7414276471856804,
+      "grad_norm": 0.19418296217918396,
+      "learning_rate": 5.695448123185956e-05,
+      "loss": 2.2173,
+      "step": 4037
+    },
+    {
+      "epoch": 1.7418589605348287,
+      "grad_norm": 0.172317773103714,
+      "learning_rate": 5.6921357728738226e-05,
+      "loss": 2.1346,
+      "step": 4038
+    },
+    {
+      "epoch": 1.7422902738839767,
+      "grad_norm": 0.1859188973903656,
+      "learning_rate": 5.6888237968997455e-05,
+      "loss": 2.1686,
+      "step": 4039
+    },
+    {
+      "epoch": 1.7427215872331248,
+      "grad_norm": 0.18734541535377502,
+      "learning_rate": 5.685512195949508e-05,
+      "loss": 2.3108,
+      "step": 4040
+    },
+    {
+      "epoch": 1.743152900582273,
+      "grad_norm": 0.18585248291492462,
+      "learning_rate": 5.682200970708815e-05,
+      "loss": 2.2407,
+      "step": 4041
+    },
+    {
+      "epoch": 1.7435842139314213,
+      "grad_norm": 0.21692951023578644,
+      "learning_rate": 5.678890121863291e-05,
+      "loss": 2.4208,
+      "step": 4042
+    },
+    {
+      "epoch": 1.7440155272805693,
+      "grad_norm": 0.18898186087608337,
+      "learning_rate": 5.6755796500984784e-05,
+      "loss": 2.0094,
+      "step": 4043
+    },
+    {
+      "epoch": 1.7444468406297173,
+      "grad_norm": 0.21968823671340942,
+      "learning_rate": 5.672269556099855e-05,
+      "loss": 2.0242,
+      "step": 4044
+    },
+    {
+      "epoch": 1.7448781539788656,
+      "grad_norm": 0.17570222914218903,
+      "learning_rate": 5.668959840552807e-05,
+      "loss": 2.1131,
+      "step": 4045
+    },
+    {
+      "epoch": 1.7453094673280138,
+      "grad_norm": 0.17798030376434326,
+      "learning_rate": 5.665650504142645e-05,
+      "loss": 2.0615,
+      "step": 4046
+    },
+    {
+      "epoch": 1.745740780677162,
+      "grad_norm": 0.18202351033687592,
+      "learning_rate": 5.6623415475546094e-05,
+      "loss": 2.1592,
+      "step": 4047
+    },
+    {
+      "epoch": 1.7461720940263101,
+      "grad_norm": 0.2013104110956192,
+      "learning_rate": 5.659032971473851e-05,
+      "loss": 2.194,
+      "step": 4048
+    },
+    {
+      "epoch": 1.7466034073754582,
+      "grad_norm": 0.1691199541091919,
+      "learning_rate": 5.6557247765854505e-05,
+      "loss": 2.3038,
+      "step": 4049
+    },
+    {
+      "epoch": 1.7470347207246064,
+      "grad_norm": 0.18335652351379395,
+      "learning_rate": 5.652416963574399e-05,
+      "loss": 2.1304,
+      "step": 4050
+    },
+    {
+      "epoch": 1.7470347207246064,
+      "eval_loss": 2.091346502304077,
+      "eval_runtime": 201.6741,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4050
+    },
+    {
+      "epoch": 1.7474660340737547,
+      "grad_norm": 0.17796850204467773,
+      "learning_rate": 5.6491095331256254e-05,
+      "loss": 2.3191,
+      "step": 4051
+    },
+    {
+      "epoch": 1.7478973474229027,
+      "grad_norm": 0.1794384866952896,
+      "learning_rate": 5.645802485923962e-05,
+      "loss": 1.8804,
+      "step": 4052
+    },
+    {
+      "epoch": 1.7483286607720507,
+      "grad_norm": 0.17685826122760773,
+      "learning_rate": 5.6424958226541694e-05,
+      "loss": 2.0451,
+      "step": 4053
+    },
+    {
+      "epoch": 1.748759974121199,
+      "grad_norm": 0.1787034571170807,
+      "learning_rate": 5.639189544000931e-05,
+      "loss": 2.0513,
+      "step": 4054
+    },
+    {
+      "epoch": 1.7491912874703472,
+      "grad_norm": 0.1818806678056717,
+      "learning_rate": 5.635883650648848e-05,
+      "loss": 2.4097,
+      "step": 4055
+    },
+    {
+      "epoch": 1.7496226008194955,
+      "grad_norm": 0.18707096576690674,
+      "learning_rate": 5.632578143282442e-05,
+      "loss": 2.195,
+      "step": 4056
+    },
+    {
+      "epoch": 1.7500539141686435,
+      "grad_norm": 0.2021271139383316,
+      "learning_rate": 5.62927302258615e-05,
+      "loss": 2.1611,
+      "step": 4057
+    },
+    {
+      "epoch": 1.7504852275177916,
+      "grad_norm": 0.18080992996692657,
+      "learning_rate": 5.625968289244338e-05,
+      "loss": 2.2268,
+      "step": 4058
+    },
+    {
+      "epoch": 1.7509165408669398,
+      "grad_norm": 0.18082956969738007,
+      "learning_rate": 5.622663943941286e-05,
+      "loss": 2.3477,
+      "step": 4059
+    },
+    {
+      "epoch": 1.751347854216088,
+      "grad_norm": 0.18515385687351227,
+      "learning_rate": 5.619359987361194e-05,
+      "loss": 2.2363,
+      "step": 4060
+    },
+    {
+      "epoch": 1.751779167565236,
+      "grad_norm": 0.17915287613868713,
+      "learning_rate": 5.616056420188182e-05,
+      "loss": 2.1316,
+      "step": 4061
+    },
+    {
+      "epoch": 1.7522104809143841,
+      "grad_norm": 0.18371932208538055,
+      "learning_rate": 5.6127532431062936e-05,
+      "loss": 2.0856,
+      "step": 4062
+    },
+    {
+      "epoch": 1.7526417942635324,
+      "grad_norm": 0.18611958622932434,
+      "learning_rate": 5.609450456799483e-05,
+      "loss": 2.0434,
+      "step": 4063
+    },
+    {
+      "epoch": 1.7530731076126806,
+      "grad_norm": 0.17234119772911072,
+      "learning_rate": 5.606148061951629e-05,
+      "loss": 1.9636,
+      "step": 4064
+    },
+    {
+      "epoch": 1.753504420961829,
+      "grad_norm": 0.1878260374069214,
+      "learning_rate": 5.6028460592465304e-05,
+      "loss": 2.1463,
+      "step": 4065
+    },
+    {
+      "epoch": 1.753935734310977,
+      "grad_norm": 0.2029743492603302,
+      "learning_rate": 5.5995444493679035e-05,
+      "loss": 2.32,
+      "step": 4066
+    },
+    {
+      "epoch": 1.754367047660125,
+      "grad_norm": 0.23618969321250916,
+      "learning_rate": 5.5962432329993817e-05,
+      "loss": 2.1887,
+      "step": 4067
+    },
+    {
+      "epoch": 1.7547983610092732,
+      "grad_norm": 0.2027716189622879,
+      "learning_rate": 5.592942410824513e-05,
+      "loss": 2.2154,
+      "step": 4068
+    },
+    {
+      "epoch": 1.7552296743584215,
+      "grad_norm": 0.23110395669937134,
+      "learning_rate": 5.589641983526779e-05,
+      "loss": 1.9544,
+      "step": 4069
+    },
+    {
+      "epoch": 1.7556609877075695,
+      "grad_norm": 0.17785879969596863,
+      "learning_rate": 5.586341951789561e-05,
+      "loss": 2.127,
+      "step": 4070
+    },
+    {
+      "epoch": 1.7560923010567175,
+      "grad_norm": 0.18353521823883057,
+      "learning_rate": 5.583042316296169e-05,
+      "loss": 2.1133,
+      "step": 4071
+    },
+    {
+      "epoch": 1.7565236144058658,
+      "grad_norm": 0.20097488164901733,
+      "learning_rate": 5.5797430777298294e-05,
+      "loss": 2.2941,
+      "step": 4072
+    },
+    {
+      "epoch": 1.756954927755014,
+      "grad_norm": 0.1747606247663498,
+      "learning_rate": 5.576444236773685e-05,
+      "loss": 2.0431,
+      "step": 4073
+    },
+    {
+      "epoch": 1.7573862411041623,
+      "grad_norm": 0.17923399806022644,
+      "learning_rate": 5.573145794110799e-05,
+      "loss": 2.0249,
+      "step": 4074
+    },
+    {
+      "epoch": 1.7578175544533103,
+      "grad_norm": 0.2070382833480835,
+      "learning_rate": 5.569847750424144e-05,
+      "loss": 2.2384,
+      "step": 4075
+    },
+    {
+      "epoch": 1.7578175544533103,
+      "eval_loss": 2.09136962890625,
+      "eval_runtime": 201.606,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4075
+    },
+    {
+      "epoch": 1.7582488678024584,
+      "grad_norm": 0.18448781967163086,
+      "learning_rate": 5.566550106396625e-05,
+      "loss": 2.0828,
+      "step": 4076
+    },
+    {
+      "epoch": 1.7586801811516066,
+      "grad_norm": 0.19364839792251587,
+      "learning_rate": 5.563252862711046e-05,
+      "loss": 2.1173,
+      "step": 4077
+    },
+    {
+      "epoch": 1.7591114945007549,
+      "grad_norm": 0.17255939543247223,
+      "learning_rate": 5.559956020050143e-05,
+      "loss": 2.0035,
+      "step": 4078
+    },
+    {
+      "epoch": 1.759542807849903,
+      "grad_norm": 0.17625278234481812,
+      "learning_rate": 5.5566595790965625e-05,
+      "loss": 2.1868,
+      "step": 4079
+    },
+    {
+      "epoch": 1.759974121199051,
+      "grad_norm": 0.18831560015678406,
+      "learning_rate": 5.553363540532868e-05,
+      "loss": 2.0214,
+      "step": 4080
+    },
+    {
+      "epoch": 1.7604054345481992,
+      "grad_norm": 0.19190707802772522,
+      "learning_rate": 5.5500679050415426e-05,
+      "loss": 2.2865,
+      "step": 4081
+    },
+    {
+      "epoch": 1.7608367478973475,
+      "grad_norm": 0.17967812716960907,
+      "learning_rate": 5.5467726733049765e-05,
+      "loss": 2.1373,
+      "step": 4082
+    },
+    {
+      "epoch": 1.7612680612464957,
+      "grad_norm": 0.19684270024299622,
+      "learning_rate": 5.543477846005493e-05,
+      "loss": 2.1682,
+      "step": 4083
+    },
+    {
+      "epoch": 1.7616993745956437,
+      "grad_norm": 0.18483242392539978,
+      "learning_rate": 5.5401834238253165e-05,
+      "loss": 2.2375,
+      "step": 4084
+    },
+    {
+      "epoch": 1.7621306879447918,
+      "grad_norm": 0.16816097497940063,
+      "learning_rate": 5.536889407446593e-05,
+      "loss": 2.146,
+      "step": 4085
+    },
+    {
+      "epoch": 1.76256200129394,
+      "grad_norm": 0.1955161839723587,
+      "learning_rate": 5.533595797551386e-05,
+      "loss": 1.9715,
+      "step": 4086
+    },
+    {
+      "epoch": 1.7629933146430883,
+      "grad_norm": 0.19922764599323273,
+      "learning_rate": 5.530302594821674e-05,
+      "loss": 2.0799,
+      "step": 4087
+    },
+    {
+      "epoch": 1.7634246279922363,
+      "grad_norm": 0.18744394183158875,
+      "learning_rate": 5.527009799939353e-05,
+      "loss": 2.2148,
+      "step": 4088
+    },
+    {
+      "epoch": 1.7638559413413846,
+      "grad_norm": 0.18498487770557404,
+      "learning_rate": 5.523717413586223e-05,
+      "loss": 2.1171,
+      "step": 4089
+    },
+    {
+      "epoch": 1.7642872546905326,
+      "grad_norm": 0.18250973522663116,
+      "learning_rate": 5.52042543644402e-05,
+      "loss": 2.2352,
+      "step": 4090
+    },
+    {
+      "epoch": 1.7647185680396809,
+      "grad_norm": 0.1901228427886963,
+      "learning_rate": 5.517133869194378e-05,
+      "loss": 2.2953,
+      "step": 4091
+    },
+    {
+      "epoch": 1.7651498813888291,
+      "grad_norm": 0.1902785301208496,
+      "learning_rate": 5.513842712518849e-05,
+      "loss": 2.2597,
+      "step": 4092
+    },
+    {
+      "epoch": 1.7655811947379771,
+      "grad_norm": 0.18489375710487366,
+      "learning_rate": 5.51055196709891e-05,
+      "loss": 1.8765,
+      "step": 4093
+    },
+    {
+      "epoch": 1.7660125080871252,
+      "grad_norm": 0.19047106802463531,
+      "learning_rate": 5.5072616336159413e-05,
+      "loss": 2.216,
+      "step": 4094
+    },
+    {
+      "epoch": 1.7664438214362734,
+      "grad_norm": 0.19863182306289673,
+      "learning_rate": 5.5039717127512456e-05,
+      "loss": 2.0272,
+      "step": 4095
+    },
+    {
+      "epoch": 1.7668751347854217,
+      "grad_norm": 0.19716544449329376,
+      "learning_rate": 5.500682205186032e-05,
+      "loss": 2.0992,
+      "step": 4096
+    },
+    {
+      "epoch": 1.7673064481345697,
+      "grad_norm": 0.1879461258649826,
+      "learning_rate": 5.497393111601433e-05,
+      "loss": 2.2904,
+      "step": 4097
+    },
+    {
+      "epoch": 1.767737761483718,
+      "grad_norm": 0.17728249728679657,
+      "learning_rate": 5.4941044326784906e-05,
+      "loss": 2.1857,
+      "step": 4098
+    },
+    {
+      "epoch": 1.768169074832866,
+      "grad_norm": 0.18926134705543518,
+      "learning_rate": 5.49081616909816e-05,
+      "loss": 2.161,
+      "step": 4099
+    },
+    {
+      "epoch": 1.7686003881820143,
+      "grad_norm": 0.18339209258556366,
+      "learning_rate": 5.487528321541315e-05,
+      "loss": 2.2346,
+      "step": 4100
+    },
+    {
+      "epoch": 1.7686003881820143,
+      "eval_loss": 2.0910797119140625,
+      "eval_runtime": 200.6694,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4100
+    },
+    {
+      "epoch": 1.7690317015311625,
+      "grad_norm": 0.17934048175811768,
+      "learning_rate": 5.4842408906887416e-05,
+      "loss": 2.1667,
+      "step": 4101
+    },
+    {
+      "epoch": 1.7694630148803105,
+      "grad_norm": 0.20105083286762238,
+      "learning_rate": 5.480953877221134e-05,
+      "loss": 2.0091,
+      "step": 4102
+    },
+    {
+      "epoch": 1.7698943282294586,
+      "grad_norm": 0.1738869994878769,
+      "learning_rate": 5.477667281819103e-05,
+      "loss": 2.0166,
+      "step": 4103
+    },
+    {
+      "epoch": 1.7703256415786068,
+      "grad_norm": 0.18088917434215546,
+      "learning_rate": 5.47438110516318e-05,
+      "loss": 2.0975,
+      "step": 4104
+    },
+    {
+      "epoch": 1.770756954927755,
+      "grad_norm": 0.17673984169960022,
+      "learning_rate": 5.4710953479338e-05,
+      "loss": 2.0978,
+      "step": 4105
+    },
+    {
+      "epoch": 1.7711882682769031,
+      "grad_norm": 0.18527352809906006,
+      "learning_rate": 5.4678100108113186e-05,
+      "loss": 2.0892,
+      "step": 4106
+    },
+    {
+      "epoch": 1.7716195816260514,
+      "grad_norm": 0.19644999504089355,
+      "learning_rate": 5.464525094475993e-05,
+      "loss": 2.1712,
+      "step": 4107
+    },
+    {
+      "epoch": 1.7720508949751994,
+      "grad_norm": 0.1868610829114914,
+      "learning_rate": 5.461240599608011e-05,
+      "loss": 2.1838,
+      "step": 4108
+    },
+    {
+      "epoch": 1.7724822083243477,
+      "grad_norm": 0.1908998340368271,
+      "learning_rate": 5.457956526887456e-05,
+      "loss": 2.2289,
+      "step": 4109
+    },
+    {
+      "epoch": 1.772913521673496,
+      "grad_norm": 0.22750572860240936,
+      "learning_rate": 5.4546728769943334e-05,
+      "loss": 2.1652,
+      "step": 4110
+    },
+    {
+      "epoch": 1.773344835022644,
+      "grad_norm": 0.2120228111743927,
+      "learning_rate": 5.451389650608559e-05,
+      "loss": 2.177,
+      "step": 4111
+    },
+    {
+      "epoch": 1.773776148371792,
+      "grad_norm": 0.19006375968456268,
+      "learning_rate": 5.448106848409962e-05,
+      "loss": 2.1953,
+      "step": 4112
+    },
+    {
+      "epoch": 1.7742074617209402,
+      "grad_norm": 0.18009048700332642,
+      "learning_rate": 5.444824471078282e-05,
+      "loss": 2.212,
+      "step": 4113
+    },
+    {
+      "epoch": 1.7746387750700885,
+      "grad_norm": 0.18991193175315857,
+      "learning_rate": 5.441542519293165e-05,
+      "loss": 2.2226,
+      "step": 4114
+    },
+    {
+      "epoch": 1.7750700884192365,
+      "grad_norm": 0.1789187490940094,
+      "learning_rate": 5.438260993734187e-05,
+      "loss": 2.1696,
+      "step": 4115
+    },
+    {
+      "epoch": 1.7755014017683848,
+      "grad_norm": 0.18606701493263245,
+      "learning_rate": 5.4349798950808124e-05,
+      "loss": 2.3047,
+      "step": 4116
+    },
+    {
+      "epoch": 1.7759327151175328,
+      "grad_norm": 0.18812114000320435,
+      "learning_rate": 5.431699224012434e-05,
+      "loss": 1.9963,
+      "step": 4117
+    },
+    {
+      "epoch": 1.776364028466681,
+      "grad_norm": 0.17936258018016815,
+      "learning_rate": 5.428418981208351e-05,
+      "loss": 2.1407,
+      "step": 4118
+    },
+    {
+      "epoch": 1.7767953418158293,
+      "grad_norm": 0.20092041790485382,
+      "learning_rate": 5.425139167347772e-05,
+      "loss": 2.2003,
+      "step": 4119
+    },
+    {
+      "epoch": 1.7772266551649774,
+      "grad_norm": 0.1748567521572113,
+      "learning_rate": 5.42185978310982e-05,
+      "loss": 2.2023,
+      "step": 4120
+    },
+    {
+      "epoch": 1.7776579685141254,
+      "grad_norm": 0.18740083277225494,
+      "learning_rate": 5.418580829173522e-05,
+      "loss": 2.1708,
+      "step": 4121
+    },
+    {
+      "epoch": 1.7780892818632736,
+      "grad_norm": 0.2029028683900833,
+      "learning_rate": 5.4153023062178294e-05,
+      "loss": 1.9945,
+      "step": 4122
+    },
+    {
+      "epoch": 1.778520595212422,
+      "grad_norm": 0.21911142766475677,
+      "learning_rate": 5.412024214921591e-05,
+      "loss": 1.8323,
+      "step": 4123
+    },
+    {
+      "epoch": 1.77895190856157,
+      "grad_norm": 0.1845363825559616,
+      "learning_rate": 5.408746555963571e-05,
+      "loss": 2.1028,
+      "step": 4124
+    },
+    {
+      "epoch": 1.7793832219107182,
+      "grad_norm": 0.2083241194486618,
+      "learning_rate": 5.4054693300224476e-05,
+      "loss": 2.2871,
+      "step": 4125
+    },
+    {
+      "epoch": 1.7793832219107182,
+      "eval_loss": 2.0913138389587402,
+      "eval_runtime": 201.3844,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4125
+    },
+    {
+      "epoch": 1.7798145352598662,
+      "grad_norm": 0.1898990422487259,
+      "learning_rate": 5.4021925377768055e-05,
+      "loss": 2.2323,
+      "step": 4126
+    },
+    {
+      "epoch": 1.7802458486090145,
+      "grad_norm": 0.182303786277771,
+      "learning_rate": 5.3989161799051405e-05,
+      "loss": 2.1606,
+      "step": 4127
+    },
+    {
+      "epoch": 1.7806771619581627,
+      "grad_norm": 0.1937088519334793,
+      "learning_rate": 5.395640257085853e-05,
+      "loss": 2.2274,
+      "step": 4128
+    },
+    {
+      "epoch": 1.7811084753073108,
+      "grad_norm": 0.18278345465660095,
+      "learning_rate": 5.3923647699972674e-05,
+      "loss": 2.0109,
+      "step": 4129
+    },
+    {
+      "epoch": 1.7815397886564588,
+      "grad_norm": 0.19565333425998688,
+      "learning_rate": 5.389089719317604e-05,
+      "loss": 2.1422,
+      "step": 4130
+    },
+    {
+      "epoch": 1.781971102005607,
+      "grad_norm": 0.1918390989303589,
+      "learning_rate": 5.385815105724997e-05,
+      "loss": 2.3274,
+      "step": 4131
+    },
+    {
+      "epoch": 1.7824024153547553,
+      "grad_norm": 0.20723852515220642,
+      "learning_rate": 5.382540929897494e-05,
+      "loss": 2.2422,
+      "step": 4132
+    },
+    {
+      "epoch": 1.7828337287039033,
+      "grad_norm": 0.17734549939632416,
+      "learning_rate": 5.379267192513048e-05,
+      "loss": 2.167,
+      "step": 4133
+    },
+    {
+      "epoch": 1.7832650420530516,
+      "grad_norm": 0.1883150041103363,
+      "learning_rate": 5.375993894249523e-05,
+      "loss": 2.1357,
+      "step": 4134
+    },
+    {
+      "epoch": 1.7836963554021996,
+      "grad_norm": 0.18035978078842163,
+      "learning_rate": 5.372721035784686e-05,
+      "loss": 2.2501,
+      "step": 4135
+    },
+    {
+      "epoch": 1.7841276687513479,
+      "grad_norm": 0.1856493055820465,
+      "learning_rate": 5.369448617796226e-05,
+      "loss": 2.2456,
+      "step": 4136
+    },
+    {
+      "epoch": 1.7845589821004961,
+      "grad_norm": 0.19670169055461884,
+      "learning_rate": 5.3661766409617275e-05,
+      "loss": 2.1014,
+      "step": 4137
+    },
+    {
+      "epoch": 1.7849902954496442,
+      "grad_norm": 0.1871079057455063,
+      "learning_rate": 5.362905105958689e-05,
+      "loss": 2.247,
+      "step": 4138
+    },
+    {
+      "epoch": 1.7854216087987922,
+      "grad_norm": 0.2041262686252594,
+      "learning_rate": 5.359634013464521e-05,
+      "loss": 2.1843,
+      "step": 4139
+    },
+    {
+      "epoch": 1.7858529221479404,
+      "grad_norm": 0.1845085173845291,
+      "learning_rate": 5.356363364156536e-05,
+      "loss": 2.3258,
+      "step": 4140
+    },
+    {
+      "epoch": 1.7862842354970887,
+      "grad_norm": 0.18137259781360626,
+      "learning_rate": 5.3530931587119624e-05,
+      "loss": 1.9705,
+      "step": 4141
+    },
+    {
+      "epoch": 1.786715548846237,
+      "grad_norm": 0.20324157178401947,
+      "learning_rate": 5.349823397807925e-05,
+      "loss": 2.3582,
+      "step": 4142
+    },
+    {
+      "epoch": 1.787146862195385,
+      "grad_norm": 0.19847539067268372,
+      "learning_rate": 5.346554082121467e-05,
+      "loss": 2.3731,
+      "step": 4143
+    },
+    {
+      "epoch": 1.787578175544533,
+      "grad_norm": 0.18964673578739166,
+      "learning_rate": 5.3432852123295366e-05,
+      "loss": 2.2977,
+      "step": 4144
+    },
+    {
+      "epoch": 1.7880094888936813,
+      "grad_norm": 0.1747114062309265,
+      "learning_rate": 5.3400167891089876e-05,
+      "loss": 2.0051,
+      "step": 4145
+    },
+    {
+      "epoch": 1.7884408022428295,
+      "grad_norm": 0.19617091119289398,
+      "learning_rate": 5.3367488131365846e-05,
+      "loss": 2.1045,
+      "step": 4146
+    },
+    {
+      "epoch": 1.7888721155919776,
+      "grad_norm": 0.18542206287384033,
+      "learning_rate": 5.333481285088999e-05,
+      "loss": 1.9911,
+      "step": 4147
+    },
+    {
+      "epoch": 1.7893034289411256,
+      "grad_norm": 0.18740200996398926,
+      "learning_rate": 5.3302142056428035e-05,
+      "loss": 2.2449,
+      "step": 4148
+    },
+    {
+      "epoch": 1.7897347422902739,
+      "grad_norm": 0.22210511565208435,
+      "learning_rate": 5.326947575474483e-05,
+      "loss": 2.0402,
+      "step": 4149
+    },
+    {
+      "epoch": 1.790166055639422,
+      "grad_norm": 0.19126732647418976,
+      "learning_rate": 5.323681395260434e-05,
+      "loss": 2.2384,
+      "step": 4150
+    },
+    {
+      "epoch": 1.790166055639422,
+      "eval_loss": 2.0913338661193848,
+      "eval_runtime": 200.6128,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 4150
+    },
+    {
+      "epoch": 1.7905973689885704,
+      "grad_norm": 0.19874311983585358,
+      "learning_rate": 5.3204156656769514e-05,
+      "loss": 2.2883,
+      "step": 4151
+    },
+    {
+      "epoch": 1.7910286823377184,
+      "grad_norm": 0.18604867160320282,
+      "learning_rate": 5.317150387400243e-05,
+      "loss": 2.143,
+      "step": 4152
+    },
+    {
+      "epoch": 1.7914599956868664,
+      "grad_norm": 0.1757567971944809,
+      "learning_rate": 5.313885561106412e-05,
+      "loss": 2.0399,
+      "step": 4153
+    },
+    {
+      "epoch": 1.7918913090360147,
+      "grad_norm": 0.19099228084087372,
+      "learning_rate": 5.3106211874714885e-05,
+      "loss": 2.1222,
+      "step": 4154
+    },
+    {
+      "epoch": 1.792322622385163,
+      "grad_norm": 0.18796351552009583,
+      "learning_rate": 5.307357267171389e-05,
+      "loss": 2.2579,
+      "step": 4155
+    },
+    {
+      "epoch": 1.792753935734311,
+      "grad_norm": 0.17198990285396576,
+      "learning_rate": 5.3040938008819435e-05,
+      "loss": 2.106,
+      "step": 4156
+    },
+    {
+      "epoch": 1.793185249083459,
+      "grad_norm": 0.16483992338180542,
+      "learning_rate": 5.3008307892788926e-05,
+      "loss": 1.8541,
+      "step": 4157
+    },
+    {
+      "epoch": 1.7936165624326073,
+      "grad_norm": 0.18197329342365265,
+      "learning_rate": 5.297568233037877e-05,
+      "loss": 2.3625,
+      "step": 4158
+    },
+    {
+      "epoch": 1.7940478757817555,
+      "grad_norm": 0.18919214606285095,
+      "learning_rate": 5.294306132834445e-05,
+      "loss": 2.1912,
+      "step": 4159
+    },
+    {
+      "epoch": 1.7944791891309038,
+      "grad_norm": 0.19854745268821716,
+      "learning_rate": 5.291044489344044e-05,
+      "loss": 2.1172,
+      "step": 4160
+    },
+    {
+      "epoch": 1.7949105024800518,
+      "grad_norm": 0.17984916269779205,
+      "learning_rate": 5.2877833032420444e-05,
+      "loss": 1.9253,
+      "step": 4161
+    },
+    {
+      "epoch": 1.7953418158291998,
+      "grad_norm": 0.1974160075187683,
+      "learning_rate": 5.284522575203702e-05,
+      "loss": 2.0063,
+      "step": 4162
+    },
+    {
+      "epoch": 1.795773129178348,
+      "grad_norm": 0.1801133155822754,
+      "learning_rate": 5.281262305904186e-05,
+      "loss": 2.0987,
+      "step": 4163
+    },
+    {
+      "epoch": 1.7962044425274963,
+      "grad_norm": 0.1818169206380844,
+      "learning_rate": 5.278002496018575e-05,
+      "loss": 2.2302,
+      "step": 4164
+    },
+    {
+      "epoch": 1.7966357558766444,
+      "grad_norm": 0.16762112081050873,
+      "learning_rate": 5.274743146221848e-05,
+      "loss": 2.4302,
+      "step": 4165
+    },
+    {
+      "epoch": 1.7970670692257924,
+      "grad_norm": 0.19300805032253265,
+      "learning_rate": 5.271484257188888e-05,
+      "loss": 2.1999,
+      "step": 4166
+    },
+    {
+      "epoch": 1.7974983825749407,
+      "grad_norm": 0.19982370734214783,
+      "learning_rate": 5.268225829594479e-05,
+      "loss": 2.5799,
+      "step": 4167
+    },
+    {
+      "epoch": 1.797929695924089,
+      "grad_norm": 0.18215444684028625,
+      "learning_rate": 5.264967864113322e-05,
+      "loss": 2.21,
+      "step": 4168
+    },
+    {
+      "epoch": 1.7983610092732372,
+      "grad_norm": 0.20935408771038055,
+      "learning_rate": 5.261710361420009e-05,
+      "loss": 2.074,
+      "step": 4169
+    },
+    {
+      "epoch": 1.7987923226223852,
+      "grad_norm": 0.20121844112873077,
+      "learning_rate": 5.258453322189043e-05,
+      "loss": 2.3008,
+      "step": 4170
+    },
+    {
+      "epoch": 1.7992236359715332,
+      "grad_norm": 0.18649135529994965,
+      "learning_rate": 5.255196747094832e-05,
+      "loss": 2.2496,
+      "step": 4171
+    },
+    {
+      "epoch": 1.7996549493206815,
+      "grad_norm": 0.18619664013385773,
+      "learning_rate": 5.2519406368116806e-05,
+      "loss": 2.0526,
+      "step": 4172
+    },
+    {
+      "epoch": 1.8000862626698297,
+      "grad_norm": 0.18811942636966705,
+      "learning_rate": 5.2486849920138085e-05,
+      "loss": 2.2005,
+      "step": 4173
+    },
+    {
+      "epoch": 1.8005175760189778,
+      "grad_norm": 0.19535639882087708,
+      "learning_rate": 5.245429813375322e-05,
+      "loss": 2.0817,
+      "step": 4174
+    },
+    {
+      "epoch": 1.8009488893681258,
+      "grad_norm": 0.19291633367538452,
+      "learning_rate": 5.2421751015702554e-05,
+      "loss": 2.1703,
+      "step": 4175
+    },
+    {
+      "epoch": 1.8009488893681258,
+      "eval_loss": 2.090921401977539,
+      "eval_runtime": 201.6385,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4175
+    },
+    {
+      "epoch": 1.801380202717274,
+      "grad_norm": 0.19690266251564026,
+      "learning_rate": 5.238920857272522e-05,
+      "loss": 2.1343,
+      "step": 4176
+    },
+    {
+      "epoch": 1.8018115160664223,
+      "grad_norm": 0.17731203138828278,
+      "learning_rate": 5.23566708115595e-05,
+      "loss": 2.0657,
+      "step": 4177
+    },
+    {
+      "epoch": 1.8022428294155706,
+      "grad_norm": 0.20055049657821655,
+      "learning_rate": 5.232413773894274e-05,
+      "loss": 2.0814,
+      "step": 4178
+    },
+    {
+      "epoch": 1.8026741427647186,
+      "grad_norm": 0.20533232390880585,
+      "learning_rate": 5.229160936161123e-05,
+      "loss": 2.1925,
+      "step": 4179
+    },
+    {
+      "epoch": 1.8031054561138666,
+      "grad_norm": 0.1787910759449005,
+      "learning_rate": 5.2259085686300356e-05,
+      "loss": 2.3697,
+      "step": 4180
+    },
+    {
+      "epoch": 1.8035367694630149,
+      "grad_norm": 0.18992182612419128,
+      "learning_rate": 5.2226566719744454e-05,
+      "loss": 2.2862,
+      "step": 4181
+    },
+    {
+      "epoch": 1.8039680828121631,
+      "grad_norm": 0.21222200989723206,
+      "learning_rate": 5.2194052468676975e-05,
+      "loss": 2.162,
+      "step": 4182
+    },
+    {
+      "epoch": 1.8043993961613112,
+      "grad_norm": 0.1847771257162094,
+      "learning_rate": 5.216154293983032e-05,
+      "loss": 2.0813,
+      "step": 4183
+    },
+    {
+      "epoch": 1.8048307095104592,
+      "grad_norm": 0.18472546339035034,
+      "learning_rate": 5.2129038139935954e-05,
+      "loss": 2.1162,
+      "step": 4184
+    },
+    {
+      "epoch": 1.8052620228596075,
+      "grad_norm": 0.16956888139247894,
+      "learning_rate": 5.209653807572437e-05,
+      "loss": 2.176,
+      "step": 4185
+    },
+    {
+      "epoch": 1.8056933362087557,
+      "grad_norm": 0.21091744303703308,
+      "learning_rate": 5.206404275392505e-05,
+      "loss": 1.8348,
+      "step": 4186
+    },
+    {
+      "epoch": 1.806124649557904,
+      "grad_norm": 0.1923355609178543,
+      "learning_rate": 5.2031552181266494e-05,
+      "loss": 2.2377,
+      "step": 4187
+    },
+    {
+      "epoch": 1.806555962907052,
+      "grad_norm": 0.19391633570194244,
+      "learning_rate": 5.199906636447622e-05,
+      "loss": 2.2979,
+      "step": 4188
+    },
+    {
+      "epoch": 1.8069872762562,
+      "grad_norm": 0.20279642939567566,
+      "learning_rate": 5.19665853102808e-05,
+      "loss": 2.104,
+      "step": 4189
+    },
+    {
+      "epoch": 1.8074185896053483,
+      "grad_norm": 0.1883164495229721,
+      "learning_rate": 5.1934109025405784e-05,
+      "loss": 2.2072,
+      "step": 4190
+    },
+    {
+      "epoch": 1.8078499029544965,
+      "grad_norm": 0.1902153640985489,
+      "learning_rate": 5.190163751657573e-05,
+      "loss": 2.2251,
+      "step": 4191
+    },
+    {
+      "epoch": 1.8082812163036446,
+      "grad_norm": 0.17991870641708374,
+      "learning_rate": 5.186917079051424e-05,
+      "loss": 2.0905,
+      "step": 4192
+    },
+    {
+      "epoch": 1.8087125296527926,
+      "grad_norm": 0.17719264328479767,
+      "learning_rate": 5.1836708853943896e-05,
+      "loss": 2.2346,
+      "step": 4193
+    },
+    {
+      "epoch": 1.8091438430019409,
+      "grad_norm": 0.1893806755542755,
+      "learning_rate": 5.1804251713586295e-05,
+      "loss": 2.0722,
+      "step": 4194
+    },
+    {
+      "epoch": 1.8095751563510891,
+      "grad_norm": 0.1863139271736145,
+      "learning_rate": 5.177179937616201e-05,
+      "loss": 2.128,
+      "step": 4195
+    },
+    {
+      "epoch": 1.8100064697002374,
+      "grad_norm": 0.1845807284116745,
+      "learning_rate": 5.173935184839071e-05,
+      "loss": 2.1744,
+      "step": 4196
+    },
+    {
+      "epoch": 1.8104377830493854,
+      "grad_norm": 0.17936205863952637,
+      "learning_rate": 5.170690913699099e-05,
+      "loss": 2.0823,
+      "step": 4197
+    },
+    {
+      "epoch": 1.8108690963985334,
+      "grad_norm": 0.17481139302253723,
+      "learning_rate": 5.167447124868043e-05,
+      "loss": 2.2196,
+      "step": 4198
+    },
+    {
+      "epoch": 1.8113004097476817,
+      "grad_norm": 0.2126890867948532,
+      "learning_rate": 5.1642038190175704e-05,
+      "loss": 1.7506,
+      "step": 4199
+    },
+    {
+      "epoch": 1.81173172309683,
+      "grad_norm": 0.1977444440126419,
+      "learning_rate": 5.1609609968192424e-05,
+      "loss": 2.1454,
+      "step": 4200
+    },
+    {
+      "epoch": 1.81173172309683,
+      "eval_loss": 2.090848684310913,
+      "eval_runtime": 201.5534,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4200
+    },
+    {
+      "epoch": 1.812163036445978,
+      "grad_norm": 0.2021864801645279,
+      "learning_rate": 5.1577186589445186e-05,
+      "loss": 2.1619,
+      "step": 4201
+    },
+    {
+      "epoch": 1.812594349795126,
+      "grad_norm": 0.1859135925769806,
+      "learning_rate": 5.154476806064758e-05,
+      "loss": 2.1467,
+      "step": 4202
+    },
+    {
+      "epoch": 1.8130256631442743,
+      "grad_norm": 0.2111302763223648,
+      "learning_rate": 5.1512354388512274e-05,
+      "loss": 2.1874,
+      "step": 4203
+    },
+    {
+      "epoch": 1.8134569764934225,
+      "grad_norm": 0.19402331113815308,
+      "learning_rate": 5.1479945579750845e-05,
+      "loss": 2.2364,
+      "step": 4204
+    },
+    {
+      "epoch": 1.8138882898425708,
+      "grad_norm": 0.19910752773284912,
+      "learning_rate": 5.1447541641073915e-05,
+      "loss": 2.2734,
+      "step": 4205
+    },
+    {
+      "epoch": 1.8143196031917188,
+      "grad_norm": 0.1855575144290924,
+      "learning_rate": 5.1415142579191016e-05,
+      "loss": 2.3032,
+      "step": 4206
+    },
+    {
+      "epoch": 1.8147509165408668,
+      "grad_norm": 0.20500385761260986,
+      "learning_rate": 5.1382748400810796e-05,
+      "loss": 2.0675,
+      "step": 4207
+    },
+    {
+      "epoch": 1.815182229890015,
+      "grad_norm": 0.22097423672676086,
+      "learning_rate": 5.135035911264078e-05,
+      "loss": 2.3183,
+      "step": 4208
+    },
+    {
+      "epoch": 1.8156135432391634,
+      "grad_norm": 0.18906009197235107,
+      "learning_rate": 5.131797472138752e-05,
+      "loss": 2.1094,
+      "step": 4209
+    },
+    {
+      "epoch": 1.8160448565883114,
+      "grad_norm": 0.1831335723400116,
+      "learning_rate": 5.128559523375661e-05,
+      "loss": 2.0186,
+      "step": 4210
+    },
+    {
+      "epoch": 1.8164761699374594,
+      "grad_norm": 0.19466155767440796,
+      "learning_rate": 5.125322065645253e-05,
+      "loss": 2.1229,
+      "step": 4211
+    },
+    {
+      "epoch": 1.8169074832866077,
+      "grad_norm": 0.4434211850166321,
+      "learning_rate": 5.122085099617882e-05,
+      "loss": 2.2268,
+      "step": 4212
+    },
+    {
+      "epoch": 1.817338796635756,
+      "grad_norm": 0.1960372030735016,
+      "learning_rate": 5.11884862596379e-05,
+      "loss": 2.1383,
+      "step": 4213
+    },
+    {
+      "epoch": 1.8177701099849042,
+      "grad_norm": 0.1842081993818283,
+      "learning_rate": 5.115612645353136e-05,
+      "loss": 2.2059,
+      "step": 4214
+    },
+    {
+      "epoch": 1.8182014233340522,
+      "grad_norm": 0.18694734573364258,
+      "learning_rate": 5.112377158455957e-05,
+      "loss": 2.1145,
+      "step": 4215
+    },
+    {
+      "epoch": 1.8186327366832002,
+      "grad_norm": 0.2071239948272705,
+      "learning_rate": 5.109142165942196e-05,
+      "loss": 2.2563,
+      "step": 4216
+    },
+    {
+      "epoch": 1.8190640500323485,
+      "grad_norm": 0.17906248569488525,
+      "learning_rate": 5.105907668481697e-05,
+      "loss": 2.1458,
+      "step": 4217
+    },
+    {
+      "epoch": 1.8194953633814968,
+      "grad_norm": 0.18535931408405304,
+      "learning_rate": 5.1026736667441975e-05,
+      "loss": 2.1868,
+      "step": 4218
+    },
+    {
+      "epoch": 1.8199266767306448,
+      "grad_norm": 0.19665364921092987,
+      "learning_rate": 5.099440161399333e-05,
+      "loss": 2.2715,
+      "step": 4219
+    },
+    {
+      "epoch": 1.8203579900797928,
+      "grad_norm": 0.20698903501033783,
+      "learning_rate": 5.0962071531166316e-05,
+      "loss": 2.2083,
+      "step": 4220
+    },
+    {
+      "epoch": 1.820789303428941,
+      "grad_norm": 0.19022543728351593,
+      "learning_rate": 5.0929746425655286e-05,
+      "loss": 1.999,
+      "step": 4221
+    },
+    {
+      "epoch": 1.8212206167780893,
+      "grad_norm": 0.18219402432441711,
+      "learning_rate": 5.089742630415349e-05,
+      "loss": 2.2458,
+      "step": 4222
+    },
+    {
+      "epoch": 1.8216519301272376,
+      "grad_norm": 0.19657008349895477,
+      "learning_rate": 5.086511117335313e-05,
+      "loss": 2.1712,
+      "step": 4223
+    },
+    {
+      "epoch": 1.8220832434763856,
+      "grad_norm": 0.19631534814834595,
+      "learning_rate": 5.0832801039945466e-05,
+      "loss": 2.0941,
+      "step": 4224
+    },
+    {
+      "epoch": 1.8225145568255336,
+      "grad_norm": 0.2034425437450409,
+      "learning_rate": 5.080049591062065e-05,
+      "loss": 2.0156,
+      "step": 4225
+    },
+    {
+      "epoch": 1.8225145568255336,
+      "eval_loss": 2.090664863586426,
+      "eval_runtime": 210.7797,
+      "eval_samples_per_second": 0.152,
+      "eval_steps_per_second": 0.152,
+      "step": 4225
+    },
+    {
+      "epoch": 1.822945870174682,
+      "grad_norm": 0.20252689719200134,
+      "learning_rate": 5.076819579206777e-05,
+      "loss": 2.1967,
+      "step": 4226
+    },
+    {
+      "epoch": 1.8233771835238302,
+      "grad_norm": 0.17087893187999725,
+      "learning_rate": 5.073590069097494e-05,
+      "loss": 1.9148,
+      "step": 4227
+    },
+    {
+      "epoch": 1.8238084968729782,
+      "grad_norm": 0.18986919522285461,
+      "learning_rate": 5.0703610614029234e-05,
+      "loss": 2.2916,
+      "step": 4228
+    },
+    {
+      "epoch": 1.8242398102221262,
+      "grad_norm": 0.20265305042266846,
+      "learning_rate": 5.0671325567916644e-05,
+      "loss": 2.177,
+      "step": 4229
+    },
+    {
+      "epoch": 1.8246711235712745,
+      "grad_norm": 0.183542400598526,
+      "learning_rate": 5.063904555932214e-05,
+      "loss": 2.1599,
+      "step": 4230
+    },
+    {
+      "epoch": 1.8251024369204227,
+      "grad_norm": 0.20302563905715942,
+      "learning_rate": 5.060677059492967e-05,
+      "loss": 2.1233,
+      "step": 4231
+    },
+    {
+      "epoch": 1.825533750269571,
+      "grad_norm": 0.15920883417129517,
+      "learning_rate": 5.057450068142214e-05,
+      "loss": 1.9773,
+      "step": 4232
+    },
+    {
+      "epoch": 1.825965063618719,
+      "grad_norm": 0.17501461505889893,
+      "learning_rate": 5.054223582548131e-05,
+      "loss": 2.1985,
+      "step": 4233
+    },
+    {
+      "epoch": 1.826396376967867,
+      "grad_norm": 0.16960765421390533,
+      "learning_rate": 5.050997603378803e-05,
+      "loss": 2.2563,
+      "step": 4234
+    },
+    {
+      "epoch": 1.8268276903170153,
+      "grad_norm": 0.19847039878368378,
+      "learning_rate": 5.047772131302204e-05,
+      "loss": 2.1497,
+      "step": 4235
+    },
+    {
+      "epoch": 1.8272590036661636,
+      "grad_norm": 0.18529687821865082,
+      "learning_rate": 5.044547166986202e-05,
+      "loss": 2.1033,
+      "step": 4236
+    },
+    {
+      "epoch": 1.8276903170153116,
+      "grad_norm": 0.1980198174715042,
+      "learning_rate": 5.0413227110985596e-05,
+      "loss": 2.0504,
+      "step": 4237
+    },
+    {
+      "epoch": 1.8281216303644596,
+      "grad_norm": 0.17270773649215698,
+      "learning_rate": 5.0380987643069394e-05,
+      "loss": 2.0891,
+      "step": 4238
+    },
+    {
+      "epoch": 1.8285529437136079,
+      "grad_norm": 0.19012048840522766,
+      "learning_rate": 5.0348753272788946e-05,
+      "loss": 2.1669,
+      "step": 4239
+    },
+    {
+      "epoch": 1.8289842570627561,
+      "grad_norm": 0.1755671203136444,
+      "learning_rate": 5.03165240068187e-05,
+      "loss": 2.0635,
+      "step": 4240
+    },
+    {
+      "epoch": 1.8294155704119044,
+      "grad_norm": 0.1877172291278839,
+      "learning_rate": 5.028429985183208e-05,
+      "loss": 2.1066,
+      "step": 4241
+    },
+    {
+      "epoch": 1.8298468837610524,
+      "grad_norm": 0.18054725229740143,
+      "learning_rate": 5.0252080814501466e-05,
+      "loss": 2.0344,
+      "step": 4242
+    },
+    {
+      "epoch": 1.8302781971102005,
+      "grad_norm": 0.17783662676811218,
+      "learning_rate": 5.0219866901498166e-05,
+      "loss": 1.9928,
+      "step": 4243
+    },
+    {
+      "epoch": 1.8307095104593487,
+      "grad_norm": 0.19931326806545258,
+      "learning_rate": 5.0187658119492394e-05,
+      "loss": 1.9977,
+      "step": 4244
+    },
+    {
+      "epoch": 1.831140823808497,
+      "grad_norm": 0.19601735472679138,
+      "learning_rate": 5.0155454475153376e-05,
+      "loss": 2.187,
+      "step": 4245
+    },
+    {
+      "epoch": 1.831572137157645,
+      "grad_norm": 0.19454285502433777,
+      "learning_rate": 5.012325597514922e-05,
+      "loss": 2.204,
+      "step": 4246
+    },
+    {
+      "epoch": 1.832003450506793,
+      "grad_norm": 0.1874435693025589,
+      "learning_rate": 5.0091062626146936e-05,
+      "loss": 1.9728,
+      "step": 4247
+    },
+    {
+      "epoch": 1.8324347638559413,
+      "grad_norm": 0.1848047524690628,
+      "learning_rate": 5.0058874434812524e-05,
+      "loss": 2.1505,
+      "step": 4248
+    },
+    {
+      "epoch": 1.8328660772050895,
+      "grad_norm": 0.2200363278388977,
+      "learning_rate": 5.0026691407810926e-05,
+      "loss": 1.9998,
+      "step": 4249
+    },
+    {
+      "epoch": 1.8332973905542378,
+      "grad_norm": 0.20545153319835663,
+      "learning_rate": 4.9994513551805976e-05,
+      "loss": 2.1764,
+      "step": 4250
+    },
+    {
+      "epoch": 1.8332973905542378,
+      "eval_loss": 2.0905096530914307,
+      "eval_runtime": 204.2211,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 4250
+    },
+    {
+      "epoch": 1.8337287039033858,
+      "grad_norm": 0.19514457881450653,
+      "learning_rate": 4.996234087346047e-05,
+      "loss": 2.2134,
+      "step": 4251
+    },
+    {
+      "epoch": 1.8341600172525339,
+      "grad_norm": 0.18694037199020386,
+      "learning_rate": 4.9930173379436043e-05,
+      "loss": 2.2111,
+      "step": 4252
+    },
+    {
+      "epoch": 1.8345913306016821,
+      "grad_norm": 0.1747187077999115,
+      "learning_rate": 4.9898011076393426e-05,
+      "loss": 2.2134,
+      "step": 4253
+    },
+    {
+      "epoch": 1.8350226439508304,
+      "grad_norm": 0.1897505521774292,
+      "learning_rate": 4.9865853970992104e-05,
+      "loss": 2.201,
+      "step": 4254
+    },
+    {
+      "epoch": 1.8354539572999784,
+      "grad_norm": 0.1944442093372345,
+      "learning_rate": 4.9833702069890575e-05,
+      "loss": 2.1418,
+      "step": 4255
+    },
+    {
+      "epoch": 1.8358852706491264,
+      "grad_norm": 0.21999591588974,
+      "learning_rate": 4.9801555379746254e-05,
+      "loss": 2.1245,
+      "step": 4256
+    },
+    {
+      "epoch": 1.8363165839982747,
+      "grad_norm": 0.19068877398967743,
+      "learning_rate": 4.976941390721546e-05,
+      "loss": 2.124,
+      "step": 4257
+    },
+    {
+      "epoch": 1.836747897347423,
+      "grad_norm": 0.19536010921001434,
+      "learning_rate": 4.973727765895345e-05,
+      "loss": 2.2138,
+      "step": 4258
+    },
+    {
+      "epoch": 1.8371792106965712,
+      "grad_norm": 0.20152810215950012,
+      "learning_rate": 4.970514664161432e-05,
+      "loss": 2.1205,
+      "step": 4259
+    },
+    {
+      "epoch": 1.8376105240457192,
+      "grad_norm": 0.18851511180400848,
+      "learning_rate": 4.9673020861851245e-05,
+      "loss": 2.0916,
+      "step": 4260
+    },
+    {
+      "epoch": 1.8380418373948673,
+      "grad_norm": 0.19676554203033447,
+      "learning_rate": 4.964090032631616e-05,
+      "loss": 2.2678,
+      "step": 4261
+    },
+    {
+      "epoch": 1.8384731507440155,
+      "grad_norm": 0.18774870038032532,
+      "learning_rate": 4.960878504165996e-05,
+      "loss": 2.3108,
+      "step": 4262
+    },
+    {
+      "epoch": 1.8389044640931638,
+      "grad_norm": 0.18802188336849213,
+      "learning_rate": 4.957667501453252e-05,
+      "loss": 2.1035,
+      "step": 4263
+    },
+    {
+      "epoch": 1.8393357774423118,
+      "grad_norm": 0.18654939532279968,
+      "learning_rate": 4.954457025158253e-05,
+      "loss": 2.0947,
+      "step": 4264
+    },
+    {
+      "epoch": 1.8397670907914598,
+      "grad_norm": 0.17369955778121948,
+      "learning_rate": 4.951247075945767e-05,
+      "loss": 2.1399,
+      "step": 4265
+    },
+    {
+      "epoch": 1.840198404140608,
+      "grad_norm": 0.1797541379928589,
+      "learning_rate": 4.948037654480443e-05,
+      "loss": 2.13,
+      "step": 4266
+    },
+    {
+      "epoch": 1.8406297174897563,
+      "grad_norm": 0.19826960563659668,
+      "learning_rate": 4.944828761426832e-05,
+      "loss": 2.2791,
+      "step": 4267
+    },
+    {
+      "epoch": 1.8410610308389046,
+      "grad_norm": 0.21253716945648193,
+      "learning_rate": 4.9416203974493684e-05,
+      "loss": 2.2161,
+      "step": 4268
+    },
+    {
+      "epoch": 1.8414923441880526,
+      "grad_norm": 0.177699014544487,
+      "learning_rate": 4.9384125632123784e-05,
+      "loss": 2.127,
+      "step": 4269
+    },
+    {
+      "epoch": 1.8419236575372007,
+      "grad_norm": 0.19456614553928375,
+      "learning_rate": 4.935205259380082e-05,
+      "loss": 2.1979,
+      "step": 4270
+    },
+    {
+      "epoch": 1.842354970886349,
+      "grad_norm": 0.21789413690567017,
+      "learning_rate": 4.9319984866165855e-05,
+      "loss": 2.1581,
+      "step": 4271
+    },
+    {
+      "epoch": 1.8427862842354972,
+      "grad_norm": 0.18620431423187256,
+      "learning_rate": 4.928792245585886e-05,
+      "loss": 2.0833,
+      "step": 4272
+    },
+    {
+      "epoch": 1.8432175975846452,
+      "grad_norm": 0.2038848102092743,
+      "learning_rate": 4.925586536951867e-05,
+      "loss": 2.199,
+      "step": 4273
+    },
+    {
+      "epoch": 1.8436489109337932,
+      "grad_norm": 0.18201008439064026,
+      "learning_rate": 4.92238136137831e-05,
+      "loss": 2.1769,
+      "step": 4274
+    },
+    {
+      "epoch": 1.8440802242829415,
+      "grad_norm": 0.18450991809368134,
+      "learning_rate": 4.9191767195288815e-05,
+      "loss": 2.3893,
+      "step": 4275
+    },
+    {
+      "epoch": 1.8440802242829415,
+      "eval_loss": 2.0907812118530273,
+      "eval_runtime": 203.8949,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 4275
+    },
+    {
+      "epoch": 1.8445115376320897,
+      "grad_norm": 0.18853482604026794,
+      "learning_rate": 4.9159726120671343e-05,
+      "loss": 2.1397,
+      "step": 4276
+    },
+    {
+      "epoch": 1.844942850981238,
+      "grad_norm": 0.22452791035175323,
+      "learning_rate": 4.912769039656519e-05,
+      "loss": 2.3544,
+      "step": 4277
+    },
+    {
+      "epoch": 1.845374164330386,
+      "grad_norm": 0.20442944765090942,
+      "learning_rate": 4.909566002960369e-05,
+      "loss": 2.2584,
+      "step": 4278
+    },
+    {
+      "epoch": 1.845805477679534,
+      "grad_norm": 0.2112889438867569,
+      "learning_rate": 4.906363502641904e-05,
+      "loss": 2.221,
+      "step": 4279
+    },
+    {
+      "epoch": 1.8462367910286823,
+      "grad_norm": 0.19078004360198975,
+      "learning_rate": 4.9031615393642375e-05,
+      "loss": 2.2339,
+      "step": 4280
+    },
+    {
+      "epoch": 1.8466681043778306,
+      "grad_norm": 0.17326442897319794,
+      "learning_rate": 4.899960113790376e-05,
+      "loss": 2.1394,
+      "step": 4281
+    },
+    {
+      "epoch": 1.8470994177269786,
+      "grad_norm": 0.2283521592617035,
+      "learning_rate": 4.8967592265832054e-05,
+      "loss": 2.2435,
+      "step": 4282
+    },
+    {
+      "epoch": 1.8475307310761266,
+      "grad_norm": 0.18705476820468903,
+      "learning_rate": 4.893558878405504e-05,
+      "loss": 2.4744,
+      "step": 4283
+    },
+    {
+      "epoch": 1.847962044425275,
+      "grad_norm": 0.1761569231748581,
+      "learning_rate": 4.8903590699199416e-05,
+      "loss": 2.0534,
+      "step": 4284
+    },
+    {
+      "epoch": 1.8483933577744232,
+      "grad_norm": 0.19114083051681519,
+      "learning_rate": 4.887159801789074e-05,
+      "loss": 2.2539,
+      "step": 4285
+    },
+    {
+      "epoch": 1.8488246711235714,
+      "grad_norm": 0.1969086229801178,
+      "learning_rate": 4.883961074675343e-05,
+      "loss": 2.2408,
+      "step": 4286
+    },
+    {
+      "epoch": 1.8492559844727194,
+      "grad_norm": 0.19349128007888794,
+      "learning_rate": 4.880762889241076e-05,
+      "loss": 2.0226,
+      "step": 4287
+    },
+    {
+      "epoch": 1.8496872978218675,
+      "grad_norm": 0.1920590102672577,
+      "learning_rate": 4.877565246148499e-05,
+      "loss": 2.2275,
+      "step": 4288
+    },
+    {
+      "epoch": 1.8501186111710157,
+      "grad_norm": 0.17939484119415283,
+      "learning_rate": 4.8743681460597164e-05,
+      "loss": 2.1999,
+      "step": 4289
+    },
+    {
+      "epoch": 1.850549924520164,
+      "grad_norm": 0.19642594456672668,
+      "learning_rate": 4.8711715896367206e-05,
+      "loss": 2.1037,
+      "step": 4290
+    },
+    {
+      "epoch": 1.850981237869312,
+      "grad_norm": 0.1994245946407318,
+      "learning_rate": 4.867975577541398e-05,
+      "loss": 2.0785,
+      "step": 4291
+    },
+    {
+      "epoch": 1.85141255121846,
+      "grad_norm": 0.22636477649211884,
+      "learning_rate": 4.8647801104355165e-05,
+      "loss": 2.1379,
+      "step": 4292
+    },
+    {
+      "epoch": 1.8518438645676083,
+      "grad_norm": 0.18629512190818787,
+      "learning_rate": 4.8615851889807294e-05,
+      "loss": 2.2186,
+      "step": 4293
+    },
+    {
+      "epoch": 1.8522751779167566,
+      "grad_norm": 0.1882411539554596,
+      "learning_rate": 4.85839081383858e-05,
+      "loss": 2.1205,
+      "step": 4294
+    },
+    {
+      "epoch": 1.8527064912659048,
+      "grad_norm": 0.17799854278564453,
+      "learning_rate": 4.855196985670504e-05,
+      "loss": 2.143,
+      "step": 4295
+    },
+    {
+      "epoch": 1.8531378046150528,
+      "grad_norm": 0.19559095799922943,
+      "learning_rate": 4.852003705137815e-05,
+      "loss": 2.1317,
+      "step": 4296
+    },
+    {
+      "epoch": 1.8535691179642009,
+      "grad_norm": 0.18567374348640442,
+      "learning_rate": 4.848810972901717e-05,
+      "loss": 2.1988,
+      "step": 4297
+    },
+    {
+      "epoch": 1.8540004313133491,
+      "grad_norm": 0.21284270286560059,
+      "learning_rate": 4.845618789623296e-05,
+      "loss": 2.1892,
+      "step": 4298
+    },
+    {
+      "epoch": 1.8544317446624974,
+      "grad_norm": 0.18262144923210144,
+      "learning_rate": 4.842427155963537e-05,
+      "loss": 2.0618,
+      "step": 4299
+    },
+    {
+      "epoch": 1.8548630580116454,
+      "grad_norm": 0.19378851354122162,
+      "learning_rate": 4.839236072583297e-05,
+      "loss": 2.1478,
+      "step": 4300
+    },
+    {
+      "epoch": 1.8548630580116454,
+      "eval_loss": 2.0903193950653076,
+      "eval_runtime": 202.772,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 4300
+    },
+    {
+      "epoch": 1.8552943713607937,
+      "grad_norm": 0.19272704422473907,
+      "learning_rate": 4.8360455401433236e-05,
+      "loss": 2.0959,
+      "step": 4301
+    },
+    {
+      "epoch": 1.8557256847099417,
+      "grad_norm": 0.21725627779960632,
+      "learning_rate": 4.832855559304256e-05,
+      "loss": 2.2652,
+      "step": 4302
+    },
+    {
+      "epoch": 1.85615699805909,
+      "grad_norm": 0.18098394572734833,
+      "learning_rate": 4.829666130726611e-05,
+      "loss": 2.2715,
+      "step": 4303
+    },
+    {
+      "epoch": 1.8565883114082382,
+      "grad_norm": 0.17119595408439636,
+      "learning_rate": 4.8264772550707994e-05,
+      "loss": 2.1291,
+      "step": 4304
+    },
+    {
+      "epoch": 1.8570196247573862,
+      "grad_norm": 0.19038599729537964,
+      "learning_rate": 4.823288932997104e-05,
+      "loss": 1.9872,
+      "step": 4305
+    },
+    {
+      "epoch": 1.8574509381065343,
+      "grad_norm": 0.2354617565870285,
+      "learning_rate": 4.82010116516571e-05,
+      "loss": 2.251,
+      "step": 4306
+    },
+    {
+      "epoch": 1.8578822514556825,
+      "grad_norm": 0.18846596777439117,
+      "learning_rate": 4.8169139522366756e-05,
+      "loss": 2.1968,
+      "step": 4307
+    },
+    {
+      "epoch": 1.8583135648048308,
+      "grad_norm": 0.1935591846704483,
+      "learning_rate": 4.813727294869949e-05,
+      "loss": 2.2544,
+      "step": 4308
+    },
+    {
+      "epoch": 1.8587448781539788,
+      "grad_norm": 0.19950206577777863,
+      "learning_rate": 4.810541193725364e-05,
+      "loss": 2.2831,
+      "step": 4309
+    },
+    {
+      "epoch": 1.859176191503127,
+      "grad_norm": 0.1986803561449051,
+      "learning_rate": 4.8073556494626366e-05,
+      "loss": 2.2389,
+      "step": 4310
+    },
+    {
+      "epoch": 1.859607504852275,
+      "grad_norm": 0.1936737447977066,
+      "learning_rate": 4.8041706627413676e-05,
+      "loss": 2.1537,
+      "step": 4311
+    },
+    {
+      "epoch": 1.8600388182014234,
+      "grad_norm": 0.17519107460975647,
+      "learning_rate": 4.800986234221041e-05,
+      "loss": 2.1312,
+      "step": 4312
+    },
+    {
+      "epoch": 1.8604701315505716,
+      "grad_norm": 0.1902090162038803,
+      "learning_rate": 4.797802364561033e-05,
+      "loss": 2.3116,
+      "step": 4313
+    },
+    {
+      "epoch": 1.8609014448997196,
+      "grad_norm": 0.1804785579442978,
+      "learning_rate": 4.794619054420597e-05,
+      "loss": 2.0768,
+      "step": 4314
+    },
+    {
+      "epoch": 1.8613327582488677,
+      "grad_norm": 0.19297802448272705,
+      "learning_rate": 4.7914363044588684e-05,
+      "loss": 1.7029,
+      "step": 4315
+    },
+    {
+      "epoch": 1.861764071598016,
+      "grad_norm": 0.19850334525108337,
+      "learning_rate": 4.788254115334876e-05,
+      "loss": 2.0629,
+      "step": 4316
+    },
+    {
+      "epoch": 1.8621953849471642,
+      "grad_norm": 0.2027156800031662,
+      "learning_rate": 4.785072487707525e-05,
+      "loss": 2.1974,
+      "step": 4317
+    },
+    {
+      "epoch": 1.8626266982963122,
+      "grad_norm": 0.2021334320306778,
+      "learning_rate": 4.7818914222356044e-05,
+      "loss": 2.285,
+      "step": 4318
+    },
+    {
+      "epoch": 1.8630580116454605,
+      "grad_norm": 0.1869433969259262,
+      "learning_rate": 4.7787109195777876e-05,
+      "loss": 2.2128,
+      "step": 4319
+    },
+    {
+      "epoch": 1.8634893249946085,
+      "grad_norm": 0.1924419105052948,
+      "learning_rate": 4.775530980392636e-05,
+      "loss": 2.1366,
+      "step": 4320
+    },
+    {
+      "epoch": 1.8639206383437568,
+      "grad_norm": 0.17339825630187988,
+      "learning_rate": 4.772351605338589e-05,
+      "loss": 2.0719,
+      "step": 4321
+    },
+    {
+      "epoch": 1.864351951692905,
+      "grad_norm": 0.1681831032037735,
+      "learning_rate": 4.76917279507397e-05,
+      "loss": 1.7802,
+      "step": 4322
+    },
+    {
+      "epoch": 1.864783265042053,
+      "grad_norm": 0.19009758532047272,
+      "learning_rate": 4.765994550256989e-05,
+      "loss": 2.1927,
+      "step": 4323
+    },
+    {
+      "epoch": 1.865214578391201,
+      "grad_norm": 0.17810070514678955,
+      "learning_rate": 4.762816871545736e-05,
+      "loss": 2.1567,
+      "step": 4324
+    },
+    {
+      "epoch": 1.8656458917403493,
+      "grad_norm": 0.17919707298278809,
+      "learning_rate": 4.759639759598183e-05,
+      "loss": 2.0385,
+      "step": 4325
+    },
+    {
+      "epoch": 1.8656458917403493,
+      "eval_loss": 2.0903007984161377,
+      "eval_runtime": 206.9154,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 4325
+    },
+    {
+      "epoch": 1.8660772050894976,
+      "grad_norm": 0.21091590821743011,
+      "learning_rate": 4.756463215072183e-05,
+      "loss": 2.3909,
+      "step": 4326
+    },
+    {
+      "epoch": 1.8665085184386456,
+      "grad_norm": 0.20307664573192596,
+      "learning_rate": 4.7532872386254786e-05,
+      "loss": 2.2274,
+      "step": 4327
+    },
+    {
+      "epoch": 1.8669398317877939,
+      "grad_norm": 0.1939563900232315,
+      "learning_rate": 4.7501118309156896e-05,
+      "loss": 2.103,
+      "step": 4328
+    },
+    {
+      "epoch": 1.867371145136942,
+      "grad_norm": 0.1929791271686554,
+      "learning_rate": 4.746936992600317e-05,
+      "loss": 2.2962,
+      "step": 4329
+    },
+    {
+      "epoch": 1.8678024584860902,
+      "grad_norm": 0.17812767624855042,
+      "learning_rate": 4.7437627243367475e-05,
+      "loss": 2.1675,
+      "step": 4330
+    },
+    {
+      "epoch": 1.8682337718352384,
+      "grad_norm": 0.18598608672618866,
+      "learning_rate": 4.74058902678225e-05,
+      "loss": 2.042,
+      "step": 4331
+    },
+    {
+      "epoch": 1.8686650851843865,
+      "grad_norm": 0.19546982645988464,
+      "learning_rate": 4.737415900593969e-05,
+      "loss": 2.2392,
+      "step": 4332
+    },
+    {
+      "epoch": 1.8690963985335345,
+      "grad_norm": 0.16199365258216858,
+      "learning_rate": 4.734243346428935e-05,
+      "loss": 2.0524,
+      "step": 4333
+    },
+    {
+      "epoch": 1.8695277118826827,
+      "grad_norm": 0.18136630952358246,
+      "learning_rate": 4.731071364944064e-05,
+      "loss": 2.1181,
+      "step": 4334
+    },
+    {
+      "epoch": 1.869959025231831,
+      "grad_norm": 0.20470373332500458,
+      "learning_rate": 4.727899956796148e-05,
+      "loss": 2.237,
+      "step": 4335
+    },
+    {
+      "epoch": 1.870390338580979,
+      "grad_norm": 0.21079720556735992,
+      "learning_rate": 4.724729122641859e-05,
+      "loss": 1.9939,
+      "step": 4336
+    },
+    {
+      "epoch": 1.8708216519301273,
+      "grad_norm": 0.19446231424808502,
+      "learning_rate": 4.721558863137756e-05,
+      "loss": 2.2238,
+      "step": 4337
+    },
+    {
+      "epoch": 1.8712529652792753,
+      "grad_norm": 0.19449461996555328,
+      "learning_rate": 4.718389178940277e-05,
+      "loss": 2.1381,
+      "step": 4338
+    },
+    {
+      "epoch": 1.8716842786284236,
+      "grad_norm": 0.18257984519004822,
+      "learning_rate": 4.715220070705737e-05,
+      "loss": 2.2377,
+      "step": 4339
+    },
+    {
+      "epoch": 1.8721155919775718,
+      "grad_norm": 0.18961714208126068,
+      "learning_rate": 4.7120515390903325e-05,
+      "loss": 2.245,
+      "step": 4340
+    },
+    {
+      "epoch": 1.8725469053267199,
+      "grad_norm": 0.20834636688232422,
+      "learning_rate": 4.708883584750147e-05,
+      "loss": 2.1548,
+      "step": 4341
+    },
+    {
+      "epoch": 1.872978218675868,
+      "grad_norm": 0.20628146827220917,
+      "learning_rate": 4.705716208341138e-05,
+      "loss": 2.2773,
+      "step": 4342
+    },
+    {
+      "epoch": 1.8734095320250161,
+      "grad_norm": 0.19520138204097748,
+      "learning_rate": 4.7025494105191486e-05,
+      "loss": 2.1877,
+      "step": 4343
+    },
+    {
+      "epoch": 1.8738408453741644,
+      "grad_norm": 0.20871102809906006,
+      "learning_rate": 4.6993831919398925e-05,
+      "loss": 2.2529,
+      "step": 4344
+    },
+    {
+      "epoch": 1.8742721587233124,
+      "grad_norm": 0.19515718519687653,
+      "learning_rate": 4.696217553258975e-05,
+      "loss": 2.2391,
+      "step": 4345
+    },
+    {
+      "epoch": 1.8747034720724607,
+      "grad_norm": 0.1878291219472885,
+      "learning_rate": 4.6930524951318735e-05,
+      "loss": 2.1282,
+      "step": 4346
+    },
+    {
+      "epoch": 1.8751347854216087,
+      "grad_norm": 0.19603882730007172,
+      "learning_rate": 4.68988801821395e-05,
+      "loss": 2.1458,
+      "step": 4347
+    },
+    {
+      "epoch": 1.875566098770757,
+      "grad_norm": 0.21738454699516296,
+      "learning_rate": 4.6867241231604425e-05,
+      "loss": 2.2133,
+      "step": 4348
+    },
+    {
+      "epoch": 1.8759974121199052,
+      "grad_norm": 0.22887803614139557,
+      "learning_rate": 4.683560810626473e-05,
+      "loss": 2.1376,
+      "step": 4349
+    },
+    {
+      "epoch": 1.8764287254690533,
+      "grad_norm": 0.18480657041072845,
+      "learning_rate": 4.680398081267037e-05,
+      "loss": 1.9866,
+      "step": 4350
+    },
+    {
+      "epoch": 1.8764287254690533,
+      "eval_loss": 2.0904784202575684,
+      "eval_runtime": 205.3133,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 4350
+    },
+    {
+      "epoch": 1.8768600388182013,
+      "grad_norm": 0.1917465478181839,
+      "learning_rate": 4.677235935737011e-05,
+      "loss": 2.1556,
+      "step": 4351
+    },
+    {
+      "epoch": 1.8772913521673495,
+      "grad_norm": 0.1883143186569214,
+      "learning_rate": 4.6740743746911554e-05,
+      "loss": 2.0558,
+      "step": 4352
+    },
+    {
+      "epoch": 1.8777226655164978,
+      "grad_norm": 0.19136305153369904,
+      "learning_rate": 4.670913398784104e-05,
+      "loss": 2.3204,
+      "step": 4353
+    },
+    {
+      "epoch": 1.878153978865646,
+      "grad_norm": 0.1742907613515854,
+      "learning_rate": 4.66775300867037e-05,
+      "loss": 2.0352,
+      "step": 4354
+    },
+    {
+      "epoch": 1.878585292214794,
+      "grad_norm": 0.18656355142593384,
+      "learning_rate": 4.66459320500435e-05,
+      "loss": 2.1635,
+      "step": 4355
+    },
+    {
+      "epoch": 1.8790166055639421,
+      "grad_norm": 0.1927529126405716,
+      "learning_rate": 4.6614339884403165e-05,
+      "loss": 2.1339,
+      "step": 4356
+    },
+    {
+      "epoch": 1.8794479189130904,
+      "grad_norm": 0.17686046659946442,
+      "learning_rate": 4.658275359632416e-05,
+      "loss": 2.1561,
+      "step": 4357
+    },
+    {
+      "epoch": 1.8798792322622386,
+      "grad_norm": 0.20081140100955963,
+      "learning_rate": 4.655117319234677e-05,
+      "loss": 2.1218,
+      "step": 4358
+    },
+    {
+      "epoch": 1.8803105456113867,
+      "grad_norm": 0.20209652185440063,
+      "learning_rate": 4.65195986790101e-05,
+      "loss": 2.2096,
+      "step": 4359
+    },
+    {
+      "epoch": 1.8807418589605347,
+      "grad_norm": 0.1781439483165741,
+      "learning_rate": 4.648803006285198e-05,
+      "loss": 2.1468,
+      "step": 4360
+    },
+    {
+      "epoch": 1.881173172309683,
+      "grad_norm": 0.2034546285867691,
+      "learning_rate": 4.645646735040902e-05,
+      "loss": 2.2163,
+      "step": 4361
+    },
+    {
+      "epoch": 1.8816044856588312,
+      "grad_norm": 0.18482056260108948,
+      "learning_rate": 4.642491054821667e-05,
+      "loss": 2.1812,
+      "step": 4362
+    },
+    {
+      "epoch": 1.8820357990079795,
+      "grad_norm": 0.2234092801809311,
+      "learning_rate": 4.639335966280908e-05,
+      "loss": 1.8202,
+      "step": 4363
+    },
+    {
+      "epoch": 1.8824671123571275,
+      "grad_norm": 0.24373753368854523,
+      "learning_rate": 4.63618147007192e-05,
+      "loss": 2.0694,
+      "step": 4364
+    },
+    {
+      "epoch": 1.8828984257062755,
+      "grad_norm": 0.20102065801620483,
+      "learning_rate": 4.633027566847875e-05,
+      "loss": 2.048,
+      "step": 4365
+    },
+    {
+      "epoch": 1.8833297390554238,
+      "grad_norm": 0.19503894448280334,
+      "learning_rate": 4.6298742572618266e-05,
+      "loss": 2.228,
+      "step": 4366
+    },
+    {
+      "epoch": 1.883761052404572,
+      "grad_norm": 0.18882206082344055,
+      "learning_rate": 4.626721541966701e-05,
+      "loss": 2.1524,
+      "step": 4367
+    },
+    {
+      "epoch": 1.88419236575372,
+      "grad_norm": 0.20258937776088715,
+      "learning_rate": 4.6235694216152995e-05,
+      "loss": 2.344,
+      "step": 4368
+    },
+    {
+      "epoch": 1.884623679102868,
+      "grad_norm": 0.1838570535182953,
+      "learning_rate": 4.620417896860307e-05,
+      "loss": 2.1242,
+      "step": 4369
+    },
+    {
+      "epoch": 1.8850549924520164,
+      "grad_norm": 0.1915067583322525,
+      "learning_rate": 4.6172669683542816e-05,
+      "loss": 2.2123,
+      "step": 4370
+    },
+    {
+      "epoch": 1.8854863058011646,
+      "grad_norm": 0.20054484903812408,
+      "learning_rate": 4.614116636749654e-05,
+      "loss": 2.2716,
+      "step": 4371
+    },
+    {
+      "epoch": 1.8859176191503129,
+      "grad_norm": 0.1780989170074463,
+      "learning_rate": 4.6109669026987336e-05,
+      "loss": 2.0959,
+      "step": 4372
+    },
+    {
+      "epoch": 1.886348932499461,
+      "grad_norm": 0.19114157557487488,
+      "learning_rate": 4.607817766853712e-05,
+      "loss": 2.2411,
+      "step": 4373
+    },
+    {
+      "epoch": 1.886780245848609,
+      "grad_norm": 0.20129813253879547,
+      "learning_rate": 4.6046692298666506e-05,
+      "loss": 2.1702,
+      "step": 4374
+    },
+    {
+      "epoch": 1.8872115591977572,
+      "grad_norm": 0.18966692686080933,
+      "learning_rate": 4.6015212923894864e-05,
+      "loss": 2.175,
+      "step": 4375
+    },
+    {
+      "epoch": 1.8872115591977572,
+      "eval_loss": 2.0902810096740723,
+      "eval_runtime": 206.0401,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 4375
+    },
+    {
+      "epoch": 1.8876428725469054,
+      "grad_norm": 0.18641971051692963,
+      "learning_rate": 4.598373955074037e-05,
+      "loss": 1.9328,
+      "step": 4376
+    },
+    {
+      "epoch": 1.8880741858960535,
+      "grad_norm": 0.17950253188610077,
+      "learning_rate": 4.595227218571994e-05,
+      "loss": 2.2209,
+      "step": 4377
+    },
+    {
+      "epoch": 1.8885054992452015,
+      "grad_norm": 0.18584321439266205,
+      "learning_rate": 4.592081083534919e-05,
+      "loss": 2.1834,
+      "step": 4378
+    },
+    {
+      "epoch": 1.8889368125943498,
+      "grad_norm": 0.18755564093589783,
+      "learning_rate": 4.588935550614255e-05,
+      "loss": 2.0961,
+      "step": 4379
+    },
+    {
+      "epoch": 1.889368125943498,
+      "grad_norm": 0.19564920663833618,
+      "learning_rate": 4.585790620461322e-05,
+      "loss": 2.3325,
+      "step": 4380
+    },
+    {
+      "epoch": 1.8897994392926463,
+      "grad_norm": 0.1829662322998047,
+      "learning_rate": 4.58264629372731e-05,
+      "loss": 2.0914,
+      "step": 4381
+    },
+    {
+      "epoch": 1.8902307526417943,
+      "grad_norm": 0.19613660871982574,
+      "learning_rate": 4.579502571063284e-05,
+      "loss": 2.2863,
+      "step": 4382
+    },
+    {
+      "epoch": 1.8906620659909423,
+      "grad_norm": 0.17947323620319366,
+      "learning_rate": 4.5763594531201914e-05,
+      "loss": 2.162,
+      "step": 4383
+    },
+    {
+      "epoch": 1.8910933793400906,
+      "grad_norm": 0.18751266598701477,
+      "learning_rate": 4.573216940548847e-05,
+      "loss": 1.9898,
+      "step": 4384
+    },
+    {
+      "epoch": 1.8915246926892388,
+      "grad_norm": 0.18965020775794983,
+      "learning_rate": 4.5700750339999416e-05,
+      "loss": 2.1875,
+      "step": 4385
+    },
+    {
+      "epoch": 1.8919560060383869,
+      "grad_norm": 0.18930767476558685,
+      "learning_rate": 4.56693373412404e-05,
+      "loss": 2.2875,
+      "step": 4386
+    },
+    {
+      "epoch": 1.892387319387535,
+      "grad_norm": 0.1917998045682907,
+      "learning_rate": 4.563793041571585e-05,
+      "loss": 2.2625,
+      "step": 4387
+    },
+    {
+      "epoch": 1.8928186327366832,
+      "grad_norm": 0.19999705255031586,
+      "learning_rate": 4.56065295699289e-05,
+      "loss": 2.2053,
+      "step": 4388
+    },
+    {
+      "epoch": 1.8932499460858314,
+      "grad_norm": 0.1889103204011917,
+      "learning_rate": 4.557513481038144e-05,
+      "loss": 2.023,
+      "step": 4389
+    },
+    {
+      "epoch": 1.8936812594349797,
+      "grad_norm": 0.19728577136993408,
+      "learning_rate": 4.5543746143574126e-05,
+      "loss": 2.0746,
+      "step": 4390
+    },
+    {
+      "epoch": 1.8941125727841277,
+      "grad_norm": 0.18072137236595154,
+      "learning_rate": 4.55123635760063e-05,
+      "loss": 1.9347,
+      "step": 4391
+    },
+    {
+      "epoch": 1.8945438861332757,
+      "grad_norm": 0.24109086394309998,
+      "learning_rate": 4.548098711417605e-05,
+      "loss": 2.2294,
+      "step": 4392
+    },
+    {
+      "epoch": 1.894975199482424,
+      "grad_norm": 0.20103009045124054,
+      "learning_rate": 4.544961676458022e-05,
+      "loss": 2.0669,
+      "step": 4393
+    },
+    {
+      "epoch": 1.8954065128315722,
+      "grad_norm": 0.21188940107822418,
+      "learning_rate": 4.541825253371442e-05,
+      "loss": 2.0499,
+      "step": 4394
+    },
+    {
+      "epoch": 1.8958378261807203,
+      "grad_norm": 0.2471272051334381,
+      "learning_rate": 4.538689442807294e-05,
+      "loss": 2.0948,
+      "step": 4395
+    },
+    {
+      "epoch": 1.8962691395298683,
+      "grad_norm": 0.18075866997241974,
+      "learning_rate": 4.5355542454148795e-05,
+      "loss": 2.2735,
+      "step": 4396
+    },
+    {
+      "epoch": 1.8967004528790166,
+      "grad_norm": 0.22322553396224976,
+      "learning_rate": 4.532419661843376e-05,
+      "loss": 2.3035,
+      "step": 4397
+    },
+    {
+      "epoch": 1.8971317662281648,
+      "grad_norm": 0.20870225131511688,
+      "learning_rate": 4.529285692741836e-05,
+      "loss": 2.199,
+      "step": 4398
+    },
+    {
+      "epoch": 1.897563079577313,
+      "grad_norm": 0.18207068741321564,
+      "learning_rate": 4.5261523387591804e-05,
+      "loss": 2.1874,
+      "step": 4399
+    },
+    {
+      "epoch": 1.897994392926461,
+      "grad_norm": 0.17733348906040192,
+      "learning_rate": 4.523019600544203e-05,
+      "loss": 2.2675,
+      "step": 4400
+    },
+    {
+      "epoch": 1.897994392926461,
+      "eval_loss": 2.0905203819274902,
+      "eval_runtime": 205.3714,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 4400
+    },
+    {
+      "epoch": 1.8984257062756091,
+      "grad_norm": 0.1851731687784195,
+      "learning_rate": 4.519887478745575e-05,
+      "loss": 2.0445,
+      "step": 4401
+    },
+    {
+      "epoch": 1.8988570196247574,
+      "grad_norm": 0.1968875527381897,
+      "learning_rate": 4.5167559740118354e-05,
+      "loss": 2.1068,
+      "step": 4402
+    },
+    {
+      "epoch": 1.8992883329739056,
+      "grad_norm": 0.19119924306869507,
+      "learning_rate": 4.5136250869913936e-05,
+      "loss": 2.074,
+      "step": 4403
+    },
+    {
+      "epoch": 1.8997196463230537,
+      "grad_norm": 0.20993638038635254,
+      "learning_rate": 4.5104948183325354e-05,
+      "loss": 2.2871,
+      "step": 4404
+    },
+    {
+      "epoch": 1.9001509596722017,
+      "grad_norm": 0.17726275324821472,
+      "learning_rate": 4.5073651686834197e-05,
+      "loss": 2.2556,
+      "step": 4405
+    },
+    {
+      "epoch": 1.90058227302135,
+      "grad_norm": 0.1941853165626526,
+      "learning_rate": 4.5042361386920714e-05,
+      "loss": 2.4722,
+      "step": 4406
+    },
+    {
+      "epoch": 1.9010135863704982,
+      "grad_norm": 0.19786939024925232,
+      "learning_rate": 4.501107729006391e-05,
+      "loss": 2.3993,
+      "step": 4407
+    },
+    {
+      "epoch": 1.9014448997196465,
+      "grad_norm": 0.21278390288352966,
+      "learning_rate": 4.4979799402741517e-05,
+      "loss": 1.9771,
+      "step": 4408
+    },
+    {
+      "epoch": 1.9018762130687945,
+      "grad_norm": 0.17766079306602478,
+      "learning_rate": 4.494852773142998e-05,
+      "loss": 2.1706,
+      "step": 4409
+    },
+    {
+      "epoch": 1.9023075264179425,
+      "grad_norm": 0.1998787224292755,
+      "learning_rate": 4.4917262282604385e-05,
+      "loss": 2.093,
+      "step": 4410
+    },
+    {
+      "epoch": 1.9027388397670908,
+      "grad_norm": 0.18379029631614685,
+      "learning_rate": 4.4886003062738604e-05,
+      "loss": 2.2796,
+      "step": 4411
+    },
+    {
+      "epoch": 1.903170153116239,
+      "grad_norm": 0.19698943197727203,
+      "learning_rate": 4.485475007830523e-05,
+      "loss": 2.2868,
+      "step": 4412
+    },
+    {
+      "epoch": 1.903601466465387,
+      "grad_norm": 0.18271850049495697,
+      "learning_rate": 4.48235033357755e-05,
+      "loss": 2.1838,
+      "step": 4413
+    },
+    {
+      "epoch": 1.9040327798145351,
+      "grad_norm": 0.19257652759552002,
+      "learning_rate": 4.479226284161941e-05,
+      "loss": 2.2262,
+      "step": 4414
+    },
+    {
+      "epoch": 1.9044640931636834,
+      "grad_norm": 0.18733306229114532,
+      "learning_rate": 4.476102860230565e-05,
+      "loss": 2.0258,
+      "step": 4415
+    },
+    {
+      "epoch": 1.9048954065128316,
+      "grad_norm": 0.19409288465976715,
+      "learning_rate": 4.472980062430162e-05,
+      "loss": 2.2714,
+      "step": 4416
+    },
+    {
+      "epoch": 1.9053267198619799,
+      "grad_norm": 0.19599054753780365,
+      "learning_rate": 4.4698578914073395e-05,
+      "loss": 2.1417,
+      "step": 4417
+    },
+    {
+      "epoch": 1.905758033211128,
+      "grad_norm": 0.18327564001083374,
+      "learning_rate": 4.466736347808575e-05,
+      "loss": 2.2224,
+      "step": 4418
+    },
+    {
+      "epoch": 1.906189346560276,
+      "grad_norm": 0.2031586766242981,
+      "learning_rate": 4.4636154322802236e-05,
+      "loss": 2.3707,
+      "step": 4419
+    },
+    {
+      "epoch": 1.9066206599094242,
+      "grad_norm": 0.21735779941082,
+      "learning_rate": 4.460495145468501e-05,
+      "loss": 2.0034,
+      "step": 4420
+    },
+    {
+      "epoch": 1.9070519732585725,
+      "grad_norm": 0.2004113644361496,
+      "learning_rate": 4.4573754880194976e-05,
+      "loss": 2.3207,
+      "step": 4421
+    },
+    {
+      "epoch": 1.9074832866077205,
+      "grad_norm": 0.19796982407569885,
+      "learning_rate": 4.4542564605791734e-05,
+      "loss": 1.9953,
+      "step": 4422
+    },
+    {
+      "epoch": 1.9079145999568685,
+      "grad_norm": 0.1805429458618164,
+      "learning_rate": 4.4511380637933595e-05,
+      "loss": 2.1355,
+      "step": 4423
+    },
+    {
+      "epoch": 1.9083459133060168,
+      "grad_norm": 0.17156881093978882,
+      "learning_rate": 4.448020298307749e-05,
+      "loss": 2.0615,
+      "step": 4424
+    },
+    {
+      "epoch": 1.908777226655165,
+      "grad_norm": 0.19201727211475372,
+      "learning_rate": 4.44490316476791e-05,
+      "loss": 2.2494,
+      "step": 4425
+    },
+    {
+      "epoch": 1.908777226655165,
+      "eval_loss": 2.0903611183166504,
+      "eval_runtime": 202.0722,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 4425
+    },
+    {
+      "epoch": 1.9092085400043133,
+      "grad_norm": 0.2967647910118103,
+      "learning_rate": 4.441786663819282e-05,
+      "loss": 2.3066,
+      "step": 4426
+    },
+    {
+      "epoch": 1.9096398533534613,
+      "grad_norm": 0.19616886973381042,
+      "learning_rate": 4.43867079610717e-05,
+      "loss": 2.0931,
+      "step": 4427
+    },
+    {
+      "epoch": 1.9100711667026093,
+      "grad_norm": 0.2792205810546875,
+      "learning_rate": 4.435555562276746e-05,
+      "loss": 2.0895,
+      "step": 4428
+    },
+    {
+      "epoch": 1.9105024800517576,
+      "grad_norm": 0.2176106572151184,
+      "learning_rate": 4.432440962973059e-05,
+      "loss": 2.2637,
+      "step": 4429
+    },
+    {
+      "epoch": 1.9109337934009059,
+      "grad_norm": 0.19296325743198395,
+      "learning_rate": 4.429326998841014e-05,
+      "loss": 2.2174,
+      "step": 4430
+    },
+    {
+      "epoch": 1.9113651067500539,
+      "grad_norm": 0.18656092882156372,
+      "learning_rate": 4.4262136705253946e-05,
+      "loss": 2.0757,
+      "step": 4431
+    },
+    {
+      "epoch": 1.911796420099202,
+      "grad_norm": 0.19575899839401245,
+      "learning_rate": 4.423100978670847e-05,
+      "loss": 2.1884,
+      "step": 4432
+    },
+    {
+      "epoch": 1.9122277334483502,
+      "grad_norm": 0.20120398700237274,
+      "learning_rate": 4.419988923921893e-05,
+      "loss": 2.3305,
+      "step": 4433
+    },
+    {
+      "epoch": 1.9126590467974984,
+      "grad_norm": 0.1956479698419571,
+      "learning_rate": 4.4168775069229156e-05,
+      "loss": 2.154,
+      "step": 4434
+    },
+    {
+      "epoch": 1.9130903601466467,
+      "grad_norm": 0.19334623217582703,
+      "learning_rate": 4.413766728318162e-05,
+      "loss": 2.1075,
+      "step": 4435
+    },
+    {
+      "epoch": 1.9135216734957947,
+      "grad_norm": 0.19544517993927002,
+      "learning_rate": 4.410656588751762e-05,
+      "loss": 2.1744,
+      "step": 4436
+    },
+    {
+      "epoch": 1.9139529868449427,
+      "grad_norm": 0.20631077885627747,
+      "learning_rate": 4.407547088867699e-05,
+      "loss": 2.272,
+      "step": 4437
+    },
+    {
+      "epoch": 1.914384300194091,
+      "grad_norm": 0.18286816775798798,
+      "learning_rate": 4.4044382293098294e-05,
+      "loss": 2.2174,
+      "step": 4438
+    },
+    {
+      "epoch": 1.9148156135432393,
+      "grad_norm": 0.20003879070281982,
+      "learning_rate": 4.401330010721876e-05,
+      "loss": 2.1827,
+      "step": 4439
+    },
+    {
+      "epoch": 1.9152469268923873,
+      "grad_norm": 0.20353899896144867,
+      "learning_rate": 4.398222433747432e-05,
+      "loss": 2.2601,
+      "step": 4440
+    },
+    {
+      "epoch": 1.9156782402415353,
+      "grad_norm": 0.1923852264881134,
+      "learning_rate": 4.395115499029954e-05,
+      "loss": 2.0675,
+      "step": 4441
+    },
+    {
+      "epoch": 1.9161095535906836,
+      "grad_norm": 0.20454512536525726,
+      "learning_rate": 4.392009207212766e-05,
+      "loss": 2.289,
+      "step": 4442
+    },
+    {
+      "epoch": 1.9165408669398318,
+      "grad_norm": 0.20002296566963196,
+      "learning_rate": 4.388903558939057e-05,
+      "loss": 2.2905,
+      "step": 4443
+    },
+    {
+      "epoch": 1.91697218028898,
+      "grad_norm": 0.2085457295179367,
+      "learning_rate": 4.385798554851891e-05,
+      "loss": 2.2252,
+      "step": 4444
+    },
+    {
+      "epoch": 1.9174034936381281,
+      "grad_norm": 0.18940648436546326,
+      "learning_rate": 4.3826941955941914e-05,
+      "loss": 2.1704,
+      "step": 4445
+    },
+    {
+      "epoch": 1.9178348069872762,
+      "grad_norm": 0.20197077095508575,
+      "learning_rate": 4.3795904818087456e-05,
+      "loss": 2.2808,
+      "step": 4446
+    },
+    {
+      "epoch": 1.9182661203364244,
+      "grad_norm": 0.1914563924074173,
+      "learning_rate": 4.376487414138219e-05,
+      "loss": 2.1631,
+      "step": 4447
+    },
+    {
+      "epoch": 1.9186974336855727,
+      "grad_norm": 0.19904504716396332,
+      "learning_rate": 4.373384993225129e-05,
+      "loss": 2.0973,
+      "step": 4448
+    },
+    {
+      "epoch": 1.9191287470347207,
+      "grad_norm": 0.1938726156949997,
+      "learning_rate": 4.3702832197118694e-05,
+      "loss": 2.1442,
+      "step": 4449
+    },
+    {
+      "epoch": 1.9195600603838687,
+      "grad_norm": 0.176131471991539,
+      "learning_rate": 4.367182094240691e-05,
+      "loss": 2.3222,
+      "step": 4450
+    },
+    {
+      "epoch": 1.9195600603838687,
+      "eval_loss": 2.089779853820801,
+      "eval_runtime": 196.8906,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 4450
+    },
+    {
+      "epoch": 1.919991373733017,
+      "grad_norm": 0.2034505158662796,
+      "learning_rate": 4.364081617453727e-05,
+      "loss": 2.1326,
+      "step": 4451
+    },
+    {
+      "epoch": 1.9204226870821652,
+      "grad_norm": 0.20447219908237457,
+      "learning_rate": 4.360981789992955e-05,
+      "loss": 2.2005,
+      "step": 4452
+    },
+    {
+      "epoch": 1.9208540004313135,
+      "grad_norm": 0.19651800394058228,
+      "learning_rate": 4.357882612500227e-05,
+      "loss": 2.1302,
+      "step": 4453
+    },
+    {
+      "epoch": 1.9212853137804615,
+      "grad_norm": 0.1718902885913849,
+      "learning_rate": 4.354784085617269e-05,
+      "loss": 2.0705,
+      "step": 4454
+    },
+    {
+      "epoch": 1.9217166271296096,
+      "grad_norm": 0.18895719945430756,
+      "learning_rate": 4.3516862099856625e-05,
+      "loss": 2.1459,
+      "step": 4455
+    },
+    {
+      "epoch": 1.9221479404787578,
+      "grad_norm": 0.20136317610740662,
+      "learning_rate": 4.348588986246857e-05,
+      "loss": 2.432,
+      "step": 4456
+    },
+    {
+      "epoch": 1.922579253827906,
+      "grad_norm": 0.21606945991516113,
+      "learning_rate": 4.345492415042161e-05,
+      "loss": 2.1322,
+      "step": 4457
+    },
+    {
+      "epoch": 1.923010567177054,
+      "grad_norm": 0.2078695446252823,
+      "learning_rate": 4.3423964970127595e-05,
+      "loss": 2.0868,
+      "step": 4458
+    },
+    {
+      "epoch": 1.9234418805262021,
+      "grad_norm": 0.1945011019706726,
+      "learning_rate": 4.339301232799694e-05,
+      "loss": 2.2264,
+      "step": 4459
+    },
+    {
+      "epoch": 1.9238731938753504,
+      "grad_norm": 0.18762318789958954,
+      "learning_rate": 4.336206623043873e-05,
+      "loss": 2.0857,
+      "step": 4460
+    },
+    {
+      "epoch": 1.9243045072244986,
+      "grad_norm": 0.21926428377628326,
+      "learning_rate": 4.3331126683860704e-05,
+      "loss": 2.0618,
+      "step": 4461
+    },
+    {
+      "epoch": 1.924735820573647,
+      "grad_norm": 0.20801591873168945,
+      "learning_rate": 4.330019369466921e-05,
+      "loss": 2.3102,
+      "step": 4462
+    },
+    {
+      "epoch": 1.925167133922795,
+      "grad_norm": 0.20444561541080475,
+      "learning_rate": 4.3269267269269286e-05,
+      "loss": 2.0529,
+      "step": 4463
+    },
+    {
+      "epoch": 1.925598447271943,
+      "grad_norm": 0.19824370741844177,
+      "learning_rate": 4.323834741406454e-05,
+      "loss": 2.2004,
+      "step": 4464
+    },
+    {
+      "epoch": 1.9260297606210912,
+      "grad_norm": 0.21752721071243286,
+      "learning_rate": 4.320743413545736e-05,
+      "loss": 2.2265,
+      "step": 4465
+    },
+    {
+      "epoch": 1.9264610739702395,
+      "grad_norm": 0.21213920414447784,
+      "learning_rate": 4.31765274398486e-05,
+      "loss": 2.2037,
+      "step": 4466
+    },
+    {
+      "epoch": 1.9268923873193875,
+      "grad_norm": 0.1710169017314911,
+      "learning_rate": 4.314562733363781e-05,
+      "loss": 2.1363,
+      "step": 4467
+    },
+    {
+      "epoch": 1.9273237006685355,
+      "grad_norm": 0.19569765031337738,
+      "learning_rate": 4.3114733823223264e-05,
+      "loss": 2.1503,
+      "step": 4468
+    },
+    {
+      "epoch": 1.9277550140176838,
+      "grad_norm": 0.17711645364761353,
+      "learning_rate": 4.308384691500178e-05,
+      "loss": 1.9612,
+      "step": 4469
+    },
+    {
+      "epoch": 1.928186327366832,
+      "grad_norm": 0.21835118532180786,
+      "learning_rate": 4.305296661536884e-05,
+      "loss": 1.9699,
+      "step": 4470
+    },
+    {
+      "epoch": 1.9286176407159803,
+      "grad_norm": 0.20168553292751312,
+      "learning_rate": 4.302209293071847e-05,
+      "loss": 2.1138,
+      "step": 4471
+    },
+    {
+      "epoch": 1.9290489540651283,
+      "grad_norm": 0.18217918276786804,
+      "learning_rate": 4.29912258674435e-05,
+      "loss": 2.1849,
+      "step": 4472
+    },
+    {
+      "epoch": 1.9294802674142764,
+      "grad_norm": 0.18976938724517822,
+      "learning_rate": 4.2960365431935257e-05,
+      "loss": 2.1309,
+      "step": 4473
+    },
+    {
+      "epoch": 1.9299115807634246,
+      "grad_norm": 0.1749623417854309,
+      "learning_rate": 4.2929511630583726e-05,
+      "loss": 1.9406,
+      "step": 4474
+    },
+    {
+      "epoch": 1.9303428941125729,
+      "grad_norm": 0.18413615226745605,
+      "learning_rate": 4.289866446977752e-05,
+      "loss": 2.1305,
+      "step": 4475
+    },
+    {
+      "epoch": 1.9303428941125729,
+      "eval_loss": 2.0894153118133545,
+      "eval_runtime": 196.8798,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 4475
+    },
+    {
+      "epoch": 1.930774207461721,
+      "grad_norm": 0.18336045742034912,
+      "learning_rate": 4.286782395590391e-05,
+      "loss": 2.1981,
+      "step": 4476
+    },
+    {
+      "epoch": 1.931205520810869,
+      "grad_norm": 0.17901524901390076,
+      "learning_rate": 4.283699009534872e-05,
+      "loss": 2.0308,
+      "step": 4477
+    },
+    {
+      "epoch": 1.9316368341600172,
+      "grad_norm": 0.17348520457744598,
+      "learning_rate": 4.2806162894496445e-05,
+      "loss": 2.0078,
+      "step": 4478
+    },
+    {
+      "epoch": 1.9320681475091654,
+      "grad_norm": 0.2009512186050415,
+      "learning_rate": 4.277534235973026e-05,
+      "loss": 1.9249,
+      "step": 4479
+    },
+    {
+      "epoch": 1.9324994608583137,
+      "grad_norm": 0.19998937845230103,
+      "learning_rate": 4.274452849743182e-05,
+      "loss": 2.1278,
+      "step": 4480
+    },
+    {
+      "epoch": 1.9329307742074617,
+      "grad_norm": 0.17956745624542236,
+      "learning_rate": 4.271372131398147e-05,
+      "loss": 1.9026,
+      "step": 4481
+    },
+    {
+      "epoch": 1.9333620875566098,
+      "grad_norm": 0.2049476057291031,
+      "learning_rate": 4.268292081575821e-05,
+      "loss": 2.2686,
+      "step": 4482
+    },
+    {
+      "epoch": 1.933793400905758,
+      "grad_norm": 0.18582724034786224,
+      "learning_rate": 4.265212700913962e-05,
+      "loss": 2.2011,
+      "step": 4483
+    },
+    {
+      "epoch": 1.9342247142549063,
+      "grad_norm": 0.18372566998004913,
+      "learning_rate": 4.2621339900501914e-05,
+      "loss": 2.0969,
+      "step": 4484
+    },
+    {
+      "epoch": 1.9346560276040543,
+      "grad_norm": 0.20212997496128082,
+      "learning_rate": 4.2590559496219805e-05,
+      "loss": 2.1321,
+      "step": 4485
+    },
+    {
+      "epoch": 1.9350873409532023,
+      "grad_norm": 0.1954246163368225,
+      "learning_rate": 4.25597858026668e-05,
+      "loss": 2.2599,
+      "step": 4486
+    },
+    {
+      "epoch": 1.9355186543023506,
+      "grad_norm": 0.22167876362800598,
+      "learning_rate": 4.2529018826214925e-05,
+      "loss": 2.2438,
+      "step": 4487
+    },
+    {
+      "epoch": 1.9359499676514988,
+      "grad_norm": 0.1907162368297577,
+      "learning_rate": 4.249825857323477e-05,
+      "loss": 2.3102,
+      "step": 4488
+    },
+    {
+      "epoch": 1.936381281000647,
+      "grad_norm": 0.20281952619552612,
+      "learning_rate": 4.246750505009563e-05,
+      "loss": 2.2093,
+      "step": 4489
+    },
+    {
+      "epoch": 1.9368125943497951,
+      "grad_norm": 0.19904431700706482,
+      "learning_rate": 4.2436758263165336e-05,
+      "loss": 2.0995,
+      "step": 4490
+    },
+    {
+      "epoch": 1.9372439076989432,
+      "grad_norm": 0.19451448321342468,
+      "learning_rate": 4.240601821881036e-05,
+      "loss": 2.2836,
+      "step": 4491
+    },
+    {
+      "epoch": 1.9376752210480914,
+      "grad_norm": 0.1946668177843094,
+      "learning_rate": 4.2375284923395736e-05,
+      "loss": 2.1484,
+      "step": 4492
+    },
+    {
+      "epoch": 1.9381065343972397,
+      "grad_norm": 0.18578791618347168,
+      "learning_rate": 4.2344558383285146e-05,
+      "loss": 2.2732,
+      "step": 4493
+    },
+    {
+      "epoch": 1.9385378477463877,
+      "grad_norm": 0.16853195428848267,
+      "learning_rate": 4.231383860484086e-05,
+      "loss": 2.1543,
+      "step": 4494
+    },
+    {
+      "epoch": 1.9389691610955357,
+      "grad_norm": 0.2038288712501526,
+      "learning_rate": 4.228312559442375e-05,
+      "loss": 2.1296,
+      "step": 4495
+    },
+    {
+      "epoch": 1.939400474444684,
+      "grad_norm": 0.28761857748031616,
+      "learning_rate": 4.225241935839323e-05,
+      "loss": 2.3585,
+      "step": 4496
+    },
+    {
+      "epoch": 1.9398317877938323,
+      "grad_norm": 0.19457559287548065,
+      "learning_rate": 4.2221719903107465e-05,
+      "loss": 2.0445,
+      "step": 4497
+    },
+    {
+      "epoch": 1.9402631011429805,
+      "grad_norm": 0.3036458492279053,
+      "learning_rate": 4.2191027234923026e-05,
+      "loss": 2.0874,
+      "step": 4498
+    },
+    {
+      "epoch": 1.9406944144921285,
+      "grad_norm": 0.19075198471546173,
+      "learning_rate": 4.216034136019516e-05,
+      "loss": 2.3152,
+      "step": 4499
+    },
+    {
+      "epoch": 1.9411257278412766,
+      "grad_norm": 0.2023703008890152,
+      "learning_rate": 4.2129662285277776e-05,
+      "loss": 2.0717,
+      "step": 4500
+    },
+    {
+      "epoch": 1.9411257278412766,
+      "eval_loss": 2.089297294616699,
+      "eval_runtime": 196.9909,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 4500
+    },
+    {
+      "epoch": 1.9415570411904248,
+      "grad_norm": 0.18044228851795197,
+      "learning_rate": 4.209899001652327e-05,
+      "loss": 2.2405,
+      "step": 4501
+    },
+    {
+      "epoch": 1.941988354539573,
+      "grad_norm": 0.18647083640098572,
+      "learning_rate": 4.206832456028272e-05,
+      "loss": 2.0873,
+      "step": 4502
+    },
+    {
+      "epoch": 1.9424196678887211,
+      "grad_norm": 0.18730871379375458,
+      "learning_rate": 4.203766592290564e-05,
+      "loss": 2.2108,
+      "step": 4503
+    },
+    {
+      "epoch": 1.9428509812378691,
+      "grad_norm": 0.19313322007656097,
+      "learning_rate": 4.200701411074034e-05,
+      "loss": 2.0227,
+      "step": 4504
+    },
+    {
+      "epoch": 1.9432822945870174,
+      "grad_norm": 0.7987073063850403,
+      "learning_rate": 4.197636913013356e-05,
+      "loss": 2.1318,
+      "step": 4505
+    },
+    {
+      "epoch": 1.9437136079361657,
+      "grad_norm": 0.19378934800624847,
+      "learning_rate": 4.1945730987430694e-05,
+      "loss": 2.1356,
+      "step": 4506
+    },
+    {
+      "epoch": 1.944144921285314,
+      "grad_norm": 0.18642853200435638,
+      "learning_rate": 4.191509968897571e-05,
+      "loss": 2.2687,
+      "step": 4507
+    },
+    {
+      "epoch": 1.944576234634462,
+      "grad_norm": 0.1928897500038147,
+      "learning_rate": 4.188447524111114e-05,
+      "loss": 2.3043,
+      "step": 4508
+    },
+    {
+      "epoch": 1.94500754798361,
+      "grad_norm": 0.18140451610088348,
+      "learning_rate": 4.185385765017811e-05,
+      "loss": 2.1565,
+      "step": 4509
+    },
+    {
+      "epoch": 1.9454388613327582,
+      "grad_norm": 0.1956416219472885,
+      "learning_rate": 4.1823246922516324e-05,
+      "loss": 2.0902,
+      "step": 4510
+    },
+    {
+      "epoch": 1.9458701746819065,
+      "grad_norm": 0.19609729945659637,
+      "learning_rate": 4.179264306446412e-05,
+      "loss": 2.3285,
+      "step": 4511
+    },
+    {
+      "epoch": 1.9463014880310545,
+      "grad_norm": 0.18456441164016724,
+      "learning_rate": 4.176204608235829e-05,
+      "loss": 2.0629,
+      "step": 4512
+    },
+    {
+      "epoch": 1.9467328013802028,
+      "grad_norm": 0.1965542435646057,
+      "learning_rate": 4.173145598253429e-05,
+      "loss": 2.2371,
+      "step": 4513
+    },
+    {
+      "epoch": 1.9471641147293508,
+      "grad_norm": 0.18580764532089233,
+      "learning_rate": 4.170087277132616e-05,
+      "loss": 2.1333,
+      "step": 4514
+    },
+    {
+      "epoch": 1.947595428078499,
+      "grad_norm": 0.18453587591648102,
+      "learning_rate": 4.167029645506648e-05,
+      "loss": 2.0632,
+      "step": 4515
+    },
+    {
+      "epoch": 1.9480267414276473,
+      "grad_norm": 0.19088754057884216,
+      "learning_rate": 4.163972704008643e-05,
+      "loss": 2.1317,
+      "step": 4516
+    },
+    {
+      "epoch": 1.9484580547767953,
+      "grad_norm": 0.20419196784496307,
+      "learning_rate": 4.160916453271567e-05,
+      "loss": 2.2218,
+      "step": 4517
+    },
+    {
+      "epoch": 1.9488893681259434,
+      "grad_norm": 0.20570674538612366,
+      "learning_rate": 4.157860893928257e-05,
+      "loss": 2.3841,
+      "step": 4518
+    },
+    {
+      "epoch": 1.9493206814750916,
+      "grad_norm": 0.18321706354618073,
+      "learning_rate": 4.154806026611399e-05,
+      "loss": 2.1808,
+      "step": 4519
+    },
+    {
+      "epoch": 1.9497519948242399,
+      "grad_norm": 0.1876799464225769,
+      "learning_rate": 4.151751851953535e-05,
+      "loss": 2.3955,
+      "step": 4520
+    },
+    {
+      "epoch": 1.950183308173388,
+      "grad_norm": 0.34781232476234436,
+      "learning_rate": 4.1486983705870656e-05,
+      "loss": 2.1793,
+      "step": 4521
+    },
+    {
+      "epoch": 1.9506146215225362,
+      "grad_norm": 0.18850237131118774,
+      "learning_rate": 4.145645583144248e-05,
+      "loss": 2.2796,
+      "step": 4522
+    },
+    {
+      "epoch": 1.9510459348716842,
+      "grad_norm": 0.18012240529060364,
+      "learning_rate": 4.1425934902571955e-05,
+      "loss": 2.2152,
+      "step": 4523
+    },
+    {
+      "epoch": 1.9514772482208325,
+      "grad_norm": 0.21512307226657867,
+      "learning_rate": 4.139542092557873e-05,
+      "loss": 2.1978,
+      "step": 4524
+    },
+    {
+      "epoch": 1.9519085615699807,
+      "grad_norm": 0.1743171215057373,
+      "learning_rate": 4.1364913906781164e-05,
+      "loss": 2.2178,
+      "step": 4525
+    },
+    {
+      "epoch": 1.9519085615699807,
+      "eval_loss": 2.089406967163086,
+      "eval_runtime": 196.9204,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 4525
+    },
+    {
+      "epoch": 1.9523398749191287,
+      "grad_norm": 0.18968987464904785,
+      "learning_rate": 4.133441385249597e-05,
+      "loss": 2.2091,
+      "step": 4526
+    },
+    {
+      "epoch": 1.9527711882682768,
+      "grad_norm": 0.1950238198041916,
+      "learning_rate": 4.130392076903851e-05,
+      "loss": 2.1411,
+      "step": 4527
+    },
+    {
+      "epoch": 1.953202501617425,
+      "grad_norm": 0.20082992315292358,
+      "learning_rate": 4.127343466272278e-05,
+      "loss": 2.1609,
+      "step": 4528
+    },
+    {
+      "epoch": 1.9536338149665733,
+      "grad_norm": 0.19735755026340485,
+      "learning_rate": 4.1242955539861235e-05,
+      "loss": 2.164,
+      "step": 4529
+    },
+    {
+      "epoch": 1.9540651283157213,
+      "grad_norm": 0.18233181536197662,
+      "learning_rate": 4.121248340676493e-05,
+      "loss": 1.7721,
+      "step": 4530
+    },
+    {
+      "epoch": 1.9544964416648696,
+      "grad_norm": 0.18331976234912872,
+      "learning_rate": 4.118201826974338e-05,
+      "loss": 2.1153,
+      "step": 4531
+    },
+    {
+      "epoch": 1.9549277550140176,
+      "grad_norm": 0.2071094512939453,
+      "learning_rate": 4.11515601351048e-05,
+      "loss": 2.2119,
+      "step": 4532
+    },
+    {
+      "epoch": 1.9553590683631659,
+      "grad_norm": 0.19779692590236664,
+      "learning_rate": 4.1121109009155857e-05,
+      "loss": 2.2707,
+      "step": 4533
+    },
+    {
+      "epoch": 1.9557903817123141,
+      "grad_norm": 0.19060441851615906,
+      "learning_rate": 4.109066489820178e-05,
+      "loss": 2.0028,
+      "step": 4534
+    },
+    {
+      "epoch": 1.9562216950614622,
+      "grad_norm": 0.19514651596546173,
+      "learning_rate": 4.106022780854636e-05,
+      "loss": 2.3064,
+      "step": 4535
+    },
+    {
+      "epoch": 1.9566530084106102,
+      "grad_norm": 0.19676922261714935,
+      "learning_rate": 4.102979774649195e-05,
+      "loss": 2.183,
+      "step": 4536
+    },
+    {
+      "epoch": 1.9570843217597584,
+      "grad_norm": 0.19450150430202484,
+      "learning_rate": 4.09993747183394e-05,
+      "loss": 2.3104,
+      "step": 4537
+    },
+    {
+      "epoch": 1.9575156351089067,
+      "grad_norm": 0.19470670819282532,
+      "learning_rate": 4.096895873038813e-05,
+      "loss": 2.3458,
+      "step": 4538
+    },
+    {
+      "epoch": 1.9579469484580547,
+      "grad_norm": 0.19570592045783997,
+      "learning_rate": 4.093854978893612e-05,
+      "loss": 2.1843,
+      "step": 4539
+    },
+    {
+      "epoch": 1.958378261807203,
+      "grad_norm": 0.19192062318325043,
+      "learning_rate": 4.0908147900279855e-05,
+      "loss": 2.0946,
+      "step": 4540
+    },
+    {
+      "epoch": 1.958809575156351,
+      "grad_norm": 0.19365684688091278,
+      "learning_rate": 4.087775307071439e-05,
+      "loss": 2.0085,
+      "step": 4541
+    },
+    {
+      "epoch": 1.9592408885054993,
+      "grad_norm": 0.20169289410114288,
+      "learning_rate": 4.084736530653329e-05,
+      "loss": 2.323,
+      "step": 4542
+    },
+    {
+      "epoch": 1.9596722018546475,
+      "grad_norm": 0.18891608715057373,
+      "learning_rate": 4.081698461402875e-05,
+      "loss": 2.1272,
+      "step": 4543
+    },
+    {
+      "epoch": 1.9601035152037956,
+      "grad_norm": 0.17692779004573822,
+      "learning_rate": 4.078661099949132e-05,
+      "loss": 2.1207,
+      "step": 4544
+    },
+    {
+      "epoch": 1.9605348285529436,
+      "grad_norm": 0.1865505576133728,
+      "learning_rate": 4.0756244469210216e-05,
+      "loss": 2.2669,
+      "step": 4545
+    },
+    {
+      "epoch": 1.9609661419020918,
+      "grad_norm": 0.19029918313026428,
+      "learning_rate": 4.07258850294732e-05,
+      "loss": 2.181,
+      "step": 4546
+    },
+    {
+      "epoch": 1.96139745525124,
+      "grad_norm": 0.19787052273750305,
+      "learning_rate": 4.069553268656652e-05,
+      "loss": 2.1802,
+      "step": 4547
+    },
+    {
+      "epoch": 1.9618287686003881,
+      "grad_norm": 0.21842846274375916,
+      "learning_rate": 4.066518744677495e-05,
+      "loss": 2.1489,
+      "step": 4548
+    },
+    {
+      "epoch": 1.9622600819495364,
+      "grad_norm": 0.18567824363708496,
+      "learning_rate": 4.0634849316381754e-05,
+      "loss": 2.2396,
+      "step": 4549
+    },
+    {
+      "epoch": 1.9626913952986844,
+      "grad_norm": 0.19945816695690155,
+      "learning_rate": 4.060451830166885e-05,
+      "loss": 2.3239,
+      "step": 4550
+    },
+    {
+      "epoch": 1.9626913952986844,
+      "eval_loss": 2.0894079208374023,
+      "eval_runtime": 196.9192,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 4550
+    },
+    {
+      "epoch": 1.9631227086478327,
+      "grad_norm": 0.16943685710430145,
+      "learning_rate": 4.0574194408916565e-05,
+      "loss": 1.8699,
+      "step": 4551
+    },
+    {
+      "epoch": 1.963554021996981,
+      "grad_norm": 0.1928304135799408,
+      "learning_rate": 4.054387764440381e-05,
+      "loss": 2.3189,
+      "step": 4552
+    },
+    {
+      "epoch": 1.963985335346129,
+      "grad_norm": 0.2010587602853775,
+      "learning_rate": 4.0513568014407993e-05,
+      "loss": 2.1832,
+      "step": 4553
+    },
+    {
+      "epoch": 1.964416648695277,
+      "grad_norm": 0.22043828666210175,
+      "learning_rate": 4.0483265525205046e-05,
+      "loss": 2.3322,
+      "step": 4554
+    },
+    {
+      "epoch": 1.9648479620444252,
+      "grad_norm": 0.1929306536912918,
+      "learning_rate": 4.045297018306945e-05,
+      "loss": 2.2782,
+      "step": 4555
+    },
+    {
+      "epoch": 1.9652792753935735,
+      "grad_norm": 0.20045654475688934,
+      "learning_rate": 4.0422681994274145e-05,
+      "loss": 2.3496,
+      "step": 4556
+    },
+    {
+      "epoch": 1.9657105887427215,
+      "grad_norm": 0.18351252377033234,
+      "learning_rate": 4.039240096509073e-05,
+      "loss": 2.1992,
+      "step": 4557
+    },
+    {
+      "epoch": 1.9661419020918698,
+      "grad_norm": 0.18495838344097137,
+      "learning_rate": 4.036212710178912e-05,
+      "loss": 2.2457,
+      "step": 4558
+    },
+    {
+      "epoch": 1.9665732154410178,
+      "grad_norm": 0.1930808424949646,
+      "learning_rate": 4.033186041063786e-05,
+      "loss": 2.1731,
+      "step": 4559
+    },
+    {
+      "epoch": 1.967004528790166,
+      "grad_norm": 0.1892196387052536,
+      "learning_rate": 4.030160089790405e-05,
+      "loss": 2.1982,
+      "step": 4560
+    },
+    {
+      "epoch": 1.9674358421393143,
+      "grad_norm": 0.27983418107032776,
+      "learning_rate": 4.0271348569853246e-05,
+      "loss": 2.2619,
+      "step": 4561
+    },
+    {
+      "epoch": 1.9678671554884624,
+      "grad_norm": 0.20122027397155762,
+      "learning_rate": 4.024110343274952e-05,
+      "loss": 2.2464,
+      "step": 4562
+    },
+    {
+      "epoch": 1.9682984688376104,
+      "grad_norm": 0.19067595899105072,
+      "learning_rate": 4.021086549285539e-05,
+      "loss": 2.0845,
+      "step": 4563
+    },
+    {
+      "epoch": 1.9687297821867586,
+      "grad_norm": 0.18311598896980286,
+      "learning_rate": 4.018063475643204e-05,
+      "loss": 2.0351,
+      "step": 4564
+    },
+    {
+      "epoch": 1.969161095535907,
+      "grad_norm": 0.1905970424413681,
+      "learning_rate": 4.015041122973904e-05,
+      "loss": 2.3344,
+      "step": 4565
+    },
+    {
+      "epoch": 1.9695924088850552,
+      "grad_norm": 0.1892702430486679,
+      "learning_rate": 4.01201949190345e-05,
+      "loss": 2.2001,
+      "step": 4566
+    },
+    {
+      "epoch": 1.9700237222342032,
+      "grad_norm": 0.17685829102993011,
+      "learning_rate": 4.0089985830575046e-05,
+      "loss": 2.0874,
+      "step": 4567
+    },
+    {
+      "epoch": 1.9704550355833512,
+      "grad_norm": 0.18338528275489807,
+      "learning_rate": 4.00597839706158e-05,
+      "loss": 2.2016,
+      "step": 4568
+    },
+    {
+      "epoch": 1.9708863489324995,
+      "grad_norm": 0.19318626821041107,
+      "learning_rate": 4.002958934541037e-05,
+      "loss": 2.2684,
+      "step": 4569
+    },
+    {
+      "epoch": 1.9713176622816477,
+      "grad_norm": 0.22403284907341003,
+      "learning_rate": 3.9999401961210906e-05,
+      "loss": 2.145,
+      "step": 4570
+    },
+    {
+      "epoch": 1.9717489756307958,
+      "grad_norm": 0.20169135928153992,
+      "learning_rate": 3.996922182426802e-05,
+      "loss": 2.148,
+      "step": 4571
+    },
+    {
+      "epoch": 1.9721802889799438,
+      "grad_norm": 0.19115835428237915,
+      "learning_rate": 3.993904894083085e-05,
+      "loss": 2.0836,
+      "step": 4572
+    },
+    {
+      "epoch": 1.972611602329092,
+      "grad_norm": 0.21998874843120575,
+      "learning_rate": 3.9908883317146984e-05,
+      "loss": 1.9864,
+      "step": 4573
+    },
+    {
+      "epoch": 1.9730429156782403,
+      "grad_norm": 0.19121584296226501,
+      "learning_rate": 3.987872495946261e-05,
+      "loss": 2.1121,
+      "step": 4574
+    },
+    {
+      "epoch": 1.9734742290273886,
+      "grad_norm": 0.1847795993089676,
+      "learning_rate": 3.984857387402234e-05,
+      "loss": 2.1973,
+      "step": 4575
+    },
+    {
+      "epoch": 1.9734742290273886,
+      "eval_loss": 2.0891201496124268,
+      "eval_runtime": 204.498,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 4575
+    },
+    {
+      "epoch": 1.9739055423765366,
+      "grad_norm": 0.19345210492610931,
+      "learning_rate": 3.9818430067069234e-05,
+      "loss": 2.0098,
+      "step": 4576
+    },
+    {
+      "epoch": 1.9743368557256846,
+      "grad_norm": 0.1937677562236786,
+      "learning_rate": 3.978829354484488e-05,
+      "loss": 2.0485,
+      "step": 4577
+    },
+    {
+      "epoch": 1.9747681690748329,
+      "grad_norm": 0.19139151275157928,
+      "learning_rate": 3.975816431358946e-05,
+      "loss": 2.3189,
+      "step": 4578
+    },
+    {
+      "epoch": 1.9751994824239811,
+      "grad_norm": 0.21577173471450806,
+      "learning_rate": 3.9728042379541516e-05,
+      "loss": 2.2798,
+      "step": 4579
+    },
+    {
+      "epoch": 1.9756307957731292,
+      "grad_norm": 0.17227891087532043,
+      "learning_rate": 3.969792774893811e-05,
+      "loss": 2.0565,
+      "step": 4580
+    },
+    {
+      "epoch": 1.9760621091222772,
+      "grad_norm": 0.19595201313495636,
+      "learning_rate": 3.966782042801483e-05,
+      "loss": 2.1484,
+      "step": 4581
+    },
+    {
+      "epoch": 1.9764934224714255,
+      "grad_norm": 0.1743748039007187,
+      "learning_rate": 3.963772042300572e-05,
+      "loss": 2.0228,
+      "step": 4582
+    },
+    {
+      "epoch": 1.9769247358205737,
+      "grad_norm": 0.20380571484565735,
+      "learning_rate": 3.960762774014331e-05,
+      "loss": 2.2646,
+      "step": 4583
+    },
+    {
+      "epoch": 1.977356049169722,
+      "grad_norm": 0.1720813810825348,
+      "learning_rate": 3.9577542385658614e-05,
+      "loss": 2.2076,
+      "step": 4584
+    },
+    {
+      "epoch": 1.97778736251887,
+      "grad_norm": 0.19041219353675842,
+      "learning_rate": 3.9547464365781145e-05,
+      "loss": 2.1829,
+      "step": 4585
+    },
+    {
+      "epoch": 1.978218675868018,
+      "grad_norm": 0.1973971128463745,
+      "learning_rate": 3.951739368673889e-05,
+      "loss": 2.3014,
+      "step": 4586
+    },
+    {
+      "epoch": 1.9786499892171663,
+      "grad_norm": 0.17989139258861542,
+      "learning_rate": 3.948733035475829e-05,
+      "loss": 2.0504,
+      "step": 4587
+    },
+    {
+      "epoch": 1.9790813025663145,
+      "grad_norm": 0.20466317236423492,
+      "learning_rate": 3.945727437606428e-05,
+      "loss": 2.2927,
+      "step": 4588
+    },
+    {
+      "epoch": 1.9795126159154626,
+      "grad_norm": 0.17659124732017517,
+      "learning_rate": 3.9427225756880353e-05,
+      "loss": 2.1761,
+      "step": 4589
+    },
+    {
+      "epoch": 1.9799439292646106,
+      "grad_norm": 0.1800772100687027,
+      "learning_rate": 3.939718450342832e-05,
+      "loss": 2.0339,
+      "step": 4590
+    },
+    {
+      "epoch": 1.9803752426137589,
+      "grad_norm": 0.20760327577590942,
+      "learning_rate": 3.9367150621928556e-05,
+      "loss": 2.0183,
+      "step": 4591
+    },
+    {
+      "epoch": 1.9808065559629071,
+      "grad_norm": 0.17776834964752197,
+      "learning_rate": 3.933712411859995e-05,
+      "loss": 2.1135,
+      "step": 4592
+    },
+    {
+      "epoch": 1.9812378693120554,
+      "grad_norm": 0.19570772349834442,
+      "learning_rate": 3.9307104999659813e-05,
+      "loss": 2.2599,
+      "step": 4593
+    },
+    {
+      "epoch": 1.9816691826612034,
+      "grad_norm": 119.74149322509766,
+      "learning_rate": 3.927709327132394e-05,
+      "loss": 2.166,
+      "step": 4594
+    },
+    {
+      "epoch": 1.9821004960103514,
+      "grad_norm": 0.22241385281085968,
+      "learning_rate": 3.92470889398065e-05,
+      "loss": 1.9044,
+      "step": 4595
+    },
+    {
+      "epoch": 1.9825318093594997,
+      "grad_norm": 0.19045329093933105,
+      "learning_rate": 3.9217092011320317e-05,
+      "loss": 2.0649,
+      "step": 4596
+    },
+    {
+      "epoch": 1.982963122708648,
+      "grad_norm": 0.19328786432743073,
+      "learning_rate": 3.918710249207655e-05,
+      "loss": 2.055,
+      "step": 4597
+    },
+    {
+      "epoch": 1.983394436057796,
+      "grad_norm": 0.21507321298122406,
+      "learning_rate": 3.915712038828485e-05,
+      "loss": 2.068,
+      "step": 4598
+    },
+    {
+      "epoch": 1.983825749406944,
+      "grad_norm": 0.19425652921199799,
+      "learning_rate": 3.912714570615335e-05,
+      "loss": 2.0588,
+      "step": 4599
+    },
+    {
+      "epoch": 1.9842570627560923,
+      "grad_norm": 0.17666107416152954,
+      "learning_rate": 3.909717845188863e-05,
+      "loss": 1.9879,
+      "step": 4600
+    },
+    {
+      "epoch": 1.9842570627560923,
+      "eval_loss": 2.0891807079315186,
+      "eval_runtime": 201.3417,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4600
+    },
+    {
+      "epoch": 1.9846883761052405,
+      "grad_norm": 0.19912296533584595,
+      "learning_rate": 3.906721863169573e-05,
+      "loss": 2.2162,
+      "step": 4601
+    },
+    {
+      "epoch": 1.9851196894543888,
+      "grad_norm": 0.1944214552640915,
+      "learning_rate": 3.903726625177815e-05,
+      "loss": 2.1721,
+      "step": 4602
+    },
+    {
+      "epoch": 1.9855510028035368,
+      "grad_norm": 0.18800340592861176,
+      "learning_rate": 3.9007321318337925e-05,
+      "loss": 1.9881,
+      "step": 4603
+    },
+    {
+      "epoch": 1.9859823161526848,
+      "grad_norm": 0.18974347412586212,
+      "learning_rate": 3.89773838375754e-05,
+      "loss": 2.3557,
+      "step": 4604
+    },
+    {
+      "epoch": 1.986413629501833,
+      "grad_norm": 0.19073079526424408,
+      "learning_rate": 3.894745381568946e-05,
+      "loss": 2.3348,
+      "step": 4605
+    },
+    {
+      "epoch": 1.9868449428509813,
+      "grad_norm": 0.20038451254367828,
+      "learning_rate": 3.89175312588775e-05,
+      "loss": 2.1509,
+      "step": 4606
+    },
+    {
+      "epoch": 1.9872762562001294,
+      "grad_norm": 0.1876484900712967,
+      "learning_rate": 3.8887616173335285e-05,
+      "loss": 2.1822,
+      "step": 4607
+    },
+    {
+      "epoch": 1.9877075695492774,
+      "grad_norm": 0.1894012689590454,
+      "learning_rate": 3.88577085652571e-05,
+      "loss": 2.16,
+      "step": 4608
+    },
+    {
+      "epoch": 1.9881388828984257,
+      "grad_norm": 0.19239047169685364,
+      "learning_rate": 3.882780844083552e-05,
+      "loss": 2.1529,
+      "step": 4609
+    },
+    {
+      "epoch": 1.988570196247574,
+      "grad_norm": 0.20243610441684723,
+      "learning_rate": 3.879791580626183e-05,
+      "loss": 2.1936,
+      "step": 4610
+    },
+    {
+      "epoch": 1.9890015095967222,
+      "grad_norm": 0.20150430500507355,
+      "learning_rate": 3.876803066772555e-05,
+      "loss": 2.1183,
+      "step": 4611
+    },
+    {
+      "epoch": 1.9894328229458702,
+      "grad_norm": 0.17391133308410645,
+      "learning_rate": 3.873815303141475e-05,
+      "loss": 2.1931,
+      "step": 4612
+    },
+    {
+      "epoch": 1.9898641362950182,
+      "grad_norm": 0.18860773742198944,
+      "learning_rate": 3.870828290351593e-05,
+      "loss": 2.1888,
+      "step": 4613
+    },
+    {
+      "epoch": 1.9902954496441665,
+      "grad_norm": 0.18391074240207672,
+      "learning_rate": 3.8678420290214e-05,
+      "loss": 1.9288,
+      "step": 4614
+    },
+    {
+      "epoch": 1.9907267629933147,
+      "grad_norm": 0.19212834537029266,
+      "learning_rate": 3.864856519769237e-05,
+      "loss": 2.1522,
+      "step": 4615
+    },
+    {
+      "epoch": 1.9911580763424628,
+      "grad_norm": 0.18585047125816345,
+      "learning_rate": 3.861871763213284e-05,
+      "loss": 2.068,
+      "step": 4616
+    },
+    {
+      "epoch": 1.9915893896916108,
+      "grad_norm": 0.25356727838516235,
+      "learning_rate": 3.858887759971568e-05,
+      "loss": 2.1914,
+      "step": 4617
+    },
+    {
+      "epoch": 1.992020703040759,
+      "grad_norm": 0.19800281524658203,
+      "learning_rate": 3.85590451066196e-05,
+      "loss": 2.313,
+      "step": 4618
+    },
+    {
+      "epoch": 1.9924520163899073,
+      "grad_norm": 0.18604665994644165,
+      "learning_rate": 3.852922015902171e-05,
+      "loss": 1.8551,
+      "step": 4619
+    },
+    {
+      "epoch": 1.9928833297390556,
+      "grad_norm": 0.18251395225524902,
+      "learning_rate": 3.849940276309766e-05,
+      "loss": 1.8771,
+      "step": 4620
+    },
+    {
+      "epoch": 1.9933146430882036,
+      "grad_norm": 0.18314236402511597,
+      "learning_rate": 3.8469592925021466e-05,
+      "loss": 2.107,
+      "step": 4621
+    },
+    {
+      "epoch": 1.9937459564373516,
+      "grad_norm": 0.18971586227416992,
+      "learning_rate": 3.8439790650965505e-05,
+      "loss": 2.347,
+      "step": 4622
+    },
+    {
+      "epoch": 1.9941772697865,
+      "grad_norm": 0.19586211442947388,
+      "learning_rate": 3.840999594710068e-05,
+      "loss": 2.1796,
+      "step": 4623
+    },
+    {
+      "epoch": 1.9946085831356481,
+      "grad_norm": 0.20417676866054535,
+      "learning_rate": 3.838020881959636e-05,
+      "loss": 2.3454,
+      "step": 4624
+    },
+    {
+      "epoch": 1.9950398964847962,
+      "grad_norm": 0.18715451657772064,
+      "learning_rate": 3.835042927462029e-05,
+      "loss": 2.125,
+      "step": 4625
+    },
+    {
+      "epoch": 1.9950398964847962,
+      "eval_loss": 2.089186429977417,
+      "eval_runtime": 200.4625,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 4625
+    },
+    {
+      "epoch": 1.9954712098339442,
+      "grad_norm": 0.17249004542827606,
+      "learning_rate": 3.832065731833861e-05,
+      "loss": 1.9698,
+      "step": 4626
+    },
+    {
+      "epoch": 1.9959025231830925,
+      "grad_norm": 0.17829373478889465,
+      "learning_rate": 3.829089295691597e-05,
+      "loss": 2.2089,
+      "step": 4627
+    },
+    {
+      "epoch": 1.9963338365322407,
+      "grad_norm": 0.1930740773677826,
+      "learning_rate": 3.826113619651537e-05,
+      "loss": 2.3168,
+      "step": 4628
+    },
+    {
+      "epoch": 1.996765149881389,
+      "grad_norm": 0.1748417466878891,
+      "learning_rate": 3.82313870432983e-05,
+      "loss": 2.3054,
+      "step": 4629
+    },
+    {
+      "epoch": 1.997196463230537,
+      "grad_norm": 0.19034349918365479,
+      "learning_rate": 3.820164550342464e-05,
+      "loss": 2.2194,
+      "step": 4630
+    },
+    {
+      "epoch": 1.997627776579685,
+      "grad_norm": 0.2013319879770279,
+      "learning_rate": 3.8171911583052686e-05,
+      "loss": 2.2437,
+      "step": 4631
+    },
+    {
+      "epoch": 1.9980590899288333,
+      "grad_norm": 0.1998211145401001,
+      "learning_rate": 3.814218528833919e-05,
+      "loss": 2.114,
+      "step": 4632
+    },
+    {
+      "epoch": 1.9984904032779816,
+      "grad_norm": 0.18663814663887024,
+      "learning_rate": 3.8112466625439294e-05,
+      "loss": 2.1535,
+      "step": 4633
+    },
+    {
+      "epoch": 1.9989217166271296,
+      "grad_norm": 0.18947145342826843,
+      "learning_rate": 3.808275560050656e-05,
+      "loss": 2.0403,
+      "step": 4634
+    },
+    {
+      "epoch": 1.9993530299762776,
+      "grad_norm": 0.20086605846881866,
+      "learning_rate": 3.805305221969306e-05,
+      "loss": 2.1506,
+      "step": 4635
+    },
+    {
+      "epoch": 1.9997843433254259,
+      "grad_norm": 0.1824001967906952,
+      "learning_rate": 3.802335648914911e-05,
+      "loss": 2.3072,
+      "step": 4636
+    },
+    {
+      "epoch": 2.0004313133491483,
+      "grad_norm": 0.3254935145378113,
+      "learning_rate": 3.7993668415023526e-05,
+      "loss": 4.2159,
+      "step": 4637
+    },
+    {
+      "epoch": 2.0008626266982965,
+      "grad_norm": 0.18431530892848969,
+      "learning_rate": 3.796398800346364e-05,
+      "loss": 2.1423,
+      "step": 4638
+    },
+    {
+      "epoch": 2.0012939400474443,
+      "grad_norm": 0.1791626363992691,
+      "learning_rate": 3.793431526061504e-05,
+      "loss": 2.0886,
+      "step": 4639
+    },
+    {
+      "epoch": 2.0017252533965926,
+      "grad_norm": 0.1967422068119049,
+      "learning_rate": 3.790465019262186e-05,
+      "loss": 2.0144,
+      "step": 4640
+    },
+    {
+      "epoch": 2.002156566745741,
+      "grad_norm": 0.19516336917877197,
+      "learning_rate": 3.7874992805626443e-05,
+      "loss": 2.1952,
+      "step": 4641
+    },
+    {
+      "epoch": 2.002587880094889,
+      "grad_norm": 0.1892661154270172,
+      "learning_rate": 3.7845343105769796e-05,
+      "loss": 2.0195,
+      "step": 4642
+    },
+    {
+      "epoch": 2.003019193444037,
+      "grad_norm": 0.1858513206243515,
+      "learning_rate": 3.7815701099191184e-05,
+      "loss": 2.1626,
+      "step": 4643
+    },
+    {
+      "epoch": 2.003450506793185,
+      "grad_norm": 0.18395952880382538,
+      "learning_rate": 3.778606679202829e-05,
+      "loss": 2.1628,
+      "step": 4644
+    },
+    {
+      "epoch": 2.0038818201423334,
+      "grad_norm": 0.18125440180301666,
+      "learning_rate": 3.775644019041723e-05,
+      "loss": 2.125,
+      "step": 4645
+    },
+    {
+      "epoch": 2.0043131334914817,
+      "grad_norm": 0.19737988710403442,
+      "learning_rate": 3.7726821300492524e-05,
+      "loss": 2.1642,
+      "step": 4646
+    },
+    {
+      "epoch": 2.00474444684063,
+      "grad_norm": 0.19009336829185486,
+      "learning_rate": 3.769721012838708e-05,
+      "loss": 1.9746,
+      "step": 4647
+    },
+    {
+      "epoch": 2.0051757601897777,
+      "grad_norm": 0.21443523466587067,
+      "learning_rate": 3.766760668023218e-05,
+      "loss": 2.2279,
+      "step": 4648
+    },
+    {
+      "epoch": 2.005607073538926,
+      "grad_norm": 0.21913030743598938,
+      "learning_rate": 3.7638010962157635e-05,
+      "loss": 2.1498,
+      "step": 4649
+    },
+    {
+      "epoch": 2.0060383868880742,
+      "grad_norm": 0.19212424755096436,
+      "learning_rate": 3.760842298029148e-05,
+      "loss": 2.1715,
+      "step": 4650
+    },
+    {
+      "epoch": 2.0060383868880742,
+      "eval_loss": 2.0894405841827393,
+      "eval_runtime": 200.6853,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4650
+    },
+    {
+      "epoch": 2.0064697002372225,
+      "grad_norm": 0.1889164000749588,
+      "learning_rate": 3.7578842740760224e-05,
+      "loss": 2.1657,
+      "step": 4651
+    },
+    {
+      "epoch": 2.0069010135863703,
+      "grad_norm": 0.1890760362148285,
+      "learning_rate": 3.754927024968885e-05,
+      "loss": 2.1499,
+      "step": 4652
+    },
+    {
+      "epoch": 2.0073323269355186,
+      "grad_norm": 0.19354365766048431,
+      "learning_rate": 3.7519705513200606e-05,
+      "loss": 2.1976,
+      "step": 4653
+    },
+    {
+      "epoch": 2.007763640284667,
+      "grad_norm": 0.19646655023097992,
+      "learning_rate": 3.7490148537417245e-05,
+      "loss": 1.9383,
+      "step": 4654
+    },
+    {
+      "epoch": 2.008194953633815,
+      "grad_norm": 0.1919821798801422,
+      "learning_rate": 3.746059932845878e-05,
+      "loss": 2.189,
+      "step": 4655
+    },
+    {
+      "epoch": 2.0086262669829633,
+      "grad_norm": 0.18532203137874603,
+      "learning_rate": 3.7431057892443777e-05,
+      "loss": 2.0868,
+      "step": 4656
+    },
+    {
+      "epoch": 2.009057580332111,
+      "grad_norm": 0.18428903818130493,
+      "learning_rate": 3.7401524235489074e-05,
+      "loss": 2.1462,
+      "step": 4657
+    },
+    {
+      "epoch": 2.0094888936812594,
+      "grad_norm": 0.20539259910583496,
+      "learning_rate": 3.737199836370996e-05,
+      "loss": 2.2161,
+      "step": 4658
+    },
+    {
+      "epoch": 2.0099202070304076,
+      "grad_norm": 0.19485358893871307,
+      "learning_rate": 3.734248028322007e-05,
+      "loss": 2.1433,
+      "step": 4659
+    },
+    {
+      "epoch": 2.010351520379556,
+      "grad_norm": 0.18940050899982452,
+      "learning_rate": 3.731297000013146e-05,
+      "loss": 1.9886,
+      "step": 4660
+    },
+    {
+      "epoch": 2.0107828337287037,
+      "grad_norm": 0.20166797935962677,
+      "learning_rate": 3.7283467520554546e-05,
+      "loss": 2.0954,
+      "step": 4661
+    },
+    {
+      "epoch": 2.011214147077852,
+      "grad_norm": 0.19941836595535278,
+      "learning_rate": 3.725397285059814e-05,
+      "loss": 2.1256,
+      "step": 4662
+    },
+    {
+      "epoch": 2.011645460427,
+      "grad_norm": 0.18807263672351837,
+      "learning_rate": 3.7224485996369445e-05,
+      "loss": 2.1517,
+      "step": 4663
+    },
+    {
+      "epoch": 2.0120767737761485,
+      "grad_norm": 0.22854174673557281,
+      "learning_rate": 3.7195006963974026e-05,
+      "loss": 2.2179,
+      "step": 4664
+    },
+    {
+      "epoch": 2.0125080871252967,
+      "grad_norm": 0.19089582562446594,
+      "learning_rate": 3.716553575951582e-05,
+      "loss": 2.1794,
+      "step": 4665
+    },
+    {
+      "epoch": 2.0129394004744445,
+      "grad_norm": 0.17721687257289886,
+      "learning_rate": 3.7136072389097204e-05,
+      "loss": 2.0641,
+      "step": 4666
+    },
+    {
+      "epoch": 2.013370713823593,
+      "grad_norm": 0.18833722174167633,
+      "learning_rate": 3.710661685881891e-05,
+      "loss": 2.0732,
+      "step": 4667
+    },
+    {
+      "epoch": 2.013802027172741,
+      "grad_norm": 1.9127404689788818,
+      "learning_rate": 3.7077169174779956e-05,
+      "loss": 2.1706,
+      "step": 4668
+    },
+    {
+      "epoch": 2.0142333405218893,
+      "grad_norm": 0.19390110671520233,
+      "learning_rate": 3.7047729343077814e-05,
+      "loss": 2.1484,
+      "step": 4669
+    },
+    {
+      "epoch": 2.014664653871037,
+      "grad_norm": 0.19157013297080994,
+      "learning_rate": 3.701829736980839e-05,
+      "loss": 1.9822,
+      "step": 4670
+    },
+    {
+      "epoch": 2.0150959672201854,
+      "grad_norm": 0.18170066177845,
+      "learning_rate": 3.698887326106584e-05,
+      "loss": 1.9444,
+      "step": 4671
+    },
+    {
+      "epoch": 2.0155272805693336,
+      "grad_norm": 0.18794336915016174,
+      "learning_rate": 3.6959457022942775e-05,
+      "loss": 2.2588,
+      "step": 4672
+    },
+    {
+      "epoch": 2.015958593918482,
+      "grad_norm": 0.21366532146930695,
+      "learning_rate": 3.6930048661530135e-05,
+      "loss": 2.2123,
+      "step": 4673
+    },
+    {
+      "epoch": 2.01638990726763,
+      "grad_norm": 0.19540277123451233,
+      "learning_rate": 3.690064818291725e-05,
+      "loss": 2.0826,
+      "step": 4674
+    },
+    {
+      "epoch": 2.016821220616778,
+      "grad_norm": 0.18262942135334015,
+      "learning_rate": 3.6871255593191794e-05,
+      "loss": 2.004,
+      "step": 4675
+    },
+    {
+      "epoch": 2.016821220616778,
+      "eval_loss": 2.089542865753174,
+      "eval_runtime": 200.436,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 4675
+    },
+    {
+      "epoch": 2.017252533965926,
+      "grad_norm": 0.1919758915901184,
+      "learning_rate": 3.6841870898439846e-05,
+      "loss": 2.3566,
+      "step": 4676
+    },
+    {
+      "epoch": 2.0176838473150744,
+      "grad_norm": 0.18473289906978607,
+      "learning_rate": 3.681249410474581e-05,
+      "loss": 1.8332,
+      "step": 4677
+    },
+    {
+      "epoch": 2.0181151606642227,
+      "grad_norm": 0.1841718554496765,
+      "learning_rate": 3.6783125218192494e-05,
+      "loss": 1.9875,
+      "step": 4678
+    },
+    {
+      "epoch": 2.0185464740133705,
+      "grad_norm": 0.1876533031463623,
+      "learning_rate": 3.6753764244861025e-05,
+      "loss": 2.0937,
+      "step": 4679
+    },
+    {
+      "epoch": 2.0189777873625188,
+      "grad_norm": 0.19570693373680115,
+      "learning_rate": 3.67244111908309e-05,
+      "loss": 2.2245,
+      "step": 4680
+    },
+    {
+      "epoch": 2.019409100711667,
+      "grad_norm": 0.1707245111465454,
+      "learning_rate": 3.669506606218007e-05,
+      "loss": 2.1095,
+      "step": 4681
+    },
+    {
+      "epoch": 2.0198404140608153,
+      "grad_norm": 0.19620443880558014,
+      "learning_rate": 3.666572886498467e-05,
+      "loss": 2.158,
+      "step": 4682
+    },
+    {
+      "epoch": 2.0202717274099635,
+      "grad_norm": 0.17991629242897034,
+      "learning_rate": 3.6636399605319307e-05,
+      "loss": 1.9504,
+      "step": 4683
+    },
+    {
+      "epoch": 2.0207030407591113,
+      "grad_norm": 0.17959769070148468,
+      "learning_rate": 3.660707828925696e-05,
+      "loss": 2.097,
+      "step": 4684
+    },
+    {
+      "epoch": 2.0211343541082596,
+      "grad_norm": 0.18759562075138092,
+      "learning_rate": 3.657776492286893e-05,
+      "loss": 2.3082,
+      "step": 4685
+    },
+    {
+      "epoch": 2.021565667457408,
+      "grad_norm": 0.18067705631256104,
+      "learning_rate": 3.654845951222486e-05,
+      "loss": 2.1922,
+      "step": 4686
+    },
+    {
+      "epoch": 2.021996980806556,
+      "grad_norm": 0.19888699054718018,
+      "learning_rate": 3.6519162063392695e-05,
+      "loss": 2.2701,
+      "step": 4687
+    },
+    {
+      "epoch": 2.022428294155704,
+      "grad_norm": 0.18732723593711853,
+      "learning_rate": 3.6489872582438883e-05,
+      "loss": 2.1285,
+      "step": 4688
+    },
+    {
+      "epoch": 2.022859607504852,
+      "grad_norm": 0.19551338255405426,
+      "learning_rate": 3.646059107542808e-05,
+      "loss": 2.1456,
+      "step": 4689
+    },
+    {
+      "epoch": 2.0232909208540004,
+      "grad_norm": 0.20126861333847046,
+      "learning_rate": 3.643131754842336e-05,
+      "loss": 2.1727,
+      "step": 4690
+    },
+    {
+      "epoch": 2.0237222342031487,
+      "grad_norm": 0.1837441325187683,
+      "learning_rate": 3.6402052007486115e-05,
+      "loss": 2.1804,
+      "step": 4691
+    },
+    {
+      "epoch": 2.024153547552297,
+      "grad_norm": 0.202824667096138,
+      "learning_rate": 3.6372794458676115e-05,
+      "loss": 2.2157,
+      "step": 4692
+    },
+    {
+      "epoch": 2.0245848609014447,
+      "grad_norm": 0.201461061835289,
+      "learning_rate": 3.6343544908051434e-05,
+      "loss": 2.1783,
+      "step": 4693
+    },
+    {
+      "epoch": 2.025016174250593,
+      "grad_norm": 0.20021897554397583,
+      "learning_rate": 3.6314303361668514e-05,
+      "loss": 2.101,
+      "step": 4694
+    },
+    {
+      "epoch": 2.0254474875997412,
+      "grad_norm": 0.18160514533519745,
+      "learning_rate": 3.6285069825582154e-05,
+      "loss": 2.0716,
+      "step": 4695
+    },
+    {
+      "epoch": 2.0258788009488895,
+      "grad_norm": 0.19927164912223816,
+      "learning_rate": 3.6255844305845464e-05,
+      "loss": 2.1218,
+      "step": 4696
+    },
+    {
+      "epoch": 2.0263101142980373,
+      "grad_norm": 0.21796950697898865,
+      "learning_rate": 3.6226626808509876e-05,
+      "loss": 2.1368,
+      "step": 4697
+    },
+    {
+      "epoch": 2.0267414276471856,
+      "grad_norm": 0.20131273567676544,
+      "learning_rate": 3.619741733962525e-05,
+      "loss": 2.2224,
+      "step": 4698
+    },
+    {
+      "epoch": 2.027172740996334,
+      "grad_norm": 0.2108275443315506,
+      "learning_rate": 3.616821590523974e-05,
+      "loss": 2.2149,
+      "step": 4699
+    },
+    {
+      "epoch": 2.027604054345482,
+      "grad_norm": 0.182052344083786,
+      "learning_rate": 3.613902251139975e-05,
+      "loss": 2.1594,
+      "step": 4700
+    },
+    {
+      "epoch": 2.027604054345482,
+      "eval_loss": 2.089719295501709,
+      "eval_runtime": 200.7819,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4700
+    },
+    {
+      "epoch": 2.0280353676946303,
+      "grad_norm": 0.20290504395961761,
+      "learning_rate": 3.6109837164150105e-05,
+      "loss": 2.1564,
+      "step": 4701
+    },
+    {
+      "epoch": 2.028466681043778,
+      "grad_norm": 0.18281979858875275,
+      "learning_rate": 3.6080659869533994e-05,
+      "loss": 1.9623,
+      "step": 4702
+    },
+    {
+      "epoch": 2.0288979943929264,
+      "grad_norm": 0.18263626098632812,
+      "learning_rate": 3.605149063359287e-05,
+      "loss": 1.9182,
+      "step": 4703
+    },
+    {
+      "epoch": 2.0293293077420747,
+      "grad_norm": 0.20167215168476105,
+      "learning_rate": 3.602232946236654e-05,
+      "loss": 2.1137,
+      "step": 4704
+    },
+    {
+      "epoch": 2.029760621091223,
+      "grad_norm": 0.19266636669635773,
+      "learning_rate": 3.5993176361893154e-05,
+      "loss": 2.1873,
+      "step": 4705
+    },
+    {
+      "epoch": 2.0301919344403707,
+      "grad_norm": 0.18675893545150757,
+      "learning_rate": 3.596403133820917e-05,
+      "loss": 2.1701,
+      "step": 4706
+    },
+    {
+      "epoch": 2.030623247789519,
+      "grad_norm": 0.20420628786087036,
+      "learning_rate": 3.5934894397349394e-05,
+      "loss": 2.0722,
+      "step": 4707
+    },
+    {
+      "epoch": 2.0310545611386672,
+      "grad_norm": 0.19602304697036743,
+      "learning_rate": 3.590576554534694e-05,
+      "loss": 2.0701,
+      "step": 4708
+    },
+    {
+      "epoch": 2.0314858744878155,
+      "grad_norm": 0.21883141994476318,
+      "learning_rate": 3.587664478823325e-05,
+      "loss": 2.2202,
+      "step": 4709
+    },
+    {
+      "epoch": 2.0319171878369637,
+      "grad_norm": 0.181867316365242,
+      "learning_rate": 3.584753213203811e-05,
+      "loss": 2.0648,
+      "step": 4710
+    },
+    {
+      "epoch": 2.0323485011861115,
+      "grad_norm": 0.2570772171020508,
+      "learning_rate": 3.581842758278957e-05,
+      "loss": 2.0124,
+      "step": 4711
+    },
+    {
+      "epoch": 2.03277981453526,
+      "grad_norm": 0.1787121295928955,
+      "learning_rate": 3.578933114651411e-05,
+      "loss": 2.1435,
+      "step": 4712
+    },
+    {
+      "epoch": 2.033211127884408,
+      "grad_norm": 0.205344557762146,
+      "learning_rate": 3.576024282923647e-05,
+      "loss": 2.4994,
+      "step": 4713
+    },
+    {
+      "epoch": 2.0336424412335563,
+      "grad_norm": 0.19113320112228394,
+      "learning_rate": 3.573116263697965e-05,
+      "loss": 2.1884,
+      "step": 4714
+    },
+    {
+      "epoch": 2.034073754582704,
+      "grad_norm": 0.19865363836288452,
+      "learning_rate": 3.5702090575765006e-05,
+      "loss": 1.9913,
+      "step": 4715
+    },
+    {
+      "epoch": 2.0345050679318524,
+      "grad_norm": 0.22159428894519806,
+      "learning_rate": 3.5673026651612295e-05,
+      "loss": 2.2952,
+      "step": 4716
+    },
+    {
+      "epoch": 2.0349363812810006,
+      "grad_norm": 0.20246855914592743,
+      "learning_rate": 3.5643970870539485e-05,
+      "loss": 2.3079,
+      "step": 4717
+    },
+    {
+      "epoch": 2.035367694630149,
+      "grad_norm": 0.18345387279987335,
+      "learning_rate": 3.561492323856289e-05,
+      "loss": 2.1188,
+      "step": 4718
+    },
+    {
+      "epoch": 2.035799007979297,
+      "grad_norm": 0.17065024375915527,
+      "learning_rate": 3.558588376169716e-05,
+      "loss": 2.0935,
+      "step": 4719
+    },
+    {
+      "epoch": 2.036230321328445,
+      "grad_norm": 0.2016269415616989,
+      "learning_rate": 3.555685244595521e-05,
+      "loss": 2.2215,
+      "step": 4720
+    },
+    {
+      "epoch": 2.036661634677593,
+      "grad_norm": 0.19801582396030426,
+      "learning_rate": 3.552782929734831e-05,
+      "loss": 2.0749,
+      "step": 4721
+    },
+    {
+      "epoch": 2.0370929480267415,
+      "grad_norm": 0.19953446090221405,
+      "learning_rate": 3.5498814321886016e-05,
+      "loss": 2.1919,
+      "step": 4722
+    },
+    {
+      "epoch": 2.0375242613758897,
+      "grad_norm": 0.18833646178245544,
+      "learning_rate": 3.546980752557619e-05,
+      "loss": 2.1415,
+      "step": 4723
+    },
+    {
+      "epoch": 2.0379555747250375,
+      "grad_norm": 0.18866832554340363,
+      "learning_rate": 3.544080891442502e-05,
+      "loss": 1.9907,
+      "step": 4724
+    },
+    {
+      "epoch": 2.0383868880741858,
+      "grad_norm": 0.1904328614473343,
+      "learning_rate": 3.541181849443694e-05,
+      "loss": 2.0671,
+      "step": 4725
+    },
+    {
+      "epoch": 2.0383868880741858,
+      "eval_loss": 2.089667320251465,
+      "eval_runtime": 200.4158,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 4725
+    },
+    {
+      "epoch": 2.038818201423334,
+      "grad_norm": 0.19026018679141998,
+      "learning_rate": 3.538283627161479e-05,
+      "loss": 2.0147,
+      "step": 4726
+    },
+    {
+      "epoch": 2.0392495147724823,
+      "grad_norm": 0.19227205216884613,
+      "learning_rate": 3.535386225195968e-05,
+      "loss": 2.2522,
+      "step": 4727
+    },
+    {
+      "epoch": 2.0396808281216305,
+      "grad_norm": 0.24319662153720856,
+      "learning_rate": 3.5324896441470916e-05,
+      "loss": 2.0963,
+      "step": 4728
+    },
+    {
+      "epoch": 2.0401121414707784,
+      "grad_norm": 0.17921534180641174,
+      "learning_rate": 3.52959388461462e-05,
+      "loss": 2.0643,
+      "step": 4729
+    },
+    {
+      "epoch": 2.0405434548199266,
+      "grad_norm": 0.21287348866462708,
+      "learning_rate": 3.526698947198157e-05,
+      "loss": 2.1059,
+      "step": 4730
+    },
+    {
+      "epoch": 2.040974768169075,
+      "grad_norm": 0.2085811197757721,
+      "learning_rate": 3.523804832497127e-05,
+      "loss": 1.9961,
+      "step": 4731
+    },
+    {
+      "epoch": 2.041406081518223,
+      "grad_norm": 0.1964728683233261,
+      "learning_rate": 3.520911541110794e-05,
+      "loss": 2.1386,
+      "step": 4732
+    },
+    {
+      "epoch": 2.041837394867371,
+      "grad_norm": 0.1720174103975296,
+      "learning_rate": 3.518019073638235e-05,
+      "loss": 2.0119,
+      "step": 4733
+    },
+    {
+      "epoch": 2.042268708216519,
+      "grad_norm": 0.17993386089801788,
+      "learning_rate": 3.515127430678375e-05,
+      "loss": 2.034,
+      "step": 4734
+    },
+    {
+      "epoch": 2.0427000215656674,
+      "grad_norm": 0.20251108705997467,
+      "learning_rate": 3.5122366128299594e-05,
+      "loss": 2.0783,
+      "step": 4735
+    },
+    {
+      "epoch": 2.0431313349148157,
+      "grad_norm": 0.1712861955165863,
+      "learning_rate": 3.509346620691561e-05,
+      "loss": 2.0154,
+      "step": 4736
+    },
+    {
+      "epoch": 2.043562648263964,
+      "grad_norm": 0.19915875792503357,
+      "learning_rate": 3.506457454861587e-05,
+      "loss": 2.2971,
+      "step": 4737
+    },
+    {
+      "epoch": 2.0439939616131118,
+      "grad_norm": 0.19039981067180634,
+      "learning_rate": 3.5035691159382696e-05,
+      "loss": 2.051,
+      "step": 4738
+    },
+    {
+      "epoch": 2.04442527496226,
+      "grad_norm": 0.19815868139266968,
+      "learning_rate": 3.5006816045196704e-05,
+      "loss": 2.2007,
+      "step": 4739
+    },
+    {
+      "epoch": 2.0448565883114083,
+      "grad_norm": 0.2041560560464859,
+      "learning_rate": 3.497794921203682e-05,
+      "loss": 2.0514,
+      "step": 4740
+    },
+    {
+      "epoch": 2.0452879016605565,
+      "grad_norm": 0.20165760815143585,
+      "learning_rate": 3.4949090665880206e-05,
+      "loss": 2.1084,
+      "step": 4741
+    },
+    {
+      "epoch": 2.0457192150097043,
+      "grad_norm": 0.19834686815738678,
+      "learning_rate": 3.492024041270238e-05,
+      "loss": 2.1179,
+      "step": 4742
+    },
+    {
+      "epoch": 2.0461505283588526,
+      "grad_norm": 0.1835087090730667,
+      "learning_rate": 3.489139845847704e-05,
+      "loss": 2.0642,
+      "step": 4743
+    },
+    {
+      "epoch": 2.046581841708001,
+      "grad_norm": 0.20027287304401398,
+      "learning_rate": 3.4862564809176296e-05,
+      "loss": 2.1164,
+      "step": 4744
+    },
+    {
+      "epoch": 2.047013155057149,
+      "grad_norm": 0.1918504238128662,
+      "learning_rate": 3.4833739470770475e-05,
+      "loss": 2.0904,
+      "step": 4745
+    },
+    {
+      "epoch": 2.0474444684062973,
+      "grad_norm": 0.173620343208313,
+      "learning_rate": 3.480492244922812e-05,
+      "loss": 2.0412,
+      "step": 4746
+    },
+    {
+      "epoch": 2.047875781755445,
+      "grad_norm": 0.1913483589887619,
+      "learning_rate": 3.477611375051612e-05,
+      "loss": 2.0591,
+      "step": 4747
+    },
+    {
+      "epoch": 2.0483070951045934,
+      "grad_norm": 0.1801992505788803,
+      "learning_rate": 3.474731338059966e-05,
+      "loss": 2.0881,
+      "step": 4748
+    },
+    {
+      "epoch": 2.0487384084537417,
+      "grad_norm": 0.19101691246032715,
+      "learning_rate": 3.4718521345442174e-05,
+      "loss": 2.0747,
+      "step": 4749
+    },
+    {
+      "epoch": 2.04916972180289,
+      "grad_norm": 0.1829538494348526,
+      "learning_rate": 3.468973765100535e-05,
+      "loss": 2.0801,
+      "step": 4750
+    },
+    {
+      "epoch": 2.04916972180289,
+      "eval_loss": 2.0897274017333984,
+      "eval_runtime": 201.0233,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4750
+    },
+    {
+      "epoch": 2.0496010351520377,
+      "grad_norm": 0.20499111711978912,
+      "learning_rate": 3.466096230324916e-05,
+      "loss": 2.1689,
+      "step": 4751
+    },
+    {
+      "epoch": 2.050032348501186,
+      "grad_norm": 0.18667557835578918,
+      "learning_rate": 3.463219530813188e-05,
+      "loss": 1.9828,
+      "step": 4752
+    },
+    {
+      "epoch": 2.0504636618503342,
+      "grad_norm": 0.20864151418209076,
+      "learning_rate": 3.4603436671610005e-05,
+      "loss": 2.1977,
+      "step": 4753
+    },
+    {
+      "epoch": 2.0508949751994825,
+      "grad_norm": 0.1942366361618042,
+      "learning_rate": 3.457468639963835e-05,
+      "loss": 2.0802,
+      "step": 4754
+    },
+    {
+      "epoch": 2.0513262885486307,
+      "grad_norm": 0.1807696372270584,
+      "learning_rate": 3.454594449816995e-05,
+      "loss": 1.9287,
+      "step": 4755
+    },
+    {
+      "epoch": 2.0517576018977786,
+      "grad_norm": 0.19067130982875824,
+      "learning_rate": 3.451721097315614e-05,
+      "loss": 2.0488,
+      "step": 4756
+    },
+    {
+      "epoch": 2.052188915246927,
+      "grad_norm": 0.21157202124595642,
+      "learning_rate": 3.448848583054648e-05,
+      "loss": 2.1799,
+      "step": 4757
+    },
+    {
+      "epoch": 2.052620228596075,
+      "grad_norm": 0.1818079799413681,
+      "learning_rate": 3.445976907628889e-05,
+      "loss": 2.0869,
+      "step": 4758
+    },
+    {
+      "epoch": 2.0530515419452233,
+      "grad_norm": 0.198555126786232,
+      "learning_rate": 3.443106071632948e-05,
+      "loss": 2.1775,
+      "step": 4759
+    },
+    {
+      "epoch": 2.0534828552943716,
+      "grad_norm": 0.19541983306407928,
+      "learning_rate": 3.440236075661258e-05,
+      "loss": 2.0928,
+      "step": 4760
+    },
+    {
+      "epoch": 2.0539141686435194,
+      "grad_norm": 0.2012958824634552,
+      "learning_rate": 3.437366920308081e-05,
+      "loss": 2.1032,
+      "step": 4761
+    },
+    {
+      "epoch": 2.0543454819926676,
+      "grad_norm": 0.2103503793478012,
+      "learning_rate": 3.434498606167515e-05,
+      "loss": 2.1628,
+      "step": 4762
+    },
+    {
+      "epoch": 2.054776795341816,
+      "grad_norm": 0.17942892014980316,
+      "learning_rate": 3.4316311338334726e-05,
+      "loss": 2.0742,
+      "step": 4763
+    },
+    {
+      "epoch": 2.055208108690964,
+      "grad_norm": 0.19537143409252167,
+      "learning_rate": 3.4287645038996946e-05,
+      "loss": 2.115,
+      "step": 4764
+    },
+    {
+      "epoch": 2.055639422040112,
+      "grad_norm": 0.18903794884681702,
+      "learning_rate": 3.425898716959748e-05,
+      "loss": 2.0925,
+      "step": 4765
+    },
+    {
+      "epoch": 2.05607073538926,
+      "grad_norm": 0.19615252315998077,
+      "learning_rate": 3.423033773607026e-05,
+      "loss": 2.2102,
+      "step": 4766
+    },
+    {
+      "epoch": 2.0565020487384085,
+      "grad_norm": 0.19369672238826752,
+      "learning_rate": 3.420169674434746e-05,
+      "loss": 2.0199,
+      "step": 4767
+    },
+    {
+      "epoch": 2.0569333620875567,
+      "grad_norm": 0.22473537921905518,
+      "learning_rate": 3.417306420035951e-05,
+      "loss": 2.0792,
+      "step": 4768
+    },
+    {
+      "epoch": 2.057364675436705,
+      "grad_norm": 0.8319330811500549,
+      "learning_rate": 3.41444401100351e-05,
+      "loss": 2.124,
+      "step": 4769
+    },
+    {
+      "epoch": 2.057795988785853,
+      "grad_norm": 0.21528014540672302,
+      "learning_rate": 3.411582447930115e-05,
+      "loss": 2.2047,
+      "step": 4770
+    },
+    {
+      "epoch": 2.058227302135001,
+      "grad_norm": 0.206075057387352,
+      "learning_rate": 3.408721731408281e-05,
+      "loss": 1.9907,
+      "step": 4771
+    },
+    {
+      "epoch": 2.0586586154841493,
+      "grad_norm": 0.20266813039779663,
+      "learning_rate": 3.405861862030357e-05,
+      "loss": 2.0341,
+      "step": 4772
+    },
+    {
+      "epoch": 2.0590899288332976,
+      "grad_norm": 0.19045396149158478,
+      "learning_rate": 3.4030028403885104e-05,
+      "loss": 2.172,
+      "step": 4773
+    },
+    {
+      "epoch": 2.0595212421824454,
+      "grad_norm": 0.18947143852710724,
+      "learning_rate": 3.400144667074727e-05,
+      "loss": 2.2121,
+      "step": 4774
+    },
+    {
+      "epoch": 2.0599525555315936,
+      "grad_norm": 0.18560044467449188,
+      "learning_rate": 3.397287342680823e-05,
+      "loss": 2.0714,
+      "step": 4775
+    },
+    {
+      "epoch": 2.0599525555315936,
+      "eval_loss": 2.089815616607666,
+      "eval_runtime": 200.7389,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 4775
+    },
+    {
+      "epoch": 2.060383868880742,
+      "grad_norm": 0.18350154161453247,
+      "learning_rate": 3.394430867798443e-05,
+      "loss": 2.0451,
+      "step": 4776
+    },
+    {
+      "epoch": 2.06081518222989,
+      "grad_norm": 0.19061952829360962,
+      "learning_rate": 3.3915752430190504e-05,
+      "loss": 2.2662,
+      "step": 4777
+    },
+    {
+      "epoch": 2.0612464955790384,
+      "grad_norm": 0.19650185108184814,
+      "learning_rate": 3.3887204689339354e-05,
+      "loss": 2.059,
+      "step": 4778
+    },
+    {
+      "epoch": 2.061677808928186,
+      "grad_norm": 0.17481555044651031,
+      "learning_rate": 3.3858665461342017e-05,
+      "loss": 2.0273,
+      "step": 4779
+    },
+    {
+      "epoch": 2.0621091222773344,
+      "grad_norm": 0.2005576342344284,
+      "learning_rate": 3.383013475210793e-05,
+      "loss": 1.9869,
+      "step": 4780
+    },
+    {
+      "epoch": 2.0625404356264827,
+      "grad_norm": 0.18315300345420837,
+      "learning_rate": 3.380161256754469e-05,
+      "loss": 2.2135,
+      "step": 4781
+    },
+    {
+      "epoch": 2.062971748975631,
+      "grad_norm": 0.2013261318206787,
+      "learning_rate": 3.377309891355808e-05,
+      "loss": 2.2487,
+      "step": 4782
+    },
+    {
+      "epoch": 2.0634030623247788,
+      "grad_norm": 0.2194448709487915,
+      "learning_rate": 3.3744593796052206e-05,
+      "loss": 1.9857,
+      "step": 4783
+    },
+    {
+      "epoch": 2.063834375673927,
+      "grad_norm": 0.20069117844104767,
+      "learning_rate": 3.3716097220929346e-05,
+      "loss": 2.0198,
+      "step": 4784
+    },
+    {
+      "epoch": 2.0642656890230753,
+      "grad_norm": 0.1908555030822754,
+      "learning_rate": 3.368760919409003e-05,
+      "loss": 2.2062,
+      "step": 4785
+    },
+    {
+      "epoch": 2.0646970023722235,
+      "grad_norm": 0.18771453201770782,
+      "learning_rate": 3.365912972143301e-05,
+      "loss": 2.0383,
+      "step": 4786
+    },
+    {
+      "epoch": 2.065128315721372,
+      "grad_norm": 0.19914081692695618,
+      "learning_rate": 3.3630658808855277e-05,
+      "loss": 2.134,
+      "step": 4787
+    },
+    {
+      "epoch": 2.0655596290705196,
+      "grad_norm": 0.18912968039512634,
+      "learning_rate": 3.3602196462252047e-05,
+      "loss": 2.1419,
+      "step": 4788
+    },
+    {
+      "epoch": 2.065990942419668,
+      "grad_norm": 0.1924545168876648,
+      "learning_rate": 3.3573742687516736e-05,
+      "loss": 2.1747,
+      "step": 4789
+    },
+    {
+      "epoch": 2.066422255768816,
+      "grad_norm": 0.18185552954673767,
+      "learning_rate": 3.354529749054105e-05,
+      "loss": 1.9784,
+      "step": 4790
+    },
+    {
+      "epoch": 2.0668535691179644,
+      "grad_norm": 0.2076537311077118,
+      "learning_rate": 3.351686087721489e-05,
+      "loss": 2.2722,
+      "step": 4791
+    },
+    {
+      "epoch": 2.067284882467112,
+      "grad_norm": 0.1932411938905716,
+      "learning_rate": 3.348843285342632e-05,
+      "loss": 2.1012,
+      "step": 4792
+    },
+    {
+      "epoch": 2.0677161958162604,
+      "grad_norm": 0.1866043508052826,
+      "learning_rate": 3.3460013425061655e-05,
+      "loss": 2.0856,
+      "step": 4793
+    },
+    {
+      "epoch": 2.0681475091654087,
+      "grad_norm": 0.1825817972421646,
+      "learning_rate": 3.3431602598005504e-05,
+      "loss": 2.1824,
+      "step": 4794
+    },
+    {
+      "epoch": 2.068578822514557,
+      "grad_norm": 0.19377149641513824,
+      "learning_rate": 3.340320037814063e-05,
+      "loss": 2.2126,
+      "step": 4795
+    },
+    {
+      "epoch": 2.069010135863705,
+      "grad_norm": 0.19704613089561462,
+      "learning_rate": 3.3374806771348006e-05,
+      "loss": 2.2377,
+      "step": 4796
+    },
+    {
+      "epoch": 2.069441449212853,
+      "grad_norm": 0.19628547132015228,
+      "learning_rate": 3.334642178350685e-05,
+      "loss": 2.0471,
+      "step": 4797
+    },
+    {
+      "epoch": 2.0698727625620013,
+      "grad_norm": 0.19294996559619904,
+      "learning_rate": 3.331804542049458e-05,
+      "loss": 2.223,
+      "step": 4798
+    },
+    {
+      "epoch": 2.0703040759111495,
+      "grad_norm": 0.20850534737110138,
+      "learning_rate": 3.328967768818683e-05,
+      "loss": 2.1281,
+      "step": 4799
+    },
+    {
+      "epoch": 2.0707353892602978,
+      "grad_norm": 0.18277223408222198,
+      "learning_rate": 3.3261318592457455e-05,
+      "loss": 2.0091,
+      "step": 4800
+    },
+    {
+      "epoch": 2.0707353892602978,
+      "eval_loss": 2.0896167755126953,
+      "eval_runtime": 197.9135,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 4800
+    },
+    {
+      "epoch": 2.0711667026094456,
+      "grad_norm": 0.1938290148973465,
+      "learning_rate": 3.323296813917851e-05,
+      "loss": 2.2023,
+      "step": 4801
+    },
+    {
+      "epoch": 2.071598015958594,
+      "grad_norm": 0.18563704192638397,
+      "learning_rate": 3.320462633422026e-05,
+      "loss": 2.1932,
+      "step": 4802
+    },
+    {
+      "epoch": 2.072029329307742,
+      "grad_norm": 0.18315260112285614,
+      "learning_rate": 3.317629318345117e-05,
+      "loss": 2.0308,
+      "step": 4803
+    },
+    {
+      "epoch": 2.0724606426568903,
+      "grad_norm": 0.18531833589076996,
+      "learning_rate": 3.314796869273799e-05,
+      "loss": 2.0022,
+      "step": 4804
+    },
+    {
+      "epoch": 2.0728919560060386,
+      "grad_norm": 0.20288553833961487,
+      "learning_rate": 3.31196528679456e-05,
+      "loss": 2.2519,
+      "step": 4805
+    },
+    {
+      "epoch": 2.0733232693551864,
+      "grad_norm": 0.19610491394996643,
+      "learning_rate": 3.3091345714937044e-05,
+      "loss": 2.1821,
+      "step": 4806
+    },
+    {
+      "epoch": 2.0737545827043347,
+      "grad_norm": 0.21335402131080627,
+      "learning_rate": 3.306304723957364e-05,
+      "loss": 2.0207,
+      "step": 4807
+    },
+    {
+      "epoch": 2.074185896053483,
+      "grad_norm": 0.20559577643871307,
+      "learning_rate": 3.303475744771495e-05,
+      "loss": 2.0895,
+      "step": 4808
+    },
+    {
+      "epoch": 2.074617209402631,
+      "grad_norm": 0.2079014629125595,
+      "learning_rate": 3.300647634521865e-05,
+      "loss": 2.2938,
+      "step": 4809
+    },
+    {
+      "epoch": 2.075048522751779,
+      "grad_norm": 0.1839982271194458,
+      "learning_rate": 3.297820393794067e-05,
+      "loss": 1.927,
+      "step": 4810
+    },
+    {
+      "epoch": 2.0754798361009272,
+      "grad_norm": 0.17540310323238373,
+      "learning_rate": 3.294994023173509e-05,
+      "loss": 1.9777,
+      "step": 4811
+    },
+    {
+      "epoch": 2.0759111494500755,
+      "grad_norm": 0.2142418920993805,
+      "learning_rate": 3.2921685232454246e-05,
+      "loss": 2.2027,
+      "step": 4812
+    },
+    {
+      "epoch": 2.0763424627992237,
+      "grad_norm": 0.18835563957691193,
+      "learning_rate": 3.289343894594864e-05,
+      "loss": 2.306,
+      "step": 4813
+    },
+    {
+      "epoch": 2.076773776148372,
+      "grad_norm": 0.2013714760541916,
+      "learning_rate": 3.286520137806696e-05,
+      "loss": 2.1524,
+      "step": 4814
+    },
+    {
+      "epoch": 2.07720508949752,
+      "grad_norm": 0.18839602172374725,
+      "learning_rate": 3.2836972534656115e-05,
+      "loss": 2.096,
+      "step": 4815
+    },
+    {
+      "epoch": 2.077636402846668,
+      "grad_norm": 0.19380220770835876,
+      "learning_rate": 3.280875242156119e-05,
+      "loss": 2.0306,
+      "step": 4816
+    },
+    {
+      "epoch": 2.0780677161958163,
+      "grad_norm": 0.19123317301273346,
+      "learning_rate": 3.278054104462545e-05,
+      "loss": 2.1462,
+      "step": 4817
+    },
+    {
+      "epoch": 2.0784990295449646,
+      "grad_norm": 0.17127066850662231,
+      "learning_rate": 3.275233840969045e-05,
+      "loss": 1.9398,
+      "step": 4818
+    },
+    {
+      "epoch": 2.0789303428941124,
+      "grad_norm": 9.493552207946777,
+      "learning_rate": 3.2724144522595755e-05,
+      "loss": 2.2694,
+      "step": 4819
+    },
+    {
+      "epoch": 2.0793616562432606,
+      "grad_norm": 0.20617228746414185,
+      "learning_rate": 3.269595938917927e-05,
+      "loss": 2.2542,
+      "step": 4820
+    },
+    {
+      "epoch": 2.079792969592409,
+      "grad_norm": 0.1898934245109558,
+      "learning_rate": 3.266778301527699e-05,
+      "loss": 2.2002,
+      "step": 4821
+    },
+    {
+      "epoch": 2.080224282941557,
+      "grad_norm": 5.428417682647705,
+      "learning_rate": 3.26396154067232e-05,
+      "loss": 1.76,
+      "step": 4822
+    },
+    {
+      "epoch": 2.0806555962907054,
+      "grad_norm": 0.19115063548088074,
+      "learning_rate": 3.2611456569350316e-05,
+      "loss": 2.2404,
+      "step": 4823
+    },
+    {
+      "epoch": 2.081086909639853,
+      "grad_norm": 0.20664536952972412,
+      "learning_rate": 3.2583306508988874e-05,
+      "loss": 2.3256,
+      "step": 4824
+    },
+    {
+      "epoch": 2.0815182229890015,
+      "grad_norm": 0.18666943907737732,
+      "learning_rate": 3.255516523146765e-05,
+      "loss": 1.9698,
+      "step": 4825
+    },
+    {
+      "epoch": 2.0815182229890015,
+      "eval_loss": 2.089388370513916,
+      "eval_runtime": 195.2277,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 4825
+    },
+    {
+      "epoch": 2.0819495363381497,
+      "grad_norm": 0.18007901310920715,
+      "learning_rate": 3.252703274261367e-05,
+      "loss": 2.1054,
+      "step": 4826
+    },
+    {
+      "epoch": 2.082380849687298,
+      "grad_norm": 0.17201073467731476,
+      "learning_rate": 3.2498909048252024e-05,
+      "loss": 2.0739,
+      "step": 4827
+    },
+    {
+      "epoch": 2.082812163036446,
+      "grad_norm": 0.1966872364282608,
+      "learning_rate": 3.247079415420606e-05,
+      "loss": 2.2652,
+      "step": 4828
+    },
+    {
+      "epoch": 2.083243476385594,
+      "grad_norm": 0.19481778144836426,
+      "learning_rate": 3.244268806629724e-05,
+      "loss": 2.1652,
+      "step": 4829
+    },
+    {
+      "epoch": 2.0836747897347423,
+      "grad_norm": 0.18113017082214355,
+      "learning_rate": 3.2414590790345266e-05,
+      "loss": 2.0232,
+      "step": 4830
+    },
+    {
+      "epoch": 2.0841061030838905,
+      "grad_norm": 0.2021803855895996,
+      "learning_rate": 3.238650233216797e-05,
+      "loss": 2.2541,
+      "step": 4831
+    },
+    {
+      "epoch": 2.084537416433039,
+      "grad_norm": 0.19255667924880981,
+      "learning_rate": 3.235842269758139e-05,
+      "loss": 2.1073,
+      "step": 4832
+    },
+    {
+      "epoch": 2.0849687297821866,
+      "grad_norm": 0.18869124352931976,
+      "learning_rate": 3.23303518923997e-05,
+      "loss": 2.0478,
+      "step": 4833
+    },
+    {
+      "epoch": 2.085400043131335,
+      "grad_norm": 1.98809814453125,
+      "learning_rate": 3.230228992243528e-05,
+      "loss": 2.1051,
+      "step": 4834
+    },
+    {
+      "epoch": 2.085831356480483,
+      "grad_norm": 0.22017183899879456,
+      "learning_rate": 3.227423679349864e-05,
+      "loss": 2.3927,
+      "step": 4835
+    },
+    {
+      "epoch": 2.0862626698296314,
+      "grad_norm": 0.18822191655635834,
+      "learning_rate": 3.224619251139854e-05,
+      "loss": 2.0913,
+      "step": 4836
+    },
+    {
+      "epoch": 2.086693983178779,
+      "grad_norm": 0.18337860703468323,
+      "learning_rate": 3.221815708194185e-05,
+      "loss": 2.2166,
+      "step": 4837
+    },
+    {
+      "epoch": 2.0871252965279274,
+      "grad_norm": 0.19429202377796173,
+      "learning_rate": 3.2190130510933564e-05,
+      "loss": 2.1312,
+      "step": 4838
+    },
+    {
+      "epoch": 2.0875566098770757,
+      "grad_norm": 0.18895988166332245,
+      "learning_rate": 3.2162112804176896e-05,
+      "loss": 2.1958,
+      "step": 4839
+    },
+    {
+      "epoch": 2.087987923226224,
+      "grad_norm": 0.18243250250816345,
+      "learning_rate": 3.213410396747326e-05,
+      "loss": 2.1509,
+      "step": 4840
+    },
+    {
+      "epoch": 2.088419236575372,
+      "grad_norm": 0.2050832360982895,
+      "learning_rate": 3.2106104006622176e-05,
+      "loss": 2.1202,
+      "step": 4841
+    },
+    {
+      "epoch": 2.08885054992452,
+      "grad_norm": 0.4805981516838074,
+      "learning_rate": 3.207811292742135e-05,
+      "loss": 2.3342,
+      "step": 4842
+    },
+    {
+      "epoch": 2.0892818632736683,
+      "grad_norm": 0.18699911236763,
+      "learning_rate": 3.205013073566661e-05,
+      "loss": 2.0611,
+      "step": 4843
+    },
+    {
+      "epoch": 2.0897131766228165,
+      "grad_norm": 0.2027273327112198,
+      "learning_rate": 3.2022157437152e-05,
+      "loss": 2.3847,
+      "step": 4844
+    },
+    {
+      "epoch": 2.090144489971965,
+      "grad_norm": 0.18583311140537262,
+      "learning_rate": 3.199419303766968e-05,
+      "loss": 2.1031,
+      "step": 4845
+    },
+    {
+      "epoch": 2.0905758033211126,
+      "grad_norm": 0.18679817020893097,
+      "learning_rate": 3.196623754301e-05,
+      "loss": 2.1334,
+      "step": 4846
+    },
+    {
+      "epoch": 2.091007116670261,
+      "grad_norm": 0.2197234332561493,
+      "learning_rate": 3.193829095896145e-05,
+      "loss": 2.1589,
+      "step": 4847
+    },
+    {
+      "epoch": 2.091438430019409,
+      "grad_norm": 0.19306516647338867,
+      "learning_rate": 3.191035329131067e-05,
+      "loss": 2.0514,
+      "step": 4848
+    },
+    {
+      "epoch": 2.0918697433685574,
+      "grad_norm": 0.19103452563285828,
+      "learning_rate": 3.1882424545842435e-05,
+      "loss": 2.2015,
+      "step": 4849
+    },
+    {
+      "epoch": 2.0923010567177056,
+      "grad_norm": 0.2152070254087448,
+      "learning_rate": 3.1854504728339754e-05,
+      "loss": 2.2573,
+      "step": 4850
+    },
+    {
+      "epoch": 2.0923010567177056,
+      "eval_loss": 2.0900402069091797,
+      "eval_runtime": 195.8281,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 4850
+    },
+    {
+      "epoch": 2.0927323700668534,
+      "grad_norm": 4.18394136428833,
+      "learning_rate": 3.182659384458372e-05,
+      "loss": 1.9889,
+      "step": 4851
+    },
+    {
+      "epoch": 2.0931636834160017,
+      "grad_norm": 0.2186499387025833,
+      "learning_rate": 3.179869190035355e-05,
+      "loss": 2.1699,
+      "step": 4852
+    },
+    {
+      "epoch": 2.09359499676515,
+      "grad_norm": 0.20570923388004303,
+      "learning_rate": 3.1770798901426624e-05,
+      "loss": 2.1021,
+      "step": 4853
+    },
+    {
+      "epoch": 2.094026310114298,
+      "grad_norm": 0.5099383592605591,
+      "learning_rate": 3.174291485357856e-05,
+      "loss": 2.1402,
+      "step": 4854
+    },
+    {
+      "epoch": 2.094457623463446,
+      "grad_norm": 0.1793668121099472,
+      "learning_rate": 3.1715039762583034e-05,
+      "loss": 2.0289,
+      "step": 4855
+    },
+    {
+      "epoch": 2.0948889368125942,
+      "grad_norm": 0.19396327435970306,
+      "learning_rate": 3.1687173634211884e-05,
+      "loss": 2.0263,
+      "step": 4856
+    },
+    {
+      "epoch": 2.0953202501617425,
+      "grad_norm": 0.2883221507072449,
+      "learning_rate": 3.165931647423509e-05,
+      "loss": 2.1575,
+      "step": 4857
+    },
+    {
+      "epoch": 2.0957515635108908,
+      "grad_norm": 0.2391921877861023,
+      "learning_rate": 3.1631468288420786e-05,
+      "loss": 2.113,
+      "step": 4858
+    },
+    {
+      "epoch": 2.096182876860039,
+      "grad_norm": 0.21463486552238464,
+      "learning_rate": 3.160362908253524e-05,
+      "loss": 2.2261,
+      "step": 4859
+    },
+    {
+      "epoch": 2.096614190209187,
+      "grad_norm": 0.20496898889541626,
+      "learning_rate": 3.157579886234287e-05,
+      "loss": 2.1229,
+      "step": 4860
+    },
+    {
+      "epoch": 2.097045503558335,
+      "grad_norm": 0.21032458543777466,
+      "learning_rate": 3.1547977633606226e-05,
+      "loss": 2.1148,
+      "step": 4861
+    },
+    {
+      "epoch": 2.0974768169074833,
+      "grad_norm": 0.20047256350517273,
+      "learning_rate": 3.152016540208599e-05,
+      "loss": 2.2359,
+      "step": 4862
+    },
+    {
+      "epoch": 2.0979081302566316,
+      "grad_norm": 0.20530466735363007,
+      "learning_rate": 3.149236217354098e-05,
+      "loss": 2.1404,
+      "step": 4863
+    },
+    {
+      "epoch": 2.0983394436057794,
+      "grad_norm": 0.1921357959508896,
+      "learning_rate": 3.146456795372822e-05,
+      "loss": 2.2775,
+      "step": 4864
+    },
+    {
+      "epoch": 2.0987707569549277,
+      "grad_norm": 0.18893566727638245,
+      "learning_rate": 3.143678274840273e-05,
+      "loss": 2.0688,
+      "step": 4865
+    },
+    {
+      "epoch": 2.099202070304076,
+      "grad_norm": 0.19544082880020142,
+      "learning_rate": 3.140900656331778e-05,
+      "loss": 2.3392,
+      "step": 4866
+    },
+    {
+      "epoch": 2.099633383653224,
+      "grad_norm": 0.20501480996608734,
+      "learning_rate": 3.138123940422469e-05,
+      "loss": 2.3066,
+      "step": 4867
+    },
+    {
+      "epoch": 2.1000646970023724,
+      "grad_norm": 0.20116248726844788,
+      "learning_rate": 3.135348127687302e-05,
+      "loss": 2.1698,
+      "step": 4868
+    },
+    {
+      "epoch": 2.1004960103515202,
+      "grad_norm": 0.18571919202804565,
+      "learning_rate": 3.1325732187010385e-05,
+      "loss": 2.0816,
+      "step": 4869
+    },
+    {
+      "epoch": 2.1009273237006685,
+      "grad_norm": 0.22345946729183197,
+      "learning_rate": 3.1297992140382486e-05,
+      "loss": 2.177,
+      "step": 4870
+    },
+    {
+      "epoch": 2.1013586370498167,
+      "grad_norm": 0.24064777791500092,
+      "learning_rate": 3.127026114273321e-05,
+      "loss": 1.808,
+      "step": 4871
+    },
+    {
+      "epoch": 2.101789950398965,
+      "grad_norm": 0.19728608429431915,
+      "learning_rate": 3.124253919980461e-05,
+      "loss": 2.3509,
+      "step": 4872
+    },
+    {
+      "epoch": 2.102221263748113,
+      "grad_norm": 0.20240743458271027,
+      "learning_rate": 3.121482631733678e-05,
+      "loss": 2.2624,
+      "step": 4873
+    },
+    {
+      "epoch": 2.102652577097261,
+      "grad_norm": 65.099853515625,
+      "learning_rate": 3.1187122501068e-05,
+      "loss": 1.9778,
+      "step": 4874
+    },
+    {
+      "epoch": 2.1030838904464093,
+      "grad_norm": 0.19699519872665405,
+      "learning_rate": 3.115942775673464e-05,
+      "loss": 2.1841,
+      "step": 4875
+    },
+    {
+      "epoch": 2.1030838904464093,
+      "eval_loss": 2.0889768600463867,
+      "eval_runtime": 195.1623,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 4875
+    },
+    {
+      "epoch": 2.1035152037955576,
+      "grad_norm": 0.18935461342334747,
+      "learning_rate": 3.113174209007118e-05,
+      "loss": 2.1658,
+      "step": 4876
+    },
+    {
+      "epoch": 2.103946517144706,
+      "grad_norm": 0.19691212475299835,
+      "learning_rate": 3.110406550681027e-05,
+      "loss": 2.2538,
+      "step": 4877
+    },
+    {
+      "epoch": 2.1043778304938536,
+      "grad_norm": 0.20997954905033112,
+      "learning_rate": 3.107639801268263e-05,
+      "loss": 2.1355,
+      "step": 4878
+    },
+    {
+      "epoch": 2.104809143843002,
+      "grad_norm": 0.17843768000602722,
+      "learning_rate": 3.104873961341712e-05,
+      "loss": 2.1078,
+      "step": 4879
+    },
+    {
+      "epoch": 2.10524045719215,
+      "grad_norm": 0.1926320642232895,
+      "learning_rate": 3.102109031474071e-05,
+      "loss": 2.4314,
+      "step": 4880
+    },
+    {
+      "epoch": 2.1056717705412984,
+      "grad_norm": 0.19593141973018646,
+      "learning_rate": 3.099345012237848e-05,
+      "loss": 2.156,
+      "step": 4881
+    },
+    {
+      "epoch": 2.106103083890446,
+      "grad_norm": 0.2088003307580948,
+      "learning_rate": 3.0965819042053674e-05,
+      "loss": 2.1124,
+      "step": 4882
+    },
+    {
+      "epoch": 2.1065343972395945,
+      "grad_norm": 0.19262035191059113,
+      "learning_rate": 3.093819707948761e-05,
+      "loss": 2.2852,
+      "step": 4883
+    },
+    {
+      "epoch": 2.1069657105887427,
+      "grad_norm": 0.22138433158397675,
+      "learning_rate": 3.0910584240399665e-05,
+      "loss": 2.2125,
+      "step": 4884
+    },
+    {
+      "epoch": 2.107397023937891,
+      "grad_norm": 0.8872026801109314,
+      "learning_rate": 3.088298053050738e-05,
+      "loss": 2.2015,
+      "step": 4885
+    },
+    {
+      "epoch": 2.1078283372870392,
+      "grad_norm": 0.1949310153722763,
+      "learning_rate": 3.085538595552646e-05,
+      "loss": 2.0707,
+      "step": 4886
+    },
+    {
+      "epoch": 2.108259650636187,
+      "grad_norm": 0.19012707471847534,
+      "learning_rate": 3.0827800521170634e-05,
+      "loss": 2.1818,
+      "step": 4887
+    },
+    {
+      "epoch": 2.1086909639853353,
+      "grad_norm": 0.20888254046440125,
+      "learning_rate": 3.080022423315177e-05,
+      "loss": 2.2013,
+      "step": 4888
+    },
+    {
+      "epoch": 2.1091222773344835,
+      "grad_norm": 0.23211535811424255,
+      "learning_rate": 3.077265709717984e-05,
+      "loss": 1.8655,
+      "step": 4889
+    },
+    {
+      "epoch": 2.109553590683632,
+      "grad_norm": 0.2039070874452591,
+      "learning_rate": 3.074509911896292e-05,
+      "loss": 2.0674,
+      "step": 4890
+    },
+    {
+      "epoch": 2.1099849040327796,
+      "grad_norm": 0.20388320088386536,
+      "learning_rate": 3.071755030420719e-05,
+      "loss": 2.1095,
+      "step": 4891
+    },
+    {
+      "epoch": 2.110416217381928,
+      "grad_norm": 0.2010328471660614,
+      "learning_rate": 3.0690010658616926e-05,
+      "loss": 2.1818,
+      "step": 4892
+    },
+    {
+      "epoch": 2.110847530731076,
+      "grad_norm": 0.20505084097385406,
+      "learning_rate": 3.066248018789454e-05,
+      "loss": 2.0214,
+      "step": 4893
+    },
+    {
+      "epoch": 2.1112788440802244,
+      "grad_norm": 0.19544564187526703,
+      "learning_rate": 3.0634958897740486e-05,
+      "loss": 1.857,
+      "step": 4894
+    },
+    {
+      "epoch": 2.1117101574293726,
+      "grad_norm": 0.20017707347869873,
+      "learning_rate": 3.0607446793853326e-05,
+      "loss": 2.2595,
+      "step": 4895
+    },
+    {
+      "epoch": 2.1121414707785204,
+      "grad_norm": 0.21018147468566895,
+      "learning_rate": 3.057994388192981e-05,
+      "loss": 2.0568,
+      "step": 4896
+    },
+    {
+      "epoch": 2.1125727841276687,
+      "grad_norm": 0.20233690738677979,
+      "learning_rate": 3.0552450167664716e-05,
+      "loss": 2.3687,
+      "step": 4897
+    },
+    {
+      "epoch": 2.113004097476817,
+      "grad_norm": 0.20755895972251892,
+      "learning_rate": 3.0524965656750845e-05,
+      "loss": 2.2019,
+      "step": 4898
+    },
+    {
+      "epoch": 2.113435410825965,
+      "grad_norm": 0.22112730145454407,
+      "learning_rate": 3.049749035487919e-05,
+      "loss": 2.1214,
+      "step": 4899
+    },
+    {
+      "epoch": 2.113866724175113,
+      "grad_norm": 0.1795358955860138,
+      "learning_rate": 3.0470024267738848e-05,
+      "loss": 2.0371,
+      "step": 4900
+    },
+    {
+      "epoch": 2.113866724175113,
+      "eval_loss": 2.0892462730407715,
+      "eval_runtime": 195.3029,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 4900
+    },
+    {
+      "epoch": 2.1142980375242613,
+      "grad_norm": 0.18432939052581787,
+      "learning_rate": 3.044256740101696e-05,
+      "loss": 2.2024,
+      "step": 4901
+    },
+    {
+      "epoch": 2.1147293508734095,
+      "grad_norm": 0.195393368601799,
+      "learning_rate": 3.041511976039876e-05,
+      "loss": 2.1915,
+      "step": 4902
+    },
+    {
+      "epoch": 2.1151606642225578,
+      "grad_norm": 0.21330732107162476,
+      "learning_rate": 3.0387681351567586e-05,
+      "loss": 2.1721,
+      "step": 4903
+    },
+    {
+      "epoch": 2.115591977571706,
+      "grad_norm": 0.1951751857995987,
+      "learning_rate": 3.036025218020486e-05,
+      "loss": 2.1533,
+      "step": 4904
+    },
+    {
+      "epoch": 2.116023290920854,
+      "grad_norm": 0.2118057757616043,
+      "learning_rate": 3.0332832251990092e-05,
+      "loss": 2.1225,
+      "step": 4905
+    },
+    {
+      "epoch": 2.116454604270002,
+      "grad_norm": 0.17883312702178955,
+      "learning_rate": 3.0305421572600876e-05,
+      "loss": 2.022,
+      "step": 4906
+    },
+    {
+      "epoch": 2.1168859176191503,
+      "grad_norm": 0.20854587852954865,
+      "learning_rate": 3.0278020147712895e-05,
+      "loss": 2.2135,
+      "step": 4907
+    },
+    {
+      "epoch": 2.1173172309682986,
+      "grad_norm": 0.1910894513130188,
+      "learning_rate": 3.0250627982999915e-05,
+      "loss": 2.1853,
+      "step": 4908
+    },
+    {
+      "epoch": 2.1177485443174464,
+      "grad_norm": 0.20413783192634583,
+      "learning_rate": 3.022324508413376e-05,
+      "loss": 2.0861,
+      "step": 4909
+    },
+    {
+      "epoch": 2.1181798576665947,
+      "grad_norm": 0.2057296335697174,
+      "learning_rate": 3.019587145678443e-05,
+      "loss": 2.227,
+      "step": 4910
+    },
+    {
+      "epoch": 2.118611171015743,
+      "grad_norm": 0.19873298704624176,
+      "learning_rate": 3.0168507106619856e-05,
+      "loss": 2.1258,
+      "step": 4911
+    },
+    {
+      "epoch": 2.119042484364891,
+      "grad_norm": 0.20596614480018616,
+      "learning_rate": 3.014115203930615e-05,
+      "loss": 2.2611,
+      "step": 4912
+    },
+    {
+      "epoch": 2.1194737977140394,
+      "grad_norm": 0.8568402528762817,
+      "learning_rate": 3.0113806260507466e-05,
+      "loss": 2.333,
+      "step": 4913
+    },
+    {
+      "epoch": 2.1199051110631872,
+      "grad_norm": 0.18790197372436523,
+      "learning_rate": 3.008646977588609e-05,
+      "loss": 2.1652,
+      "step": 4914
+    },
+    {
+      "epoch": 2.1203364244123355,
+      "grad_norm": 0.19961890578269958,
+      "learning_rate": 3.0059142591102328e-05,
+      "loss": 2.083,
+      "step": 4915
+    },
+    {
+      "epoch": 2.1207677377614838,
+      "grad_norm": 0.19002991914749146,
+      "learning_rate": 3.0031824711814517e-05,
+      "loss": 2.2128,
+      "step": 4916
+    },
+    {
+      "epoch": 2.121199051110632,
+      "grad_norm": 0.1990480273962021,
+      "learning_rate": 3.000451614367918e-05,
+      "loss": 2.1486,
+      "step": 4917
+    },
+    {
+      "epoch": 2.12163036445978,
+      "grad_norm": 0.1983197182416916,
+      "learning_rate": 2.9977216892350845e-05,
+      "loss": 2.1523,
+      "step": 4918
+    },
+    {
+      "epoch": 2.122061677808928,
+      "grad_norm": 0.20927980542182922,
+      "learning_rate": 2.9949926963482106e-05,
+      "loss": 2.2411,
+      "step": 4919
+    },
+    {
+      "epoch": 2.1224929911580763,
+      "grad_norm": 1.3745379447937012,
+      "learning_rate": 2.9922646362723658e-05,
+      "loss": 2.1764,
+      "step": 4920
+    },
+    {
+      "epoch": 2.1229243045072246,
+      "grad_norm": 0.20675300061702728,
+      "learning_rate": 2.9895375095724223e-05,
+      "loss": 2.1202,
+      "step": 4921
+    },
+    {
+      "epoch": 2.123355617856373,
+      "grad_norm": 0.18410125374794006,
+      "learning_rate": 2.986811316813064e-05,
+      "loss": 2.2244,
+      "step": 4922
+    },
+    {
+      "epoch": 2.1237869312055206,
+      "grad_norm": 0.4287169575691223,
+      "learning_rate": 2.9840860585587783e-05,
+      "loss": 2.3137,
+      "step": 4923
+    },
+    {
+      "epoch": 2.124218244554669,
+      "grad_norm": 0.19607892632484436,
+      "learning_rate": 2.9813617353738596e-05,
+      "loss": 2.0012,
+      "step": 4924
+    },
+    {
+      "epoch": 2.124649557903817,
+      "grad_norm": 0.19588974118232727,
+      "learning_rate": 2.9786383478224072e-05,
+      "loss": 2.2361,
+      "step": 4925
+    },
+    {
+      "epoch": 2.124649557903817,
+      "eval_loss": 2.089700222015381,
+      "eval_runtime": 195.2735,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 4925
+    },
+    {
+      "epoch": 2.1250808712529654,
+      "grad_norm": 0.1861879676580429,
+      "learning_rate": 2.975915896468331e-05,
+      "loss": 2.1908,
+      "step": 4926
+    },
+    {
+      "epoch": 2.1255121846021137,
+      "grad_norm": 0.1903456449508667,
+      "learning_rate": 2.9731943818753394e-05,
+      "loss": 2.1286,
+      "step": 4927
+    },
+    {
+      "epoch": 2.1259434979512615,
+      "grad_norm": 0.19927555322647095,
+      "learning_rate": 2.9704738046069584e-05,
+      "loss": 2.1186,
+      "step": 4928
+    },
+    {
+      "epoch": 2.1263748113004097,
+      "grad_norm": 0.20145562291145325,
+      "learning_rate": 2.967754165226513e-05,
+      "loss": 2.1554,
+      "step": 4929
+    },
+    {
+      "epoch": 2.126806124649558,
+      "grad_norm": 0.18136072158813477,
+      "learning_rate": 2.9650354642971286e-05,
+      "loss": 1.9959,
+      "step": 4930
+    },
+    {
+      "epoch": 2.1272374379987062,
+      "grad_norm": 0.1880529671907425,
+      "learning_rate": 2.9623177023817423e-05,
+      "loss": 2.0683,
+      "step": 4931
+    },
+    {
+      "epoch": 2.127668751347854,
+      "grad_norm": 0.1958688646554947,
+      "learning_rate": 2.959600880043102e-05,
+      "loss": 2.3095,
+      "step": 4932
+    },
+    {
+      "epoch": 2.1281000646970023,
+      "grad_norm": 0.19369158148765564,
+      "learning_rate": 2.9568849978437517e-05,
+      "loss": 2.072,
+      "step": 4933
+    },
+    {
+      "epoch": 2.1285313780461506,
+      "grad_norm": 0.19699695706367493,
+      "learning_rate": 2.9541700563460457e-05,
+      "loss": 2.2225,
+      "step": 4934
+    },
+    {
+      "epoch": 2.128962691395299,
+      "grad_norm": 0.21904636919498444,
+      "learning_rate": 2.951456056112142e-05,
+      "loss": 1.9931,
+      "step": 4935
+    },
+    {
+      "epoch": 2.129394004744447,
+      "grad_norm": 0.18061836063861847,
+      "learning_rate": 2.9487429977040034e-05,
+      "loss": 2.3713,
+      "step": 4936
+    },
+    {
+      "epoch": 2.129825318093595,
+      "grad_norm": 0.23203131556510925,
+      "learning_rate": 2.9460308816833988e-05,
+      "loss": 2.1424,
+      "step": 4937
+    },
+    {
+      "epoch": 2.130256631442743,
+      "grad_norm": 0.19963587820529938,
+      "learning_rate": 2.9433197086119022e-05,
+      "loss": 2.0433,
+      "step": 4938
+    },
+    {
+      "epoch": 2.1306879447918914,
+      "grad_norm": 0.1829756796360016,
+      "learning_rate": 2.9406094790508904e-05,
+      "loss": 2.0641,
+      "step": 4939
+    },
+    {
+      "epoch": 2.1311192581410396,
+      "grad_norm": 0.21752561628818512,
+      "learning_rate": 2.937900193561547e-05,
+      "loss": 2.1851,
+      "step": 4940
+    },
+    {
+      "epoch": 2.1315505714901875,
+      "grad_norm": 0.1959228664636612,
+      "learning_rate": 2.9351918527048557e-05,
+      "loss": 2.3427,
+      "step": 4941
+    },
+    {
+      "epoch": 2.1319818848393357,
+      "grad_norm": 0.17727304995059967,
+      "learning_rate": 2.932484457041617e-05,
+      "loss": 2.0664,
+      "step": 4942
+    },
+    {
+      "epoch": 2.132413198188484,
+      "grad_norm": 0.18500760197639465,
+      "learning_rate": 2.9297780071324184e-05,
+      "loss": 2.2685,
+      "step": 4943
+    },
+    {
+      "epoch": 2.132844511537632,
+      "grad_norm": 0.19489331543445587,
+      "learning_rate": 2.9270725035376625e-05,
+      "loss": 2.0761,
+      "step": 4944
+    },
+    {
+      "epoch": 2.1332758248867805,
+      "grad_norm": 0.19658870995044708,
+      "learning_rate": 2.92436794681755e-05,
+      "loss": 2.1817,
+      "step": 4945
+    },
+    {
+      "epoch": 2.1337071382359283,
+      "grad_norm": 0.1881716102361679,
+      "learning_rate": 2.921664337532095e-05,
+      "loss": 2.1831,
+      "step": 4946
+    },
+    {
+      "epoch": 2.1341384515850765,
+      "grad_norm": 0.19964499771595,
+      "learning_rate": 2.918961676241111e-05,
+      "loss": 2.2812,
+      "step": 4947
+    },
+    {
+      "epoch": 2.134569764934225,
+      "grad_norm": 0.19984838366508484,
+      "learning_rate": 2.9162599635042025e-05,
+      "loss": 2.214,
+      "step": 4948
+    },
+    {
+      "epoch": 2.135001078283373,
+      "grad_norm": 0.3285719156265259,
+      "learning_rate": 2.9135591998807985e-05,
+      "loss": 2.0155,
+      "step": 4949
+    },
+    {
+      "epoch": 2.135432391632521,
+      "grad_norm": 0.2052600383758545,
+      "learning_rate": 2.9108593859301185e-05,
+      "loss": 2.1134,
+      "step": 4950
+    },
+    {
+      "epoch": 2.135432391632521,
+      "eval_loss": 2.0897464752197266,
+      "eval_runtime": 195.252,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 4950
+    },
+    {
+      "epoch": 2.135863704981669,
+      "grad_norm": 0.18723013997077942,
+      "learning_rate": 2.9081605222111893e-05,
+      "loss": 1.8742,
+      "step": 4951
+    },
+    {
+      "epoch": 2.1362950183308174,
+      "grad_norm": 0.20071329176425934,
+      "learning_rate": 2.9054626092828386e-05,
+      "loss": 2.1228,
+      "step": 4952
+    },
+    {
+      "epoch": 2.1367263316799656,
+      "grad_norm": 0.19112932682037354,
+      "learning_rate": 2.9027656477037002e-05,
+      "loss": 2.1832,
+      "step": 4953
+    },
+    {
+      "epoch": 2.137157645029114,
+      "grad_norm": 0.1887974590063095,
+      "learning_rate": 2.900069638032209e-05,
+      "loss": 2.0068,
+      "step": 4954
+    },
+    {
+      "epoch": 2.1375889583782617,
+      "grad_norm": 0.2175447791814804,
+      "learning_rate": 2.8973745808266e-05,
+      "loss": 2.4054,
+      "step": 4955
+    },
+    {
+      "epoch": 2.13802027172741,
+      "grad_norm": 0.19409482181072235,
+      "learning_rate": 2.8946804766449234e-05,
+      "loss": 2.2914,
+      "step": 4956
+    },
+    {
+      "epoch": 2.138451585076558,
+      "grad_norm": 0.20012657344341278,
+      "learning_rate": 2.8919873260450146e-05,
+      "loss": 2.3076,
+      "step": 4957
+    },
+    {
+      "epoch": 2.1388828984257064,
+      "grad_norm": 0.19830457866191864,
+      "learning_rate": 2.8892951295845222e-05,
+      "loss": 2.336,
+      "step": 4958
+    },
+    {
+      "epoch": 2.1393142117748543,
+      "grad_norm": 0.1941172331571579,
+      "learning_rate": 2.8866038878208915e-05,
+      "loss": 2.2341,
+      "step": 4959
+    },
+    {
+      "epoch": 2.1397455251240025,
+      "grad_norm": 0.20140814781188965,
+      "learning_rate": 2.883913601311381e-05,
+      "loss": 2.249,
+      "step": 4960
+    },
+    {
+      "epoch": 2.1401768384731508,
+      "grad_norm": 0.19238823652267456,
+      "learning_rate": 2.881224270613042e-05,
+      "loss": 2.0708,
+      "step": 4961
+    },
+    {
+      "epoch": 2.140608151822299,
+      "grad_norm": 0.19633318483829498,
+      "learning_rate": 2.8785358962827222e-05,
+      "loss": 2.0966,
+      "step": 4962
+    },
+    {
+      "epoch": 2.1410394651714473,
+      "grad_norm": 0.2094806581735611,
+      "learning_rate": 2.8758484788770876e-05,
+      "loss": 2.0274,
+      "step": 4963
+    },
+    {
+      "epoch": 2.141470778520595,
+      "grad_norm": 0.197383850812912,
+      "learning_rate": 2.873162018952595e-05,
+      "loss": 2.2224,
+      "step": 4964
+    },
+    {
+      "epoch": 2.1419020918697433,
+      "grad_norm": 0.2031429558992386,
+      "learning_rate": 2.8704765170655032e-05,
+      "loss": 2.0326,
+      "step": 4965
+    },
+    {
+      "epoch": 2.1423334052188916,
+      "grad_norm": 0.18839189410209656,
+      "learning_rate": 2.867791973771877e-05,
+      "loss": 2.1396,
+      "step": 4966
+    },
+    {
+      "epoch": 2.14276471856804,
+      "grad_norm": 0.18205758929252625,
+      "learning_rate": 2.8651083896275798e-05,
+      "loss": 2.1926,
+      "step": 4967
+    },
+    {
+      "epoch": 2.1431960319171877,
+      "grad_norm": 0.20114341378211975,
+      "learning_rate": 2.8624257651882755e-05,
+      "loss": 1.9477,
+      "step": 4968
+    },
+    {
+      "epoch": 2.143627345266336,
+      "grad_norm": 0.21320831775665283,
+      "learning_rate": 2.8597441010094336e-05,
+      "loss": 2.3524,
+      "step": 4969
+    },
+    {
+      "epoch": 2.144058658615484,
+      "grad_norm": 0.17399685084819794,
+      "learning_rate": 2.85706339764632e-05,
+      "loss": 2.1856,
+      "step": 4970
+    },
+    {
+      "epoch": 2.1444899719646324,
+      "grad_norm": 0.20471583306789398,
+      "learning_rate": 2.8543836556540045e-05,
+      "loss": 2.1719,
+      "step": 4971
+    },
+    {
+      "epoch": 2.1449212853137807,
+      "grad_norm": 0.19019953906536102,
+      "learning_rate": 2.8517048755873564e-05,
+      "loss": 2.2953,
+      "step": 4972
+    },
+    {
+      "epoch": 2.1453525986629285,
+      "grad_norm": 0.18899403512477875,
+      "learning_rate": 2.8490270580010448e-05,
+      "loss": 2.0092,
+      "step": 4973
+    },
+    {
+      "epoch": 2.1457839120120767,
+      "grad_norm": 0.18942353129386902,
+      "learning_rate": 2.846350203449545e-05,
+      "loss": 2.15,
+      "step": 4974
+    },
+    {
+      "epoch": 2.146215225361225,
+      "grad_norm": 0.2105664610862732,
+      "learning_rate": 2.843674312487129e-05,
+      "loss": 2.2613,
+      "step": 4975
+    },
+    {
+      "epoch": 2.146215225361225,
+      "eval_loss": 2.0892958641052246,
+      "eval_runtime": 212.7334,
+      "eval_samples_per_second": 0.15,
+      "eval_steps_per_second": 0.15,
+      "step": 4975
+    },
+    {
+      "epoch": 2.1466465387103733,
+      "grad_norm": 0.19566898047924042,
+      "learning_rate": 2.8409993856678655e-05,
+      "loss": 2.0619,
+      "step": 4976
+    },
+    {
+      "epoch": 2.147077852059521,
+      "grad_norm": 0.18931123614311218,
+      "learning_rate": 2.8383254235456268e-05,
+      "loss": 2.1251,
+      "step": 4977
+    },
+    {
+      "epoch": 2.1475091654086693,
+      "grad_norm": 0.20515230298042297,
+      "learning_rate": 2.8356524266740908e-05,
+      "loss": 2.0915,
+      "step": 4978
+    },
+    {
+      "epoch": 2.1479404787578176,
+      "grad_norm": 0.20855127274990082,
+      "learning_rate": 2.832980395606728e-05,
+      "loss": 2.108,
+      "step": 4979
+    },
+    {
+      "epoch": 2.148371792106966,
+      "grad_norm": 0.18840979039669037,
+      "learning_rate": 2.8303093308968126e-05,
+      "loss": 2.2457,
+      "step": 4980
+    },
+    {
+      "epoch": 2.148803105456114,
+      "grad_norm": 0.20065249502658844,
+      "learning_rate": 2.8276392330974164e-05,
+      "loss": 2.1103,
+      "step": 4981
+    },
+    {
+      "epoch": 2.149234418805262,
+      "grad_norm": 0.19785119593143463,
+      "learning_rate": 2.824970102761413e-05,
+      "loss": 2.1526,
+      "step": 4982
+    },
+    {
+      "epoch": 2.14966573215441,
+      "grad_norm": 0.18779438734054565,
+      "learning_rate": 2.8223019404414755e-05,
+      "loss": 1.9152,
+      "step": 4983
+    },
+    {
+      "epoch": 2.1500970455035584,
+      "grad_norm": 0.20784412324428558,
+      "learning_rate": 2.8196347466900757e-05,
+      "loss": 2.136,
+      "step": 4984
+    },
+    {
+      "epoch": 2.1505283588527067,
+      "grad_norm": 0.19135190546512604,
+      "learning_rate": 2.8169685220594852e-05,
+      "loss": 2.1212,
+      "step": 4985
+    },
+    {
+      "epoch": 2.1509596722018545,
+      "grad_norm": 0.20130488276481628,
+      "learning_rate": 2.8143032671017748e-05,
+      "loss": 2.4026,
+      "step": 4986
+    },
+    {
+      "epoch": 2.1513909855510027,
+      "grad_norm": 0.17856507003307343,
+      "learning_rate": 2.8116389823688117e-05,
+      "loss": 2.0381,
+      "step": 4987
+    },
+    {
+      "epoch": 2.151822298900151,
+      "grad_norm": 0.1963197886943817,
+      "learning_rate": 2.8089756684122748e-05,
+      "loss": 2.0304,
+      "step": 4988
+    },
+    {
+      "epoch": 2.1522536122492992,
+      "grad_norm": 0.22759243845939636,
+      "learning_rate": 2.8063133257836222e-05,
+      "loss": 2.085,
+      "step": 4989
+    },
+    {
+      "epoch": 2.1526849255984475,
+      "grad_norm": 0.1714658886194229,
+      "learning_rate": 2.8036519550341253e-05,
+      "loss": 1.9922,
+      "step": 4990
+    },
+    {
+      "epoch": 2.1531162389475953,
+      "grad_norm": 0.20330412685871124,
+      "learning_rate": 2.8009915567148466e-05,
+      "loss": 2.0577,
+      "step": 4991
+    },
+    {
+      "epoch": 2.1535475522967436,
+      "grad_norm": 0.20327334105968475,
+      "learning_rate": 2.7983321313766568e-05,
+      "loss": 2.1033,
+      "step": 4992
+    },
+    {
+      "epoch": 2.153978865645892,
+      "grad_norm": 0.1952381581068039,
+      "learning_rate": 2.795673679570218e-05,
+      "loss": 2.2632,
+      "step": 4993
+    },
+    {
+      "epoch": 2.15441017899504,
+      "grad_norm": 0.2106030136346817,
+      "learning_rate": 2.793016201845984e-05,
+      "loss": 2.3056,
+      "step": 4994
+    },
+    {
+      "epoch": 2.154841492344188,
+      "grad_norm": 0.2100643515586853,
+      "learning_rate": 2.7903596987542226e-05,
+      "loss": 2.1765,
+      "step": 4995
+    },
+    {
+      "epoch": 2.155272805693336,
+      "grad_norm": 0.18079394102096558,
+      "learning_rate": 2.78770417084499e-05,
+      "loss": 2.2167,
+      "step": 4996
+    },
+    {
+      "epoch": 2.1557041190424844,
+      "grad_norm": 0.19508136808872223,
+      "learning_rate": 2.7850496186681413e-05,
+      "loss": 2.1174,
+      "step": 4997
+    },
+    {
+      "epoch": 2.1561354323916326,
+      "grad_norm": 0.19507291913032532,
+      "learning_rate": 2.782396042773331e-05,
+      "loss": 2.1595,
+      "step": 4998
+    },
+    {
+      "epoch": 2.156566745740781,
+      "grad_norm": 0.20600342750549316,
+      "learning_rate": 2.779743443710011e-05,
+      "loss": 2.1292,
+      "step": 4999
+    },
+    {
+      "epoch": 2.1569980590899287,
+      "grad_norm": 0.1863345503807068,
+      "learning_rate": 2.77709182202743e-05,
+      "loss": 2.0938,
+      "step": 5000
+    },
+    {
+      "epoch": 2.1569980590899287,
+      "eval_loss": 2.0890560150146484,
+      "eval_runtime": 206.7094,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 5000
+    },
+    {
+      "epoch": 2.157429372439077,
+      "grad_norm": 0.22767731547355652,
+      "learning_rate": 2.7744411782746343e-05,
+      "loss": 2.0735,
+      "step": 5001
+    },
+    {
+      "epoch": 2.157860685788225,
+      "grad_norm": 0.20078809559345245,
+      "learning_rate": 2.7717915130004757e-05,
+      "loss": 2.2704,
+      "step": 5002
+    },
+    {
+      "epoch": 2.1582919991373735,
+      "grad_norm": 0.18619324266910553,
+      "learning_rate": 2.7691428267535885e-05,
+      "loss": 2.1069,
+      "step": 5003
+    },
+    {
+      "epoch": 2.1587233124865213,
+      "grad_norm": 0.21010684967041016,
+      "learning_rate": 2.766495120082415e-05,
+      "loss": 2.1341,
+      "step": 5004
+    },
+    {
+      "epoch": 2.1591546258356695,
+      "grad_norm": 0.18977265059947968,
+      "learning_rate": 2.763848393535188e-05,
+      "loss": 1.9988,
+      "step": 5005
+    },
+    {
+      "epoch": 2.159585939184818,
+      "grad_norm": 0.22023966908454895,
+      "learning_rate": 2.761202647659948e-05,
+      "loss": 2.1059,
+      "step": 5006
+    },
+    {
+      "epoch": 2.160017252533966,
+      "grad_norm": 0.18467730283737183,
+      "learning_rate": 2.7585578830045253e-05,
+      "loss": 2.072,
+      "step": 5007
+    },
+    {
+      "epoch": 2.1604485658831143,
+      "grad_norm": 0.19137078523635864,
+      "learning_rate": 2.7559141001165396e-05,
+      "loss": 2.2661,
+      "step": 5008
+    },
+    {
+      "epoch": 2.160879879232262,
+      "grad_norm": 0.19556930661201477,
+      "learning_rate": 2.753271299543422e-05,
+      "loss": 2.017,
+      "step": 5009
+    },
+    {
+      "epoch": 2.1613111925814104,
+      "grad_norm": 0.19731999933719635,
+      "learning_rate": 2.750629481832391e-05,
+      "loss": 2.1438,
+      "step": 5010
+    },
+    {
+      "epoch": 2.1617425059305586,
+      "grad_norm": 0.17040055990219116,
+      "learning_rate": 2.747988647530464e-05,
+      "loss": 2.0037,
+      "step": 5011
+    },
+    {
+      "epoch": 2.162173819279707,
+      "grad_norm": 0.20169128477573395,
+      "learning_rate": 2.7453487971844548e-05,
+      "loss": 2.1116,
+      "step": 5012
+    },
+    {
+      "epoch": 2.1626051326288547,
+      "grad_norm": 0.19329191744327545,
+      "learning_rate": 2.7427099313409733e-05,
+      "loss": 2.0255,
+      "step": 5013
+    },
+    {
+      "epoch": 2.163036445978003,
+      "grad_norm": 0.19396258890628815,
+      "learning_rate": 2.7400720505464253e-05,
+      "loss": 2.1873,
+      "step": 5014
+    },
+    {
+      "epoch": 2.163467759327151,
+      "grad_norm": 0.1850919872522354,
+      "learning_rate": 2.7374351553470124e-05,
+      "loss": 2.2272,
+      "step": 5015
+    },
+    {
+      "epoch": 2.1638990726762994,
+      "grad_norm": 0.1951063722372055,
+      "learning_rate": 2.7347992462887335e-05,
+      "loss": 2.1948,
+      "step": 5016
+    },
+    {
+      "epoch": 2.1643303860254477,
+      "grad_norm": 0.19742482900619507,
+      "learning_rate": 2.7321643239173817e-05,
+      "loss": 2.3527,
+      "step": 5017
+    },
+    {
+      "epoch": 2.1647616993745955,
+      "grad_norm": 0.19653549790382385,
+      "learning_rate": 2.7295303887785475e-05,
+      "loss": 2.1226,
+      "step": 5018
+    },
+    {
+      "epoch": 2.1651930127237438,
+      "grad_norm": 0.19391842186450958,
+      "learning_rate": 2.7268974414176124e-05,
+      "loss": 2.3314,
+      "step": 5019
+    },
+    {
+      "epoch": 2.165624326072892,
+      "grad_norm": 0.2106139212846756,
+      "learning_rate": 2.7242654823797627e-05,
+      "loss": 2.0381,
+      "step": 5020
+    },
+    {
+      "epoch": 2.1660556394220403,
+      "grad_norm": 0.19162026047706604,
+      "learning_rate": 2.7216345122099737e-05,
+      "loss": 2.1074,
+      "step": 5021
+    },
+    {
+      "epoch": 2.166486952771188,
+      "grad_norm": 0.18334653973579407,
+      "learning_rate": 2.7190045314530112e-05,
+      "loss": 2.1011,
+      "step": 5022
+    },
+    {
+      "epoch": 2.1669182661203363,
+      "grad_norm": 0.20370136201381683,
+      "learning_rate": 2.7163755406534425e-05,
+      "loss": 2.2357,
+      "step": 5023
+    },
+    {
+      "epoch": 2.1673495794694846,
+      "grad_norm": 0.19202132523059845,
+      "learning_rate": 2.7137475403556333e-05,
+      "loss": 2.2211,
+      "step": 5024
+    },
+    {
+      "epoch": 2.167780892818633,
+      "grad_norm": 0.1901710033416748,
+      "learning_rate": 2.711120531103738e-05,
+      "loss": 2.1074,
+      "step": 5025
+    },
+    {
+      "epoch": 2.167780892818633,
+      "eval_loss": 2.0886192321777344,
+      "eval_runtime": 207.6955,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 5025
+    },
+    {
+      "epoch": 2.168212206167781,
+      "grad_norm": 0.1857885867357254,
+      "learning_rate": 2.708494513441706e-05,
+      "loss": 1.9152,
+      "step": 5026
+    },
+    {
+      "epoch": 2.168643519516929,
+      "grad_norm": 0.19882744550704956,
+      "learning_rate": 2.7058694879132852e-05,
+      "loss": 2.0951,
+      "step": 5027
+    },
+    {
+      "epoch": 2.169074832866077,
+      "grad_norm": 0.3018729090690613,
+      "learning_rate": 2.703245455062014e-05,
+      "loss": 2.1891,
+      "step": 5028
+    },
+    {
+      "epoch": 2.1695061462152254,
+      "grad_norm": 0.20267686247825623,
+      "learning_rate": 2.7006224154312274e-05,
+      "loss": 1.928,
+      "step": 5029
+    },
+    {
+      "epoch": 2.1699374595643737,
+      "grad_norm": 0.19706034660339355,
+      "learning_rate": 2.698000369564055e-05,
+      "loss": 2.3663,
+      "step": 5030
+    },
+    {
+      "epoch": 2.1703687729135215,
+      "grad_norm": 0.19120123982429504,
+      "learning_rate": 2.6953793180034207e-05,
+      "loss": 2.2033,
+      "step": 5031
+    },
+    {
+      "epoch": 2.1708000862626697,
+      "grad_norm": 0.19281716644763947,
+      "learning_rate": 2.6927592612920404e-05,
+      "loss": 2.2885,
+      "step": 5032
+    },
+    {
+      "epoch": 2.171231399611818,
+      "grad_norm": 0.22012881934642792,
+      "learning_rate": 2.690140199972423e-05,
+      "loss": 2.158,
+      "step": 5033
+    },
+    {
+      "epoch": 2.1716627129609662,
+      "grad_norm": 0.1995842605829239,
+      "learning_rate": 2.687522134586883e-05,
+      "loss": 2.039,
+      "step": 5034
+    },
+    {
+      "epoch": 2.1720940263101145,
+      "grad_norm": 0.18780255317687988,
+      "learning_rate": 2.6849050656775107e-05,
+      "loss": 2.103,
+      "step": 5035
+    },
+    {
+      "epoch": 2.1725253396592623,
+      "grad_norm": 0.20365270972251892,
+      "learning_rate": 2.6822889937862015e-05,
+      "loss": 2.2397,
+      "step": 5036
+    },
+    {
+      "epoch": 2.1729566530084106,
+      "grad_norm": 0.19723571836948395,
+      "learning_rate": 2.679673919454639e-05,
+      "loss": 2.0369,
+      "step": 5037
+    },
+    {
+      "epoch": 2.173387966357559,
+      "grad_norm": 0.20004676282405853,
+      "learning_rate": 2.677059843224308e-05,
+      "loss": 2.171,
+      "step": 5038
+    },
+    {
+      "epoch": 2.173819279706707,
+      "grad_norm": 0.1994907110929489,
+      "learning_rate": 2.674446765636482e-05,
+      "loss": 2.2445,
+      "step": 5039
+    },
+    {
+      "epoch": 2.174250593055855,
+      "grad_norm": 0.1878870576620102,
+      "learning_rate": 2.6718346872322197e-05,
+      "loss": 2.123,
+      "step": 5040
+    },
+    {
+      "epoch": 2.174681906405003,
+      "grad_norm": 0.20285697281360626,
+      "learning_rate": 2.6692236085523873e-05,
+      "loss": 2.0787,
+      "step": 5041
+    },
+    {
+      "epoch": 2.1751132197541514,
+      "grad_norm": 0.2116626501083374,
+      "learning_rate": 2.666613530137636e-05,
+      "loss": 2.1716,
+      "step": 5042
+    },
+    {
+      "epoch": 2.1755445331032996,
+      "grad_norm": 0.19868099689483643,
+      "learning_rate": 2.6640044525284097e-05,
+      "loss": 2.1836,
+      "step": 5043
+    },
+    {
+      "epoch": 2.175975846452448,
+      "grad_norm": 0.18697194755077362,
+      "learning_rate": 2.661396376264947e-05,
+      "loss": 2.1,
+      "step": 5044
+    },
+    {
+      "epoch": 2.1764071598015957,
+      "grad_norm": 0.1909298449754715,
+      "learning_rate": 2.658789301887279e-05,
+      "loss": 2.0179,
+      "step": 5045
+    },
+    {
+      "epoch": 2.176838473150744,
+      "grad_norm": 0.18372386693954468,
+      "learning_rate": 2.656183229935229e-05,
+      "loss": 1.861,
+      "step": 5046
+    },
+    {
+      "epoch": 2.1772697864998922,
+      "grad_norm": 0.20448540151119232,
+      "learning_rate": 2.653578160948409e-05,
+      "loss": 2.2202,
+      "step": 5047
+    },
+    {
+      "epoch": 2.1777010998490405,
+      "grad_norm": 0.19497567415237427,
+      "learning_rate": 2.6509740954662355e-05,
+      "loss": 2.1244,
+      "step": 5048
+    },
+    {
+      "epoch": 2.1781324131981883,
+      "grad_norm": 0.20127350091934204,
+      "learning_rate": 2.6483710340279013e-05,
+      "loss": 2.2618,
+      "step": 5049
+    },
+    {
+      "epoch": 2.1785637265473365,
+      "grad_norm": 0.1878756284713745,
+      "learning_rate": 2.645768977172401e-05,
+      "loss": 2.1859,
+      "step": 5050
+    },
+    {
+      "epoch": 2.1785637265473365,
+      "eval_loss": 2.088935375213623,
+      "eval_runtime": 207.6666,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 5050
+    },
+    {
+      "epoch": 2.178995039896485,
+      "grad_norm": 0.19722066819667816,
+      "learning_rate": 2.6431679254385162e-05,
+      "loss": 2.2203,
+      "step": 5051
+    },
+    {
+      "epoch": 2.179426353245633,
+      "grad_norm": 0.19405268132686615,
+      "learning_rate": 2.6405678793648293e-05,
+      "loss": 2.1512,
+      "step": 5052
+    },
+    {
+      "epoch": 2.1798576665947813,
+      "grad_norm": 0.20881058275699615,
+      "learning_rate": 2.637968839489706e-05,
+      "loss": 2.262,
+      "step": 5053
+    },
+    {
+      "epoch": 2.180288979943929,
+      "grad_norm": 0.1833294779062271,
+      "learning_rate": 2.635370806351301e-05,
+      "loss": 2.1621,
+      "step": 5054
+    },
+    {
+      "epoch": 2.1807202932930774,
+      "grad_norm": 0.20712466537952423,
+      "learning_rate": 2.632773780487571e-05,
+      "loss": 2.1567,
+      "step": 5055
+    },
+    {
+      "epoch": 2.1811516066422256,
+      "grad_norm": 0.2082417905330658,
+      "learning_rate": 2.6301777624362564e-05,
+      "loss": 2.3652,
+      "step": 5056
+    },
+    {
+      "epoch": 2.181582919991374,
+      "grad_norm": 0.19520720839500427,
+      "learning_rate": 2.627582752734893e-05,
+      "loss": 2.1002,
+      "step": 5057
+    },
+    {
+      "epoch": 2.1820142333405217,
+      "grad_norm": 0.2027909755706787,
+      "learning_rate": 2.624988751920803e-05,
+      "loss": 2.0569,
+      "step": 5058
+    },
+    {
+      "epoch": 2.18244554668967,
+      "grad_norm": 0.18182145059108734,
+      "learning_rate": 2.622395760531105e-05,
+      "loss": 1.9879,
+      "step": 5059
+    },
+    {
+      "epoch": 2.182876860038818,
+      "grad_norm": 0.25678521394729614,
+      "learning_rate": 2.6198037791027054e-05,
+      "loss": 2.3149,
+      "step": 5060
+    },
+    {
+      "epoch": 2.1833081733879665,
+      "grad_norm": 0.19661037623882294,
+      "learning_rate": 2.617212808172302e-05,
+      "loss": 2.1367,
+      "step": 5061
+    },
+    {
+      "epoch": 2.1837394867371147,
+      "grad_norm": 0.19917482137680054,
+      "learning_rate": 2.614622848276384e-05,
+      "loss": 2.2148,
+      "step": 5062
+    },
+    {
+      "epoch": 2.1841708000862625,
+      "grad_norm": 0.18143093585968018,
+      "learning_rate": 2.6120338999512307e-05,
+      "loss": 2.1091,
+      "step": 5063
+    },
+    {
+      "epoch": 2.1846021134354108,
+      "grad_norm": 0.1831212192773819,
+      "learning_rate": 2.609445963732912e-05,
+      "loss": 1.9578,
+      "step": 5064
+    },
+    {
+      "epoch": 2.185033426784559,
+      "grad_norm": 0.2225896567106247,
+      "learning_rate": 2.6068590401572868e-05,
+      "loss": 2.2214,
+      "step": 5065
+    },
+    {
+      "epoch": 2.1854647401337073,
+      "grad_norm": 0.19103094935417175,
+      "learning_rate": 2.6042731297600125e-05,
+      "loss": 2.0882,
+      "step": 5066
+    },
+    {
+      "epoch": 2.185896053482855,
+      "grad_norm": 0.1908239871263504,
+      "learning_rate": 2.601688233076523e-05,
+      "loss": 2.089,
+      "step": 5067
+    },
+    {
+      "epoch": 2.1863273668320033,
+      "grad_norm": 0.18195314705371857,
+      "learning_rate": 2.599104350642052e-05,
+      "loss": 2.1344,
+      "step": 5068
+    },
+    {
+      "epoch": 2.1867586801811516,
+      "grad_norm": 0.20526288449764252,
+      "learning_rate": 2.5965214829916175e-05,
+      "loss": 2.174,
+      "step": 5069
+    },
+    {
+      "epoch": 2.1871899935303,
+      "grad_norm": 0.17393609881401062,
+      "learning_rate": 2.5939396306600367e-05,
+      "loss": 2.0965,
+      "step": 5070
+    },
+    {
+      "epoch": 2.187621306879448,
+      "grad_norm": 0.19621096551418304,
+      "learning_rate": 2.591358794181909e-05,
+      "loss": 2.0679,
+      "step": 5071
+    },
+    {
+      "epoch": 2.188052620228596,
+      "grad_norm": 0.1995396614074707,
+      "learning_rate": 2.5887789740916188e-05,
+      "loss": 2.2099,
+      "step": 5072
+    },
+    {
+      "epoch": 2.188483933577744,
+      "grad_norm": 0.19627083837985992,
+      "learning_rate": 2.5862001709233524e-05,
+      "loss": 2.2104,
+      "step": 5073
+    },
+    {
+      "epoch": 2.1889152469268924,
+      "grad_norm": 0.1899205595254898,
+      "learning_rate": 2.5836223852110762e-05,
+      "loss": 1.9811,
+      "step": 5074
+    },
+    {
+      "epoch": 2.1893465602760407,
+      "grad_norm": 0.20641696453094482,
+      "learning_rate": 2.581045617488551e-05,
+      "loss": 2.1019,
+      "step": 5075
+    },
+    {
+      "epoch": 2.1893465602760407,
+      "eval_loss": 2.0890145301818848,
+      "eval_runtime": 203.979,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5075
+    },
+    {
+      "epoch": 2.1897778736251885,
+      "grad_norm": 0.19259873032569885,
+      "learning_rate": 2.5784698682893224e-05,
+      "loss": 2.1706,
+      "step": 5076
+    },
+    {
+      "epoch": 2.1902091869743368,
+      "grad_norm": 0.18554507195949554,
+      "learning_rate": 2.5758951381467293e-05,
+      "loss": 2.0585,
+      "step": 5077
+    },
+    {
+      "epoch": 2.190640500323485,
+      "grad_norm": 0.18829064071178436,
+      "learning_rate": 2.5733214275938965e-05,
+      "loss": 1.8606,
+      "step": 5078
+    },
+    {
+      "epoch": 2.1910718136726333,
+      "grad_norm": 0.20728611946105957,
+      "learning_rate": 2.5707487371637368e-05,
+      "loss": 2.187,
+      "step": 5079
+    },
+    {
+      "epoch": 2.1915031270217815,
+      "grad_norm": 0.1915161907672882,
+      "learning_rate": 2.568177067388961e-05,
+      "loss": 2.2266,
+      "step": 5080
+    },
+    {
+      "epoch": 2.1919344403709293,
+      "grad_norm": 0.20191590487957,
+      "learning_rate": 2.565606418802056e-05,
+      "loss": 2.0212,
+      "step": 5081
+    },
+    {
+      "epoch": 2.1923657537200776,
+      "grad_norm": 0.18292774260044098,
+      "learning_rate": 2.5630367919353024e-05,
+      "loss": 2.0412,
+      "step": 5082
+    },
+    {
+      "epoch": 2.192797067069226,
+      "grad_norm": 0.16870975494384766,
+      "learning_rate": 2.5604681873207682e-05,
+      "loss": 2.0399,
+      "step": 5083
+    },
+    {
+      "epoch": 2.193228380418374,
+      "grad_norm": 0.21954478323459625,
+      "learning_rate": 2.5579006054903168e-05,
+      "loss": 1.985,
+      "step": 5084
+    },
+    {
+      "epoch": 2.193659693767522,
+      "grad_norm": 0.19486844539642334,
+      "learning_rate": 2.5553340469755922e-05,
+      "loss": 2.1811,
+      "step": 5085
+    },
+    {
+      "epoch": 2.19409100711667,
+      "grad_norm": 0.2029792219400406,
+      "learning_rate": 2.5527685123080232e-05,
+      "loss": 2.189,
+      "step": 5086
+    },
+    {
+      "epoch": 2.1945223204658184,
+      "grad_norm": 0.19896060228347778,
+      "learning_rate": 2.550204002018838e-05,
+      "loss": 2.2319,
+      "step": 5087
+    },
+    {
+      "epoch": 2.1949536338149667,
+      "grad_norm": 0.1982891857624054,
+      "learning_rate": 2.5476405166390443e-05,
+      "loss": 2.1252,
+      "step": 5088
+    },
+    {
+      "epoch": 2.195384947164115,
+      "grad_norm": 0.20024412870407104,
+      "learning_rate": 2.54507805669944e-05,
+      "loss": 2.3929,
+      "step": 5089
+    },
+    {
+      "epoch": 2.1958162605132627,
+      "grad_norm": 0.1865210235118866,
+      "learning_rate": 2.542516622730609e-05,
+      "loss": 2.2408,
+      "step": 5090
+    },
+    {
+      "epoch": 2.196247573862411,
+      "grad_norm": 0.19232657551765442,
+      "learning_rate": 2.539956215262926e-05,
+      "loss": 2.0633,
+      "step": 5091
+    },
+    {
+      "epoch": 2.1966788872115592,
+      "grad_norm": 0.17881444096565247,
+      "learning_rate": 2.537396834826551e-05,
+      "loss": 1.7351,
+      "step": 5092
+    },
+    {
+      "epoch": 2.1971102005607075,
+      "grad_norm": 0.17570246756076813,
+      "learning_rate": 2.5348384819514296e-05,
+      "loss": 2.2295,
+      "step": 5093
+    },
+    {
+      "epoch": 2.1975415139098553,
+      "grad_norm": 0.19344256818294525,
+      "learning_rate": 2.5322811571673027e-05,
+      "loss": 2.1898,
+      "step": 5094
+    },
+    {
+      "epoch": 2.1979728272590036,
+      "grad_norm": 0.19992877542972565,
+      "learning_rate": 2.5297248610036863e-05,
+      "loss": 2.0582,
+      "step": 5095
+    },
+    {
+      "epoch": 2.198404140608152,
+      "grad_norm": 0.1935880184173584,
+      "learning_rate": 2.5271695939898902e-05,
+      "loss": 2.1934,
+      "step": 5096
+    },
+    {
+      "epoch": 2.1988354539573,
+      "grad_norm": 0.1977873146533966,
+      "learning_rate": 2.5246153566550107e-05,
+      "loss": 2.2696,
+      "step": 5097
+    },
+    {
+      "epoch": 2.1992667673064483,
+      "grad_norm": 0.19314149022102356,
+      "learning_rate": 2.5220621495279332e-05,
+      "loss": 2.1523,
+      "step": 5098
+    },
+    {
+      "epoch": 2.199698080655596,
+      "grad_norm": 0.19658662378787994,
+      "learning_rate": 2.519509973137328e-05,
+      "loss": 1.8741,
+      "step": 5099
+    },
+    {
+      "epoch": 2.2001293940047444,
+      "grad_norm": 0.23600396513938904,
+      "learning_rate": 2.5169588280116436e-05,
+      "loss": 2.2851,
+      "step": 5100
+    },
+    {
+      "epoch": 2.2001293940047444,
+      "eval_loss": 2.0888924598693848,
+      "eval_runtime": 204.5932,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 5100
+    },
+    {
+      "epoch": 2.2005607073538926,
+      "grad_norm": 0.21375681459903717,
+      "learning_rate": 2.5144087146791304e-05,
+      "loss": 2.239,
+      "step": 5101
+    },
+    {
+      "epoch": 2.200992020703041,
+      "grad_norm": 0.184476837515831,
+      "learning_rate": 2.5118596336678137e-05,
+      "loss": 2.1957,
+      "step": 5102
+    },
+    {
+      "epoch": 2.2014233340521887,
+      "grad_norm": 0.20138037204742432,
+      "learning_rate": 2.5093115855055094e-05,
+      "loss": 2.2173,
+      "step": 5103
+    },
+    {
+      "epoch": 2.201854647401337,
+      "grad_norm": 0.20391976833343506,
+      "learning_rate": 2.5067645707198182e-05,
+      "loss": 2.025,
+      "step": 5104
+    },
+    {
+      "epoch": 2.202285960750485,
+      "grad_norm": 0.18930117785930634,
+      "learning_rate": 2.5042185898381282e-05,
+      "loss": 2.1973,
+      "step": 5105
+    },
+    {
+      "epoch": 2.2027172740996335,
+      "grad_norm": 0.19269056618213654,
+      "learning_rate": 2.501673643387612e-05,
+      "loss": 2.0268,
+      "step": 5106
+    },
+    {
+      "epoch": 2.2031485874487817,
+      "grad_norm": 0.19102050364017487,
+      "learning_rate": 2.499129731895229e-05,
+      "loss": 2.0892,
+      "step": 5107
+    },
+    {
+      "epoch": 2.2035799007979295,
+      "grad_norm": 0.18067455291748047,
+      "learning_rate": 2.4965868558877235e-05,
+      "loss": 2.0446,
+      "step": 5108
+    },
+    {
+      "epoch": 2.204011214147078,
+      "grad_norm": 0.1875220686197281,
+      "learning_rate": 2.494045015891626e-05,
+      "loss": 1.9844,
+      "step": 5109
+    },
+    {
+      "epoch": 2.204442527496226,
+      "grad_norm": 0.18868571519851685,
+      "learning_rate": 2.4915042124332517e-05,
+      "loss": 2.0551,
+      "step": 5110
+    },
+    {
+      "epoch": 2.2048738408453743,
+      "grad_norm": 0.21553297340869904,
+      "learning_rate": 2.4889644460386995e-05,
+      "loss": 2.1265,
+      "step": 5111
+    },
+    {
+      "epoch": 2.205305154194522,
+      "grad_norm": 2.6596360206604004,
+      "learning_rate": 2.4864257172338642e-05,
+      "loss": 2.1501,
+      "step": 5112
+    },
+    {
+      "epoch": 2.2057364675436704,
+      "grad_norm": 0.21748773753643036,
+      "learning_rate": 2.4838880265444092e-05,
+      "loss": 1.5991,
+      "step": 5113
+    },
+    {
+      "epoch": 2.2061677808928186,
+      "grad_norm": 0.1903078258037567,
+      "learning_rate": 2.4813513744957933e-05,
+      "loss": 2.1071,
+      "step": 5114
+    },
+    {
+      "epoch": 2.206599094241967,
+      "grad_norm": 0.19003809988498688,
+      "learning_rate": 2.4788157616132555e-05,
+      "loss": 2.0368,
+      "step": 5115
+    },
+    {
+      "epoch": 2.207030407591115,
+      "grad_norm": 0.22406871616840363,
+      "learning_rate": 2.476281188421828e-05,
+      "loss": 1.8828,
+      "step": 5116
+    },
+    {
+      "epoch": 2.207461720940263,
+      "grad_norm": 0.18773460388183594,
+      "learning_rate": 2.473747655446321e-05,
+      "loss": 2.2092,
+      "step": 5117
+    },
+    {
+      "epoch": 2.207893034289411,
+      "grad_norm": 0.1880549192428589,
+      "learning_rate": 2.4712151632113238e-05,
+      "loss": 2.1944,
+      "step": 5118
+    },
+    {
+      "epoch": 2.2083243476385594,
+      "grad_norm": 0.1941085010766983,
+      "learning_rate": 2.4686837122412222e-05,
+      "loss": 2.1148,
+      "step": 5119
+    },
+    {
+      "epoch": 2.2087556609877077,
+      "grad_norm": 0.18133896589279175,
+      "learning_rate": 2.466153303060181e-05,
+      "loss": 1.8703,
+      "step": 5120
+    },
+    {
+      "epoch": 2.2091869743368555,
+      "grad_norm": 0.19666768610477448,
+      "learning_rate": 2.4636239361921464e-05,
+      "loss": 2.1024,
+      "step": 5121
+    },
+    {
+      "epoch": 2.2096182876860038,
+      "grad_norm": 0.20931586623191833,
+      "learning_rate": 2.461095612160853e-05,
+      "loss": 2.0844,
+      "step": 5122
+    },
+    {
+      "epoch": 2.210049601035152,
+      "grad_norm": 0.20741429924964905,
+      "learning_rate": 2.4585683314898177e-05,
+      "loss": 2.087,
+      "step": 5123
+    },
+    {
+      "epoch": 2.2104809143843003,
+      "grad_norm": 0.18422767519950867,
+      "learning_rate": 2.4560420947023422e-05,
+      "loss": 2.2925,
+      "step": 5124
+    },
+    {
+      "epoch": 2.2109122277334485,
+      "grad_norm": 0.19024157524108887,
+      "learning_rate": 2.4535169023215076e-05,
+      "loss": 1.8757,
+      "step": 5125
+    },
+    {
+      "epoch": 2.2109122277334485,
+      "eval_loss": 2.088559865951538,
+      "eval_runtime": 204.4158,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5125
+    },
+    {
+      "epoch": 2.2113435410825963,
+      "grad_norm": 0.1930663138628006,
+      "learning_rate": 2.450992754870191e-05,
+      "loss": 2.0434,
+      "step": 5126
+    },
+    {
+      "epoch": 2.2117748544317446,
+      "grad_norm": 0.17984342575073242,
+      "learning_rate": 2.4484696528710368e-05,
+      "loss": 2.0912,
+      "step": 5127
+    },
+    {
+      "epoch": 2.212206167780893,
+      "grad_norm": 0.27934229373931885,
+      "learning_rate": 2.4459475968464838e-05,
+      "loss": 2.2206,
+      "step": 5128
+    },
+    {
+      "epoch": 2.212637481130041,
+      "grad_norm": 0.18810495734214783,
+      "learning_rate": 2.443426587318748e-05,
+      "loss": 2.1079,
+      "step": 5129
+    },
+    {
+      "epoch": 2.213068794479189,
+      "grad_norm": 0.19722598791122437,
+      "learning_rate": 2.4409066248098383e-05,
+      "loss": 2.1985,
+      "step": 5130
+    },
+    {
+      "epoch": 2.213500107828337,
+      "grad_norm": 0.1870979517698288,
+      "learning_rate": 2.4383877098415395e-05,
+      "loss": 2.1231,
+      "step": 5131
+    },
+    {
+      "epoch": 2.2139314211774854,
+      "grad_norm": 0.20809242129325867,
+      "learning_rate": 2.4358698429354125e-05,
+      "loss": 1.837,
+      "step": 5132
+    },
+    {
+      "epoch": 2.2143627345266337,
+      "grad_norm": 0.18958339095115662,
+      "learning_rate": 2.4333530246128183e-05,
+      "loss": 2.2733,
+      "step": 5133
+    },
+    {
+      "epoch": 2.214794047875782,
+      "grad_norm": 0.1951441764831543,
+      "learning_rate": 2.430837255394888e-05,
+      "loss": 2.177,
+      "step": 5134
+    },
+    {
+      "epoch": 2.2152253612249297,
+      "grad_norm": 0.18995796144008636,
+      "learning_rate": 2.4283225358025397e-05,
+      "loss": 2.0873,
+      "step": 5135
+    },
+    {
+      "epoch": 2.215656674574078,
+      "grad_norm": 0.20475205779075623,
+      "learning_rate": 2.4258088663564724e-05,
+      "loss": 2.1105,
+      "step": 5136
+    },
+    {
+      "epoch": 2.2160879879232263,
+      "grad_norm": 0.19895505905151367,
+      "learning_rate": 2.4232962475771704e-05,
+      "loss": 2.1126,
+      "step": 5137
+    },
+    {
+      "epoch": 2.2165193012723745,
+      "grad_norm": 0.19223228096961975,
+      "learning_rate": 2.4207846799848985e-05,
+      "loss": 1.8898,
+      "step": 5138
+    },
+    {
+      "epoch": 2.2169506146215223,
+      "grad_norm": 0.19018810987472534,
+      "learning_rate": 2.418274164099701e-05,
+      "loss": 2.0493,
+      "step": 5139
+    },
+    {
+      "epoch": 2.2173819279706706,
+      "grad_norm": 0.19304291903972626,
+      "learning_rate": 2.415764700441416e-05,
+      "loss": 2.143,
+      "step": 5140
+    },
+    {
+      "epoch": 2.217813241319819,
+      "grad_norm": 0.2021966278553009,
+      "learning_rate": 2.4132562895296474e-05,
+      "loss": 2.301,
+      "step": 5141
+    },
+    {
+      "epoch": 2.218244554668967,
+      "grad_norm": 0.1942233294248581,
+      "learning_rate": 2.410748931883792e-05,
+      "loss": 2.1455,
+      "step": 5142
+    },
+    {
+      "epoch": 2.2186758680181153,
+      "grad_norm": 0.20182685554027557,
+      "learning_rate": 2.408242628023022e-05,
+      "loss": 2.3461,
+      "step": 5143
+    },
+    {
+      "epoch": 2.219107181367263,
+      "grad_norm": 0.1861187219619751,
+      "learning_rate": 2.4057373784663015e-05,
+      "loss": 2.2172,
+      "step": 5144
+    },
+    {
+      "epoch": 2.2195384947164114,
+      "grad_norm": 0.19665570557117462,
+      "learning_rate": 2.403233183732369e-05,
+      "loss": 2.1557,
+      "step": 5145
+    },
+    {
+      "epoch": 2.2199698080655597,
+      "grad_norm": 0.20461268723011017,
+      "learning_rate": 2.4007300443397384e-05,
+      "loss": 2.0571,
+      "step": 5146
+    },
+    {
+      "epoch": 2.220401121414708,
+      "grad_norm": 0.1917303204536438,
+      "learning_rate": 2.3982279608067194e-05,
+      "loss": 2.1758,
+      "step": 5147
+    },
+    {
+      "epoch": 2.2208324347638557,
+      "grad_norm": 0.18870019912719727,
+      "learning_rate": 2.395726933651393e-05,
+      "loss": 2.1837,
+      "step": 5148
+    },
+    {
+      "epoch": 2.221263748113004,
+      "grad_norm": 0.20907212793827057,
+      "learning_rate": 2.3932269633916235e-05,
+      "loss": 2.2427,
+      "step": 5149
+    },
+    {
+      "epoch": 2.2216950614621522,
+      "grad_norm": 0.1932532638311386,
+      "learning_rate": 2.3907280505450593e-05,
+      "loss": 2.2975,
+      "step": 5150
+    },
+    {
+      "epoch": 2.2216950614621522,
+      "eval_loss": 2.088294267654419,
+      "eval_runtime": 204.373,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5150
+    },
+    {
+      "epoch": 2.2221263748113005,
+      "grad_norm": 0.1814073622226715,
+      "learning_rate": 2.388230195629125e-05,
+      "loss": 1.9633,
+      "step": 5151
+    },
+    {
+      "epoch": 2.2225576881604487,
+      "grad_norm": 0.2006046324968338,
+      "learning_rate": 2.3857333991610307e-05,
+      "loss": 2.2333,
+      "step": 5152
+    },
+    {
+      "epoch": 2.2229890015095966,
+      "grad_norm": 0.20529131591320038,
+      "learning_rate": 2.383237661657765e-05,
+      "loss": 2.3329,
+      "step": 5153
+    },
+    {
+      "epoch": 2.223420314858745,
+      "grad_norm": 0.20727939903736115,
+      "learning_rate": 2.3807429836360974e-05,
+      "loss": 2.2073,
+      "step": 5154
+    },
+    {
+      "epoch": 2.223851628207893,
+      "grad_norm": 0.19187138974666595,
+      "learning_rate": 2.3782493656125784e-05,
+      "loss": 2.1682,
+      "step": 5155
+    },
+    {
+      "epoch": 2.2242829415570413,
+      "grad_norm": 0.1747562140226364,
+      "learning_rate": 2.375756808103538e-05,
+      "loss": 2.1047,
+      "step": 5156
+    },
+    {
+      "epoch": 2.224714254906189,
+      "grad_norm": 0.2382543832063675,
+      "learning_rate": 2.373265311625087e-05,
+      "loss": 1.7174,
+      "step": 5157
+    },
+    {
+      "epoch": 2.2251455682553374,
+      "grad_norm": 0.19224894046783447,
+      "learning_rate": 2.370774876693122e-05,
+      "loss": 2.1815,
+      "step": 5158
+    },
+    {
+      "epoch": 2.2255768816044856,
+      "grad_norm": 0.22981807589530945,
+      "learning_rate": 2.3682855038233092e-05,
+      "loss": 2.1429,
+      "step": 5159
+    },
+    {
+      "epoch": 2.226008194953634,
+      "grad_norm": 0.19596794247627258,
+      "learning_rate": 2.365797193531102e-05,
+      "loss": 2.0008,
+      "step": 5160
+    },
+    {
+      "epoch": 2.226439508302782,
+      "grad_norm": 0.2190484255552292,
+      "learning_rate": 2.3633099463317305e-05,
+      "loss": 2.1318,
+      "step": 5161
+    },
+    {
+      "epoch": 2.22687082165193,
+      "grad_norm": 0.19309651851654053,
+      "learning_rate": 2.3608237627402105e-05,
+      "loss": 2.1676,
+      "step": 5162
+    },
+    {
+      "epoch": 2.227302135001078,
+      "grad_norm": 0.20434337854385376,
+      "learning_rate": 2.3583386432713337e-05,
+      "loss": 2.2481,
+      "step": 5163
+    },
+    {
+      "epoch": 2.2277334483502265,
+      "grad_norm": 0.19345074892044067,
+      "learning_rate": 2.355854588439665e-05,
+      "loss": 2.238,
+      "step": 5164
+    },
+    {
+      "epoch": 2.2281647616993747,
+      "grad_norm": 0.18643070757389069,
+      "learning_rate": 2.353371598759561e-05,
+      "loss": 2.1621,
+      "step": 5165
+    },
+    {
+      "epoch": 2.2285960750485225,
+      "grad_norm": 0.182948037981987,
+      "learning_rate": 2.3508896747451505e-05,
+      "loss": 2.2342,
+      "step": 5166
+    },
+    {
+      "epoch": 2.229027388397671,
+      "grad_norm": 0.2127479761838913,
+      "learning_rate": 2.3484088169103428e-05,
+      "loss": 2.2865,
+      "step": 5167
+    },
+    {
+      "epoch": 2.229458701746819,
+      "grad_norm": 0.1867491900920868,
+      "learning_rate": 2.3459290257688256e-05,
+      "loss": 2.1871,
+      "step": 5168
+    },
+    {
+      "epoch": 2.2298900150959673,
+      "grad_norm": 0.20324353873729706,
+      "learning_rate": 2.343450301834068e-05,
+      "loss": 2.2684,
+      "step": 5169
+    },
+    {
+      "epoch": 2.2303213284451155,
+      "grad_norm": 0.21033866703510284,
+      "learning_rate": 2.340972645619317e-05,
+      "loss": 2.2214,
+      "step": 5170
+    },
+    {
+      "epoch": 2.2307526417942634,
+      "grad_norm": 0.21309709548950195,
+      "learning_rate": 2.3384960576375947e-05,
+      "loss": 2.328,
+      "step": 5171
+    },
+    {
+      "epoch": 2.2311839551434116,
+      "grad_norm": 0.1905982345342636,
+      "learning_rate": 2.3360205384017143e-05,
+      "loss": 2.0782,
+      "step": 5172
+    },
+    {
+      "epoch": 2.23161526849256,
+      "grad_norm": 0.19004592299461365,
+      "learning_rate": 2.3335460884242506e-05,
+      "loss": 2.2175,
+      "step": 5173
+    },
+    {
+      "epoch": 2.232046581841708,
+      "grad_norm": 0.2126663327217102,
+      "learning_rate": 2.3310727082175695e-05,
+      "loss": 2.0913,
+      "step": 5174
+    },
+    {
+      "epoch": 2.232477895190856,
+      "grad_norm": 0.17693957686424255,
+      "learning_rate": 2.3286003982938074e-05,
+      "loss": 2.0777,
+      "step": 5175
+    },
+    {
+      "epoch": 2.232477895190856,
+      "eval_loss": 2.088388442993164,
+      "eval_runtime": 207.049,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 5175
+    },
+    {
+      "epoch": 2.232909208540004,
+      "grad_norm": 0.18512453138828278,
+      "learning_rate": 2.326129159164889e-05,
+      "loss": 2.118,
+      "step": 5176
+    },
+    {
+      "epoch": 2.2333405218891524,
+      "grad_norm": 0.1941794455051422,
+      "learning_rate": 2.3236589913425118e-05,
+      "loss": 2.0438,
+      "step": 5177
+    },
+    {
+      "epoch": 2.2337718352383007,
+      "grad_norm": 0.20930613577365875,
+      "learning_rate": 2.3211898953381422e-05,
+      "loss": 2.0802,
+      "step": 5178
+    },
+    {
+      "epoch": 2.234203148587449,
+      "grad_norm": 0.18975000083446503,
+      "learning_rate": 2.3187218716630422e-05,
+      "loss": 2.1545,
+      "step": 5179
+    },
+    {
+      "epoch": 2.2346344619365968,
+      "grad_norm": 0.19006623327732086,
+      "learning_rate": 2.31625492082824e-05,
+      "loss": 2.131,
+      "step": 5180
+    },
+    {
+      "epoch": 2.235065775285745,
+      "grad_norm": 0.2047644406557083,
+      "learning_rate": 2.3137890433445448e-05,
+      "loss": 2.2678,
+      "step": 5181
+    },
+    {
+      "epoch": 2.2354970886348933,
+      "grad_norm": 0.49673357605934143,
+      "learning_rate": 2.3113242397225437e-05,
+      "loss": 2.162,
+      "step": 5182
+    },
+    {
+      "epoch": 2.2359284019840415,
+      "grad_norm": 0.18441151082515717,
+      "learning_rate": 2.3088605104726017e-05,
+      "loss": 2.0678,
+      "step": 5183
+    },
+    {
+      "epoch": 2.2363597153331893,
+      "grad_norm": 0.19082053005695343,
+      "learning_rate": 2.3063978561048594e-05,
+      "loss": 1.9763,
+      "step": 5184
+    },
+    {
+      "epoch": 2.2367910286823376,
+      "grad_norm": 0.190688356757164,
+      "learning_rate": 2.3039362771292374e-05,
+      "loss": 2.1109,
+      "step": 5185
+    },
+    {
+      "epoch": 2.237222342031486,
+      "grad_norm": 0.21213744580745697,
+      "learning_rate": 2.301475774055433e-05,
+      "loss": 2.3109,
+      "step": 5186
+    },
+    {
+      "epoch": 2.237653655380634,
+      "grad_norm": 0.19175684452056885,
+      "learning_rate": 2.29901634739292e-05,
+      "loss": 2.1973,
+      "step": 5187
+    },
+    {
+      "epoch": 2.2380849687297824,
+      "grad_norm": 0.20844250917434692,
+      "learning_rate": 2.2965579976509484e-05,
+      "loss": 2.0611,
+      "step": 5188
+    },
+    {
+      "epoch": 2.23851628207893,
+      "grad_norm": 0.19638632237911224,
+      "learning_rate": 2.2941007253385453e-05,
+      "loss": 2.1502,
+      "step": 5189
+    },
+    {
+      "epoch": 2.2389475954280784,
+      "grad_norm": 0.2016148567199707,
+      "learning_rate": 2.291644530964522e-05,
+      "loss": 2.3361,
+      "step": 5190
+    },
+    {
+      "epoch": 2.2393789087772267,
+      "grad_norm": 0.20059362053871155,
+      "learning_rate": 2.289189415037454e-05,
+      "loss": 2.2438,
+      "step": 5191
+    },
+    {
+      "epoch": 2.239810222126375,
+      "grad_norm": 0.2173454612493515,
+      "learning_rate": 2.2867353780657e-05,
+      "loss": 2.2661,
+      "step": 5192
+    },
+    {
+      "epoch": 2.2402415354755227,
+      "grad_norm": 0.19429714977741241,
+      "learning_rate": 2.2842824205574002e-05,
+      "loss": 2.1804,
+      "step": 5193
+    },
+    {
+      "epoch": 2.240672848824671,
+      "grad_norm": 0.19609315693378448,
+      "learning_rate": 2.2818305430204624e-05,
+      "loss": 2.1481,
+      "step": 5194
+    },
+    {
+      "epoch": 2.2411041621738192,
+      "grad_norm": 0.2272697538137436,
+      "learning_rate": 2.2793797459625783e-05,
+      "loss": 2.2735,
+      "step": 5195
+    },
+    {
+      "epoch": 2.2415354755229675,
+      "grad_norm": 0.18201479315757751,
+      "learning_rate": 2.2769300298912044e-05,
+      "loss": 2.1451,
+      "step": 5196
+    },
+    {
+      "epoch": 2.2419667888721158,
+      "grad_norm": 0.19466306269168854,
+      "learning_rate": 2.2744813953135894e-05,
+      "loss": 2.1083,
+      "step": 5197
+    },
+    {
+      "epoch": 2.2423981022212636,
+      "grad_norm": 0.19931024312973022,
+      "learning_rate": 2.2720338427367465e-05,
+      "loss": 2.1513,
+      "step": 5198
+    },
+    {
+      "epoch": 2.242829415570412,
+      "grad_norm": 0.2232944369316101,
+      "learning_rate": 2.269587372667468e-05,
+      "loss": 2.1877,
+      "step": 5199
+    },
+    {
+      "epoch": 2.24326072891956,
+      "grad_norm": 0.1860596090555191,
+      "learning_rate": 2.2671419856123233e-05,
+      "loss": 2.2027,
+      "step": 5200
+    },
+    {
+      "epoch": 2.24326072891956,
+      "eval_loss": 2.0881733894348145,
+      "eval_runtime": 203.8998,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5200
+    },
+    {
+      "epoch": 2.2436920422687083,
+      "grad_norm": 0.2147083878517151,
+      "learning_rate": 2.2646976820776565e-05,
+      "loss": 2.3301,
+      "step": 5201
+    },
+    {
+      "epoch": 2.244123355617856,
+      "grad_norm": 0.20780740678310394,
+      "learning_rate": 2.2622544625695858e-05,
+      "loss": 2.2091,
+      "step": 5202
+    },
+    {
+      "epoch": 2.2445546689670044,
+      "grad_norm": 0.18065021932125092,
+      "learning_rate": 2.2598123275940067e-05,
+      "loss": 2.0219,
+      "step": 5203
+    },
+    {
+      "epoch": 2.2449859823161527,
+      "grad_norm": 0.19713814556598663,
+      "learning_rate": 2.257371277656595e-05,
+      "loss": 1.9993,
+      "step": 5204
+    },
+    {
+      "epoch": 2.245417295665301,
+      "grad_norm": 0.21324028074741364,
+      "learning_rate": 2.25493131326279e-05,
+      "loss": 2.0939,
+      "step": 5205
+    },
+    {
+      "epoch": 2.245848609014449,
+      "grad_norm": 0.20643126964569092,
+      "learning_rate": 2.252492434917816e-05,
+      "loss": 2.1377,
+      "step": 5206
+    },
+    {
+      "epoch": 2.246279922363597,
+      "grad_norm": 0.2044958770275116,
+      "learning_rate": 2.250054643126666e-05,
+      "loss": 2.2487,
+      "step": 5207
+    },
+    {
+      "epoch": 2.2467112357127452,
+      "grad_norm": 0.19290834665298462,
+      "learning_rate": 2.247617938394118e-05,
+      "loss": 2.0619,
+      "step": 5208
+    },
+    {
+      "epoch": 2.2471425490618935,
+      "grad_norm": 0.1928108036518097,
+      "learning_rate": 2.2451823212247165e-05,
+      "loss": 2.1128,
+      "step": 5209
+    },
+    {
+      "epoch": 2.2475738624110417,
+      "grad_norm": 0.20691651105880737,
+      "learning_rate": 2.2427477921227774e-05,
+      "loss": 2.2613,
+      "step": 5210
+    },
+    {
+      "epoch": 2.2480051757601895,
+      "grad_norm": 0.20217694342136383,
+      "learning_rate": 2.2403143515924017e-05,
+      "loss": 2.1959,
+      "step": 5211
+    },
+    {
+      "epoch": 2.248436489109338,
+      "grad_norm": 0.19633910059928894,
+      "learning_rate": 2.2378820001374587e-05,
+      "loss": 2.0647,
+      "step": 5212
+    },
+    {
+      "epoch": 2.248867802458486,
+      "grad_norm": 0.1810738444328308,
+      "learning_rate": 2.2354507382615933e-05,
+      "loss": 2.0933,
+      "step": 5213
+    },
+    {
+      "epoch": 2.2492991158076343,
+      "grad_norm": 0.21027354896068573,
+      "learning_rate": 2.2330205664682246e-05,
+      "loss": 2.051,
+      "step": 5214
+    },
+    {
+      "epoch": 2.2497304291567826,
+      "grad_norm": 0.1850845068693161,
+      "learning_rate": 2.230591485260546e-05,
+      "loss": 2.1125,
+      "step": 5215
+    },
+    {
+      "epoch": 2.2501617425059304,
+      "grad_norm": 0.22473980486392975,
+      "learning_rate": 2.228163495141526e-05,
+      "loss": 2.1118,
+      "step": 5216
+    },
+    {
+      "epoch": 2.2505930558550786,
+      "grad_norm": 0.22738492488861084,
+      "learning_rate": 2.2257365966139043e-05,
+      "loss": 1.7314,
+      "step": 5217
+    },
+    {
+      "epoch": 2.251024369204227,
+      "grad_norm": 0.19673076272010803,
+      "learning_rate": 2.2233107901802028e-05,
+      "loss": 2.1071,
+      "step": 5218
+    },
+    {
+      "epoch": 2.251455682553375,
+      "grad_norm": 0.20075039565563202,
+      "learning_rate": 2.220886076342705e-05,
+      "loss": 1.973,
+      "step": 5219
+    },
+    {
+      "epoch": 2.251886995902523,
+      "grad_norm": 0.17662428319454193,
+      "learning_rate": 2.218462455603475e-05,
+      "loss": 2.1947,
+      "step": 5220
+    },
+    {
+      "epoch": 2.252318309251671,
+      "grad_norm": 0.19919851422309875,
+      "learning_rate": 2.2160399284643514e-05,
+      "loss": 2.1702,
+      "step": 5221
+    },
+    {
+      "epoch": 2.2527496226008195,
+      "grad_norm": 0.17526623606681824,
+      "learning_rate": 2.2136184954269456e-05,
+      "loss": 2.0352,
+      "step": 5222
+    },
+    {
+      "epoch": 2.2531809359499677,
+      "grad_norm": 0.20351293683052063,
+      "learning_rate": 2.211198156992645e-05,
+      "loss": 2.384,
+      "step": 5223
+    },
+    {
+      "epoch": 2.253612249299116,
+      "grad_norm": 0.19407595694065094,
+      "learning_rate": 2.2087789136625973e-05,
+      "loss": 2.0531,
+      "step": 5224
+    },
+    {
+      "epoch": 2.2540435626482638,
+      "grad_norm": 0.19197069108486176,
+      "learning_rate": 2.206360765937743e-05,
+      "loss": 2.1488,
+      "step": 5225
+    },
+    {
+      "epoch": 2.2540435626482638,
+      "eval_loss": 2.0879340171813965,
+      "eval_runtime": 203.9602,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5225
+    },
+    {
+      "epoch": 2.254474875997412,
+      "grad_norm": 0.19059327244758606,
+      "learning_rate": 2.2039437143187812e-05,
+      "loss": 2.1152,
+      "step": 5226
+    },
+    {
+      "epoch": 2.2549061893465603,
+      "grad_norm": 0.19949564337730408,
+      "learning_rate": 2.2015277593061908e-05,
+      "loss": 2.1029,
+      "step": 5227
+    },
+    {
+      "epoch": 2.2553375026957085,
+      "grad_norm": 0.190524160861969,
+      "learning_rate": 2.1991129014002203e-05,
+      "loss": 2.0127,
+      "step": 5228
+    },
+    {
+      "epoch": 2.2557688160448564,
+      "grad_norm": 0.18100754916667938,
+      "learning_rate": 2.1966991411008938e-05,
+      "loss": 1.8489,
+      "step": 5229
+    },
+    {
+      "epoch": 2.2562001293940046,
+      "grad_norm": 0.21011635661125183,
+      "learning_rate": 2.1942864789080055e-05,
+      "loss": 2.23,
+      "step": 5230
+    },
+    {
+      "epoch": 2.256631442743153,
+      "grad_norm": 0.18916666507720947,
+      "learning_rate": 2.1918749153211243e-05,
+      "loss": 2.0753,
+      "step": 5231
+    },
+    {
+      "epoch": 2.257062756092301,
+      "grad_norm": 0.18899627029895782,
+      "learning_rate": 2.1894644508395906e-05,
+      "loss": 2.1395,
+      "step": 5232
+    },
+    {
+      "epoch": 2.2574940694414494,
+      "grad_norm": 0.20524995028972626,
+      "learning_rate": 2.1870550859625176e-05,
+      "loss": 2.226,
+      "step": 5233
+    },
+    {
+      "epoch": 2.257925382790597,
+      "grad_norm": 0.20154394209384918,
+      "learning_rate": 2.1846468211887898e-05,
+      "loss": 2.0271,
+      "step": 5234
+    },
+    {
+      "epoch": 2.2583566961397454,
+      "grad_norm": 0.20304295420646667,
+      "learning_rate": 2.1822396570170642e-05,
+      "loss": 2.0985,
+      "step": 5235
+    },
+    {
+      "epoch": 2.2587880094888937,
+      "grad_norm": 0.18911685049533844,
+      "learning_rate": 2.179833593945775e-05,
+      "loss": 2.1121,
+      "step": 5236
+    },
+    {
+      "epoch": 2.259219322838042,
+      "grad_norm": 0.20598070323467255,
+      "learning_rate": 2.1774286324731193e-05,
+      "loss": 1.992,
+      "step": 5237
+    },
+    {
+      "epoch": 2.2596506361871898,
+      "grad_norm": 0.19132232666015625,
+      "learning_rate": 2.175024773097069e-05,
+      "loss": 2.0416,
+      "step": 5238
+    },
+    {
+      "epoch": 2.260081949536338,
+      "grad_norm": 0.18360275030136108,
+      "learning_rate": 2.172622016315374e-05,
+      "loss": 2.1088,
+      "step": 5239
+    },
+    {
+      "epoch": 2.2605132628854863,
+      "grad_norm": 0.23842813074588776,
+      "learning_rate": 2.170220362625551e-05,
+      "loss": 2.1172,
+      "step": 5240
+    },
+    {
+      "epoch": 2.2609445762346345,
+      "grad_norm": 0.22701770067214966,
+      "learning_rate": 2.1678198125248895e-05,
+      "loss": 2.3083,
+      "step": 5241
+    },
+    {
+      "epoch": 2.2613758895837828,
+      "grad_norm": 0.18772007524967194,
+      "learning_rate": 2.165420366510442e-05,
+      "loss": 2.1169,
+      "step": 5242
+    },
+    {
+      "epoch": 2.2618072029329306,
+      "grad_norm": 0.9131419062614441,
+      "learning_rate": 2.1630220250790487e-05,
+      "loss": 2.2628,
+      "step": 5243
+    },
+    {
+      "epoch": 2.262238516282079,
+      "grad_norm": 0.213560551404953,
+      "learning_rate": 2.160624788727309e-05,
+      "loss": 2.1191,
+      "step": 5244
+    },
+    {
+      "epoch": 2.262669829631227,
+      "grad_norm": 0.1981581449508667,
+      "learning_rate": 2.1582286579515975e-05,
+      "loss": 2.2597,
+      "step": 5245
+    },
+    {
+      "epoch": 2.2631011429803753,
+      "grad_norm": 0.19422860443592072,
+      "learning_rate": 2.1558336332480593e-05,
+      "loss": 2.2362,
+      "step": 5246
+    },
+    {
+      "epoch": 2.263532456329523,
+      "grad_norm": 0.3238939046859741,
+      "learning_rate": 2.1534397151126094e-05,
+      "loss": 2.2472,
+      "step": 5247
+    },
+    {
+      "epoch": 2.2639637696786714,
+      "grad_norm": 0.19412267208099365,
+      "learning_rate": 2.1510469040409355e-05,
+      "loss": 2.1139,
+      "step": 5248
+    },
+    {
+      "epoch": 2.2643950830278197,
+      "grad_norm": 0.1949174553155899,
+      "learning_rate": 2.1486552005284934e-05,
+      "loss": 2.297,
+      "step": 5249
+    },
+    {
+      "epoch": 2.264826396376968,
+      "grad_norm": 0.20700618624687195,
+      "learning_rate": 2.146264605070518e-05,
+      "loss": 2.2371,
+      "step": 5250
+    },
+    {
+      "epoch": 2.264826396376968,
+      "eval_loss": 2.087846279144287,
+      "eval_runtime": 204.2358,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5250
+    },
+    {
+      "epoch": 2.265257709726116,
+      "grad_norm": 0.19957131147384644,
+      "learning_rate": 2.1438751181620013e-05,
+      "loss": 2.1992,
+      "step": 5251
+    },
+    {
+      "epoch": 2.265689023075264,
+      "grad_norm": 0.20005294680595398,
+      "learning_rate": 2.1414867402977118e-05,
+      "loss": 2.3806,
+      "step": 5252
+    },
+    {
+      "epoch": 2.2661203364244122,
+      "grad_norm": 0.19896568357944489,
+      "learning_rate": 2.1390994719721957e-05,
+      "loss": 2.1262,
+      "step": 5253
+    },
+    {
+      "epoch": 2.2665516497735605,
+      "grad_norm": 0.1914672702550888,
+      "learning_rate": 2.1367133136797594e-05,
+      "loss": 1.9649,
+      "step": 5254
+    },
+    {
+      "epoch": 2.2669829631227087,
+      "grad_norm": 0.20051509141921997,
+      "learning_rate": 2.1343282659144855e-05,
+      "loss": 2.0603,
+      "step": 5255
+    },
+    {
+      "epoch": 2.2674142764718566,
+      "grad_norm": 0.2132277637720108,
+      "learning_rate": 2.1319443291702178e-05,
+      "loss": 2.3511,
+      "step": 5256
+    },
+    {
+      "epoch": 2.267845589821005,
+      "grad_norm": 0.1911388635635376,
+      "learning_rate": 2.129561503940583e-05,
+      "loss": 2.0434,
+      "step": 5257
+    },
+    {
+      "epoch": 2.268276903170153,
+      "grad_norm": 0.19434180855751038,
+      "learning_rate": 2.12717979071897e-05,
+      "loss": 2.23,
+      "step": 5258
+    },
+    {
+      "epoch": 2.2687082165193013,
+      "grad_norm": 0.19305378198623657,
+      "learning_rate": 2.124799189998536e-05,
+      "loss": 2.3032,
+      "step": 5259
+    },
+    {
+      "epoch": 2.2691395298684496,
+      "grad_norm": 0.23718416690826416,
+      "learning_rate": 2.1224197022722124e-05,
+      "loss": 2.1086,
+      "step": 5260
+    },
+    {
+      "epoch": 2.269570843217598,
+      "grad_norm": 0.18516016006469727,
+      "learning_rate": 2.120041328032698e-05,
+      "loss": 2.1842,
+      "step": 5261
+    },
+    {
+      "epoch": 2.2700021565667456,
+      "grad_norm": 0.18666745722293854,
+      "learning_rate": 2.1176640677724603e-05,
+      "loss": 2.1297,
+      "step": 5262
+    },
+    {
+      "epoch": 2.270433469915894,
+      "grad_norm": 0.27829509973526,
+      "learning_rate": 2.1152879219837357e-05,
+      "loss": 2.0394,
+      "step": 5263
+    },
+    {
+      "epoch": 2.270864783265042,
+      "grad_norm": 0.22595706582069397,
+      "learning_rate": 2.1129128911585376e-05,
+      "loss": 2.1602,
+      "step": 5264
+    },
+    {
+      "epoch": 2.27129609661419,
+      "grad_norm": 0.20288696885108948,
+      "learning_rate": 2.1105389757886344e-05,
+      "loss": 2.1171,
+      "step": 5265
+    },
+    {
+      "epoch": 2.271727409963338,
+      "grad_norm": 0.2027527093887329,
+      "learning_rate": 2.1081661763655756e-05,
+      "loss": 2.1885,
+      "step": 5266
+    },
+    {
+      "epoch": 2.2721587233124865,
+      "grad_norm": 0.1916697472333908,
+      "learning_rate": 2.1057944933806717e-05,
+      "loss": 2.0902,
+      "step": 5267
+    },
+    {
+      "epoch": 2.2725900366616347,
+      "grad_norm": 0.2065373808145523,
+      "learning_rate": 2.1034239273250094e-05,
+      "loss": 2.231,
+      "step": 5268
+    },
+    {
+      "epoch": 2.273021350010783,
+      "grad_norm": 0.436259388923645,
+      "learning_rate": 2.101054478689442e-05,
+      "loss": 2.1565,
+      "step": 5269
+    },
+    {
+      "epoch": 2.2734526633599312,
+      "grad_norm": 0.21986541152000427,
+      "learning_rate": 2.098686147964582e-05,
+      "loss": 2.0916,
+      "step": 5270
+    },
+    {
+      "epoch": 2.273883976709079,
+      "grad_norm": 0.2018526792526245,
+      "learning_rate": 2.096318935640825e-05,
+      "loss": 2.1801,
+      "step": 5271
+    },
+    {
+      "epoch": 2.2743152900582273,
+      "grad_norm": 0.19900456070899963,
+      "learning_rate": 2.0939528422083258e-05,
+      "loss": 2.2055,
+      "step": 5272
+    },
+    {
+      "epoch": 2.2747466034073756,
+      "grad_norm": 2.7183778285980225,
+      "learning_rate": 2.0915878681570107e-05,
+      "loss": 2.0523,
+      "step": 5273
+    },
+    {
+      "epoch": 2.2751779167565234,
+      "grad_norm": 0.18772058188915253,
+      "learning_rate": 2.0892240139765724e-05,
+      "loss": 2.1065,
+      "step": 5274
+    },
+    {
+      "epoch": 2.2756092301056716,
+      "grad_norm": 0.2923305630683899,
+      "learning_rate": 2.0868612801564735e-05,
+      "loss": 2.213,
+      "step": 5275
+    },
+    {
+      "epoch": 2.2756092301056716,
+      "eval_loss": 2.088007926940918,
+      "eval_runtime": 203.9558,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5275
+    },
+    {
+      "epoch": 2.27604054345482,
+      "grad_norm": 0.18926513195037842,
+      "learning_rate": 2.0844996671859436e-05,
+      "loss": 2.1041,
+      "step": 5276
+    },
+    {
+      "epoch": 2.276471856803968,
+      "grad_norm": 0.1900864690542221,
+      "learning_rate": 2.08213917555398e-05,
+      "loss": 2.1807,
+      "step": 5277
+    },
+    {
+      "epoch": 2.2769031701531164,
+      "grad_norm": 0.21080556511878967,
+      "learning_rate": 2.07977980574935e-05,
+      "loss": 2.1454,
+      "step": 5278
+    },
+    {
+      "epoch": 2.2773344835022646,
+      "grad_norm": 0.20365285873413086,
+      "learning_rate": 2.0774215582605852e-05,
+      "loss": 2.0932,
+      "step": 5279
+    },
+    {
+      "epoch": 2.2777657968514124,
+      "grad_norm": 0.2221464365720749,
+      "learning_rate": 2.0750644335759867e-05,
+      "loss": 2.1837,
+      "step": 5280
+    },
+    {
+      "epoch": 2.2781971102005607,
+      "grad_norm": 0.17998287081718445,
+      "learning_rate": 2.0727084321836206e-05,
+      "loss": 2.2303,
+      "step": 5281
+    },
+    {
+      "epoch": 2.278628423549709,
+      "grad_norm": 0.18822330236434937,
+      "learning_rate": 2.070353554571331e-05,
+      "loss": 2.0179,
+      "step": 5282
+    },
+    {
+      "epoch": 2.279059736898857,
+      "grad_norm": 0.21984271705150604,
+      "learning_rate": 2.067999801226713e-05,
+      "loss": 2.2341,
+      "step": 5283
+    },
+    {
+      "epoch": 2.279491050248005,
+      "grad_norm": 0.20923283696174622,
+      "learning_rate": 2.0656471726371378e-05,
+      "loss": 2.2797,
+      "step": 5284
+    },
+    {
+      "epoch": 2.2799223635971533,
+      "grad_norm": 0.19424287974834442,
+      "learning_rate": 2.0632956692897457e-05,
+      "loss": 2.2956,
+      "step": 5285
+    },
+    {
+      "epoch": 2.2803536769463015,
+      "grad_norm": 0.19724376499652863,
+      "learning_rate": 2.0609452916714402e-05,
+      "loss": 2.3226,
+      "step": 5286
+    },
+    {
+      "epoch": 2.28078499029545,
+      "grad_norm": 0.2070743590593338,
+      "learning_rate": 2.0585960402688956e-05,
+      "loss": 2.0051,
+      "step": 5287
+    },
+    {
+      "epoch": 2.281216303644598,
+      "grad_norm": 0.21037553250789642,
+      "learning_rate": 2.0562479155685422e-05,
+      "loss": 1.9465,
+      "step": 5288
+    },
+    {
+      "epoch": 2.281647616993746,
+      "grad_norm": 0.19721102714538574,
+      "learning_rate": 2.053900918056592e-05,
+      "loss": 2.4951,
+      "step": 5289
+    },
+    {
+      "epoch": 2.282078930342894,
+      "grad_norm": 0.20401695370674133,
+      "learning_rate": 2.0515550482190145e-05,
+      "loss": 2.3082,
+      "step": 5290
+    },
+    {
+      "epoch": 2.2825102436920424,
+      "grad_norm": 0.197083979845047,
+      "learning_rate": 2.0492103065415465e-05,
+      "loss": 2.41,
+      "step": 5291
+    },
+    {
+      "epoch": 2.2829415570411906,
+      "grad_norm": 0.1975667029619217,
+      "learning_rate": 2.0468666935096943e-05,
+      "loss": 2.0927,
+      "step": 5292
+    },
+    {
+      "epoch": 2.2833728703903384,
+      "grad_norm": 0.18257242441177368,
+      "learning_rate": 2.0445242096087272e-05,
+      "loss": 1.9528,
+      "step": 5293
+    },
+    {
+      "epoch": 2.2838041837394867,
+      "grad_norm": 0.20901605486869812,
+      "learning_rate": 2.0421828553236833e-05,
+      "loss": 1.9898,
+      "step": 5294
+    },
+    {
+      "epoch": 2.284235497088635,
+      "grad_norm": 0.19688163697719574,
+      "learning_rate": 2.0398426311393613e-05,
+      "loss": 2.0778,
+      "step": 5295
+    },
+    {
+      "epoch": 2.284666810437783,
+      "grad_norm": 0.21484370529651642,
+      "learning_rate": 2.0375035375403393e-05,
+      "loss": 2.1694,
+      "step": 5296
+    },
+    {
+      "epoch": 2.2850981237869314,
+      "grad_norm": 0.2258901596069336,
+      "learning_rate": 2.035165575010944e-05,
+      "loss": 2.1767,
+      "step": 5297
+    },
+    {
+      "epoch": 2.2855294371360793,
+      "grad_norm": 0.20959137380123138,
+      "learning_rate": 2.0328287440352772e-05,
+      "loss": 2.2924,
+      "step": 5298
+    },
+    {
+      "epoch": 2.2859607504852275,
+      "grad_norm": 0.1976427137851715,
+      "learning_rate": 2.030493045097209e-05,
+      "loss": 2.2265,
+      "step": 5299
+    },
+    {
+      "epoch": 2.2863920638343758,
+      "grad_norm": 0.1976931393146515,
+      "learning_rate": 2.0281584786803686e-05,
+      "loss": 2.1814,
+      "step": 5300
+    },
+    {
+      "epoch": 2.2863920638343758,
+      "eval_loss": 2.087761640548706,
+      "eval_runtime": 203.8094,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5300
+    },
+    {
+      "epoch": 2.286823377183524,
+      "grad_norm": 0.19544154405593872,
+      "learning_rate": 2.0258250452681572e-05,
+      "loss": 2.2168,
+      "step": 5301
+    },
+    {
+      "epoch": 2.287254690532672,
+      "grad_norm": 0.2057940512895584,
+      "learning_rate": 2.02349274534373e-05,
+      "loss": 2.2899,
+      "step": 5302
+    },
+    {
+      "epoch": 2.28768600388182,
+      "grad_norm": 0.1918468326330185,
+      "learning_rate": 2.0211615793900223e-05,
+      "loss": 2.1694,
+      "step": 5303
+    },
+    {
+      "epoch": 2.2881173172309683,
+      "grad_norm": 0.19360606372356415,
+      "learning_rate": 2.018831547889725e-05,
+      "loss": 2.2603,
+      "step": 5304
+    },
+    {
+      "epoch": 2.2885486305801166,
+      "grad_norm": 0.19668936729431152,
+      "learning_rate": 2.016502651325297e-05,
+      "loss": 2.1752,
+      "step": 5305
+    },
+    {
+      "epoch": 2.288979943929265,
+      "grad_norm": 0.19086293876171112,
+      "learning_rate": 2.0141748901789608e-05,
+      "loss": 1.9608,
+      "step": 5306
+    },
+    {
+      "epoch": 2.2894112572784127,
+      "grad_norm": 0.19300664961338043,
+      "learning_rate": 2.0118482649327063e-05,
+      "loss": 2.2325,
+      "step": 5307
+    },
+    {
+      "epoch": 2.289842570627561,
+      "grad_norm": 0.17769969999790192,
+      "learning_rate": 2.0095227760682862e-05,
+      "loss": 2.0467,
+      "step": 5308
+    },
+    {
+      "epoch": 2.290273883976709,
+      "grad_norm": 0.2028777003288269,
+      "learning_rate": 2.0071984240672174e-05,
+      "loss": 2.0266,
+      "step": 5309
+    },
+    {
+      "epoch": 2.2907051973258574,
+      "grad_norm": 0.1918572634458542,
+      "learning_rate": 2.0048752094107833e-05,
+      "loss": 2.0798,
+      "step": 5310
+    },
+    {
+      "epoch": 2.2911365106750052,
+      "grad_norm": 0.20298618078231812,
+      "learning_rate": 2.0025531325800315e-05,
+      "loss": 2.2868,
+      "step": 5311
+    },
+    {
+      "epoch": 2.2915678240241535,
+      "grad_norm": 0.20322071015834808,
+      "learning_rate": 2.0002321940557718e-05,
+      "loss": 2.1486,
+      "step": 5312
+    },
+    {
+      "epoch": 2.2919991373733017,
+      "grad_norm": 0.2008322775363922,
+      "learning_rate": 1.997912394318579e-05,
+      "loss": 2.2418,
+      "step": 5313
+    },
+    {
+      "epoch": 2.29243045072245,
+      "grad_norm": 0.21370355784893036,
+      "learning_rate": 1.9955937338487994e-05,
+      "loss": 2.282,
+      "step": 5314
+    },
+    {
+      "epoch": 2.2928617640715983,
+      "grad_norm": 0.18719002604484558,
+      "learning_rate": 1.9932762131265295e-05,
+      "loss": 2.1381,
+      "step": 5315
+    },
+    {
+      "epoch": 2.293293077420746,
+      "grad_norm": 0.19345025718212128,
+      "learning_rate": 1.990959832631638e-05,
+      "loss": 2.295,
+      "step": 5316
+    },
+    {
+      "epoch": 2.2937243907698943,
+      "grad_norm": 0.21962137520313263,
+      "learning_rate": 1.988644592843762e-05,
+      "loss": 2.0284,
+      "step": 5317
+    },
+    {
+      "epoch": 2.2941557041190426,
+      "grad_norm": 0.18697109818458557,
+      "learning_rate": 1.9863304942422932e-05,
+      "loss": 2.0799,
+      "step": 5318
+    },
+    {
+      "epoch": 2.294587017468191,
+      "grad_norm": 0.18803496658802032,
+      "learning_rate": 1.9840175373063947e-05,
+      "loss": 2.1393,
+      "step": 5319
+    },
+    {
+      "epoch": 2.2950183308173386,
+      "grad_norm": 0.1989171802997589,
+      "learning_rate": 1.981705722514983e-05,
+      "loss": 2.0179,
+      "step": 5320
+    },
+    {
+      "epoch": 2.295449644166487,
+      "grad_norm": 0.19127577543258667,
+      "learning_rate": 1.9793950503467504e-05,
+      "loss": 2.3421,
+      "step": 5321
+    },
+    {
+      "epoch": 2.295880957515635,
+      "grad_norm": 0.19992578029632568,
+      "learning_rate": 1.9770855212801444e-05,
+      "loss": 2.2596,
+      "step": 5322
+    },
+    {
+      "epoch": 2.2963122708647834,
+      "grad_norm": 0.807466447353363,
+      "learning_rate": 1.9747771357933794e-05,
+      "loss": 2.0612,
+      "step": 5323
+    },
+    {
+      "epoch": 2.2967435842139317,
+      "grad_norm": 0.18776492774486542,
+      "learning_rate": 1.972469894364431e-05,
+      "loss": 2.1945,
+      "step": 5324
+    },
+    {
+      "epoch": 2.2971748975630795,
+      "grad_norm": 0.19602049887180328,
+      "learning_rate": 1.9701637974710393e-05,
+      "loss": 2.2251,
+      "step": 5325
+    },
+    {
+      "epoch": 2.2971748975630795,
+      "eval_loss": 2.0878446102142334,
+      "eval_runtime": 203.056,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 5325
+    },
+    {
+      "epoch": 2.2976062109122277,
+      "grad_norm": 0.19274592399597168,
+      "learning_rate": 1.967858845590706e-05,
+      "loss": 2.0118,
+      "step": 5326
+    },
+    {
+      "epoch": 2.298037524261376,
+      "grad_norm": 0.1943180114030838,
+      "learning_rate": 1.965555039200695e-05,
+      "loss": 2.1096,
+      "step": 5327
+    },
+    {
+      "epoch": 2.2984688376105242,
+      "grad_norm": 0.18350040912628174,
+      "learning_rate": 1.9632523787780423e-05,
+      "loss": 2.0139,
+      "step": 5328
+    },
+    {
+      "epoch": 2.298900150959672,
+      "grad_norm": 0.1943226009607315,
+      "learning_rate": 1.9609508647995303e-05,
+      "loss": 2.1024,
+      "step": 5329
+    },
+    {
+      "epoch": 2.2993314643088203,
+      "grad_norm": 0.2782226800918579,
+      "learning_rate": 1.9586504977417136e-05,
+      "loss": 2.1859,
+      "step": 5330
+    },
+    {
+      "epoch": 2.2997627776579685,
+      "grad_norm": 0.23133975267410278,
+      "learning_rate": 1.9563512780809126e-05,
+      "loss": 2.2521,
+      "step": 5331
+    },
+    {
+      "epoch": 2.300194091007117,
+      "grad_norm": 0.19578418135643005,
+      "learning_rate": 1.9540532062932042e-05,
+      "loss": 2.154,
+      "step": 5332
+    },
+    {
+      "epoch": 2.300625404356265,
+      "grad_norm": 0.1878523975610733,
+      "learning_rate": 1.9517562828544316e-05,
+      "loss": 2.2197,
+      "step": 5333
+    },
+    {
+      "epoch": 2.301056717705413,
+      "grad_norm": 0.21165266633033752,
+      "learning_rate": 1.9494605082401894e-05,
+      "loss": 2.1544,
+      "step": 5334
+    },
+    {
+      "epoch": 2.301488031054561,
+      "grad_norm": 0.19851690530776978,
+      "learning_rate": 1.9471658829258517e-05,
+      "loss": 1.9271,
+      "step": 5335
+    },
+    {
+      "epoch": 2.3019193444037094,
+      "grad_norm": 0.21829640865325928,
+      "learning_rate": 1.9448724073865425e-05,
+      "loss": 2.1683,
+      "step": 5336
+    },
+    {
+      "epoch": 2.3023506577528576,
+      "grad_norm": 0.19689638912677765,
+      "learning_rate": 1.9425800820971508e-05,
+      "loss": 2.2477,
+      "step": 5337
+    },
+    {
+      "epoch": 2.3027819711020054,
+      "grad_norm": 0.19567584991455078,
+      "learning_rate": 1.9402889075323293e-05,
+      "loss": 1.9999,
+      "step": 5338
+    },
+    {
+      "epoch": 2.3032132844511537,
+      "grad_norm": 0.18084855377674103,
+      "learning_rate": 1.937998884166488e-05,
+      "loss": 1.9266,
+      "step": 5339
+    },
+    {
+      "epoch": 2.303644597800302,
+      "grad_norm": 0.20621658861637115,
+      "learning_rate": 1.9357100124738037e-05,
+      "loss": 2.0463,
+      "step": 5340
+    },
+    {
+      "epoch": 2.30407591114945,
+      "grad_norm": 0.237623929977417,
+      "learning_rate": 1.9334222929282088e-05,
+      "loss": 1.9552,
+      "step": 5341
+    },
+    {
+      "epoch": 2.3045072244985985,
+      "grad_norm": 0.19974792003631592,
+      "learning_rate": 1.9311357260034075e-05,
+      "loss": 2.1163,
+      "step": 5342
+    },
+    {
+      "epoch": 2.3049385378477463,
+      "grad_norm": 0.20718558132648468,
+      "learning_rate": 1.9288503121728523e-05,
+      "loss": 2.1902,
+      "step": 5343
+    },
+    {
+      "epoch": 2.3053698511968945,
+      "grad_norm": 0.21341772377490997,
+      "learning_rate": 1.9265660519097624e-05,
+      "loss": 2.2041,
+      "step": 5344
+    },
+    {
+      "epoch": 2.305801164546043,
+      "grad_norm": 0.20624461770057678,
+      "learning_rate": 1.9242829456871246e-05,
+      "loss": 2.0728,
+      "step": 5345
+    },
+    {
+      "epoch": 2.306232477895191,
+      "grad_norm": 0.21290695667266846,
+      "learning_rate": 1.9220009939776774e-05,
+      "loss": 2.233,
+      "step": 5346
+    },
+    {
+      "epoch": 2.306663791244339,
+      "grad_norm": 0.2021273970603943,
+      "learning_rate": 1.9197201972539265e-05,
+      "loss": 2.2376,
+      "step": 5347
+    },
+    {
+      "epoch": 2.307095104593487,
+      "grad_norm": 0.19745802879333496,
+      "learning_rate": 1.9174405559881288e-05,
+      "loss": 2.2321,
+      "step": 5348
+    },
+    {
+      "epoch": 2.3075264179426354,
+      "grad_norm": 0.2031654715538025,
+      "learning_rate": 1.9151620706523167e-05,
+      "loss": 2.0205,
+      "step": 5349
+    },
+    {
+      "epoch": 2.3079577312917836,
+      "grad_norm": 0.20771507918834686,
+      "learning_rate": 1.9128847417182718e-05,
+      "loss": 2.0798,
+      "step": 5350
+    },
+    {
+      "epoch": 2.3079577312917836,
+      "eval_loss": 2.0877935886383057,
+      "eval_runtime": 203.6267,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5350
+    },
+    {
+      "epoch": 2.308389044640932,
+      "grad_norm": 0.189523383975029,
+      "learning_rate": 1.910608569657542e-05,
+      "loss": 2.2223,
+      "step": 5351
+    },
+    {
+      "epoch": 2.3088203579900797,
+      "grad_norm": 0.21872511506080627,
+      "learning_rate": 1.9083335549414318e-05,
+      "loss": 2.1221,
+      "step": 5352
+    },
+    {
+      "epoch": 2.309251671339228,
+      "grad_norm": 0.20272767543792725,
+      "learning_rate": 1.9060596980410086e-05,
+      "loss": 2.2688,
+      "step": 5353
+    },
+    {
+      "epoch": 2.309682984688376,
+      "grad_norm": 0.20737117528915405,
+      "learning_rate": 1.903786999427099e-05,
+      "loss": 2.1833,
+      "step": 5354
+    },
+    {
+      "epoch": 2.3101142980375244,
+      "grad_norm": 0.2046089470386505,
+      "learning_rate": 1.90151545957029e-05,
+      "loss": 2.2038,
+      "step": 5355
+    },
+    {
+      "epoch": 2.3105456113866722,
+      "grad_norm": 0.21469274163246155,
+      "learning_rate": 1.8992450789409294e-05,
+      "loss": 1.8162,
+      "step": 5356
+    },
+    {
+      "epoch": 2.3109769247358205,
+      "grad_norm": 0.19832271337509155,
+      "learning_rate": 1.8969758580091243e-05,
+      "loss": 2.0864,
+      "step": 5357
+    },
+    {
+      "epoch": 2.3114082380849688,
+      "grad_norm": 0.19137664139270782,
+      "learning_rate": 1.8947077972447414e-05,
+      "loss": 1.9514,
+      "step": 5358
+    },
+    {
+      "epoch": 2.311839551434117,
+      "grad_norm": 0.1882704496383667,
+      "learning_rate": 1.892440897117405e-05,
+      "loss": 2.0432,
+      "step": 5359
+    },
+    {
+      "epoch": 2.3122708647832653,
+      "grad_norm": 0.18820488452911377,
+      "learning_rate": 1.8901751580965084e-05,
+      "loss": 2.1626,
+      "step": 5360
+    },
+    {
+      "epoch": 2.312702178132413,
+      "grad_norm": 0.21664069592952728,
+      "learning_rate": 1.887910580651191e-05,
+      "loss": 2.1695,
+      "step": 5361
+    },
+    {
+      "epoch": 2.3131334914815613,
+      "grad_norm": 0.19514897465705872,
+      "learning_rate": 1.8856471652503597e-05,
+      "loss": 2.0783,
+      "step": 5362
+    },
+    {
+      "epoch": 2.3135648048307096,
+      "grad_norm": 0.22643904387950897,
+      "learning_rate": 1.8833849123626813e-05,
+      "loss": 2.2025,
+      "step": 5363
+    },
+    {
+      "epoch": 2.313996118179858,
+      "grad_norm": 0.17537686228752136,
+      "learning_rate": 1.8811238224565796e-05,
+      "loss": 2.1062,
+      "step": 5364
+    },
+    {
+      "epoch": 2.3144274315290057,
+      "grad_norm": 0.19371572136878967,
+      "learning_rate": 1.8788638960002398e-05,
+      "loss": 2.0133,
+      "step": 5365
+    },
+    {
+      "epoch": 2.314858744878154,
+      "grad_norm": 0.18847690522670746,
+      "learning_rate": 1.876605133461597e-05,
+      "loss": 1.9666,
+      "step": 5366
+    },
+    {
+      "epoch": 2.315290058227302,
+      "grad_norm": 0.1891636997461319,
+      "learning_rate": 1.8743475353083596e-05,
+      "loss": 2.1242,
+      "step": 5367
+    },
+    {
+      "epoch": 2.3157213715764504,
+      "grad_norm": 0.18955299258232117,
+      "learning_rate": 1.8720911020079877e-05,
+      "loss": 2.1174,
+      "step": 5368
+    },
+    {
+      "epoch": 2.3161526849255987,
+      "grad_norm": 0.18408434092998505,
+      "learning_rate": 1.8698358340276976e-05,
+      "loss": 1.9551,
+      "step": 5369
+    },
+    {
+      "epoch": 2.3165839982747465,
+      "grad_norm": 0.20017340779304504,
+      "learning_rate": 1.8675817318344687e-05,
+      "loss": 2.141,
+      "step": 5370
+    },
+    {
+      "epoch": 2.3170153116238947,
+      "grad_norm": 0.18992261588573456,
+      "learning_rate": 1.865328795895037e-05,
+      "loss": 2.0916,
+      "step": 5371
+    },
+    {
+      "epoch": 2.317446624973043,
+      "grad_norm": 0.2128012776374817,
+      "learning_rate": 1.863077026675899e-05,
+      "loss": 2.2291,
+      "step": 5372
+    },
+    {
+      "epoch": 2.3178779383221912,
+      "grad_norm": 0.20145027339458466,
+      "learning_rate": 1.8608264246433036e-05,
+      "loss": 2.3127,
+      "step": 5373
+    },
+    {
+      "epoch": 2.318309251671339,
+      "grad_norm": 0.18959259986877441,
+      "learning_rate": 1.8585769902632714e-05,
+      "loss": 2.2089,
+      "step": 5374
+    },
+    {
+      "epoch": 2.3187405650204873,
+      "grad_norm": 0.19251097738742828,
+      "learning_rate": 1.8563287240015635e-05,
+      "loss": 2.143,
+      "step": 5375
+    },
+    {
+      "epoch": 2.3187405650204873,
+      "eval_loss": 2.08791184425354,
+      "eval_runtime": 203.5372,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5375
+    },
+    {
+      "epoch": 2.3191718783696356,
+      "grad_norm": 0.1923358142375946,
+      "learning_rate": 1.85408162632371e-05,
+      "loss": 1.9816,
+      "step": 5376
+    },
+    {
+      "epoch": 2.319603191718784,
+      "grad_norm": 0.17821118235588074,
+      "learning_rate": 1.8518356976950012e-05,
+      "loss": 1.9172,
+      "step": 5377
+    },
+    {
+      "epoch": 2.320034505067932,
+      "grad_norm": 0.22167186439037323,
+      "learning_rate": 1.8495909385804776e-05,
+      "loss": 2.3065,
+      "step": 5378
+    },
+    {
+      "epoch": 2.32046581841708,
+      "grad_norm": 0.18814527988433838,
+      "learning_rate": 1.847347349444944e-05,
+      "loss": 2.1164,
+      "step": 5379
+    },
+    {
+      "epoch": 2.320897131766228,
+      "grad_norm": 0.1985539346933365,
+      "learning_rate": 1.845104930752954e-05,
+      "loss": 2.2795,
+      "step": 5380
+    },
+    {
+      "epoch": 2.3213284451153764,
+      "grad_norm": 0.20663782954216003,
+      "learning_rate": 1.842863682968831e-05,
+      "loss": 2.1005,
+      "step": 5381
+    },
+    {
+      "epoch": 2.3217597584645246,
+      "grad_norm": 0.2148127406835556,
+      "learning_rate": 1.8406236065566488e-05,
+      "loss": 2.0653,
+      "step": 5382
+    },
+    {
+      "epoch": 2.3221910718136725,
+      "grad_norm": 0.19326218962669373,
+      "learning_rate": 1.838384701980238e-05,
+      "loss": 2.132,
+      "step": 5383
+    },
+    {
+      "epoch": 2.3226223851628207,
+      "grad_norm": 0.19003017246723175,
+      "learning_rate": 1.836146969703189e-05,
+      "loss": 1.9125,
+      "step": 5384
+    },
+    {
+      "epoch": 2.323053698511969,
+      "grad_norm": 0.17955321073532104,
+      "learning_rate": 1.8339104101888488e-05,
+      "loss": 2.1892,
+      "step": 5385
+    },
+    {
+      "epoch": 2.3234850118611172,
+      "grad_norm": 0.18654711544513702,
+      "learning_rate": 1.8316750239003218e-05,
+      "loss": 2.2338,
+      "step": 5386
+    },
+    {
+      "epoch": 2.3239163252102655,
+      "grad_norm": 0.2069181203842163,
+      "learning_rate": 1.829440811300466e-05,
+      "loss": 2.2528,
+      "step": 5387
+    },
+    {
+      "epoch": 2.3243476385594133,
+      "grad_norm": 0.22007881104946136,
+      "learning_rate": 1.8272077728519077e-05,
+      "loss": 2.3733,
+      "step": 5388
+    },
+    {
+      "epoch": 2.3247789519085615,
+      "grad_norm": 0.2045450210571289,
+      "learning_rate": 1.8249759090170134e-05,
+      "loss": 2.2287,
+      "step": 5389
+    },
+    {
+      "epoch": 2.32521026525771,
+      "grad_norm": 0.19702892005443573,
+      "learning_rate": 1.8227452202579168e-05,
+      "loss": 2.1169,
+      "step": 5390
+    },
+    {
+      "epoch": 2.325641578606858,
+      "grad_norm": 0.17602592706680298,
+      "learning_rate": 1.82051570703651e-05,
+      "loss": 2.1293,
+      "step": 5391
+    },
+    {
+      "epoch": 2.326072891956006,
+      "grad_norm": 0.20260587334632874,
+      "learning_rate": 1.818287369814435e-05,
+      "loss": 2.2672,
+      "step": 5392
+    },
+    {
+      "epoch": 2.326504205305154,
+      "grad_norm": 0.20817983150482178,
+      "learning_rate": 1.8160602090530972e-05,
+      "loss": 2.1941,
+      "step": 5393
+    },
+    {
+      "epoch": 2.3269355186543024,
+      "grad_norm": 0.18919779360294342,
+      "learning_rate": 1.8138342252136467e-05,
+      "loss": 2.1362,
+      "step": 5394
+    },
+    {
+      "epoch": 2.3273668320034506,
+      "grad_norm": 0.19633108377456665,
+      "learning_rate": 1.811609418757005e-05,
+      "loss": 2.2055,
+      "step": 5395
+    },
+    {
+      "epoch": 2.327798145352599,
+      "grad_norm": 0.215801402926445,
+      "learning_rate": 1.8093857901438404e-05,
+      "loss": 2.1827,
+      "step": 5396
+    },
+    {
+      "epoch": 2.3282294587017467,
+      "grad_norm": 0.19851261377334595,
+      "learning_rate": 1.8071633398345804e-05,
+      "loss": 1.9047,
+      "step": 5397
+    },
+    {
+      "epoch": 2.328660772050895,
+      "grad_norm": 0.20020222663879395,
+      "learning_rate": 1.8049420682894054e-05,
+      "loss": 2.3035,
+      "step": 5398
+    },
+    {
+      "epoch": 2.329092085400043,
+      "grad_norm": 0.21065831184387207,
+      "learning_rate": 1.802721975968256e-05,
+      "loss": 2.1154,
+      "step": 5399
+    },
+    {
+      "epoch": 2.3295233987491915,
+      "grad_norm": 0.20882493257522583,
+      "learning_rate": 1.8005030633308252e-05,
+      "loss": 2.1965,
+      "step": 5400
+    },
+    {
+      "epoch": 2.3295233987491915,
+      "eval_loss": 2.0880227088928223,
+      "eval_runtime": 203.6302,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5400
+    },
+    {
+      "epoch": 2.3299547120983393,
+      "grad_norm": 0.1875244826078415,
+      "learning_rate": 1.798285330836564e-05,
+      "loss": 2.3193,
+      "step": 5401
+    },
+    {
+      "epoch": 2.3303860254474875,
+      "grad_norm": 0.19934353232383728,
+      "learning_rate": 1.7960687789446782e-05,
+      "loss": 2.0483,
+      "step": 5402
+    },
+    {
+      "epoch": 2.3308173387966358,
+      "grad_norm": 0.20099350810050964,
+      "learning_rate": 1.7938534081141288e-05,
+      "loss": 1.9604,
+      "step": 5403
+    },
+    {
+      "epoch": 2.331248652145784,
+      "grad_norm": 0.195001482963562,
+      "learning_rate": 1.791639218803633e-05,
+      "loss": 2.0237,
+      "step": 5404
+    },
+    {
+      "epoch": 2.3316799654949323,
+      "grad_norm": 0.19568021595478058,
+      "learning_rate": 1.789426211471659e-05,
+      "loss": 2.1974,
+      "step": 5405
+    },
+    {
+      "epoch": 2.33211127884408,
+      "grad_norm": 0.1914471983909607,
+      "learning_rate": 1.7872143865764436e-05,
+      "loss": 2.2391,
+      "step": 5406
+    },
+    {
+      "epoch": 2.3325425921932283,
+      "grad_norm": 0.19289901852607727,
+      "learning_rate": 1.7850037445759608e-05,
+      "loss": 2.1496,
+      "step": 5407
+    },
+    {
+      "epoch": 2.3329739055423766,
+      "grad_norm": 0.18522243201732635,
+      "learning_rate": 1.7827942859279482e-05,
+      "loss": 2.1157,
+      "step": 5408
+    },
+    {
+      "epoch": 2.333405218891525,
+      "grad_norm": 0.18957099318504333,
+      "learning_rate": 1.780586011089904e-05,
+      "loss": 1.9777,
+      "step": 5409
+    },
+    {
+      "epoch": 2.3338365322406727,
+      "grad_norm": 0.19008250534534454,
+      "learning_rate": 1.7783789205190717e-05,
+      "loss": 2.056,
+      "step": 5410
+    },
+    {
+      "epoch": 2.334267845589821,
+      "grad_norm": 0.19285881519317627,
+      "learning_rate": 1.776173014672457e-05,
+      "loss": 2.3684,
+      "step": 5411
+    },
+    {
+      "epoch": 2.334699158938969,
+      "grad_norm": 0.20712214708328247,
+      "learning_rate": 1.77396829400681e-05,
+      "loss": 2.3285,
+      "step": 5412
+    },
+    {
+      "epoch": 2.3351304722881174,
+      "grad_norm": 0.20356346666812897,
+      "learning_rate": 1.7717647589786486e-05,
+      "loss": 1.9902,
+      "step": 5413
+    },
+    {
+      "epoch": 2.3355617856372657,
+      "grad_norm": 0.19335150718688965,
+      "learning_rate": 1.7695624100442372e-05,
+      "loss": 2.13,
+      "step": 5414
+    },
+    {
+      "epoch": 2.3359930989864135,
+      "grad_norm": 0.2026914805173874,
+      "learning_rate": 1.7673612476595962e-05,
+      "loss": 2.1992,
+      "step": 5415
+    },
+    {
+      "epoch": 2.3364244123355618,
+      "grad_norm": 0.20388589799404144,
+      "learning_rate": 1.7651612722804986e-05,
+      "loss": 2.1595,
+      "step": 5416
+    },
+    {
+      "epoch": 2.33685572568471,
+      "grad_norm": 0.1827160269021988,
+      "learning_rate": 1.7629624843624758e-05,
+      "loss": 2.1954,
+      "step": 5417
+    },
+    {
+      "epoch": 2.3372870390338583,
+      "grad_norm": 0.22113530337810516,
+      "learning_rate": 1.7607648843608097e-05,
+      "loss": 2.1259,
+      "step": 5418
+    },
+    {
+      "epoch": 2.337718352383006,
+      "grad_norm": 0.2118062674999237,
+      "learning_rate": 1.7585684727305348e-05,
+      "loss": 2.0906,
+      "step": 5419
+    },
+    {
+      "epoch": 2.3381496657321543,
+      "grad_norm": 0.18365009129047394,
+      "learning_rate": 1.7563732499264492e-05,
+      "loss": 2.0491,
+      "step": 5420
+    },
+    {
+      "epoch": 2.3385809790813026,
+      "grad_norm": 0.2063271403312683,
+      "learning_rate": 1.754179216403092e-05,
+      "loss": 2.1538,
+      "step": 5421
+    },
+    {
+      "epoch": 2.339012292430451,
+      "grad_norm": 0.20388810336589813,
+      "learning_rate": 1.7519863726147607e-05,
+      "loss": 2.0918,
+      "step": 5422
+    },
+    {
+      "epoch": 2.339443605779599,
+      "grad_norm": 0.21293078362941742,
+      "learning_rate": 1.7497947190155118e-05,
+      "loss": 2.2654,
+      "step": 5423
+    },
+    {
+      "epoch": 2.339874919128747,
+      "grad_norm": 0.19700607657432556,
+      "learning_rate": 1.7476042560591505e-05,
+      "loss": 2.1123,
+      "step": 5424
+    },
+    {
+      "epoch": 2.340306232477895,
+      "grad_norm": 0.19425487518310547,
+      "learning_rate": 1.7454149841992366e-05,
+      "loss": 2.2007,
+      "step": 5425
+    },
+    {
+      "epoch": 2.340306232477895,
+      "eval_loss": 2.087836265563965,
+      "eval_runtime": 200.5705,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 5425
+    },
+    {
+      "epoch": 2.3407375458270434,
+      "grad_norm": 0.2275095134973526,
+      "learning_rate": 1.7432269038890773e-05,
+      "loss": 2.0679,
+      "step": 5426
+    },
+    {
+      "epoch": 2.3411688591761917,
+      "grad_norm": 0.23057441413402557,
+      "learning_rate": 1.7410400155817445e-05,
+      "loss": 2.1185,
+      "step": 5427
+    },
+    {
+      "epoch": 2.3416001725253395,
+      "grad_norm": 0.17289194464683533,
+      "learning_rate": 1.738854319730055e-05,
+      "loss": 1.9679,
+      "step": 5428
+    },
+    {
+      "epoch": 2.3420314858744877,
+      "grad_norm": 0.18307892978191376,
+      "learning_rate": 1.736669816786582e-05,
+      "loss": 1.9135,
+      "step": 5429
+    },
+    {
+      "epoch": 2.342462799223636,
+      "grad_norm": 0.19872839748859406,
+      "learning_rate": 1.73448650720365e-05,
+      "loss": 2.0873,
+      "step": 5430
+    },
+    {
+      "epoch": 2.3428941125727842,
+      "grad_norm": 0.19778376817703247,
+      "learning_rate": 1.732304391433337e-05,
+      "loss": 2.2195,
+      "step": 5431
+    },
+    {
+      "epoch": 2.3433254259219325,
+      "grad_norm": 0.2171076387166977,
+      "learning_rate": 1.7301234699274747e-05,
+      "loss": 2.2742,
+      "step": 5432
+    },
+    {
+      "epoch": 2.3437567392710803,
+      "grad_norm": 0.18240311741828918,
+      "learning_rate": 1.727943743137646e-05,
+      "loss": 2.073,
+      "step": 5433
+    },
+    {
+      "epoch": 2.3441880526202286,
+      "grad_norm": 0.19008196890354156,
+      "learning_rate": 1.725765211515188e-05,
+      "loss": 2.0548,
+      "step": 5434
+    },
+    {
+      "epoch": 2.344619365969377,
+      "grad_norm": 18.04897117614746,
+      "learning_rate": 1.7235878755111883e-05,
+      "loss": 2.2369,
+      "step": 5435
+    },
+    {
+      "epoch": 2.345050679318525,
+      "grad_norm": 0.18793705105781555,
+      "learning_rate": 1.7214117355764868e-05,
+      "loss": 1.9845,
+      "step": 5436
+    },
+    {
+      "epoch": 2.345481992667673,
+      "grad_norm": 0.1938720941543579,
+      "learning_rate": 1.7192367921616805e-05,
+      "loss": 2.145,
+      "step": 5437
+    },
+    {
+      "epoch": 2.345913306016821,
+      "grad_norm": 0.21334202587604523,
+      "learning_rate": 1.7170630457171166e-05,
+      "loss": 2.3754,
+      "step": 5438
+    },
+    {
+      "epoch": 2.3463446193659694,
+      "grad_norm": 0.1999683380126953,
+      "learning_rate": 1.714890496692888e-05,
+      "loss": 2.2846,
+      "step": 5439
+    },
+    {
+      "epoch": 2.3467759327151176,
+      "grad_norm": 0.2468402087688446,
+      "learning_rate": 1.712719145538845e-05,
+      "loss": 2.1985,
+      "step": 5440
+    },
+    {
+      "epoch": 2.347207246064266,
+      "grad_norm": 0.19367870688438416,
+      "learning_rate": 1.710548992704593e-05,
+      "loss": 2.2734,
+      "step": 5441
+    },
+    {
+      "epoch": 2.3476385594134137,
+      "grad_norm": 0.1968536078929901,
+      "learning_rate": 1.7083800386394845e-05,
+      "loss": 1.9915,
+      "step": 5442
+    },
+    {
+      "epoch": 2.348069872762562,
+      "grad_norm": 0.20344178378582,
+      "learning_rate": 1.7062122837926263e-05,
+      "loss": 2.057,
+      "step": 5443
+    },
+    {
+      "epoch": 2.34850118611171,
+      "grad_norm": 0.19281743466854095,
+      "learning_rate": 1.7040457286128732e-05,
+      "loss": 2.0647,
+      "step": 5444
+    },
+    {
+      "epoch": 2.3489324994608585,
+      "grad_norm": 0.1950780153274536,
+      "learning_rate": 1.7018803735488356e-05,
+      "loss": 2.2951,
+      "step": 5445
+    },
+    {
+      "epoch": 2.3493638128100063,
+      "grad_norm": 0.2115524411201477,
+      "learning_rate": 1.6997162190488745e-05,
+      "loss": 2.0425,
+      "step": 5446
+    },
+    {
+      "epoch": 2.3497951261591545,
+      "grad_norm": 0.18657691776752472,
+      "learning_rate": 1.6975532655611014e-05,
+      "loss": 2.0386,
+      "step": 5447
+    },
+    {
+      "epoch": 2.350226439508303,
+      "grad_norm": 0.19606566429138184,
+      "learning_rate": 1.695391513533378e-05,
+      "loss": 2.0809,
+      "step": 5448
+    },
+    {
+      "epoch": 2.350657752857451,
+      "grad_norm": 0.18347160518169403,
+      "learning_rate": 1.6932309634133203e-05,
+      "loss": 2.3827,
+      "step": 5449
+    },
+    {
+      "epoch": 2.3510890662065993,
+      "grad_norm": 0.19450697302818298,
+      "learning_rate": 1.6910716156482934e-05,
+      "loss": 2.3068,
+      "step": 5450
+    },
+    {
+      "epoch": 2.3510890662065993,
+      "eval_loss": 2.088000774383545,
+      "eval_runtime": 201.0248,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 5450
+    },
+    {
+      "epoch": 2.351520379555747,
+      "grad_norm": 0.19642473757266998,
+      "learning_rate": 1.6889134706854118e-05,
+      "loss": 2.1511,
+      "step": 5451
+    },
+    {
+      "epoch": 2.3519516929048954,
+      "grad_norm": 0.19885718822479248,
+      "learning_rate": 1.686756528971549e-05,
+      "loss": 2.2092,
+      "step": 5452
+    },
+    {
+      "epoch": 2.3523830062540436,
+      "grad_norm": 0.20507009327411652,
+      "learning_rate": 1.6846007909533178e-05,
+      "loss": 2.1743,
+      "step": 5453
+    },
+    {
+      "epoch": 2.352814319603192,
+      "grad_norm": 0.20062431693077087,
+      "learning_rate": 1.6824462570770865e-05,
+      "loss": 2.1195,
+      "step": 5454
+    },
+    {
+      "epoch": 2.3532456329523397,
+      "grad_norm": 0.2013835459947586,
+      "learning_rate": 1.68029292778898e-05,
+      "loss": 2.0451,
+      "step": 5455
+    },
+    {
+      "epoch": 2.353676946301488,
+      "grad_norm": 0.18703527748584747,
+      "learning_rate": 1.6781408035348657e-05,
+      "loss": 2.0735,
+      "step": 5456
+    },
+    {
+      "epoch": 2.354108259650636,
+      "grad_norm": 0.19562098383903503,
+      "learning_rate": 1.6759898847603668e-05,
+      "loss": 2.13,
+      "step": 5457
+    },
+    {
+      "epoch": 2.3545395729997844,
+      "grad_norm": 0.2325151264667511,
+      "learning_rate": 1.6738401719108477e-05,
+      "loss": 2.1173,
+      "step": 5458
+    },
+    {
+      "epoch": 2.3549708863489327,
+      "grad_norm": 0.19891884922981262,
+      "learning_rate": 1.6716916654314374e-05,
+      "loss": 2.1029,
+      "step": 5459
+    },
+    {
+      "epoch": 2.3554021996980805,
+      "grad_norm": 0.18463841080665588,
+      "learning_rate": 1.669544365767006e-05,
+      "loss": 2.0109,
+      "step": 5460
+    },
+    {
+      "epoch": 2.3558335130472288,
+      "grad_norm": 0.20649510622024536,
+      "learning_rate": 1.6673982733621735e-05,
+      "loss": 2.1988,
+      "step": 5461
+    },
+    {
+      "epoch": 2.356264826396377,
+      "grad_norm": 0.185276597738266,
+      "learning_rate": 1.6652533886613127e-05,
+      "loss": 2.1077,
+      "step": 5462
+    },
+    {
+      "epoch": 2.3566961397455253,
+      "grad_norm": 0.19789348542690277,
+      "learning_rate": 1.6631097121085463e-05,
+      "loss": 2.0926,
+      "step": 5463
+    },
+    {
+      "epoch": 2.357127453094673,
+      "grad_norm": 0.19963330030441284,
+      "learning_rate": 1.6609672441477446e-05,
+      "loss": 1.9511,
+      "step": 5464
+    },
+    {
+      "epoch": 2.3575587664438213,
+      "grad_norm": 0.2180710732936859,
+      "learning_rate": 1.658825985222527e-05,
+      "loss": 2.1496,
+      "step": 5465
+    },
+    {
+      "epoch": 2.3579900797929696,
+      "grad_norm": 0.19674089550971985,
+      "learning_rate": 1.6566859357762722e-05,
+      "loss": 2.0358,
+      "step": 5466
+    },
+    {
+      "epoch": 2.358421393142118,
+      "grad_norm": 0.19395115971565247,
+      "learning_rate": 1.6545470962520933e-05,
+      "loss": 2.1949,
+      "step": 5467
+    },
+    {
+      "epoch": 2.358852706491266,
+      "grad_norm": 0.19646000862121582,
+      "learning_rate": 1.6524094670928608e-05,
+      "loss": 2.1848,
+      "step": 5468
+    },
+    {
+      "epoch": 2.359284019840414,
+      "grad_norm": 0.17927643656730652,
+      "learning_rate": 1.6502730487411986e-05,
+      "loss": 2.1109,
+      "step": 5469
+    },
+    {
+      "epoch": 2.359715333189562,
+      "grad_norm": 0.19817306101322174,
+      "learning_rate": 1.6481378416394733e-05,
+      "loss": 2.2828,
+      "step": 5470
+    },
+    {
+      "epoch": 2.3601466465387104,
+      "grad_norm": 0.20632949471473694,
+      "learning_rate": 1.646003846229805e-05,
+      "loss": 2.2486,
+      "step": 5471
+    },
+    {
+      "epoch": 2.3605779598878587,
+      "grad_norm": 0.1893763691186905,
+      "learning_rate": 1.6438710629540534e-05,
+      "loss": 2.1597,
+      "step": 5472
+    },
+    {
+      "epoch": 2.3610092732370065,
+      "grad_norm": 0.20030063390731812,
+      "learning_rate": 1.6417394922538423e-05,
+      "loss": 1.9663,
+      "step": 5473
+    },
+    {
+      "epoch": 2.3614405865861547,
+      "grad_norm": 0.18441982567310333,
+      "learning_rate": 1.6396091345705346e-05,
+      "loss": 2.0616,
+      "step": 5474
+    },
+    {
+      "epoch": 2.361871899935303,
+      "grad_norm": 0.2071865200996399,
+      "learning_rate": 1.6374799903452438e-05,
+      "loss": 2.1528,
+      "step": 5475
+    },
+    {
+      "epoch": 2.361871899935303,
+      "eval_loss": 2.0878982543945312,
+      "eval_runtime": 204.4375,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5475
+    },
+    {
+      "epoch": 2.3623032132844513,
+      "grad_norm": 0.21171632409095764,
+      "learning_rate": 1.6353520600188315e-05,
+      "loss": 2.2473,
+      "step": 5476
+    },
+    {
+      "epoch": 2.3627345266335995,
+      "grad_norm": 0.21455693244934082,
+      "learning_rate": 1.633225344031911e-05,
+      "loss": 2.3628,
+      "step": 5477
+    },
+    {
+      "epoch": 2.3631658399827473,
+      "grad_norm": 0.195206418633461,
+      "learning_rate": 1.6310998428248404e-05,
+      "loss": 2.0825,
+      "step": 5478
+    },
+    {
+      "epoch": 2.3635971533318956,
+      "grad_norm": 0.18683667480945587,
+      "learning_rate": 1.6289755568377286e-05,
+      "loss": 1.8842,
+      "step": 5479
+    },
+    {
+      "epoch": 2.364028466681044,
+      "grad_norm": 0.2660259008407593,
+      "learning_rate": 1.626852486510432e-05,
+      "loss": 2.1353,
+      "step": 5480
+    },
+    {
+      "epoch": 2.364459780030192,
+      "grad_norm": 0.20735685527324677,
+      "learning_rate": 1.624730632282556e-05,
+      "loss": 2.182,
+      "step": 5481
+    },
+    {
+      "epoch": 2.36489109337934,
+      "grad_norm": 0.20878545939922333,
+      "learning_rate": 1.622609994593451e-05,
+      "loss": 2.1041,
+      "step": 5482
+    },
+    {
+      "epoch": 2.365322406728488,
+      "grad_norm": 0.19899587333202362,
+      "learning_rate": 1.6204905738822233e-05,
+      "loss": 2.0607,
+      "step": 5483
+    },
+    {
+      "epoch": 2.3657537200776364,
+      "grad_norm": 0.19157008826732635,
+      "learning_rate": 1.618372370587721e-05,
+      "loss": 2.1185,
+      "step": 5484
+    },
+    {
+      "epoch": 2.3661850334267847,
+      "grad_norm": 0.19703416526317596,
+      "learning_rate": 1.616255385148538e-05,
+      "loss": 2.1078,
+      "step": 5485
+    },
+    {
+      "epoch": 2.366616346775933,
+      "grad_norm": 0.1869412213563919,
+      "learning_rate": 1.6141396180030192e-05,
+      "loss": 2.1147,
+      "step": 5486
+    },
+    {
+      "epoch": 2.3670476601250807,
+      "grad_norm": 0.2072449028491974,
+      "learning_rate": 1.612025069589261e-05,
+      "loss": 2.1778,
+      "step": 5487
+    },
+    {
+      "epoch": 2.367478973474229,
+      "grad_norm": 0.2161872535943985,
+      "learning_rate": 1.609911740345102e-05,
+      "loss": 2.189,
+      "step": 5488
+    },
+    {
+      "epoch": 2.3679102868233772,
+      "grad_norm": 0.19158300757408142,
+      "learning_rate": 1.6077996307081302e-05,
+      "loss": 2.1283,
+      "step": 5489
+    },
+    {
+      "epoch": 2.3683416001725255,
+      "grad_norm": 0.19557763636112213,
+      "learning_rate": 1.605688741115681e-05,
+      "loss": 2.0445,
+      "step": 5490
+    },
+    {
+      "epoch": 2.3687729135216733,
+      "grad_norm": 0.19908276200294495,
+      "learning_rate": 1.603579072004837e-05,
+      "loss": 2.1983,
+      "step": 5491
+    },
+    {
+      "epoch": 2.3692042268708216,
+      "grad_norm": 0.21621151268482208,
+      "learning_rate": 1.6014706238124276e-05,
+      "loss": 2.148,
+      "step": 5492
+    },
+    {
+      "epoch": 2.36963554021997,
+      "grad_norm": 0.18872621655464172,
+      "learning_rate": 1.599363396975031e-05,
+      "loss": 2.0592,
+      "step": 5493
+    },
+    {
+      "epoch": 2.370066853569118,
+      "grad_norm": 0.20846761763095856,
+      "learning_rate": 1.597257391928972e-05,
+      "loss": 2.1044,
+      "step": 5494
+    },
+    {
+      "epoch": 2.3704981669182663,
+      "grad_norm": 0.18615426123142242,
+      "learning_rate": 1.595152609110321e-05,
+      "loss": 1.8943,
+      "step": 5495
+    },
+    {
+      "epoch": 2.370929480267414,
+      "grad_norm": 0.18413949012756348,
+      "learning_rate": 1.5930490489548962e-05,
+      "loss": 2.2685,
+      "step": 5496
+    },
+    {
+      "epoch": 2.3713607936165624,
+      "grad_norm": 0.21032410860061646,
+      "learning_rate": 1.5909467118982617e-05,
+      "loss": 2.0688,
+      "step": 5497
+    },
+    {
+      "epoch": 2.3717921069657106,
+      "grad_norm": 0.19261309504508972,
+      "learning_rate": 1.5888455983757345e-05,
+      "loss": 2.1473,
+      "step": 5498
+    },
+    {
+      "epoch": 2.372223420314859,
+      "grad_norm": 0.19529931247234344,
+      "learning_rate": 1.5867457088223675e-05,
+      "loss": 2.1682,
+      "step": 5499
+    },
+    {
+      "epoch": 2.3726547336640067,
+      "grad_norm": 0.20554889738559723,
+      "learning_rate": 1.5846470436729655e-05,
+      "loss": 2.2578,
+      "step": 5500
+    },
+    {
+      "epoch": 2.3726547336640067,
+      "eval_loss": 2.087747573852539,
+      "eval_runtime": 204.2922,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5500
+    },
+    {
+      "epoch": 2.373086047013155,
+      "grad_norm": 0.21087957918643951,
+      "learning_rate": 1.582549603362085e-05,
+      "loss": 2.0516,
+      "step": 5501
+    },
+    {
+      "epoch": 2.373517360362303,
+      "grad_norm": 0.21421092748641968,
+      "learning_rate": 1.580453388324021e-05,
+      "loss": 2.0853,
+      "step": 5502
+    },
+    {
+      "epoch": 2.3739486737114515,
+      "grad_norm": 0.20026853680610657,
+      "learning_rate": 1.5783583989928194e-05,
+      "loss": 2.1586,
+      "step": 5503
+    },
+    {
+      "epoch": 2.3743799870605997,
+      "grad_norm": 0.2020842432975769,
+      "learning_rate": 1.5762646358022653e-05,
+      "loss": 2.2867,
+      "step": 5504
+    },
+    {
+      "epoch": 2.3748113004097475,
+      "grad_norm": 0.20774178206920624,
+      "learning_rate": 1.574172099185902e-05,
+      "loss": 2.273,
+      "step": 5505
+    },
+    {
+      "epoch": 2.375242613758896,
+      "grad_norm": 0.20067383348941803,
+      "learning_rate": 1.572080789577009e-05,
+      "loss": 2.1054,
+      "step": 5506
+    },
+    {
+      "epoch": 2.375673927108044,
+      "grad_norm": 0.17484228312969208,
+      "learning_rate": 1.5699907074086155e-05,
+      "loss": 2.2669,
+      "step": 5507
+    },
+    {
+      "epoch": 2.3761052404571923,
+      "grad_norm": 0.1788690686225891,
+      "learning_rate": 1.567901853113495e-05,
+      "loss": 1.8956,
+      "step": 5508
+    },
+    {
+      "epoch": 2.37653655380634,
+      "grad_norm": 0.23119118809700012,
+      "learning_rate": 1.565814227124168e-05,
+      "loss": 2.1177,
+      "step": 5509
+    },
+    {
+      "epoch": 2.3769678671554884,
+      "grad_norm": 0.1931457668542862,
+      "learning_rate": 1.5637278298729008e-05,
+      "loss": 2.0814,
+      "step": 5510
+    },
+    {
+      "epoch": 2.3773991805046366,
+      "grad_norm": 0.20546475052833557,
+      "learning_rate": 1.5616426617917026e-05,
+      "loss": 2.2013,
+      "step": 5511
+    },
+    {
+      "epoch": 2.377830493853785,
+      "grad_norm": 0.19993987679481506,
+      "learning_rate": 1.5595587233123357e-05,
+      "loss": 2.2371,
+      "step": 5512
+    },
+    {
+      "epoch": 2.378261807202933,
+      "grad_norm": 0.16741925477981567,
+      "learning_rate": 1.5574760148662965e-05,
+      "loss": 2.0317,
+      "step": 5513
+    },
+    {
+      "epoch": 2.378693120552081,
+      "grad_norm": 0.19004949927330017,
+      "learning_rate": 1.5553945368848334e-05,
+      "loss": 2.3056,
+      "step": 5514
+    },
+    {
+      "epoch": 2.379124433901229,
+      "grad_norm": 0.20242708921432495,
+      "learning_rate": 1.5533142897989413e-05,
+      "loss": 2.2824,
+      "step": 5515
+    },
+    {
+      "epoch": 2.3795557472503774,
+      "grad_norm": 0.20056506991386414,
+      "learning_rate": 1.551235274039359e-05,
+      "loss": 2.1467,
+      "step": 5516
+    },
+    {
+      "epoch": 2.3799870605995257,
+      "grad_norm": 0.24377302825450897,
+      "learning_rate": 1.5491574900365687e-05,
+      "loss": 2.2118,
+      "step": 5517
+    },
+    {
+      "epoch": 2.3804183739486735,
+      "grad_norm": 0.18310534954071045,
+      "learning_rate": 1.547080938220794e-05,
+      "loss": 1.7325,
+      "step": 5518
+    },
+    {
+      "epoch": 2.3808496872978218,
+      "grad_norm": 0.17511674761772156,
+      "learning_rate": 1.5450056190220123e-05,
+      "loss": 1.9583,
+      "step": 5519
+    },
+    {
+      "epoch": 2.38128100064697,
+      "grad_norm": 0.19512279331684113,
+      "learning_rate": 1.5429315328699408e-05,
+      "loss": 2.0862,
+      "step": 5520
+    },
+    {
+      "epoch": 2.3817123139961183,
+      "grad_norm": 0.20123524963855743,
+      "learning_rate": 1.54085868019404e-05,
+      "loss": 1.8113,
+      "step": 5521
+    },
+    {
+      "epoch": 2.3821436273452665,
+      "grad_norm": 0.1939847469329834,
+      "learning_rate": 1.5387870614235178e-05,
+      "loss": 2.1032,
+      "step": 5522
+    },
+    {
+      "epoch": 2.3825749406944143,
+      "grad_norm": 0.20574967563152313,
+      "learning_rate": 1.5367166769873255e-05,
+      "loss": 2.0414,
+      "step": 5523
+    },
+    {
+      "epoch": 2.3830062540435626,
+      "grad_norm": 0.20114852488040924,
+      "learning_rate": 1.534647527314159e-05,
+      "loss": 2.2412,
+      "step": 5524
+    },
+    {
+      "epoch": 2.383437567392711,
+      "grad_norm": 0.1988348662853241,
+      "learning_rate": 1.5325796128324574e-05,
+      "loss": 2.0585,
+      "step": 5525
+    },
+    {
+      "epoch": 2.383437567392711,
+      "eval_loss": 2.0880467891693115,
+      "eval_runtime": 203.5213,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5525
+    },
+    {
+      "epoch": 2.383868880741859,
+      "grad_norm": 0.20394539833068848,
+      "learning_rate": 1.5305129339704057e-05,
+      "loss": 2.1355,
+      "step": 5526
+    },
+    {
+      "epoch": 2.384300194091007,
+      "grad_norm": 0.20129942893981934,
+      "learning_rate": 1.528447491155932e-05,
+      "loss": 2.1748,
+      "step": 5527
+    },
+    {
+      "epoch": 2.384731507440155,
+      "grad_norm": 0.2347383201122284,
+      "learning_rate": 1.526383284816707e-05,
+      "loss": 1.9031,
+      "step": 5528
+    },
+    {
+      "epoch": 2.3851628207893034,
+      "grad_norm": 0.1985783576965332,
+      "learning_rate": 1.5243203153801512e-05,
+      "loss": 1.986,
+      "step": 5529
+    },
+    {
+      "epoch": 2.3855941341384517,
+      "grad_norm": 0.20156444609165192,
+      "learning_rate": 1.5222585832734258e-05,
+      "loss": 2.1152,
+      "step": 5530
+    },
+    {
+      "epoch": 2.3860254474876,
+      "grad_norm": 0.2207607924938202,
+      "learning_rate": 1.5201980889234292e-05,
+      "loss": 2.1955,
+      "step": 5531
+    },
+    {
+      "epoch": 2.3864567608367477,
+      "grad_norm": 0.1918589174747467,
+      "learning_rate": 1.5181388327568109e-05,
+      "loss": 2.2195,
+      "step": 5532
+    },
+    {
+      "epoch": 2.386888074185896,
+      "grad_norm": 0.20215705037117004,
+      "learning_rate": 1.5160808151999657e-05,
+      "loss": 2.0159,
+      "step": 5533
+    },
+    {
+      "epoch": 2.3873193875350442,
+      "grad_norm": 0.19671401381492615,
+      "learning_rate": 1.5140240366790268e-05,
+      "loss": 2.1159,
+      "step": 5534
+    },
+    {
+      "epoch": 2.3877507008841925,
+      "grad_norm": 0.18804560601711273,
+      "learning_rate": 1.511968497619873e-05,
+      "loss": 2.1538,
+      "step": 5535
+    },
+    {
+      "epoch": 2.3881820142333403,
+      "grad_norm": 0.19151148200035095,
+      "learning_rate": 1.509914198448125e-05,
+      "loss": 2.1489,
+      "step": 5536
+    },
+    {
+      "epoch": 2.3886133275824886,
+      "grad_norm": 0.19355329871177673,
+      "learning_rate": 1.5078611395891488e-05,
+      "loss": 2.0815,
+      "step": 5537
+    },
+    {
+      "epoch": 2.389044640931637,
+      "grad_norm": 0.19582180678844452,
+      "learning_rate": 1.5058093214680533e-05,
+      "loss": 1.8665,
+      "step": 5538
+    },
+    {
+      "epoch": 2.389475954280785,
+      "grad_norm": 0.20649458467960358,
+      "learning_rate": 1.503758744509688e-05,
+      "loss": 2.1585,
+      "step": 5539
+    },
+    {
+      "epoch": 2.3899072676299333,
+      "grad_norm": 0.21218176186084747,
+      "learning_rate": 1.5017094091386497e-05,
+      "loss": 2.1526,
+      "step": 5540
+    },
+    {
+      "epoch": 2.390338580979081,
+      "grad_norm": 0.19842740893363953,
+      "learning_rate": 1.4996613157792733e-05,
+      "loss": 2.0773,
+      "step": 5541
+    },
+    {
+      "epoch": 2.3907698943282294,
+      "grad_norm": 0.1913827359676361,
+      "learning_rate": 1.4976144648556405e-05,
+      "loss": 2.2597,
+      "step": 5542
+    },
+    {
+      "epoch": 2.3912012076773776,
+      "grad_norm": 0.19794151186943054,
+      "learning_rate": 1.4955688567915718e-05,
+      "loss": 2.1475,
+      "step": 5543
+    },
+    {
+      "epoch": 2.391632521026526,
+      "grad_norm": 0.2209303081035614,
+      "learning_rate": 1.493524492010639e-05,
+      "loss": 2.0839,
+      "step": 5544
+    },
+    {
+      "epoch": 2.3920638343756737,
+      "grad_norm": 0.21582293510437012,
+      "learning_rate": 1.4914813709361432e-05,
+      "loss": 2.1666,
+      "step": 5545
+    },
+    {
+      "epoch": 2.392495147724822,
+      "grad_norm": 0.191584050655365,
+      "learning_rate": 1.4894394939911358e-05,
+      "loss": 2.0405,
+      "step": 5546
+    },
+    {
+      "epoch": 2.3929264610739702,
+      "grad_norm": 0.19593264162540436,
+      "learning_rate": 1.4873988615984139e-05,
+      "loss": 2.1676,
+      "step": 5547
+    },
+    {
+      "epoch": 2.3933577744231185,
+      "grad_norm": 0.19111260771751404,
+      "learning_rate": 1.4853594741805102e-05,
+      "loss": 2.2233,
+      "step": 5548
+    },
+    {
+      "epoch": 2.3937890877722667,
+      "grad_norm": 0.2119603306055069,
+      "learning_rate": 1.4833213321597046e-05,
+      "loss": 2.1493,
+      "step": 5549
+    },
+    {
+      "epoch": 2.3942204011214145,
+      "grad_norm": 0.2001449316740036,
+      "learning_rate": 1.4812844359580101e-05,
+      "loss": 2.0279,
+      "step": 5550
+    },
+    {
+      "epoch": 2.3942204011214145,
+      "eval_loss": 2.087592601776123,
+      "eval_runtime": 203.9108,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5550
+    },
+    {
+      "epoch": 2.394651714470563,
+      "grad_norm": 0.2072679102420807,
+      "learning_rate": 1.4792487859971944e-05,
+      "loss": 2.2792,
+      "step": 5551
+    },
+    {
+      "epoch": 2.395083027819711,
+      "grad_norm": 0.18107175827026367,
+      "learning_rate": 1.4772143826987595e-05,
+      "loss": 1.9677,
+      "step": 5552
+    },
+    {
+      "epoch": 2.3955143411688593,
+      "grad_norm": 0.19043400883674622,
+      "learning_rate": 1.475181226483951e-05,
+      "loss": 2.1645,
+      "step": 5553
+    },
+    {
+      "epoch": 2.395945654518007,
+      "grad_norm": 0.18325147032737732,
+      "learning_rate": 1.4731493177737564e-05,
+      "loss": 2.2081,
+      "step": 5554
+    },
+    {
+      "epoch": 2.3963769678671554,
+      "grad_norm": 0.26620277762413025,
+      "learning_rate": 1.471118656988903e-05,
+      "loss": 2.0327,
+      "step": 5555
+    },
+    {
+      "epoch": 2.3968082812163036,
+      "grad_norm": 0.19313527643680573,
+      "learning_rate": 1.469089244549863e-05,
+      "loss": 2.1078,
+      "step": 5556
+    },
+    {
+      "epoch": 2.397239594565452,
+      "grad_norm": 0.20311668515205383,
+      "learning_rate": 1.467061080876848e-05,
+      "loss": 2.1917,
+      "step": 5557
+    },
+    {
+      "epoch": 2.3976709079146,
+      "grad_norm": 0.21538524329662323,
+      "learning_rate": 1.4650341663898107e-05,
+      "loss": 2.2499,
+      "step": 5558
+    },
+    {
+      "epoch": 2.398102221263748,
+      "grad_norm": 0.1952304244041443,
+      "learning_rate": 1.4630085015084467e-05,
+      "loss": 2.0227,
+      "step": 5559
+    },
+    {
+      "epoch": 2.398533534612896,
+      "grad_norm": 0.21069559454917908,
+      "learning_rate": 1.4609840866521901e-05,
+      "loss": 2.2244,
+      "step": 5560
+    },
+    {
+      "epoch": 2.3989648479620445,
+      "grad_norm": 0.2014789879322052,
+      "learning_rate": 1.4589609222402219e-05,
+      "loss": 2.1381,
+      "step": 5561
+    },
+    {
+      "epoch": 2.3993961613111927,
+      "grad_norm": 0.21519064903259277,
+      "learning_rate": 1.4569390086914605e-05,
+      "loss": 2.3563,
+      "step": 5562
+    },
+    {
+      "epoch": 2.3998274746603405,
+      "grad_norm": 0.1834069937467575,
+      "learning_rate": 1.4549183464245609e-05,
+      "loss": 1.9828,
+      "step": 5563
+    },
+    {
+      "epoch": 2.4002587880094888,
+      "grad_norm": 0.2046196162700653,
+      "learning_rate": 1.4528989358579237e-05,
+      "loss": 2.2115,
+      "step": 5564
+    },
+    {
+      "epoch": 2.400690101358637,
+      "grad_norm": 0.1972675323486328,
+      "learning_rate": 1.4508807774096939e-05,
+      "loss": 2.0984,
+      "step": 5565
+    },
+    {
+      "epoch": 2.4011214147077853,
+      "grad_norm": 0.19684778153896332,
+      "learning_rate": 1.4488638714977513e-05,
+      "loss": 2.2921,
+      "step": 5566
+    },
+    {
+      "epoch": 2.4015527280569335,
+      "grad_norm": 0.18879768252372742,
+      "learning_rate": 1.4468482185397184e-05,
+      "loss": 2.1441,
+      "step": 5567
+    },
+    {
+      "epoch": 2.4019840414060813,
+      "grad_norm": 0.22304823994636536,
+      "learning_rate": 1.4448338189529574e-05,
+      "loss": 2.0291,
+      "step": 5568
+    },
+    {
+      "epoch": 2.4024153547552296,
+      "grad_norm": 0.21063990890979767,
+      "learning_rate": 1.4428206731545727e-05,
+      "loss": 2.1963,
+      "step": 5569
+    },
+    {
+      "epoch": 2.402846668104378,
+      "grad_norm": 0.22029930353164673,
+      "learning_rate": 1.4408087815614077e-05,
+      "loss": 1.9112,
+      "step": 5570
+    },
+    {
+      "epoch": 2.403277981453526,
+      "grad_norm": 0.2134680449962616,
+      "learning_rate": 1.4387981445900468e-05,
+      "loss": 2.0876,
+      "step": 5571
+    },
+    {
+      "epoch": 2.403709294802674,
+      "grad_norm": 0.20396561920642853,
+      "learning_rate": 1.4367887626568141e-05,
+      "loss": 2.2803,
+      "step": 5572
+    },
+    {
+      "epoch": 2.404140608151822,
+      "grad_norm": 0.2179805487394333,
+      "learning_rate": 1.434780636177774e-05,
+      "loss": 2.1741,
+      "step": 5573
+    },
+    {
+      "epoch": 2.4045719215009704,
+      "grad_norm": 0.18951210379600525,
+      "learning_rate": 1.43277376556873e-05,
+      "loss": 2.1052,
+      "step": 5574
+    },
+    {
+      "epoch": 2.4050032348501187,
+      "grad_norm": 0.2312512844800949,
+      "learning_rate": 1.430768151245229e-05,
+      "loss": 2.1362,
+      "step": 5575
+    },
+    {
+      "epoch": 2.4050032348501187,
+      "eval_loss": 2.087423324584961,
+      "eval_runtime": 197.2855,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 5575
+    },
+    {
+      "epoch": 2.405434548199267,
+      "grad_norm": 0.1995340883731842,
+      "learning_rate": 1.4287637936225565e-05,
+      "loss": 2.1059,
+      "step": 5576
+    },
+    {
+      "epoch": 2.4058658615484148,
+      "grad_norm": 0.19406473636627197,
+      "learning_rate": 1.426760693115733e-05,
+      "loss": 2.1551,
+      "step": 5577
+    },
+    {
+      "epoch": 2.406297174897563,
+      "grad_norm": 0.22575275599956512,
+      "learning_rate": 1.4247588501395215e-05,
+      "loss": 2.0501,
+      "step": 5578
+    },
+    {
+      "epoch": 2.4067284882467113,
+      "grad_norm": 0.21095532178878784,
+      "learning_rate": 1.4227582651084299e-05,
+      "loss": 2.275,
+      "step": 5579
+    },
+    {
+      "epoch": 2.4071598015958595,
+      "grad_norm": 0.19748379290103912,
+      "learning_rate": 1.4207589384366991e-05,
+      "loss": 2.2486,
+      "step": 5580
+    },
+    {
+      "epoch": 2.4075911149450073,
+      "grad_norm": 0.237355038523674,
+      "learning_rate": 1.4187608705383114e-05,
+      "loss": 2.0254,
+      "step": 5581
+    },
+    {
+      "epoch": 2.4080224282941556,
+      "grad_norm": 0.19868335127830505,
+      "learning_rate": 1.4167640618269891e-05,
+      "loss": 2.1226,
+      "step": 5582
+    },
+    {
+      "epoch": 2.408453741643304,
+      "grad_norm": 0.1926330029964447,
+      "learning_rate": 1.4147685127161923e-05,
+      "loss": 2.0953,
+      "step": 5583
+    },
+    {
+      "epoch": 2.408885054992452,
+      "grad_norm": 0.20621483027935028,
+      "learning_rate": 1.4127742236191223e-05,
+      "loss": 2.2536,
+      "step": 5584
+    },
+    {
+      "epoch": 2.4093163683416003,
+      "grad_norm": 0.21812713146209717,
+      "learning_rate": 1.4107811949487185e-05,
+      "loss": 2.2392,
+      "step": 5585
+    },
+    {
+      "epoch": 2.409747681690748,
+      "grad_norm": 0.20326219499111176,
+      "learning_rate": 1.4087894271176586e-05,
+      "loss": 2.1634,
+      "step": 5586
+    },
+    {
+      "epoch": 2.4101789950398964,
+      "grad_norm": 0.19478273391723633,
+      "learning_rate": 1.4067989205383592e-05,
+      "loss": 2.1454,
+      "step": 5587
+    },
+    {
+      "epoch": 2.4106103083890447,
+      "grad_norm": 0.19656997919082642,
+      "learning_rate": 1.4048096756229766e-05,
+      "loss": 2.1551,
+      "step": 5588
+    },
+    {
+      "epoch": 2.411041621738193,
+      "grad_norm": 0.1959657073020935,
+      "learning_rate": 1.4028216927834084e-05,
+      "loss": 2.1672,
+      "step": 5589
+    },
+    {
+      "epoch": 2.4114729350873407,
+      "grad_norm": 0.1876540184020996,
+      "learning_rate": 1.4008349724312876e-05,
+      "loss": 1.9919,
+      "step": 5590
+    },
+    {
+      "epoch": 2.411904248436489,
+      "grad_norm": 0.1929643601179123,
+      "learning_rate": 1.3988495149779838e-05,
+      "loss": 2.2164,
+      "step": 5591
+    },
+    {
+      "epoch": 2.4123355617856372,
+      "grad_norm": 0.19551007449626923,
+      "learning_rate": 1.3968653208346067e-05,
+      "loss": 2.1125,
+      "step": 5592
+    },
+    {
+      "epoch": 2.4127668751347855,
+      "grad_norm": 0.2112320214509964,
+      "learning_rate": 1.3948823904120102e-05,
+      "loss": 2.1652,
+      "step": 5593
+    },
+    {
+      "epoch": 2.4131981884839337,
+      "grad_norm": 0.22940433025360107,
+      "learning_rate": 1.3929007241207783e-05,
+      "loss": 2.4004,
+      "step": 5594
+    },
+    {
+      "epoch": 2.4136295018330816,
+      "grad_norm": 0.2000400274991989,
+      "learning_rate": 1.3909203223712401e-05,
+      "loss": 2.2189,
+      "step": 5595
+    },
+    {
+      "epoch": 2.41406081518223,
+      "grad_norm": 0.18464751541614532,
+      "learning_rate": 1.3889411855734528e-05,
+      "loss": 2.1399,
+      "step": 5596
+    },
+    {
+      "epoch": 2.414492128531378,
+      "grad_norm": 0.20219942927360535,
+      "learning_rate": 1.3869633141372244e-05,
+      "loss": 2.0755,
+      "step": 5597
+    },
+    {
+      "epoch": 2.4149234418805263,
+      "grad_norm": 0.1929488629102707,
+      "learning_rate": 1.384986708472093e-05,
+      "loss": 2.342,
+      "step": 5598
+    },
+    {
+      "epoch": 2.415354755229674,
+      "grad_norm": 0.21046137809753418,
+      "learning_rate": 1.3830113689873357e-05,
+      "loss": 2.1805,
+      "step": 5599
+    },
+    {
+      "epoch": 2.4157860685788224,
+      "grad_norm": 0.18207894265651703,
+      "learning_rate": 1.38103729609197e-05,
+      "loss": 2.1979,
+      "step": 5600
+    },
+    {
+      "epoch": 2.4157860685788224,
+      "eval_loss": 2.0876383781433105,
+      "eval_runtime": 194.9901,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5600
+    },
+    {
+      "epoch": 2.4162173819279706,
+      "grad_norm": 0.18475405871868134,
+      "learning_rate": 1.3790644901947464e-05,
+      "loss": 2.2139,
+      "step": 5601
+    },
+    {
+      "epoch": 2.416648695277119,
+      "grad_norm": 0.1960173100233078,
+      "learning_rate": 1.3770929517041584e-05,
+      "loss": 2.1189,
+      "step": 5602
+    },
+    {
+      "epoch": 2.417080008626267,
+      "grad_norm": 0.1998952180147171,
+      "learning_rate": 1.3751226810284336e-05,
+      "loss": 2.0716,
+      "step": 5603
+    },
+    {
+      "epoch": 2.417511321975415,
+      "grad_norm": 0.1847987174987793,
+      "learning_rate": 1.3731536785755388e-05,
+      "loss": 2.0077,
+      "step": 5604
+    },
+    {
+      "epoch": 2.417942635324563,
+      "grad_norm": 0.19390426576137543,
+      "learning_rate": 1.3711859447531761e-05,
+      "loss": 2.2725,
+      "step": 5605
+    },
+    {
+      "epoch": 2.4183739486737115,
+      "grad_norm": 0.18967781960964203,
+      "learning_rate": 1.369219479968785e-05,
+      "loss": 2.0619,
+      "step": 5606
+    },
+    {
+      "epoch": 2.4188052620228597,
+      "grad_norm": 0.19128675758838654,
+      "learning_rate": 1.3672542846295486e-05,
+      "loss": 2.194,
+      "step": 5607
+    },
+    {
+      "epoch": 2.4192365753720075,
+      "grad_norm": 0.18843933939933777,
+      "learning_rate": 1.36529035914238e-05,
+      "loss": 1.9821,
+      "step": 5608
+    },
+    {
+      "epoch": 2.419667888721156,
+      "grad_norm": 0.19080249965190887,
+      "learning_rate": 1.3633277039139296e-05,
+      "loss": 2.0477,
+      "step": 5609
+    },
+    {
+      "epoch": 2.420099202070304,
+      "grad_norm": 0.2069847136735916,
+      "learning_rate": 1.3613663193505848e-05,
+      "loss": 2.2744,
+      "step": 5610
+    },
+    {
+      "epoch": 2.4205305154194523,
+      "grad_norm": 0.19362397491931915,
+      "learning_rate": 1.3594062058584773e-05,
+      "loss": 2.1301,
+      "step": 5611
+    },
+    {
+      "epoch": 2.4209618287686006,
+      "grad_norm": 0.19271761178970337,
+      "learning_rate": 1.3574473638434668e-05,
+      "loss": 2.2115,
+      "step": 5612
+    },
+    {
+      "epoch": 2.4213931421177484,
+      "grad_norm": 0.19851896166801453,
+      "learning_rate": 1.3554897937111531e-05,
+      "loss": 2.1409,
+      "step": 5613
+    },
+    {
+      "epoch": 2.4218244554668966,
+      "grad_norm": 0.19920331239700317,
+      "learning_rate": 1.3535334958668734e-05,
+      "loss": 2.1274,
+      "step": 5614
+    },
+    {
+      "epoch": 2.422255768816045,
+      "grad_norm": 0.19172550737857819,
+      "learning_rate": 1.3515784707157004e-05,
+      "loss": 2.1971,
+      "step": 5615
+    },
+    {
+      "epoch": 2.422687082165193,
+      "grad_norm": 0.1874234974384308,
+      "learning_rate": 1.3496247186624421e-05,
+      "loss": 2.076,
+      "step": 5616
+    },
+    {
+      "epoch": 2.423118395514341,
+      "grad_norm": 0.1872202306985855,
+      "learning_rate": 1.3476722401116455e-05,
+      "loss": 2.0465,
+      "step": 5617
+    },
+    {
+      "epoch": 2.423549708863489,
+      "grad_norm": 0.20311930775642395,
+      "learning_rate": 1.3457210354675924e-05,
+      "loss": 2.2531,
+      "step": 5618
+    },
+    {
+      "epoch": 2.4239810222126374,
+      "grad_norm": 0.18927481770515442,
+      "learning_rate": 1.343771105134301e-05,
+      "loss": 2.1877,
+      "step": 5619
+    },
+    {
+      "epoch": 2.4244123355617857,
+      "grad_norm": 0.20810562372207642,
+      "learning_rate": 1.341822449515524e-05,
+      "loss": 2.1992,
+      "step": 5620
+    },
+    {
+      "epoch": 2.424843648910934,
+      "grad_norm": 0.23007936775684357,
+      "learning_rate": 1.3398750690147544e-05,
+      "loss": 1.6996,
+      "step": 5621
+    },
+    {
+      "epoch": 2.4252749622600818,
+      "grad_norm": 0.20400209724903107,
+      "learning_rate": 1.3379289640352203e-05,
+      "loss": 1.9667,
+      "step": 5622
+    },
+    {
+      "epoch": 2.42570627560923,
+      "grad_norm": 0.194600448012352,
+      "learning_rate": 1.33598413497988e-05,
+      "loss": 2.0099,
+      "step": 5623
+    },
+    {
+      "epoch": 2.4261375889583783,
+      "grad_norm": 0.20893391966819763,
+      "learning_rate": 1.3340405822514305e-05,
+      "loss": 2.2339,
+      "step": 5624
+    },
+    {
+      "epoch": 2.4265689023075265,
+      "grad_norm": 0.1931038498878479,
+      "learning_rate": 1.3320983062523109e-05,
+      "loss": 2.1979,
+      "step": 5625
+    },
+    {
+      "epoch": 2.4265689023075265,
+      "eval_loss": 2.0875062942504883,
+      "eval_runtime": 194.7639,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5625
+    },
+    {
+      "epoch": 2.4270002156566743,
+      "grad_norm": 0.216264545917511,
+      "learning_rate": 1.3301573073846874e-05,
+      "loss": 2.0019,
+      "step": 5626
+    },
+    {
+      "epoch": 2.4274315290058226,
+      "grad_norm": 0.21702730655670166,
+      "learning_rate": 1.3282175860504657e-05,
+      "loss": 1.9417,
+      "step": 5627
+    },
+    {
+      "epoch": 2.427862842354971,
+      "grad_norm": 0.19851179420948029,
+      "learning_rate": 1.3262791426512868e-05,
+      "loss": 2.2475,
+      "step": 5628
+    },
+    {
+      "epoch": 2.428294155704119,
+      "grad_norm": 0.19763562083244324,
+      "learning_rate": 1.3243419775885252e-05,
+      "loss": 2.1318,
+      "step": 5629
+    },
+    {
+      "epoch": 2.4287254690532674,
+      "grad_norm": 0.20032916963100433,
+      "learning_rate": 1.3224060912632938e-05,
+      "loss": 2.3122,
+      "step": 5630
+    },
+    {
+      "epoch": 2.429156782402415,
+      "grad_norm": 0.20838148891925812,
+      "learning_rate": 1.3204714840764377e-05,
+      "loss": 2.0032,
+      "step": 5631
+    },
+    {
+      "epoch": 2.4295880957515634,
+      "grad_norm": 0.1888265162706375,
+      "learning_rate": 1.3185381564285396e-05,
+      "loss": 2.052,
+      "step": 5632
+    },
+    {
+      "epoch": 2.4300194091007117,
+      "grad_norm": 0.18382272124290466,
+      "learning_rate": 1.3166061087199153e-05,
+      "loss": 2.1631,
+      "step": 5633
+    },
+    {
+      "epoch": 2.43045072244986,
+      "grad_norm": 0.19664019346237183,
+      "learning_rate": 1.3146753413506147e-05,
+      "loss": 2.1096,
+      "step": 5634
+    },
+    {
+      "epoch": 2.4308820357990077,
+      "grad_norm": 0.21699444949626923,
+      "learning_rate": 1.3127458547204281e-05,
+      "loss": 2.0513,
+      "step": 5635
+    },
+    {
+      "epoch": 2.431313349148156,
+      "grad_norm": 0.22892074286937714,
+      "learning_rate": 1.3108176492288766e-05,
+      "loss": 1.7689,
+      "step": 5636
+    },
+    {
+      "epoch": 2.4317446624973043,
+      "grad_norm": 0.19827568531036377,
+      "learning_rate": 1.3088907252752126e-05,
+      "loss": 2.2414,
+      "step": 5637
+    },
+    {
+      "epoch": 2.4321759758464525,
+      "grad_norm": 0.19449590146541595,
+      "learning_rate": 1.3069650832584269e-05,
+      "loss": 2.0536,
+      "step": 5638
+    },
+    {
+      "epoch": 2.4326072891956008,
+      "grad_norm": 0.2000982165336609,
+      "learning_rate": 1.3050407235772489e-05,
+      "loss": 2.1,
+      "step": 5639
+    },
+    {
+      "epoch": 2.4330386025447486,
+      "grad_norm": 0.26171600818634033,
+      "learning_rate": 1.3031176466301357e-05,
+      "loss": 2.1844,
+      "step": 5640
+    },
+    {
+      "epoch": 2.433469915893897,
+      "grad_norm": 0.19738814234733582,
+      "learning_rate": 1.3011958528152839e-05,
+      "loss": 2.2074,
+      "step": 5641
+    },
+    {
+      "epoch": 2.433901229243045,
+      "grad_norm": 0.19321668148040771,
+      "learning_rate": 1.2992753425306158e-05,
+      "loss": 2.2208,
+      "step": 5642
+    },
+    {
+      "epoch": 2.4343325425921933,
+      "grad_norm": 0.2076299637556076,
+      "learning_rate": 1.2973561161738e-05,
+      "loss": 2.1897,
+      "step": 5643
+    },
+    {
+      "epoch": 2.434763855941341,
+      "grad_norm": 0.20294933021068573,
+      "learning_rate": 1.2954381741422312e-05,
+      "loss": 2.2548,
+      "step": 5644
+    },
+    {
+      "epoch": 2.4351951692904894,
+      "grad_norm": 0.1980472356081009,
+      "learning_rate": 1.2935215168330409e-05,
+      "loss": 2.2052,
+      "step": 5645
+    },
+    {
+      "epoch": 2.4356264826396377,
+      "grad_norm": 0.20628444850444794,
+      "learning_rate": 1.2916061446430925e-05,
+      "loss": 2.2023,
+      "step": 5646
+    },
+    {
+      "epoch": 2.436057795988786,
+      "grad_norm": 0.1949494630098343,
+      "learning_rate": 1.2896920579689862e-05,
+      "loss": 2.0623,
+      "step": 5647
+    },
+    {
+      "epoch": 2.436489109337934,
+      "grad_norm": 0.2031056135892868,
+      "learning_rate": 1.2877792572070542e-05,
+      "loss": 2.1009,
+      "step": 5648
+    },
+    {
+      "epoch": 2.436920422687082,
+      "grad_norm": 0.20041726529598236,
+      "learning_rate": 1.285867742753362e-05,
+      "loss": 2.0863,
+      "step": 5649
+    },
+    {
+      "epoch": 2.4373517360362302,
+      "grad_norm": 0.19893936812877655,
+      "learning_rate": 1.2839575150037116e-05,
+      "loss": 2.0612,
+      "step": 5650
+    },
+    {
+      "epoch": 2.4373517360362302,
+      "eval_loss": 2.0876355171203613,
+      "eval_runtime": 195.172,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5650
+    },
+    {
+      "epoch": 2.4377830493853785,
+      "grad_norm": 0.20607835054397583,
+      "learning_rate": 1.2820485743536339e-05,
+      "loss": 2.1069,
+      "step": 5651
+    },
+    {
+      "epoch": 2.4382143627345267,
+      "grad_norm": 0.19930028915405273,
+      "learning_rate": 1.280140921198396e-05,
+      "loss": 2.1847,
+      "step": 5652
+    },
+    {
+      "epoch": 2.4386456760836746,
+      "grad_norm": 0.17793206870555878,
+      "learning_rate": 1.2782345559330012e-05,
+      "loss": 2.171,
+      "step": 5653
+    },
+    {
+      "epoch": 2.439076989432823,
+      "grad_norm": 0.17830494046211243,
+      "learning_rate": 1.2763294789521838e-05,
+      "loss": 2.0727,
+      "step": 5654
+    },
+    {
+      "epoch": 2.439508302781971,
+      "grad_norm": 0.20324087142944336,
+      "learning_rate": 1.2744256906504066e-05,
+      "loss": 2.311,
+      "step": 5655
+    },
+    {
+      "epoch": 2.4399396161311193,
+      "grad_norm": 0.2175140231847763,
+      "learning_rate": 1.27252319142187e-05,
+      "loss": 2.1746,
+      "step": 5656
+    },
+    {
+      "epoch": 2.4403709294802676,
+      "grad_norm": 0.2338789403438568,
+      "learning_rate": 1.2706219816605116e-05,
+      "loss": 2.2366,
+      "step": 5657
+    },
+    {
+      "epoch": 2.4408022428294154,
+      "grad_norm": 0.19201385974884033,
+      "learning_rate": 1.2687220617599944e-05,
+      "loss": 1.9177,
+      "step": 5658
+    },
+    {
+      "epoch": 2.4412335561785636,
+      "grad_norm": 0.22386674582958221,
+      "learning_rate": 1.2668234321137192e-05,
+      "loss": 2.2145,
+      "step": 5659
+    },
+    {
+      "epoch": 2.441664869527712,
+      "grad_norm": 0.2086491733789444,
+      "learning_rate": 1.2649260931148169e-05,
+      "loss": 2.1575,
+      "step": 5660
+    },
+    {
+      "epoch": 2.44209618287686,
+      "grad_norm": 0.18864378333091736,
+      "learning_rate": 1.2630300451561521e-05,
+      "loss": 2.0536,
+      "step": 5661
+    },
+    {
+      "epoch": 2.442527496226008,
+      "grad_norm": 0.21808907389640808,
+      "learning_rate": 1.2611352886303236e-05,
+      "loss": 2.2387,
+      "step": 5662
+    },
+    {
+      "epoch": 2.442958809575156,
+      "grad_norm": 0.2047359198331833,
+      "learning_rate": 1.2592418239296604e-05,
+      "loss": 2.1705,
+      "step": 5663
+    },
+    {
+      "epoch": 2.4433901229243045,
+      "grad_norm": 0.18208691477775574,
+      "learning_rate": 1.2573496514462254e-05,
+      "loss": 1.9052,
+      "step": 5664
+    },
+    {
+      "epoch": 2.4438214362734527,
+      "grad_norm": 0.1993609517812729,
+      "learning_rate": 1.2554587715718125e-05,
+      "loss": 2.3345,
+      "step": 5665
+    },
+    {
+      "epoch": 2.444252749622601,
+      "grad_norm": 0.20080235600471497,
+      "learning_rate": 1.2535691846979497e-05,
+      "loss": 2.0826,
+      "step": 5666
+    },
+    {
+      "epoch": 2.444684062971749,
+      "grad_norm": 0.20008035004138947,
+      "learning_rate": 1.2516808912158982e-05,
+      "loss": 2.1286,
+      "step": 5667
+    },
+    {
+      "epoch": 2.445115376320897,
+      "grad_norm": 0.2622080445289612,
+      "learning_rate": 1.2497938915166509e-05,
+      "loss": 2.1557,
+      "step": 5668
+    },
+    {
+      "epoch": 2.4455466896700453,
+      "grad_norm": 0.1986854374408722,
+      "learning_rate": 1.2479081859909268e-05,
+      "loss": 2.2945,
+      "step": 5669
+    },
+    {
+      "epoch": 2.4459780030191935,
+      "grad_norm": 0.32015666365623474,
+      "learning_rate": 1.246023775029183e-05,
+      "loss": 2.0321,
+      "step": 5670
+    },
+    {
+      "epoch": 2.4464093163683414,
+      "grad_norm": 0.20417790114879608,
+      "learning_rate": 1.2441406590216112e-05,
+      "loss": 2.1611,
+      "step": 5671
+    },
+    {
+      "epoch": 2.4468406297174896,
+      "grad_norm": 0.1966090053319931,
+      "learning_rate": 1.2422588383581283e-05,
+      "loss": 1.9855,
+      "step": 5672
+    },
+    {
+      "epoch": 2.447271943066638,
+      "grad_norm": 0.20060087740421295,
+      "learning_rate": 1.2403783134283874e-05,
+      "loss": 2.2288,
+      "step": 5673
+    },
+    {
+      "epoch": 2.447703256415786,
+      "grad_norm": 0.1950266808271408,
+      "learning_rate": 1.2384990846217693e-05,
+      "loss": 2.2091,
+      "step": 5674
+    },
+    {
+      "epoch": 2.4481345697649344,
+      "grad_norm": 0.18725168704986572,
+      "learning_rate": 1.2366211523273914e-05,
+      "loss": 2.0898,
+      "step": 5675
+    },
+    {
+      "epoch": 2.4481345697649344,
+      "eval_loss": 2.08742094039917,
+      "eval_runtime": 194.5775,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5675
+    },
+    {
+      "epoch": 2.448565883114082,
+      "grad_norm": 0.19744376838207245,
+      "learning_rate": 1.2347445169340991e-05,
+      "loss": 2.3429,
+      "step": 5676
+    },
+    {
+      "epoch": 2.4489971964632304,
+      "grad_norm": 0.19756796956062317,
+      "learning_rate": 1.23286917883047e-05,
+      "loss": 2.0573,
+      "step": 5677
+    },
+    {
+      "epoch": 2.4494285098123787,
+      "grad_norm": 0.1958743929862976,
+      "learning_rate": 1.2309951384048137e-05,
+      "loss": 2.1837,
+      "step": 5678
+    },
+    {
+      "epoch": 2.449859823161527,
+      "grad_norm": 0.191923588514328,
+      "learning_rate": 1.2291223960451702e-05,
+      "loss": 2.2266,
+      "step": 5679
+    },
+    {
+      "epoch": 2.4502911365106748,
+      "grad_norm": 0.20301900804042816,
+      "learning_rate": 1.2272509521393104e-05,
+      "loss": 1.9793,
+      "step": 5680
+    },
+    {
+      "epoch": 2.450722449859823,
+      "grad_norm": 0.20013652741909027,
+      "learning_rate": 1.2253808070747417e-05,
+      "loss": 2.1545,
+      "step": 5681
+    },
+    {
+      "epoch": 2.4511537632089713,
+      "grad_norm": 0.20575332641601562,
+      "learning_rate": 1.2235119612386938e-05,
+      "loss": 2.2977,
+      "step": 5682
+    },
+    {
+      "epoch": 2.4515850765581195,
+      "grad_norm": 0.18771253526210785,
+      "learning_rate": 1.2216444150181326e-05,
+      "loss": 1.9519,
+      "step": 5683
+    },
+    {
+      "epoch": 2.452016389907268,
+      "grad_norm": 0.20944516360759735,
+      "learning_rate": 1.2197781687997524e-05,
+      "loss": 2.1642,
+      "step": 5684
+    },
+    {
+      "epoch": 2.452447703256416,
+      "grad_norm": 0.18974487483501434,
+      "learning_rate": 1.217913222969983e-05,
+      "loss": 2.1065,
+      "step": 5685
+    },
+    {
+      "epoch": 2.452879016605564,
+      "grad_norm": 0.19545859098434448,
+      "learning_rate": 1.2160495779149834e-05,
+      "loss": 2.1656,
+      "step": 5686
+    },
+    {
+      "epoch": 2.453310329954712,
+      "grad_norm": 0.19517193734645844,
+      "learning_rate": 1.2141872340206363e-05,
+      "loss": 2.0571,
+      "step": 5687
+    },
+    {
+      "epoch": 2.4537416433038604,
+      "grad_norm": 0.19417819380760193,
+      "learning_rate": 1.2123261916725606e-05,
+      "loss": 2.0085,
+      "step": 5688
+    },
+    {
+      "epoch": 2.454172956653008,
+      "grad_norm": 0.18139328062534332,
+      "learning_rate": 1.2104664512561101e-05,
+      "loss": 1.9799,
+      "step": 5689
+    },
+    {
+      "epoch": 2.4546042700021564,
+      "grad_norm": 0.19426609575748444,
+      "learning_rate": 1.2086080131563625e-05,
+      "loss": 2.116,
+      "step": 5690
+    },
+    {
+      "epoch": 2.4550355833513047,
+      "grad_norm": 0.182246133685112,
+      "learning_rate": 1.2067508777581265e-05,
+      "loss": 2.2792,
+      "step": 5691
+    },
+    {
+      "epoch": 2.455466896700453,
+      "grad_norm": 0.19076038897037506,
+      "learning_rate": 1.2048950454459435e-05,
+      "loss": 2.0988,
+      "step": 5692
+    },
+    {
+      "epoch": 2.455898210049601,
+      "grad_norm": 0.1936006098985672,
+      "learning_rate": 1.2030405166040832e-05,
+      "loss": 1.9843,
+      "step": 5693
+    },
+    {
+      "epoch": 2.4563295233987494,
+      "grad_norm": 0.19579117000102997,
+      "learning_rate": 1.2011872916165465e-05,
+      "loss": 2.1017,
+      "step": 5694
+    },
+    {
+      "epoch": 2.4567608367478972,
+      "grad_norm": 0.20319832861423492,
+      "learning_rate": 1.1993353708670641e-05,
+      "loss": 2.1175,
+      "step": 5695
+    },
+    {
+      "epoch": 2.4571921500970455,
+      "grad_norm": 0.20473158359527588,
+      "learning_rate": 1.1974847547390959e-05,
+      "loss": 2.0309,
+      "step": 5696
+    },
+    {
+      "epoch": 2.4576234634461938,
+      "grad_norm": 0.20871326327323914,
+      "learning_rate": 1.1956354436158319e-05,
+      "loss": 2.1344,
+      "step": 5697
+    },
+    {
+      "epoch": 2.4580547767953416,
+      "grad_norm": 0.18608273565769196,
+      "learning_rate": 1.193787437880191e-05,
+      "loss": 2.1024,
+      "step": 5698
+    },
+    {
+      "epoch": 2.45848609014449,
+      "grad_norm": 0.19599929451942444,
+      "learning_rate": 1.1919407379148268e-05,
+      "loss": 2.111,
+      "step": 5699
+    },
+    {
+      "epoch": 2.458917403493638,
+      "grad_norm": 0.1737334430217743,
+      "learning_rate": 1.190095344102118e-05,
+      "loss": 2.0196,
+      "step": 5700
+    },
+    {
+      "epoch": 2.458917403493638,
+      "eval_loss": 2.087368965148926,
+      "eval_runtime": 194.2931,
+      "eval_samples_per_second": 0.165,
+      "eval_steps_per_second": 0.165,
+      "step": 5700
+    },
+    {
+      "epoch": 2.4593487168427863,
+      "grad_norm": 0.16876773536205292,
+      "learning_rate": 1.1882512568241703e-05,
+      "loss": 1.9346,
+      "step": 5701
+    },
+    {
+      "epoch": 2.4597800301919346,
+      "grad_norm": 0.19252726435661316,
+      "learning_rate": 1.1864084764628226e-05,
+      "loss": 2.1173,
+      "step": 5702
+    },
+    {
+      "epoch": 2.460211343541083,
+      "grad_norm": 0.20373648405075073,
+      "learning_rate": 1.1845670033996454e-05,
+      "loss": 2.195,
+      "step": 5703
+    },
+    {
+      "epoch": 2.4606426568902307,
+      "grad_norm": 0.2033834308385849,
+      "learning_rate": 1.1827268380159354e-05,
+      "loss": 2.0791,
+      "step": 5704
+    },
+    {
+      "epoch": 2.461073970239379,
+      "grad_norm": 0.1925143599510193,
+      "learning_rate": 1.1808879806927171e-05,
+      "loss": 1.907,
+      "step": 5705
+    },
+    {
+      "epoch": 2.461505283588527,
+      "grad_norm": 0.18729040026664734,
+      "learning_rate": 1.1790504318107475e-05,
+      "loss": 2.0964,
+      "step": 5706
+    },
+    {
+      "epoch": 2.4619365969376754,
+      "grad_norm": 0.20501752197742462,
+      "learning_rate": 1.1772141917505112e-05,
+      "loss": 2.1624,
+      "step": 5707
+    },
+    {
+      "epoch": 2.4623679102868232,
+      "grad_norm": 0.27427372336387634,
+      "learning_rate": 1.1753792608922202e-05,
+      "loss": 2.1858,
+      "step": 5708
+    },
+    {
+      "epoch": 2.4627992236359715,
+      "grad_norm": 0.1857091635465622,
+      "learning_rate": 1.1735456396158188e-05,
+      "loss": 2.1267,
+      "step": 5709
+    },
+    {
+      "epoch": 2.4632305369851197,
+      "grad_norm": 0.1786196380853653,
+      "learning_rate": 1.1717133283009783e-05,
+      "loss": 1.9905,
+      "step": 5710
+    },
+    {
+      "epoch": 2.463661850334268,
+      "grad_norm": 2.388453722000122,
+      "learning_rate": 1.1698823273270969e-05,
+      "loss": 2.2965,
+      "step": 5711
+    },
+    {
+      "epoch": 2.4640931636834162,
+      "grad_norm": 0.26211005449295044,
+      "learning_rate": 1.1680526370733037e-05,
+      "loss": 2.1418,
+      "step": 5712
+    },
+    {
+      "epoch": 2.464524477032564,
+      "grad_norm": 0.21328003704547882,
+      "learning_rate": 1.1662242579184583e-05,
+      "loss": 2.1827,
+      "step": 5713
+    },
+    {
+      "epoch": 2.4649557903817123,
+      "grad_norm": 0.19279971718788147,
+      "learning_rate": 1.1643971902411478e-05,
+      "loss": 2.079,
+      "step": 5714
+    },
+    {
+      "epoch": 2.4653871037308606,
+      "grad_norm": 0.19928252696990967,
+      "learning_rate": 1.1625714344196813e-05,
+      "loss": 1.957,
+      "step": 5715
+    },
+    {
+      "epoch": 2.465818417080009,
+      "grad_norm": 0.1971615105867386,
+      "learning_rate": 1.160746990832103e-05,
+      "loss": 2.0119,
+      "step": 5716
+    },
+    {
+      "epoch": 2.4662497304291566,
+      "grad_norm": 0.21007204055786133,
+      "learning_rate": 1.1589238598561873e-05,
+      "loss": 1.7142,
+      "step": 5717
+    },
+    {
+      "epoch": 2.466681043778305,
+      "grad_norm": 0.1840398609638214,
+      "learning_rate": 1.1571020418694308e-05,
+      "loss": 1.8995,
+      "step": 5718
+    },
+    {
+      "epoch": 2.467112357127453,
+      "grad_norm": 0.19897425174713135,
+      "learning_rate": 1.1552815372490606e-05,
+      "loss": 2.1401,
+      "step": 5719
+    },
+    {
+      "epoch": 2.4675436704766014,
+      "grad_norm": 0.18006531894207,
+      "learning_rate": 1.1534623463720338e-05,
+      "loss": 1.9013,
+      "step": 5720
+    },
+    {
+      "epoch": 2.4679749838257496,
+      "grad_norm": 0.21226273477077484,
+      "learning_rate": 1.1516444696150324e-05,
+      "loss": 1.9415,
+      "step": 5721
+    },
+    {
+      "epoch": 2.4684062971748975,
+      "grad_norm": 0.21781636774539948,
+      "learning_rate": 1.1498279073544683e-05,
+      "loss": 2.1623,
+      "step": 5722
+    },
+    {
+      "epoch": 2.4688376105240457,
+      "grad_norm": 0.19267801940441132,
+      "learning_rate": 1.1480126599664801e-05,
+      "loss": 2.1478,
+      "step": 5723
+    },
+    {
+      "epoch": 2.469268923873194,
+      "grad_norm": 0.1948312222957611,
+      "learning_rate": 1.1461987278269347e-05,
+      "loss": 2.1864,
+      "step": 5724
+    },
+    {
+      "epoch": 2.469700237222342,
+      "grad_norm": 0.20160384476184845,
+      "learning_rate": 1.1443861113114264e-05,
+      "loss": 2.0665,
+      "step": 5725
+    },
+    {
+      "epoch": 2.469700237222342,
+      "eval_loss": 2.0871713161468506,
+      "eval_runtime": 194.8679,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5725
+    },
+    {
+      "epoch": 2.47013155057149,
+      "grad_norm": 0.20347172021865845,
+      "learning_rate": 1.142574810795276e-05,
+      "loss": 2.4612,
+      "step": 5726
+    },
+    {
+      "epoch": 2.4705628639206383,
+      "grad_norm": 0.2064550369977951,
+      "learning_rate": 1.1407648266535382e-05,
+      "loss": 2.165,
+      "step": 5727
+    },
+    {
+      "epoch": 2.4709941772697865,
+      "grad_norm": 0.20214228332042694,
+      "learning_rate": 1.1389561592609852e-05,
+      "loss": 1.9555,
+      "step": 5728
+    },
+    {
+      "epoch": 2.471425490618935,
+      "grad_norm": 0.2005394995212555,
+      "learning_rate": 1.1371488089921224e-05,
+      "loss": 2.1927,
+      "step": 5729
+    },
+    {
+      "epoch": 2.471856803968083,
+      "grad_norm": 0.20218129456043243,
+      "learning_rate": 1.1353427762211794e-05,
+      "loss": 2.0424,
+      "step": 5730
+    },
+    {
+      "epoch": 2.472288117317231,
+      "grad_norm": 0.19005270302295685,
+      "learning_rate": 1.1335380613221198e-05,
+      "loss": 2.1325,
+      "step": 5731
+    },
+    {
+      "epoch": 2.472719430666379,
+      "grad_norm": 0.19264422357082367,
+      "learning_rate": 1.1317346646686279e-05,
+      "loss": 2.1046,
+      "step": 5732
+    },
+    {
+      "epoch": 2.4731507440155274,
+      "grad_norm": 0.2074297070503235,
+      "learning_rate": 1.129932586634115e-05,
+      "loss": 2.0958,
+      "step": 5733
+    },
+    {
+      "epoch": 2.4735820573646756,
+      "grad_norm": 0.18325988948345184,
+      "learning_rate": 1.1281318275917192e-05,
+      "loss": 1.9128,
+      "step": 5734
+    },
+    {
+      "epoch": 2.4740133707138234,
+      "grad_norm": 0.21392813324928284,
+      "learning_rate": 1.1263323879143123e-05,
+      "loss": 1.9925,
+      "step": 5735
+    },
+    {
+      "epoch": 2.4744446840629717,
+      "grad_norm": 0.18119731545448303,
+      "learning_rate": 1.1245342679744866e-05,
+      "loss": 2.2641,
+      "step": 5736
+    },
+    {
+      "epoch": 2.47487599741212,
+      "grad_norm": 0.2063588947057724,
+      "learning_rate": 1.1227374681445603e-05,
+      "loss": 2.1409,
+      "step": 5737
+    },
+    {
+      "epoch": 2.475307310761268,
+      "grad_norm": 0.20988424122333527,
+      "learning_rate": 1.1209419887965828e-05,
+      "loss": 2.1957,
+      "step": 5738
+    },
+    {
+      "epoch": 2.4757386241104165,
+      "grad_norm": 0.1982715129852295,
+      "learning_rate": 1.119147830302326e-05,
+      "loss": 2.2277,
+      "step": 5739
+    },
+    {
+      "epoch": 2.4761699374595643,
+      "grad_norm": 0.19500423967838287,
+      "learning_rate": 1.117354993033291e-05,
+      "loss": 2.0229,
+      "step": 5740
+    },
+    {
+      "epoch": 2.4766012508087125,
+      "grad_norm": 0.21447238326072693,
+      "learning_rate": 1.1155634773607039e-05,
+      "loss": 2.1768,
+      "step": 5741
+    },
+    {
+      "epoch": 2.4770325641578608,
+      "grad_norm": 0.19308245182037354,
+      "learning_rate": 1.1137732836555183e-05,
+      "loss": 2.0162,
+      "step": 5742
+    },
+    {
+      "epoch": 2.477463877507009,
+      "grad_norm": 0.21386805176734924,
+      "learning_rate": 1.1119844122884125e-05,
+      "loss": 1.9618,
+      "step": 5743
+    },
+    {
+      "epoch": 2.477895190856157,
+      "grad_norm": 0.17448431253433228,
+      "learning_rate": 1.1101968636297912e-05,
+      "loss": 1.9715,
+      "step": 5744
+    },
+    {
+      "epoch": 2.478326504205305,
+      "grad_norm": 0.21260018646717072,
+      "learning_rate": 1.1084106380497882e-05,
+      "loss": 2.2935,
+      "step": 5745
+    },
+    {
+      "epoch": 2.4787578175544533,
+      "grad_norm": 0.2187669724225998,
+      "learning_rate": 1.1066257359182609e-05,
+      "loss": 1.9428,
+      "step": 5746
+    },
+    {
+      "epoch": 2.4791891309036016,
+      "grad_norm": 0.19976875185966492,
+      "learning_rate": 1.1048421576047914e-05,
+      "loss": 1.994,
+      "step": 5747
+    },
+    {
+      "epoch": 2.47962044425275,
+      "grad_norm": 0.21492761373519897,
+      "learning_rate": 1.1030599034786867e-05,
+      "loss": 2.0978,
+      "step": 5748
+    },
+    {
+      "epoch": 2.4800517576018977,
+      "grad_norm": 0.23321105539798737,
+      "learning_rate": 1.1012789739089864e-05,
+      "loss": 2.105,
+      "step": 5749
+    },
+    {
+      "epoch": 2.480483070951046,
+      "grad_norm": 0.2062074840068817,
+      "learning_rate": 1.0994993692644508e-05,
+      "loss": 2.1688,
+      "step": 5750
+    },
+    {
+      "epoch": 2.480483070951046,
+      "eval_loss": 2.0872256755828857,
+      "eval_runtime": 194.9368,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 5750
+    },
+    {
+      "epoch": 2.480914384300194,
+      "grad_norm": 0.2005544900894165,
+      "learning_rate": 1.097721089913565e-05,
+      "loss": 2.1628,
+      "step": 5751
+    },
+    {
+      "epoch": 2.4813456976493424,
+      "grad_norm": 0.21592171490192413,
+      "learning_rate": 1.0959441362245411e-05,
+      "loss": 2.1939,
+      "step": 5752
+    },
+    {
+      "epoch": 2.4817770109984902,
+      "grad_norm": 0.19873322546482086,
+      "learning_rate": 1.0941685085653182e-05,
+      "loss": 2.1372,
+      "step": 5753
+    },
+    {
+      "epoch": 2.4822083243476385,
+      "grad_norm": 0.20747192203998566,
+      "learning_rate": 1.0923942073035579e-05,
+      "loss": 2.5073,
+      "step": 5754
+    },
+    {
+      "epoch": 2.4826396376967867,
+      "grad_norm": 0.19865496456623077,
+      "learning_rate": 1.09062123280665e-05,
+      "loss": 2.0926,
+      "step": 5755
+    },
+    {
+      "epoch": 2.483070951045935,
+      "grad_norm": 0.1956738829612732,
+      "learning_rate": 1.0888495854417071e-05,
+      "loss": 2.0335,
+      "step": 5756
+    },
+    {
+      "epoch": 2.4835022643950833,
+      "grad_norm": 0.2046099156141281,
+      "learning_rate": 1.0870792655755683e-05,
+      "loss": 2.4057,
+      "step": 5757
+    },
+    {
+      "epoch": 2.483933577744231,
+      "grad_norm": 0.1916872262954712,
+      "learning_rate": 1.0853102735747973e-05,
+      "loss": 2.1253,
+      "step": 5758
+    },
+    {
+      "epoch": 2.4843648910933793,
+      "grad_norm": 0.2286677360534668,
+      "learning_rate": 1.0835426098056851e-05,
+      "loss": 2.2194,
+      "step": 5759
+    },
+    {
+      "epoch": 2.4847962044425276,
+      "grad_norm": 0.23386746644973755,
+      "learning_rate": 1.0817762746342473e-05,
+      "loss": 2.3312,
+      "step": 5760
+    },
+    {
+      "epoch": 2.485227517791676,
+      "grad_norm": 0.1910087913274765,
+      "learning_rate": 1.080011268426218e-05,
+      "loss": 2.1249,
+      "step": 5761
+    },
+    {
+      "epoch": 2.4856588311408236,
+      "grad_norm": 0.18502314388751984,
+      "learning_rate": 1.0782475915470615e-05,
+      "loss": 2.1537,
+      "step": 5762
+    },
+    {
+      "epoch": 2.486090144489972,
+      "grad_norm": 0.19793559610843658,
+      "learning_rate": 1.0764852443619697e-05,
+      "loss": 2.0208,
+      "step": 5763
+    },
+    {
+      "epoch": 2.48652145783912,
+      "grad_norm": 0.21486476063728333,
+      "learning_rate": 1.0747242272358553e-05,
+      "loss": 2.2666,
+      "step": 5764
+    },
+    {
+      "epoch": 2.4869527711882684,
+      "grad_norm": 0.211890310049057,
+      "learning_rate": 1.0729645405333545e-05,
+      "loss": 1.9514,
+      "step": 5765
+    },
+    {
+      "epoch": 2.4873840845374167,
+      "grad_norm": 0.19332849979400635,
+      "learning_rate": 1.071206184618831e-05,
+      "loss": 2.2484,
+      "step": 5766
+    },
+    {
+      "epoch": 2.4878153978865645,
+      "grad_norm": 0.20185452699661255,
+      "learning_rate": 1.0694491598563712e-05,
+      "loss": 2.0449,
+      "step": 5767
+    },
+    {
+      "epoch": 2.4882467112357127,
+      "grad_norm": 0.1992376446723938,
+      "learning_rate": 1.0676934666097856e-05,
+      "loss": 2.0498,
+      "step": 5768
+    },
+    {
+      "epoch": 2.488678024584861,
+      "grad_norm": 0.21259760856628418,
+      "learning_rate": 1.0659391052426109e-05,
+      "loss": 2.1275,
+      "step": 5769
+    },
+    {
+      "epoch": 2.4891093379340092,
+      "grad_norm": 0.20208117365837097,
+      "learning_rate": 1.0641860761181057e-05,
+      "loss": 2.0495,
+      "step": 5770
+    },
+    {
+      "epoch": 2.489540651283157,
+      "grad_norm": 0.20526215434074402,
+      "learning_rate": 1.0624343795992546e-05,
+      "loss": 2.1591,
+      "step": 5771
+    },
+    {
+      "epoch": 2.4899719646323053,
+      "grad_norm": 0.20042036473751068,
+      "learning_rate": 1.0606840160487633e-05,
+      "loss": 2.0136,
+      "step": 5772
+    },
+    {
+      "epoch": 2.4904032779814536,
+      "grad_norm": 0.20826835930347443,
+      "learning_rate": 1.0589349858290689e-05,
+      "loss": 2.3665,
+      "step": 5773
+    },
+    {
+      "epoch": 2.490834591330602,
+      "grad_norm": 0.20930518209934235,
+      "learning_rate": 1.0571872893023228e-05,
+      "loss": 2.1048,
+      "step": 5774
+    },
+    {
+      "epoch": 2.49126590467975,
+      "grad_norm": 0.21557918190956116,
+      "learning_rate": 1.055440926830405e-05,
+      "loss": 2.1528,
+      "step": 5775
+    },
+    {
+      "epoch": 2.49126590467975,
+      "eval_loss": 2.087198495864868,
+      "eval_runtime": 208.443,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 5775
+    },
+    {
+      "epoch": 2.491697218028898,
+      "grad_norm": 0.19226939976215363,
+      "learning_rate": 1.0536958987749188e-05,
+      "loss": 2.0221,
+      "step": 5776
+    },
+    {
+      "epoch": 2.492128531378046,
+      "grad_norm": 0.1970117837190628,
+      "learning_rate": 1.0519522054971944e-05,
+      "loss": 2.0414,
+      "step": 5777
+    },
+    {
+      "epoch": 2.4925598447271944,
+      "grad_norm": 0.18709838390350342,
+      "learning_rate": 1.0502098473582809e-05,
+      "loss": 2.1164,
+      "step": 5778
+    },
+    {
+      "epoch": 2.4929911580763426,
+      "grad_norm": 0.18943005800247192,
+      "learning_rate": 1.0484688247189502e-05,
+      "loss": 2.104,
+      "step": 5779
+    },
+    {
+      "epoch": 2.4934224714254904,
+      "grad_norm": 0.18927139043807983,
+      "learning_rate": 1.0467291379397027e-05,
+      "loss": 2.1364,
+      "step": 5780
+    },
+    {
+      "epoch": 2.4938537847746387,
+      "grad_norm": 0.19848081469535828,
+      "learning_rate": 1.0449907873807598e-05,
+      "loss": 2.0313,
+      "step": 5781
+    },
+    {
+      "epoch": 2.494285098123787,
+      "grad_norm": 0.17562636733055115,
+      "learning_rate": 1.043253773402065e-05,
+      "loss": 1.9745,
+      "step": 5782
+    },
+    {
+      "epoch": 2.494716411472935,
+      "grad_norm": 0.21291068196296692,
+      "learning_rate": 1.0415180963632855e-05,
+      "loss": 2.1383,
+      "step": 5783
+    },
+    {
+      "epoch": 2.4951477248220835,
+      "grad_norm": 0.2445984035730362,
+      "learning_rate": 1.039783756623814e-05,
+      "loss": 2.106,
+      "step": 5784
+    },
+    {
+      "epoch": 2.4955790381712313,
+      "grad_norm": 0.18510930240154266,
+      "learning_rate": 1.0380507545427624e-05,
+      "loss": 2.2387,
+      "step": 5785
+    },
+    {
+      "epoch": 2.4960103515203795,
+      "grad_norm": 0.19964301586151123,
+      "learning_rate": 1.0363190904789695e-05,
+      "loss": 2.045,
+      "step": 5786
+    },
+    {
+      "epoch": 2.496441664869528,
+      "grad_norm": 0.19906377792358398,
+      "learning_rate": 1.0345887647909943e-05,
+      "loss": 2.2741,
+      "step": 5787
+    },
+    {
+      "epoch": 2.496872978218676,
+      "grad_norm": 0.20326967537403107,
+      "learning_rate": 1.0328597778371201e-05,
+      "loss": 2.1149,
+      "step": 5788
+    },
+    {
+      "epoch": 2.497304291567824,
+      "grad_norm": 0.20548272132873535,
+      "learning_rate": 1.0311321299753513e-05,
+      "loss": 1.996,
+      "step": 5789
+    },
+    {
+      "epoch": 2.497735604916972,
+      "grad_norm": 0.23548679053783417,
+      "learning_rate": 1.0294058215634172e-05,
+      "loss": 1.7426,
+      "step": 5790
+    },
+    {
+      "epoch": 2.4981669182661204,
+      "grad_norm": 0.19880113005638123,
+      "learning_rate": 1.0276808529587696e-05,
+      "loss": 2.1825,
+      "step": 5791
+    },
+    {
+      "epoch": 2.4985982316152686,
+      "grad_norm": 0.18782949447631836,
+      "learning_rate": 1.0259572245185832e-05,
+      "loss": 2.3359,
+      "step": 5792
+    },
+    {
+      "epoch": 2.499029544964417,
+      "grad_norm": 0.20282316207885742,
+      "learning_rate": 1.0242349365997515e-05,
+      "loss": 2.2454,
+      "step": 5793
+    },
+    {
+      "epoch": 2.4994608583135647,
+      "grad_norm": 0.19014997780323029,
+      "learning_rate": 1.022513989558891e-05,
+      "loss": 1.987,
+      "step": 5794
+    },
+    {
+      "epoch": 2.499892171662713,
+      "grad_norm": 0.21561653912067413,
+      "learning_rate": 1.0207943837523483e-05,
+      "loss": 2.307,
+      "step": 5795
+    },
+    {
+      "epoch": 2.500323485011861,
+      "grad_norm": 0.21344530582427979,
+      "learning_rate": 1.0190761195361835e-05,
+      "loss": 2.0094,
+      "step": 5796
+    },
+    {
+      "epoch": 2.500754798361009,
+      "grad_norm": 0.2258821278810501,
+      "learning_rate": 1.017359197266182e-05,
+      "loss": 2.1,
+      "step": 5797
+    },
+    {
+      "epoch": 2.5011861117101573,
+      "grad_norm": 0.18700018525123596,
+      "learning_rate": 1.0156436172978522e-05,
+      "loss": 2.1812,
+      "step": 5798
+    },
+    {
+      "epoch": 2.5016174250593055,
+      "grad_norm": 0.18775956332683563,
+      "learning_rate": 1.0139293799864229e-05,
+      "loss": 1.9226,
+      "step": 5799
+    },
+    {
+      "epoch": 2.5020487384084538,
+      "grad_norm": 0.2064802497625351,
+      "learning_rate": 1.0122164856868461e-05,
+      "loss": 2.0509,
+      "step": 5800
+    },
+    {
+      "epoch": 2.5020487384084538,
+      "eval_loss": 2.087212562561035,
+      "eval_runtime": 202.1354,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 5800
+    },
+    {
+      "epoch": 2.502480051757602,
+      "grad_norm": 0.1930222511291504,
+      "learning_rate": 1.010504934753795e-05,
+      "loss": 2.1301,
+      "step": 5801
+    },
+    {
+      "epoch": 2.5029113651067503,
+      "grad_norm": 0.19863606989383698,
+      "learning_rate": 1.008794727541665e-05,
+      "loss": 2.1475,
+      "step": 5802
+    },
+    {
+      "epoch": 2.503342678455898,
+      "grad_norm": 0.20644447207450867,
+      "learning_rate": 1.0070858644045737e-05,
+      "loss": 2.1981,
+      "step": 5803
+    },
+    {
+      "epoch": 2.5037739918050463,
+      "grad_norm": 0.20835015177726746,
+      "learning_rate": 1.0053783456963574e-05,
+      "loss": 2.2185,
+      "step": 5804
+    },
+    {
+      "epoch": 2.5042053051541946,
+      "grad_norm": 0.19597108662128448,
+      "learning_rate": 1.0036721717705823e-05,
+      "loss": 2.0874,
+      "step": 5805
+    },
+    {
+      "epoch": 2.5046366185033424,
+      "grad_norm": 0.1970120519399643,
+      "learning_rate": 1.001967342980525e-05,
+      "loss": 2.2048,
+      "step": 5806
+    },
+    {
+      "epoch": 2.5050679318524907,
+      "grad_norm": 0.22588270902633667,
+      "learning_rate": 1.0002638596791898e-05,
+      "loss": 2.1817,
+      "step": 5807
+    },
+    {
+      "epoch": 2.505499245201639,
+      "grad_norm": 0.21643772721290588,
+      "learning_rate": 9.985617222193013e-06,
+      "loss": 2.2974,
+      "step": 5808
+    },
+    {
+      "epoch": 2.505930558550787,
+      "grad_norm": 0.20602551102638245,
+      "learning_rate": 9.96860930953309e-06,
+      "loss": 2.1928,
+      "step": 5809
+    },
+    {
+      "epoch": 2.5063618718999354,
+      "grad_norm": 0.20020756125450134,
+      "learning_rate": 9.951614862333789e-06,
+      "loss": 2.0934,
+      "step": 5810
+    },
+    {
+      "epoch": 2.5067931852490837,
+      "grad_norm": 0.1873340904712677,
+      "learning_rate": 9.934633884113954e-06,
+      "loss": 2.037,
+      "step": 5811
+    },
+    {
+      "epoch": 2.5072244985982315,
+      "grad_norm": 0.19135379791259766,
+      "learning_rate": 9.917666378389738e-06,
+      "loss": 2.1021,
+      "step": 5812
+    },
+    {
+      "epoch": 2.5076558119473797,
+      "grad_norm": 0.18421195447444916,
+      "learning_rate": 9.900712348674422e-06,
+      "loss": 1.9513,
+      "step": 5813
+    },
+    {
+      "epoch": 2.508087125296528,
+      "grad_norm": 0.20990701019763947,
+      "learning_rate": 9.883771798478533e-06,
+      "loss": 2.3694,
+      "step": 5814
+    },
+    {
+      "epoch": 2.508518438645676,
+      "grad_norm": 0.18025539815425873,
+      "learning_rate": 9.866844731309787e-06,
+      "loss": 2.0232,
+      "step": 5815
+    },
+    {
+      "epoch": 2.508949751994824,
+      "grad_norm": 0.20541740953922272,
+      "learning_rate": 9.849931150673133e-06,
+      "loss": 2.1529,
+      "step": 5816
+    },
+    {
+      "epoch": 2.5093810653439723,
+      "grad_norm": 0.19457145035266876,
+      "learning_rate": 9.833031060070693e-06,
+      "loss": 2.011,
+      "step": 5817
+    },
+    {
+      "epoch": 2.5098123786931206,
+      "grad_norm": 0.20473068952560425,
+      "learning_rate": 9.81614446300181e-06,
+      "loss": 2.179,
+      "step": 5818
+    },
+    {
+      "epoch": 2.510243692042269,
+      "grad_norm": 0.2168092429637909,
+      "learning_rate": 9.799271362963096e-06,
+      "loss": 2.2023,
+      "step": 5819
+    },
+    {
+      "epoch": 2.510675005391417,
+      "grad_norm": 0.19788049161434174,
+      "learning_rate": 9.782411763448247e-06,
+      "loss": 2.1279,
+      "step": 5820
+    },
+    {
+      "epoch": 2.511106318740565,
+      "grad_norm": 0.1992284208536148,
+      "learning_rate": 9.765565667948247e-06,
+      "loss": 2.1279,
+      "step": 5821
+    },
+    {
+      "epoch": 2.511537632089713,
+      "grad_norm": 0.18312112987041473,
+      "learning_rate": 9.748733079951254e-06,
+      "loss": 1.8849,
+      "step": 5822
+    },
+    {
+      "epoch": 2.5119689454388614,
+      "grad_norm": 0.17599177360534668,
+      "learning_rate": 9.731914002942664e-06,
+      "loss": 2.0223,
+      "step": 5823
+    },
+    {
+      "epoch": 2.5124002587880097,
+      "grad_norm": 0.20038534700870514,
+      "learning_rate": 9.715108440405067e-06,
+      "loss": 2.1895,
+      "step": 5824
+    },
+    {
+      "epoch": 2.5128315721371575,
+      "grad_norm": 0.19765359163284302,
+      "learning_rate": 9.69831639581817e-06,
+      "loss": 2.1974,
+      "step": 5825
+    },
+    {
+      "epoch": 2.5128315721371575,
+      "eval_loss": 2.087245225906372,
+      "eval_runtime": 202.6707,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 5825
+    },
+    {
+      "epoch": 2.5132628854863057,
+      "grad_norm": 4.632514476776123,
+      "learning_rate": 9.681537872659009e-06,
+      "loss": 2.2227,
+      "step": 5826
+    },
+    {
+      "epoch": 2.513694198835454,
+      "grad_norm": 0.20792263746261597,
+      "learning_rate": 9.664772874401738e-06,
+      "loss": 2.1268,
+      "step": 5827
+    },
+    {
+      "epoch": 2.5141255121846022,
+      "grad_norm": 0.18867172300815582,
+      "learning_rate": 9.648021404517747e-06,
+      "loss": 1.9849,
+      "step": 5828
+    },
+    {
+      "epoch": 2.5145568255337505,
+      "grad_norm": 0.19542069733142853,
+      "learning_rate": 9.631283466475595e-06,
+      "loss": 2.1466,
+      "step": 5829
+    },
+    {
+      "epoch": 2.5149881388828983,
+      "grad_norm": 0.20394869148731232,
+      "learning_rate": 9.61455906374106e-06,
+      "loss": 2.0995,
+      "step": 5830
+    },
+    {
+      "epoch": 2.5154194522320465,
+      "grad_norm": 0.1910642683506012,
+      "learning_rate": 9.597848199777111e-06,
+      "loss": 2.2057,
+      "step": 5831
+    },
+    {
+      "epoch": 2.515850765581195,
+      "grad_norm": 0.19179588556289673,
+      "learning_rate": 9.581150878043925e-06,
+      "loss": 2.0844,
+      "step": 5832
+    },
+    {
+      "epoch": 2.516282078930343,
+      "grad_norm": 0.20327067375183105,
+      "learning_rate": 9.56446710199886e-06,
+      "loss": 2.2818,
+      "step": 5833
+    },
+    {
+      "epoch": 2.516713392279491,
+      "grad_norm": 0.18520274758338928,
+      "learning_rate": 9.54779687509647e-06,
+      "loss": 1.811,
+      "step": 5834
+    },
+    {
+      "epoch": 2.517144705628639,
+      "grad_norm": 0.18724271655082703,
+      "learning_rate": 9.531140200788508e-06,
+      "loss": 1.8034,
+      "step": 5835
+    },
+    {
+      "epoch": 2.5175760189777874,
+      "grad_norm": 0.19156400859355927,
+      "learning_rate": 9.514497082523903e-06,
+      "loss": 2.2527,
+      "step": 5836
+    },
+    {
+      "epoch": 2.5180073323269356,
+      "grad_norm": 0.20256634056568146,
+      "learning_rate": 9.497867523748837e-06,
+      "loss": 2.0685,
+      "step": 5837
+    },
+    {
+      "epoch": 2.518438645676084,
+      "grad_norm": 0.1926616132259369,
+      "learning_rate": 9.48125152790663e-06,
+      "loss": 2.0477,
+      "step": 5838
+    },
+    {
+      "epoch": 2.5188699590252317,
+      "grad_norm": 0.19700340926647186,
+      "learning_rate": 9.464649098437781e-06,
+      "loss": 2.3059,
+      "step": 5839
+    },
+    {
+      "epoch": 2.51930127237438,
+      "grad_norm": 0.209873229265213,
+      "learning_rate": 9.448060238780008e-06,
+      "loss": 2.318,
+      "step": 5840
+    },
+    {
+      "epoch": 2.519732585723528,
+      "grad_norm": 0.1966349333524704,
+      "learning_rate": 9.431484952368237e-06,
+      "loss": 2.0184,
+      "step": 5841
+    },
+    {
+      "epoch": 2.5201638990726765,
+      "grad_norm": 0.23189082741737366,
+      "learning_rate": 9.414923242634556e-06,
+      "loss": 2.1466,
+      "step": 5842
+    },
+    {
+      "epoch": 2.5205952124218243,
+      "grad_norm": 0.18889522552490234,
+      "learning_rate": 9.398375113008239e-06,
+      "loss": 1.9915,
+      "step": 5843
+    },
+    {
+      "epoch": 2.5210265257709725,
+      "grad_norm": 0.2166232466697693,
+      "learning_rate": 9.38184056691576e-06,
+      "loss": 2.1035,
+      "step": 5844
+    },
+    {
+      "epoch": 2.521457839120121,
+      "grad_norm": 0.2198508232831955,
+      "learning_rate": 9.365319607780792e-06,
+      "loss": 2.3479,
+      "step": 5845
+    },
+    {
+      "epoch": 2.521889152469269,
+      "grad_norm": 0.2350691556930542,
+      "learning_rate": 9.348812239024153e-06,
+      "loss": 2.1984,
+      "step": 5846
+    },
+    {
+      "epoch": 2.5223204658184173,
+      "grad_norm": 0.19254909455776215,
+      "learning_rate": 9.332318464063895e-06,
+      "loss": 2.0115,
+      "step": 5847
+    },
+    {
+      "epoch": 2.522751779167565,
+      "grad_norm": 0.18990060687065125,
+      "learning_rate": 9.315838286315236e-06,
+      "loss": 2.1486,
+      "step": 5848
+    },
+    {
+      "epoch": 2.5231830925167134,
+      "grad_norm": 0.19294965267181396,
+      "learning_rate": 9.299371709190561e-06,
+      "loss": 2.0948,
+      "step": 5849
+    },
+    {
+      "epoch": 2.5236144058658616,
+      "grad_norm": 0.23233257234096527,
+      "learning_rate": 9.282918736099446e-06,
+      "loss": 2.141,
+      "step": 5850
+    },
+    {
+      "epoch": 2.5236144058658616,
+      "eval_loss": 2.0872998237609863,
+      "eval_runtime": 202.8781,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 5850
+    },
+    {
+      "epoch": 2.52404571921501,
+      "grad_norm": 0.19171443581581116,
+      "learning_rate": 9.266479370448712e-06,
+      "loss": 2.1621,
+      "step": 5851
+    },
+    {
+      "epoch": 2.5244770325641577,
+      "grad_norm": 17.882905960083008,
+      "learning_rate": 9.250053615642263e-06,
+      "loss": 1.9856,
+      "step": 5852
+    },
+    {
+      "epoch": 2.524908345913306,
+      "grad_norm": 0.20517498254776,
+      "learning_rate": 9.233641475081239e-06,
+      "loss": 2.2057,
+      "step": 5853
+    },
+    {
+      "epoch": 2.525339659262454,
+      "grad_norm": 0.2134210765361786,
+      "learning_rate": 9.217242952163934e-06,
+      "loss": 2.1409,
+      "step": 5854
+    },
+    {
+      "epoch": 2.5257709726116024,
+      "grad_norm": 0.19968686997890472,
+      "learning_rate": 9.200858050285883e-06,
+      "loss": 2.1377,
+      "step": 5855
+    },
+    {
+      "epoch": 2.5262022859607507,
+      "grad_norm": 0.1992197185754776,
+      "learning_rate": 9.184486772839758e-06,
+      "loss": 2.04,
+      "step": 5856
+    },
+    {
+      "epoch": 2.5266335993098985,
+      "grad_norm": 0.2095516175031662,
+      "learning_rate": 9.168129123215363e-06,
+      "loss": 2.1054,
+      "step": 5857
+    },
+    {
+      "epoch": 2.5270649126590468,
+      "grad_norm": 0.18523921072483063,
+      "learning_rate": 9.15178510479977e-06,
+      "loss": 2.0172,
+      "step": 5858
+    },
+    {
+      "epoch": 2.527496226008195,
+      "grad_norm": 0.20448660850524902,
+      "learning_rate": 9.13545472097718e-06,
+      "loss": 2.0324,
+      "step": 5859
+    },
+    {
+      "epoch": 2.5279275393573433,
+      "grad_norm": 0.19547265768051147,
+      "learning_rate": 9.119137975128967e-06,
+      "loss": 2.2357,
+      "step": 5860
+    },
+    {
+      "epoch": 2.528358852706491,
+      "grad_norm": 0.20918558537960052,
+      "learning_rate": 9.102834870633703e-06,
+      "loss": 2.2514,
+      "step": 5861
+    },
+    {
+      "epoch": 2.5287901660556393,
+      "grad_norm": 0.2057720124721527,
+      "learning_rate": 9.086545410867107e-06,
+      "loss": 2.1962,
+      "step": 5862
+    },
+    {
+      "epoch": 2.5292214794047876,
+      "grad_norm": 0.19434405863285065,
+      "learning_rate": 9.070269599202112e-06,
+      "loss": 2.141,
+      "step": 5863
+    },
+    {
+      "epoch": 2.529652792753936,
+      "grad_norm": 0.20333506166934967,
+      "learning_rate": 9.05400743900877e-06,
+      "loss": 2.2222,
+      "step": 5864
+    },
+    {
+      "epoch": 2.530084106103084,
+      "grad_norm": 0.19405965507030487,
+      "learning_rate": 9.037758933654396e-06,
+      "loss": 2.1248,
+      "step": 5865
+    },
+    {
+      "epoch": 2.530515419452232,
+      "grad_norm": 0.2062096744775772,
+      "learning_rate": 9.021524086503374e-06,
+      "loss": 2.1761,
+      "step": 5866
+    },
+    {
+      "epoch": 2.53094673280138,
+      "grad_norm": 0.1919679492712021,
+      "learning_rate": 9.005302900917313e-06,
+      "loss": 2.0342,
+      "step": 5867
+    },
+    {
+      "epoch": 2.5313780461505284,
+      "grad_norm": 0.20065893232822418,
+      "learning_rate": 8.98909538025497e-06,
+      "loss": 2.2201,
+      "step": 5868
+    },
+    {
+      "epoch": 2.5318093594996767,
+      "grad_norm": 0.2016979157924652,
+      "learning_rate": 8.972901527872337e-06,
+      "loss": 2.0871,
+      "step": 5869
+    },
+    {
+      "epoch": 2.5322406728488245,
+      "grad_norm": 0.2010689228773117,
+      "learning_rate": 8.956721347122519e-06,
+      "loss": 2.2679,
+      "step": 5870
+    },
+    {
+      "epoch": 2.5326719861979727,
+      "grad_norm": 0.18522782623767853,
+      "learning_rate": 8.94055484135575e-06,
+      "loss": 2.1558,
+      "step": 5871
+    },
+    {
+      "epoch": 2.533103299547121,
+      "grad_norm": 0.21023201942443848,
+      "learning_rate": 8.92440201391952e-06,
+      "loss": 2.0731,
+      "step": 5872
+    },
+    {
+      "epoch": 2.5335346128962692,
+      "grad_norm": 0.21523456275463104,
+      "learning_rate": 8.908262868158452e-06,
+      "loss": 1.936,
+      "step": 5873
+    },
+    {
+      "epoch": 2.5339659262454175,
+      "grad_norm": 0.1987065076828003,
+      "learning_rate": 8.892137407414329e-06,
+      "loss": 2.1175,
+      "step": 5874
+    },
+    {
+      "epoch": 2.5343972395945653,
+      "grad_norm": 0.19248445332050323,
+      "learning_rate": 8.876025635026096e-06,
+      "loss": 2.0221,
+      "step": 5875
+    },
+    {
+      "epoch": 2.5343972395945653,
+      "eval_loss": 2.087226390838623,
+      "eval_runtime": 200.8487,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 5875
+    },
+    {
+      "epoch": 2.5348285529437136,
+      "grad_norm": 0.20776793360710144,
+      "learning_rate": 8.859927554329877e-06,
+      "loss": 1.9483,
+      "step": 5876
+    },
+    {
+      "epoch": 2.535259866292862,
+      "grad_norm": 0.19198954105377197,
+      "learning_rate": 8.843843168658945e-06,
+      "loss": 2.2548,
+      "step": 5877
+    },
+    {
+      "epoch": 2.53569117964201,
+      "grad_norm": 0.1943345069885254,
+      "learning_rate": 8.827772481343764e-06,
+      "loss": 2.2774,
+      "step": 5878
+    },
+    {
+      "epoch": 2.536122492991158,
+      "grad_norm": 0.21345116198062897,
+      "learning_rate": 8.811715495711938e-06,
+      "loss": 2.2552,
+      "step": 5879
+    },
+    {
+      "epoch": 2.536553806340306,
+      "grad_norm": 0.1998181790113449,
+      "learning_rate": 8.795672215088231e-06,
+      "loss": 2.098,
+      "step": 5880
+    },
+    {
+      "epoch": 2.5369851196894544,
+      "grad_norm": 0.1961486041545868,
+      "learning_rate": 8.779642642794604e-06,
+      "loss": 2.0678,
+      "step": 5881
+    },
+    {
+      "epoch": 2.5374164330386026,
+      "grad_norm": 0.2047472447156906,
+      "learning_rate": 8.763626782150118e-06,
+      "loss": 2.1216,
+      "step": 5882
+    },
+    {
+      "epoch": 2.537847746387751,
+      "grad_norm": 0.19389478862285614,
+      "learning_rate": 8.747624636471071e-06,
+      "loss": 2.2829,
+      "step": 5883
+    },
+    {
+      "epoch": 2.5382790597368987,
+      "grad_norm": 0.1919451355934143,
+      "learning_rate": 8.731636209070886e-06,
+      "loss": 2.0559,
+      "step": 5884
+    },
+    {
+      "epoch": 2.538710373086047,
+      "grad_norm": 0.1973874270915985,
+      "learning_rate": 8.715661503260108e-06,
+      "loss": 2.1969,
+      "step": 5885
+    },
+    {
+      "epoch": 2.5391416864351952,
+      "grad_norm": 0.19648849964141846,
+      "learning_rate": 8.699700522346464e-06,
+      "loss": 2.1255,
+      "step": 5886
+    },
+    {
+      "epoch": 2.5395729997843435,
+      "grad_norm": 0.22413647174835205,
+      "learning_rate": 8.683753269634905e-06,
+      "loss": 2.1444,
+      "step": 5887
+    },
+    {
+      "epoch": 2.5400043131334913,
+      "grad_norm": 0.212081179022789,
+      "learning_rate": 8.667819748427437e-06,
+      "loss": 2.0756,
+      "step": 5888
+    },
+    {
+      "epoch": 2.5404356264826395,
+      "grad_norm": 0.19365866482257843,
+      "learning_rate": 8.651899962023297e-06,
+      "loss": 2.2299,
+      "step": 5889
+    },
+    {
+      "epoch": 2.540866939831788,
+      "grad_norm": 3.2413103580474854,
+      "learning_rate": 8.635993913718842e-06,
+      "loss": 1.843,
+      "step": 5890
+    },
+    {
+      "epoch": 2.541298253180936,
+      "grad_norm": 0.19114452600479126,
+      "learning_rate": 8.620101606807583e-06,
+      "loss": 2.1101,
+      "step": 5891
+    },
+    {
+      "epoch": 2.5417295665300843,
+      "grad_norm": 0.20034754276275635,
+      "learning_rate": 8.604223044580216e-06,
+      "loss": 2.2439,
+      "step": 5892
+    },
+    {
+      "epoch": 2.542160879879232,
+      "grad_norm": 0.21234995126724243,
+      "learning_rate": 8.588358230324553e-06,
+      "loss": 2.153,
+      "step": 5893
+    },
+    {
+      "epoch": 2.5425921932283804,
+      "grad_norm": 0.18674694001674652,
+      "learning_rate": 8.572507167325579e-06,
+      "loss": 2.0226,
+      "step": 5894
+    },
+    {
+      "epoch": 2.5430235065775286,
+      "grad_norm": 0.20961590111255646,
+      "learning_rate": 8.556669858865448e-06,
+      "loss": 2.3047,
+      "step": 5895
+    },
+    {
+      "epoch": 2.543454819926677,
+      "grad_norm": 0.2046174705028534,
+      "learning_rate": 8.54084630822341e-06,
+      "loss": 2.214,
+      "step": 5896
+    },
+    {
+      "epoch": 2.5438861332758247,
+      "grad_norm": 0.21066071093082428,
+      "learning_rate": 8.525036518675959e-06,
+      "loss": 2.3135,
+      "step": 5897
+    },
+    {
+      "epoch": 2.544317446624973,
+      "grad_norm": 0.22002047300338745,
+      "learning_rate": 8.50924049349664e-06,
+      "loss": 2.0497,
+      "step": 5898
+    },
+    {
+      "epoch": 2.544748759974121,
+      "grad_norm": 0.19885514676570892,
+      "learning_rate": 8.493458235956212e-06,
+      "loss": 2.1107,
+      "step": 5899
+    },
+    {
+      "epoch": 2.5451800733232695,
+      "grad_norm": 0.17786090075969696,
+      "learning_rate": 8.477689749322537e-06,
+      "loss": 2.3082,
+      "step": 5900
+    },
+    {
+      "epoch": 2.5451800733232695,
+      "eval_loss": 2.0872507095336914,
+      "eval_runtime": 202.5459,
+      "eval_samples_per_second": 0.158,
+      "eval_steps_per_second": 0.158,
+      "step": 5900
+    },
+    {
+      "epoch": 2.5456113866724177,
+      "grad_norm": 0.20288458466529846,
+      "learning_rate": 8.461935036860695e-06,
+      "loss": 2.2552,
+      "step": 5901
+    },
+    {
+      "epoch": 2.5460427000215655,
+      "grad_norm": 0.19387395679950714,
+      "learning_rate": 8.446194101832865e-06,
+      "loss": 2.0872,
+      "step": 5902
+    },
+    {
+      "epoch": 2.5464740133707138,
+      "grad_norm": 0.20319326221942902,
+      "learning_rate": 8.430466947498338e-06,
+      "loss": 2.4172,
+      "step": 5903
+    },
+    {
+      "epoch": 2.546905326719862,
+      "grad_norm": 0.6388072967529297,
+      "learning_rate": 8.414753577113634e-06,
+      "loss": 2.148,
+      "step": 5904
+    },
+    {
+      "epoch": 2.5473366400690103,
+      "grad_norm": 0.20338557660579681,
+      "learning_rate": 8.399053993932367e-06,
+      "loss": 2.1059,
+      "step": 5905
+    },
+    {
+      "epoch": 2.547767953418158,
+      "grad_norm": 0.19849160313606262,
+      "learning_rate": 8.383368201205312e-06,
+      "loss": 2.194,
+      "step": 5906
+    },
+    {
+      "epoch": 2.5481992667673063,
+      "grad_norm": 0.22318509221076965,
+      "learning_rate": 8.367696202180371e-06,
+      "loss": 2.0911,
+      "step": 5907
+    },
+    {
+      "epoch": 2.5486305801164546,
+      "grad_norm": 0.19940035045146942,
+      "learning_rate": 8.352038000102614e-06,
+      "loss": 1.9267,
+      "step": 5908
+    },
+    {
+      "epoch": 2.549061893465603,
+      "grad_norm": 0.211517333984375,
+      "learning_rate": 8.336393598214233e-06,
+      "loss": 2.2814,
+      "step": 5909
+    },
+    {
+      "epoch": 2.549493206814751,
+      "grad_norm": 0.186599001288414,
+      "learning_rate": 8.320762999754568e-06,
+      "loss": 1.8915,
+      "step": 5910
+    },
+    {
+      "epoch": 2.549924520163899,
+      "grad_norm": 0.2223532497882843,
+      "learning_rate": 8.305146207960148e-06,
+      "loss": 2.2837,
+      "step": 5911
+    },
+    {
+      "epoch": 2.550355833513047,
+      "grad_norm": 0.20304465293884277,
+      "learning_rate": 8.289543226064552e-06,
+      "loss": 2.0781,
+      "step": 5912
+    },
+    {
+      "epoch": 2.5507871468621954,
+      "grad_norm": 0.20744968950748444,
+      "learning_rate": 8.273954057298562e-06,
+      "loss": 1.9754,
+      "step": 5913
+    },
+    {
+      "epoch": 2.5512184602113437,
+      "grad_norm": 0.20294539630413055,
+      "learning_rate": 8.258378704890073e-06,
+      "loss": 2.0599,
+      "step": 5914
+    },
+    {
+      "epoch": 2.5516497735604915,
+      "grad_norm": 0.1947573870420456,
+      "learning_rate": 8.242817172064162e-06,
+      "loss": 2.1859,
+      "step": 5915
+    },
+    {
+      "epoch": 2.5520810869096398,
+      "grad_norm": 0.19360491633415222,
+      "learning_rate": 8.22726946204301e-06,
+      "loss": 2.2329,
+      "step": 5916
+    },
+    {
+      "epoch": 2.552512400258788,
+      "grad_norm": 0.21240995824337006,
+      "learning_rate": 8.211735578045909e-06,
+      "loss": 2.154,
+      "step": 5917
+    },
+    {
+      "epoch": 2.5529437136079363,
+      "grad_norm": 0.20330595970153809,
+      "learning_rate": 8.19621552328935e-06,
+      "loss": 1.878,
+      "step": 5918
+    },
+    {
+      "epoch": 2.5533750269570845,
+      "grad_norm": 0.20137812197208405,
+      "learning_rate": 8.180709300986926e-06,
+      "loss": 2.1407,
+      "step": 5919
+    },
+    {
+      "epoch": 2.5538063403062323,
+      "grad_norm": 0.19222082197666168,
+      "learning_rate": 8.165216914349357e-06,
+      "loss": 2.2948,
+      "step": 5920
+    },
+    {
+      "epoch": 2.5542376536553806,
+      "grad_norm": 0.22950433194637299,
+      "learning_rate": 8.149738366584526e-06,
+      "loss": 2.049,
+      "step": 5921
+    },
+    {
+      "epoch": 2.554668967004529,
+      "grad_norm": 0.22426775097846985,
+      "learning_rate": 8.134273660897422e-06,
+      "loss": 2.1904,
+      "step": 5922
+    },
+    {
+      "epoch": 2.555100280353677,
+      "grad_norm": 0.2124083787202835,
+      "learning_rate": 8.118822800490199e-06,
+      "loss": 2.2236,
+      "step": 5923
+    },
+    {
+      "epoch": 2.555531593702825,
+      "grad_norm": 0.21372617781162262,
+      "learning_rate": 8.103385788562105e-06,
+      "loss": 2.3067,
+      "step": 5924
+    },
+    {
+      "epoch": 2.555962907051973,
+      "grad_norm": 0.20108765363693237,
+      "learning_rate": 8.087962628309552e-06,
+      "loss": 2.2048,
+      "step": 5925
+    },
+    {
+      "epoch": 2.555962907051973,
+      "eval_loss": 2.087303638458252,
+      "eval_runtime": 204.4825,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 5925
+    },
+    {
+      "epoch": 2.5563942204011214,
+      "grad_norm": 0.20634593069553375,
+      "learning_rate": 8.072553322926076e-06,
+      "loss": 2.1873,
+      "step": 5926
+    },
+    {
+      "epoch": 2.5568255337502697,
+      "grad_norm": 0.17352814972400665,
+      "learning_rate": 8.05715787560235e-06,
+      "loss": 2.1897,
+      "step": 5927
+    },
+    {
+      "epoch": 2.557256847099418,
+      "grad_norm": 0.1895429790019989,
+      "learning_rate": 8.04177628952613e-06,
+      "loss": 2.0911,
+      "step": 5928
+    },
+    {
+      "epoch": 2.5576881604485657,
+      "grad_norm": 0.21185369789600372,
+      "learning_rate": 8.026408567882415e-06,
+      "loss": 2.1121,
+      "step": 5929
+    },
+    {
+      "epoch": 2.558119473797714,
+      "grad_norm": 0.214601069688797,
+      "learning_rate": 8.011054713853196e-06,
+      "loss": 2.0906,
+      "step": 5930
+    },
+    {
+      "epoch": 2.5585507871468622,
+      "grad_norm": 0.2063516229391098,
+      "learning_rate": 7.995714730617674e-06,
+      "loss": 2.3176,
+      "step": 5931
+    },
+    {
+      "epoch": 2.5589821004960105,
+      "grad_norm": 0.20094339549541473,
+      "learning_rate": 7.980388621352151e-06,
+      "loss": 2.2627,
+      "step": 5932
+    },
+    {
+      "epoch": 2.5594134138451583,
+      "grad_norm": 0.190909743309021,
+      "learning_rate": 7.965076389230096e-06,
+      "loss": 2.267,
+      "step": 5933
+    },
+    {
+      "epoch": 2.5598447271943066,
+      "grad_norm": 0.19475963711738586,
+      "learning_rate": 7.949778037422073e-06,
+      "loss": 1.9725,
+      "step": 5934
+    },
+    {
+      "epoch": 2.560276040543455,
+      "grad_norm": 0.2023172229528427,
+      "learning_rate": 7.93449356909572e-06,
+      "loss": 2.0805,
+      "step": 5935
+    },
+    {
+      "epoch": 2.560707353892603,
+      "grad_norm": 0.22337819635868073,
+      "learning_rate": 7.919222987415911e-06,
+      "loss": 2.2469,
+      "step": 5936
+    },
+    {
+      "epoch": 2.5611386672417513,
+      "grad_norm": 0.1987045407295227,
+      "learning_rate": 7.903966295544562e-06,
+      "loss": 2.2534,
+      "step": 5937
+    },
+    {
+      "epoch": 2.561569980590899,
+      "grad_norm": 0.19819551706314087,
+      "learning_rate": 7.888723496640739e-06,
+      "loss": 1.9927,
+      "step": 5938
+    },
+    {
+      "epoch": 2.5620012939400474,
+      "grad_norm": 0.19501759111881256,
+      "learning_rate": 7.873494593860636e-06,
+      "loss": 2.1534,
+      "step": 5939
+    },
+    {
+      "epoch": 2.5624326072891956,
+      "grad_norm": 0.19448330998420715,
+      "learning_rate": 7.858279590357553e-06,
+      "loss": 2.2013,
+      "step": 5940
+    },
+    {
+      "epoch": 2.562863920638344,
+      "grad_norm": 0.19504505395889282,
+      "learning_rate": 7.843078489281937e-06,
+      "loss": 2.1991,
+      "step": 5941
+    },
+    {
+      "epoch": 2.5632952339874917,
+      "grad_norm": 0.21552002429962158,
+      "learning_rate": 7.827891293781322e-06,
+      "loss": 2.1397,
+      "step": 5942
+    },
+    {
+      "epoch": 2.56372654733664,
+      "grad_norm": 0.19737036526203156,
+      "learning_rate": 7.812718007000424e-06,
+      "loss": 2.0697,
+      "step": 5943
+    },
+    {
+      "epoch": 2.564157860685788,
+      "grad_norm": 0.19582600891590118,
+      "learning_rate": 7.797558632080996e-06,
+      "loss": 2.1335,
+      "step": 5944
+    },
+    {
+      "epoch": 2.5645891740349365,
+      "grad_norm": 0.19773516058921814,
+      "learning_rate": 7.782413172161963e-06,
+      "loss": 2.0992,
+      "step": 5945
+    },
+    {
+      "epoch": 2.5650204873840847,
+      "grad_norm": 0.20045964419841766,
+      "learning_rate": 7.767281630379349e-06,
+      "loss": 2.2479,
+      "step": 5946
+    },
+    {
+      "epoch": 2.5654518007332325,
+      "grad_norm": 0.19492270052433014,
+      "learning_rate": 7.752164009866338e-06,
+      "loss": 2.1007,
+      "step": 5947
+    },
+    {
+      "epoch": 2.565883114082381,
+      "grad_norm": 0.19753959774971008,
+      "learning_rate": 7.737060313753185e-06,
+      "loss": 2.1354,
+      "step": 5948
+    },
+    {
+      "epoch": 2.566314427431529,
+      "grad_norm": 0.19580677151679993,
+      "learning_rate": 7.721970545167247e-06,
+      "loss": 2.1234,
+      "step": 5949
+    },
+    {
+      "epoch": 2.5667457407806773,
+      "grad_norm": 0.23202303051948547,
+      "learning_rate": 7.706894707233057e-06,
+      "loss": 2.0789,
+      "step": 5950
+    },
+    {
+      "epoch": 2.5667457407806773,
+      "eval_loss": 2.087388277053833,
+      "eval_runtime": 203.6764,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 5950
+    },
+    {
+      "epoch": 2.567177054129825,
+      "grad_norm": 0.1929044872522354,
+      "learning_rate": 7.691832803072221e-06,
+      "loss": 2.1219,
+      "step": 5951
+    },
+    {
+      "epoch": 2.5676083674789734,
+      "grad_norm": 0.18711425364017487,
+      "learning_rate": 7.676784835803481e-06,
+      "loss": 2.2567,
+      "step": 5952
+    },
+    {
+      "epoch": 2.5680396808281216,
+      "grad_norm": 0.19768792390823364,
+      "learning_rate": 7.661750808542666e-06,
+      "loss": 2.0054,
+      "step": 5953
+    },
+    {
+      "epoch": 2.56847099417727,
+      "grad_norm": 0.18793371319770813,
+      "learning_rate": 7.646730724402749e-06,
+      "loss": 2.1189,
+      "step": 5954
+    },
+    {
+      "epoch": 2.568902307526418,
+      "grad_norm": 0.1859693080186844,
+      "learning_rate": 7.631724586493804e-06,
+      "loss": 2.0841,
+      "step": 5955
+    },
+    {
+      "epoch": 2.569333620875566,
+      "grad_norm": 0.19793583452701569,
+      "learning_rate": 7.616732397922992e-06,
+      "loss": 2.1784,
+      "step": 5956
+    },
+    {
+      "epoch": 2.569764934224714,
+      "grad_norm": 0.1949821561574936,
+      "learning_rate": 7.601754161794657e-06,
+      "loss": 1.89,
+      "step": 5957
+    },
+    {
+      "epoch": 2.5701962475738624,
+      "grad_norm": 0.1898735910654068,
+      "learning_rate": 7.586789881210159e-06,
+      "loss": 2.122,
+      "step": 5958
+    },
+    {
+      "epoch": 2.5706275609230107,
+      "grad_norm": 0.20294371247291565,
+      "learning_rate": 7.571839559268036e-06,
+      "loss": 2.1028,
+      "step": 5959
+    },
+    {
+      "epoch": 2.5710588742721585,
+      "grad_norm": 0.1875556856393814,
+      "learning_rate": 7.556903199063904e-06,
+      "loss": 2.0433,
+      "step": 5960
+    },
+    {
+      "epoch": 2.5714901876213068,
+      "grad_norm": 0.21623234450817108,
+      "learning_rate": 7.541980803690522e-06,
+      "loss": 2.1431,
+      "step": 5961
+    },
+    {
+      "epoch": 2.571921500970455,
+      "grad_norm": 0.20514778792858124,
+      "learning_rate": 7.527072376237736e-06,
+      "loss": 2.0767,
+      "step": 5962
+    },
+    {
+      "epoch": 2.5723528143196033,
+      "grad_norm": 0.1848430186510086,
+      "learning_rate": 7.5121779197924585e-06,
+      "loss": 2.1939,
+      "step": 5963
+    },
+    {
+      "epoch": 2.5727841276687515,
+      "grad_norm": 0.18685248494148254,
+      "learning_rate": 7.49729743743879e-06,
+      "loss": 2.0598,
+      "step": 5964
+    },
+    {
+      "epoch": 2.5732154410178993,
+      "grad_norm": 0.19710910320281982,
+      "learning_rate": 7.482430932257882e-06,
+      "loss": 1.9915,
+      "step": 5965
+    },
+    {
+      "epoch": 2.5736467543670476,
+      "grad_norm": 0.19875414669513702,
+      "learning_rate": 7.467578407328015e-06,
+      "loss": 1.9632,
+      "step": 5966
+    },
+    {
+      "epoch": 2.574078067716196,
+      "grad_norm": 0.2088191658258438,
+      "learning_rate": 7.452739865724553e-06,
+      "loss": 2.0285,
+      "step": 5967
+    },
+    {
+      "epoch": 2.574509381065344,
+      "grad_norm": 0.19341960549354553,
+      "learning_rate": 7.4379153105199955e-06,
+      "loss": 1.9608,
+      "step": 5968
+    },
+    {
+      "epoch": 2.574940694414492,
+      "grad_norm": 0.2098260521888733,
+      "learning_rate": 7.423104744783914e-06,
+      "loss": 2.1184,
+      "step": 5969
+    },
+    {
+      "epoch": 2.57537200776364,
+      "grad_norm": 0.20200322568416595,
+      "learning_rate": 7.408308171583005e-06,
+      "loss": 2.1119,
+      "step": 5970
+    },
+    {
+      "epoch": 2.5758033211127884,
+      "grad_norm": 0.2155592143535614,
+      "learning_rate": 7.3935255939810586e-06,
+      "loss": 2.1972,
+      "step": 5971
+    },
+    {
+      "epoch": 2.5762346344619367,
+      "grad_norm": 0.2060846984386444,
+      "learning_rate": 7.3787570150389695e-06,
+      "loss": 2.0123,
+      "step": 5972
+    },
+    {
+      "epoch": 2.576665947811085,
+      "grad_norm": 0.21392014622688293,
+      "learning_rate": 7.364002437814731e-06,
+      "loss": 2.2761,
+      "step": 5973
+    },
+    {
+      "epoch": 2.5770972611602327,
+      "grad_norm": 0.2115173488855362,
+      "learning_rate": 7.349261865363434e-06,
+      "loss": 2.2229,
+      "step": 5974
+    },
+    {
+      "epoch": 2.577528574509381,
+      "grad_norm": 0.19911466538906097,
+      "learning_rate": 7.334535300737313e-06,
+      "loss": 2.1165,
+      "step": 5975
+    },
+    {
+      "epoch": 2.577528574509381,
+      "eval_loss": 2.087284564971924,
+      "eval_runtime": 204.9153,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 5975
+    },
+    {
+      "epoch": 2.5779598878585293,
+      "grad_norm": 0.17648504674434662,
+      "learning_rate": 7.319822746985618e-06,
+      "loss": 1.9316,
+      "step": 5976
+    },
+    {
+      "epoch": 2.5783912012076775,
+      "grad_norm": 0.21495161950588226,
+      "learning_rate": 7.3051242071547554e-06,
+      "loss": 2.0624,
+      "step": 5977
+    },
+    {
+      "epoch": 2.5788225145568253,
+      "grad_norm": 0.18364490568637848,
+      "learning_rate": 7.290439684288216e-06,
+      "loss": 1.6425,
+      "step": 5978
+    },
+    {
+      "epoch": 2.5792538279059736,
+      "grad_norm": 0.20838278532028198,
+      "learning_rate": 7.275769181426616e-06,
+      "loss": 2.07,
+      "step": 5979
+    },
+    {
+      "epoch": 2.579685141255122,
+      "grad_norm": 0.19302137196063995,
+      "learning_rate": 7.261112701607635e-06,
+      "loss": 1.8957,
+      "step": 5980
+    },
+    {
+      "epoch": 2.58011645460427,
+      "grad_norm": 0.1876315027475357,
+      "learning_rate": 7.2464702478660295e-06,
+      "loss": 1.8388,
+      "step": 5981
+    },
+    {
+      "epoch": 2.5805477679534183,
+      "grad_norm": 0.2408015877008438,
+      "learning_rate": 7.2318418232337054e-06,
+      "loss": 1.9432,
+      "step": 5982
+    },
+    {
+      "epoch": 2.5809790813025666,
+      "grad_norm": 0.20141193270683289,
+      "learning_rate": 7.217227430739633e-06,
+      "loss": 2.0929,
+      "step": 5983
+    },
+    {
+      "epoch": 2.5814103946517144,
+      "grad_norm": 0.204561248421669,
+      "learning_rate": 7.202627073409883e-06,
+      "loss": 2.039,
+      "step": 5984
+    },
+    {
+      "epoch": 2.5818417080008627,
+      "grad_norm": 0.20142017304897308,
+      "learning_rate": 7.188040754267613e-06,
+      "loss": 2.4515,
+      "step": 5985
+    },
+    {
+      "epoch": 2.582273021350011,
+      "grad_norm": 0.1931556612253189,
+      "learning_rate": 7.1734684763330885e-06,
+      "loss": 2.1068,
+      "step": 5986
+    },
+    {
+      "epoch": 2.5827043346991587,
+      "grad_norm": 0.20126812160015106,
+      "learning_rate": 7.158910242623656e-06,
+      "loss": 2.0849,
+      "step": 5987
+    },
+    {
+      "epoch": 2.583135648048307,
+      "grad_norm": 0.21254923939704895,
+      "learning_rate": 7.144366056153736e-06,
+      "loss": 2.0209,
+      "step": 5988
+    },
+    {
+      "epoch": 2.5835669613974552,
+      "grad_norm": 0.21428455412387848,
+      "learning_rate": 7.129835919934912e-06,
+      "loss": 2.0088,
+      "step": 5989
+    },
+    {
+      "epoch": 2.5839982747466035,
+      "grad_norm": 0.19039230048656464,
+      "learning_rate": 7.115319836975761e-06,
+      "loss": 2.1215,
+      "step": 5990
+    },
+    {
+      "epoch": 2.5844295880957517,
+      "grad_norm": 0.20700082182884216,
+      "learning_rate": 7.100817810282011e-06,
+      "loss": 1.9611,
+      "step": 5991
+    },
+    {
+      "epoch": 2.5848609014449,
+      "grad_norm": 0.2018066942691803,
+      "learning_rate": 7.086329842856453e-06,
+      "loss": 2.1965,
+      "step": 5992
+    },
+    {
+      "epoch": 2.585292214794048,
+      "grad_norm": 0.20085200667381287,
+      "learning_rate": 7.071855937699003e-06,
+      "loss": 2.2475,
+      "step": 5993
+    },
+    {
+      "epoch": 2.585723528143196,
+      "grad_norm": 0.22047682106494904,
+      "learning_rate": 7.057396097806647e-06,
+      "loss": 2.1593,
+      "step": 5994
+    },
+    {
+      "epoch": 2.5861548414923443,
+      "grad_norm": 0.19113773107528687,
+      "learning_rate": 7.042950326173416e-06,
+      "loss": 2.0633,
+      "step": 5995
+    },
+    {
+      "epoch": 2.586586154841492,
+      "grad_norm": 0.19312886893749237,
+      "learning_rate": 7.028518625790492e-06,
+      "loss": 2.1159,
+      "step": 5996
+    },
+    {
+      "epoch": 2.5870174681906404,
+      "grad_norm": 0.20536403357982635,
+      "learning_rate": 7.014100999646108e-06,
+      "loss": 2.1573,
+      "step": 5997
+    },
+    {
+      "epoch": 2.5874487815397886,
+      "grad_norm": 0.1941017359495163,
+      "learning_rate": 6.999697450725602e-06,
+      "loss": 2.1259,
+      "step": 5998
+    },
+    {
+      "epoch": 2.587880094888937,
+      "grad_norm": 0.19312000274658203,
+      "learning_rate": 6.985307982011373e-06,
+      "loss": 1.9693,
+      "step": 5999
+    },
+    {
+      "epoch": 2.588311408238085,
+      "grad_norm": 0.2647634744644165,
+      "learning_rate": 6.97093259648292e-06,
+      "loss": 2.0193,
+      "step": 6000
+    },
+    {
+      "epoch": 2.588311408238085,
+      "eval_loss": 2.0873632431030273,
+      "eval_runtime": 211.403,
+      "eval_samples_per_second": 0.151,
+      "eval_steps_per_second": 0.151,
+      "step": 6000
+    },
+    {
+      "epoch": 2.5887427215872334,
+      "grad_norm": 0.18285487592220306,
+      "learning_rate": 6.956571297116828e-06,
+      "loss": 2.0634,
+      "step": 6001
+    },
+    {
+      "epoch": 2.589174034936381,
+      "grad_norm": 0.1905910074710846,
+      "learning_rate": 6.942224086886744e-06,
+      "loss": 2.114,
+      "step": 6002
+    },
+    {
+      "epoch": 2.5896053482855295,
+      "grad_norm": 0.19421325623989105,
+      "learning_rate": 6.927890968763458e-06,
+      "loss": 2.1351,
+      "step": 6003
+    },
+    {
+      "epoch": 2.5900366616346777,
+      "grad_norm": 0.18619950115680695,
+      "learning_rate": 6.913571945714763e-06,
+      "loss": 1.9864,
+      "step": 6004
+    },
+    {
+      "epoch": 2.5904679749838255,
+      "grad_norm": 0.21644152700901031,
+      "learning_rate": 6.89926702070557e-06,
+      "loss": 2.1049,
+      "step": 6005
+    },
+    {
+      "epoch": 2.590899288332974,
+      "grad_norm": 0.20959539711475372,
+      "learning_rate": 6.884976196697869e-06,
+      "loss": 2.0024,
+      "step": 6006
+    },
+    {
+      "epoch": 2.591330601682122,
+      "grad_norm": 0.20113292336463928,
+      "learning_rate": 6.870699476650751e-06,
+      "loss": 2.1691,
+      "step": 6007
+    },
+    {
+      "epoch": 2.5917619150312703,
+      "grad_norm": 0.18709880113601685,
+      "learning_rate": 6.856436863520368e-06,
+      "loss": 1.9913,
+      "step": 6008
+    },
+    {
+      "epoch": 2.5921932283804185,
+      "grad_norm": 0.20955413579940796,
+      "learning_rate": 6.842188360259915e-06,
+      "loss": 2.1659,
+      "step": 6009
+    },
+    {
+      "epoch": 2.592624541729567,
+      "grad_norm": 0.188461035490036,
+      "learning_rate": 6.827953969819733e-06,
+      "loss": 2.1667,
+      "step": 6010
+    },
+    {
+      "epoch": 2.5930558550787146,
+      "grad_norm": 0.18821631371974945,
+      "learning_rate": 6.813733695147197e-06,
+      "loss": 2.0207,
+      "step": 6011
+    },
+    {
+      "epoch": 2.593487168427863,
+      "grad_norm": 0.20827440917491913,
+      "learning_rate": 6.7995275391867775e-06,
+      "loss": 2.1734,
+      "step": 6012
+    },
+    {
+      "epoch": 2.593918481777011,
+      "grad_norm": 0.24497582018375397,
+      "learning_rate": 6.785335504880005e-06,
+      "loss": 1.9533,
+      "step": 6013
+    },
+    {
+      "epoch": 2.594349795126159,
+      "grad_norm": 0.19103097915649414,
+      "learning_rate": 6.771157595165494e-06,
+      "loss": 2.1005,
+      "step": 6014
+    },
+    {
+      "epoch": 2.594781108475307,
+      "grad_norm": 0.19511078298091888,
+      "learning_rate": 6.756993812978947e-06,
+      "loss": 2.2424,
+      "step": 6015
+    },
+    {
+      "epoch": 2.5952124218244554,
+      "grad_norm": 0.1991245299577713,
+      "learning_rate": 6.742844161253128e-06,
+      "loss": 2.1054,
+      "step": 6016
+    },
+    {
+      "epoch": 2.5956437351736037,
+      "grad_norm": 0.18926672637462616,
+      "learning_rate": 6.7287086429178655e-06,
+      "loss": 2.1717,
+      "step": 6017
+    },
+    {
+      "epoch": 2.596075048522752,
+      "grad_norm": 0.20850855112075806,
+      "learning_rate": 6.714587260900087e-06,
+      "loss": 2.1144,
+      "step": 6018
+    },
+    {
+      "epoch": 2.5965063618719,
+      "grad_norm": 0.18856045603752136,
+      "learning_rate": 6.700480018123769e-06,
+      "loss": 1.949,
+      "step": 6019
+    },
+    {
+      "epoch": 2.596937675221048,
+      "grad_norm": 0.22177739441394806,
+      "learning_rate": 6.686386917509959e-06,
+      "loss": 2.0571,
+      "step": 6020
+    },
+    {
+      "epoch": 2.5973689885701963,
+      "grad_norm": 0.22452549636363983,
+      "learning_rate": 6.672307961976831e-06,
+      "loss": 2.1356,
+      "step": 6021
+    },
+    {
+      "epoch": 2.5978003019193445,
+      "grad_norm": 0.24075448513031006,
+      "learning_rate": 6.658243154439544e-06,
+      "loss": 2.0627,
+      "step": 6022
+    },
+    {
+      "epoch": 2.5982316152684923,
+      "grad_norm": 0.2034800946712494,
+      "learning_rate": 6.644192497810394e-06,
+      "loss": 2.2097,
+      "step": 6023
+    },
+    {
+      "epoch": 2.5986629286176406,
+      "grad_norm": 0.1853140890598297,
+      "learning_rate": 6.630155994998687e-06,
+      "loss": 1.7731,
+      "step": 6024
+    },
+    {
+      "epoch": 2.599094241966789,
+      "grad_norm": 0.2146259993314743,
+      "learning_rate": 6.616133648910882e-06,
+      "loss": 2.0087,
+      "step": 6025
+    },
+    {
+      "epoch": 2.599094241966789,
+      "eval_loss": 2.0873498916625977,
+      "eval_runtime": 204.7135,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 6025
+    },
+    {
+      "epoch": 2.599525555315937,
+      "grad_norm": 0.2032996118068695,
+      "learning_rate": 6.602125462450447e-06,
+      "loss": 2.3627,
+      "step": 6026
+    },
+    {
+      "epoch": 2.5999568686650854,
+      "grad_norm": 0.19560416042804718,
+      "learning_rate": 6.588131438517904e-06,
+      "loss": 2.2585,
+      "step": 6027
+    },
+    {
+      "epoch": 2.6003881820142336,
+      "grad_norm": 0.21011295914649963,
+      "learning_rate": 6.574151580010894e-06,
+      "loss": 1.9346,
+      "step": 6028
+    },
+    {
+      "epoch": 2.6008194953633814,
+      "grad_norm": 0.1834384649991989,
+      "learning_rate": 6.560185889824085e-06,
+      "loss": 2.059,
+      "step": 6029
+    },
+    {
+      "epoch": 2.6012508087125297,
+      "grad_norm": 0.2010362297296524,
+      "learning_rate": 6.546234370849246e-06,
+      "loss": 2.3003,
+      "step": 6030
+    },
+    {
+      "epoch": 2.601682122061678,
+      "grad_norm": 0.1939825564622879,
+      "learning_rate": 6.532297025975167e-06,
+      "loss": 1.987,
+      "step": 6031
+    },
+    {
+      "epoch": 2.6021134354108257,
+      "grad_norm": 0.19909532368183136,
+      "learning_rate": 6.518373858087747e-06,
+      "loss": 2.2438,
+      "step": 6032
+    },
+    {
+      "epoch": 2.602544748759974,
+      "grad_norm": 0.20319907367229462,
+      "learning_rate": 6.504464870069914e-06,
+      "loss": 2.1522,
+      "step": 6033
+    },
+    {
+      "epoch": 2.6029760621091222,
+      "grad_norm": 0.20442801713943481,
+      "learning_rate": 6.490570064801673e-06,
+      "loss": 2.0656,
+      "step": 6034
+    },
+    {
+      "epoch": 2.6034073754582705,
+      "grad_norm": 0.19795440137386322,
+      "learning_rate": 6.476689445160138e-06,
+      "loss": 2.0074,
+      "step": 6035
+    },
+    {
+      "epoch": 2.6038386888074188,
+      "grad_norm": 0.19856740534305573,
+      "learning_rate": 6.4628230140193945e-06,
+      "loss": 2.1488,
+      "step": 6036
+    },
+    {
+      "epoch": 2.604270002156567,
+      "grad_norm": 0.2064681053161621,
+      "learning_rate": 6.448970774250661e-06,
+      "loss": 2.1495,
+      "step": 6037
+    },
+    {
+      "epoch": 2.604701315505715,
+      "grad_norm": 0.22349761426448822,
+      "learning_rate": 6.43513272872217e-06,
+      "loss": 2.004,
+      "step": 6038
+    },
+    {
+      "epoch": 2.605132628854863,
+      "grad_norm": 0.21363207697868347,
+      "learning_rate": 6.421308880299278e-06,
+      "loss": 2.256,
+      "step": 6039
+    },
+    {
+      "epoch": 2.6055639422040113,
+      "grad_norm": 0.19845212996006012,
+      "learning_rate": 6.407499231844354e-06,
+      "loss": 2.0137,
+      "step": 6040
+    },
+    {
+      "epoch": 2.605995255553159,
+      "grad_norm": 0.1993948072195053,
+      "learning_rate": 6.393703786216811e-06,
+      "loss": 2.1983,
+      "step": 6041
+    },
+    {
+      "epoch": 2.6064265689023074,
+      "grad_norm": 0.19329705834388733,
+      "learning_rate": 6.379922546273172e-06,
+      "loss": 2.2284,
+      "step": 6042
+    },
+    {
+      "epoch": 2.6068578822514556,
+      "grad_norm": 0.2110152691602707,
+      "learning_rate": 6.3661555148669885e-06,
+      "loss": 2.1247,
+      "step": 6043
+    },
+    {
+      "epoch": 2.607289195600604,
+      "grad_norm": 0.20873045921325684,
+      "learning_rate": 6.352402694848871e-06,
+      "loss": 2.236,
+      "step": 6044
+    },
+    {
+      "epoch": 2.607720508949752,
+      "grad_norm": 0.20056168735027313,
+      "learning_rate": 6.338664089066492e-06,
+      "loss": 2.3545,
+      "step": 6045
+    },
+    {
+      "epoch": 2.6081518222989004,
+      "grad_norm": 0.19856448471546173,
+      "learning_rate": 6.324939700364584e-06,
+      "loss": 2.1218,
+      "step": 6046
+    },
+    {
+      "epoch": 2.6085831356480482,
+      "grad_norm": 0.1984233558177948,
+      "learning_rate": 6.311229531584924e-06,
+      "loss": 2.1644,
+      "step": 6047
+    },
+    {
+      "epoch": 2.6090144489971965,
+      "grad_norm": 0.2085246741771698,
+      "learning_rate": 6.297533585566358e-06,
+      "loss": 1.9952,
+      "step": 6048
+    },
+    {
+      "epoch": 2.6094457623463447,
+      "grad_norm": 0.19457945227622986,
+      "learning_rate": 6.2838518651447845e-06,
+      "loss": 2.0665,
+      "step": 6049
+    },
+    {
+      "epoch": 2.6098770756954925,
+      "grad_norm": 0.20151954889297485,
+      "learning_rate": 6.270184373153153e-06,
+      "loss": 2.1981,
+      "step": 6050
+    },
+    {
+      "epoch": 2.6098770756954925,
+      "eval_loss": 2.0872039794921875,
+      "eval_runtime": 207.3196,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 6050
+    },
+    {
+      "epoch": 2.610308389044641,
+      "grad_norm": 0.2253008335828781,
+      "learning_rate": 6.256531112421459e-06,
+      "loss": 2.1381,
+      "step": 6051
+    },
+    {
+      "epoch": 2.610739702393789,
+      "grad_norm": 0.2223806083202362,
+      "learning_rate": 6.242892085776749e-06,
+      "loss": 2.1014,
+      "step": 6052
+    },
+    {
+      "epoch": 2.6111710157429373,
+      "grad_norm": 0.1943701058626175,
+      "learning_rate": 6.229267296043172e-06,
+      "loss": 1.8895,
+      "step": 6053
+    },
+    {
+      "epoch": 2.6116023290920856,
+      "grad_norm": 0.20236046612262726,
+      "learning_rate": 6.215656746041855e-06,
+      "loss": 2.3317,
+      "step": 6054
+    },
+    {
+      "epoch": 2.612033642441234,
+      "grad_norm": 0.2128019779920578,
+      "learning_rate": 6.2020604385910115e-06,
+      "loss": 2.4217,
+      "step": 6055
+    },
+    {
+      "epoch": 2.6124649557903816,
+      "grad_norm": 0.1914263665676117,
+      "learning_rate": 6.188478376505929e-06,
+      "loss": 2.1676,
+      "step": 6056
+    },
+    {
+      "epoch": 2.61289626913953,
+      "grad_norm": 0.2062346190214157,
+      "learning_rate": 6.174910562598909e-06,
+      "loss": 2.1603,
+      "step": 6057
+    },
+    {
+      "epoch": 2.613327582488678,
+      "grad_norm": 0.19567060470581055,
+      "learning_rate": 6.161356999679321e-06,
+      "loss": 1.9346,
+      "step": 6058
+    },
+    {
+      "epoch": 2.613758895837826,
+      "grad_norm": 0.19446727633476257,
+      "learning_rate": 6.147817690553552e-06,
+      "loss": 2.1298,
+      "step": 6059
+    },
+    {
+      "epoch": 2.614190209186974,
+      "grad_norm": 0.2038094699382782,
+      "learning_rate": 6.1342926380250955e-06,
+      "loss": 2.2005,
+      "step": 6060
+    },
+    {
+      "epoch": 2.6146215225361225,
+      "grad_norm": 0.21893729269504547,
+      "learning_rate": 6.12078184489446e-06,
+      "loss": 1.9707,
+      "step": 6061
+    },
+    {
+      "epoch": 2.6150528358852707,
+      "grad_norm": 0.19223694503307343,
+      "learning_rate": 6.107285313959181e-06,
+      "loss": 2.085,
+      "step": 6062
+    },
+    {
+      "epoch": 2.615484149234419,
+      "grad_norm": 0.20330168306827545,
+      "learning_rate": 6.093803048013885e-06,
+      "loss": 2.2486,
+      "step": 6063
+    },
+    {
+      "epoch": 2.615915462583567,
+      "grad_norm": 0.1966131627559662,
+      "learning_rate": 6.08033504985021e-06,
+      "loss": 2.2113,
+      "step": 6064
+    },
+    {
+      "epoch": 2.616346775932715,
+      "grad_norm": 0.20934173464775085,
+      "learning_rate": 6.066881322256853e-06,
+      "loss": 2.0675,
+      "step": 6065
+    },
+    {
+      "epoch": 2.6167780892818633,
+      "grad_norm": 0.19852526485919952,
+      "learning_rate": 6.053441868019551e-06,
+      "loss": 2.1583,
+      "step": 6066
+    },
+    {
+      "epoch": 2.6172094026310115,
+      "grad_norm": 0.2178497314453125,
+      "learning_rate": 6.040016689921121e-06,
+      "loss": 2.054,
+      "step": 6067
+    },
+    {
+      "epoch": 2.6176407159801593,
+      "grad_norm": 0.2043694406747818,
+      "learning_rate": 6.026605790741354e-06,
+      "loss": 2.245,
+      "step": 6068
+    },
+    {
+      "epoch": 2.6180720293293076,
+      "grad_norm": 0.23877695202827454,
+      "learning_rate": 6.0132091732571385e-06,
+      "loss": 1.8492,
+      "step": 6069
+    },
+    {
+      "epoch": 2.618503342678456,
+      "grad_norm": 0.1892527937889099,
+      "learning_rate": 5.999826840242372e-06,
+      "loss": 2.038,
+      "step": 6070
+    },
+    {
+      "epoch": 2.618934656027604,
+      "grad_norm": 0.2118130773305893,
+      "learning_rate": 5.986458794468055e-06,
+      "loss": 2.2015,
+      "step": 6071
+    },
+    {
+      "epoch": 2.6193659693767524,
+      "grad_norm": 0.20288275182247162,
+      "learning_rate": 5.973105038702172e-06,
+      "loss": 2.3318,
+      "step": 6072
+    },
+    {
+      "epoch": 2.6197972827259006,
+      "grad_norm": 0.21088698506355286,
+      "learning_rate": 5.959765575709729e-06,
+      "loss": 2.1951,
+      "step": 6073
+    },
+    {
+      "epoch": 2.6202285960750484,
+      "grad_norm": 0.201625794172287,
+      "learning_rate": 5.946440408252851e-06,
+      "loss": 2.1579,
+      "step": 6074
+    },
+    {
+      "epoch": 2.6206599094241967,
+      "grad_norm": 0.19988271594047546,
+      "learning_rate": 5.9331295390906364e-06,
+      "loss": 1.9267,
+      "step": 6075
+    },
+    {
+      "epoch": 2.6206599094241967,
+      "eval_loss": 2.0870893001556396,
+      "eval_runtime": 206.5147,
+      "eval_samples_per_second": 0.155,
+      "eval_steps_per_second": 0.155,
+      "step": 6075
+    },
+    {
+      "epoch": 2.621091222773345,
+      "grad_norm": 0.2041642665863037,
+      "learning_rate": 5.919832970979265e-06,
+      "loss": 2.2081,
+      "step": 6076
+    },
+    {
+      "epoch": 2.6215225361224928,
+      "grad_norm": 0.18532341718673706,
+      "learning_rate": 5.90655070667192e-06,
+      "loss": 2.3406,
+      "step": 6077
+    },
+    {
+      "epoch": 2.621953849471641,
+      "grad_norm": 0.1991955190896988,
+      "learning_rate": 5.89328274891884e-06,
+      "loss": 2.1091,
+      "step": 6078
+    },
+    {
+      "epoch": 2.6223851628207893,
+      "grad_norm": 0.21103163063526154,
+      "learning_rate": 5.8800291004673036e-06,
+      "loss": 2.0537,
+      "step": 6079
+    },
+    {
+      "epoch": 2.6228164761699375,
+      "grad_norm": 0.18268458545207977,
+      "learning_rate": 5.866789764061617e-06,
+      "loss": 1.9784,
+      "step": 6080
+    },
+    {
+      "epoch": 2.6232477895190858,
+      "grad_norm": 0.4187362492084503,
+      "learning_rate": 5.853564742443167e-06,
+      "loss": 2.0681,
+      "step": 6081
+    },
+    {
+      "epoch": 2.623679102868234,
+      "grad_norm": 0.21114923059940338,
+      "learning_rate": 5.8403540383502925e-06,
+      "loss": 2.072,
+      "step": 6082
+    },
+    {
+      "epoch": 2.624110416217382,
+      "grad_norm": 0.1981240063905716,
+      "learning_rate": 5.827157654518427e-06,
+      "loss": 2.131,
+      "step": 6083
+    },
+    {
+      "epoch": 2.62454172956653,
+      "grad_norm": 0.21558067202568054,
+      "learning_rate": 5.813975593680029e-06,
+      "loss": 2.1768,
+      "step": 6084
+    },
+    {
+      "epoch": 2.6249730429156783,
+      "grad_norm": 0.1967075616121292,
+      "learning_rate": 5.8008078585645965e-06,
+      "loss": 2.2239,
+      "step": 6085
+    },
+    {
+      "epoch": 2.625404356264826,
+      "grad_norm": 0.2138763815164566,
+      "learning_rate": 5.787654451898658e-06,
+      "loss": 2.0824,
+      "step": 6086
+    },
+    {
+      "epoch": 2.6258356696139744,
+      "grad_norm": 0.20435772836208344,
+      "learning_rate": 5.774515376405733e-06,
+      "loss": 2.2836,
+      "step": 6087
+    },
+    {
+      "epoch": 2.6262669829631227,
+      "grad_norm": 0.1953764408826828,
+      "learning_rate": 5.761390634806448e-06,
+      "loss": 2.1566,
+      "step": 6088
+    },
+    {
+      "epoch": 2.626698296312271,
+      "grad_norm": 0.22136853635311127,
+      "learning_rate": 5.748280229818419e-06,
+      "loss": 2.1489,
+      "step": 6089
+    },
+    {
+      "epoch": 2.627129609661419,
+      "grad_norm": 0.22649671137332916,
+      "learning_rate": 5.735184164156284e-06,
+      "loss": 2.0366,
+      "step": 6090
+    },
+    {
+      "epoch": 2.6275609230105674,
+      "grad_norm": 0.21080105006694794,
+      "learning_rate": 5.7221024405317395e-06,
+      "loss": 2.2014,
+      "step": 6091
+    },
+    {
+      "epoch": 2.6279922363597152,
+      "grad_norm": 0.180626779794693,
+      "learning_rate": 5.709035061653494e-06,
+      "loss": 2.1699,
+      "step": 6092
+    },
+    {
+      "epoch": 2.6284235497088635,
+      "grad_norm": 0.191581591963768,
+      "learning_rate": 5.6959820302272916e-06,
+      "loss": 2.0196,
+      "step": 6093
+    },
+    {
+      "epoch": 2.6288548630580117,
+      "grad_norm": 0.20116713643074036,
+      "learning_rate": 5.682943348955913e-06,
+      "loss": 2.0793,
+      "step": 6094
+    },
+    {
+      "epoch": 2.6292861764071596,
+      "grad_norm": 0.19673985242843628,
+      "learning_rate": 5.669919020539146e-06,
+      "loss": 1.968,
+      "step": 6095
+    },
+    {
+      "epoch": 2.629717489756308,
+      "grad_norm": 0.18929420411586761,
+      "learning_rate": 5.656909047673827e-06,
+      "loss": 2.1299,
+      "step": 6096
+    },
+    {
+      "epoch": 2.630148803105456,
+      "grad_norm": 0.20557215809822083,
+      "learning_rate": 5.64391343305381e-06,
+      "loss": 2.0828,
+      "step": 6097
+    },
+    {
+      "epoch": 2.6305801164546043,
+      "grad_norm": 0.19969946146011353,
+      "learning_rate": 5.6309321793699735e-06,
+      "loss": 2.2417,
+      "step": 6098
+    },
+    {
+      "epoch": 2.6310114298037526,
+      "grad_norm": 0.21544784307479858,
+      "learning_rate": 5.617965289310261e-06,
+      "loss": 2.1295,
+      "step": 6099
+    },
+    {
+      "epoch": 2.631442743152901,
+      "grad_norm": 0.20157448947429657,
+      "learning_rate": 5.605012765559575e-06,
+      "loss": 2.1707,
+      "step": 6100
+    },
+    {
+      "epoch": 2.631442743152901,
+      "eval_loss": 2.0871706008911133,
+      "eval_runtime": 204.4249,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 6100
+    },
+    {
+      "epoch": 2.6318740565020486,
+      "grad_norm": 0.19672569632530212,
+      "learning_rate": 5.592074610799868e-06,
+      "loss": 2.2779,
+      "step": 6101
+    },
+    {
+      "epoch": 2.632305369851197,
+      "grad_norm": 0.21693246066570282,
+      "learning_rate": 5.579150827710158e-06,
+      "loss": 2.2466,
+      "step": 6102
+    },
+    {
+      "epoch": 2.632736683200345,
+      "grad_norm": 0.28595659136772156,
+      "learning_rate": 5.566241418966452e-06,
+      "loss": 2.1764,
+      "step": 6103
+    },
+    {
+      "epoch": 2.633167996549493,
+      "grad_norm": 0.19356441497802734,
+      "learning_rate": 5.55334638724178e-06,
+      "loss": 2.4061,
+      "step": 6104
+    },
+    {
+      "epoch": 2.633599309898641,
+      "grad_norm": 0.19937746226787567,
+      "learning_rate": 5.540465735206165e-06,
+      "loss": 2.0587,
+      "step": 6105
+    },
+    {
+      "epoch": 2.6340306232477895,
+      "grad_norm": 0.17437097430229187,
+      "learning_rate": 5.527599465526739e-06,
+      "loss": 1.831,
+      "step": 6106
+    },
+    {
+      "epoch": 2.6344619365969377,
+      "grad_norm": 0.20675235986709595,
+      "learning_rate": 5.51474758086757e-06,
+      "loss": 2.1508,
+      "step": 6107
+    },
+    {
+      "epoch": 2.634893249946086,
+      "grad_norm": 0.19855298101902008,
+      "learning_rate": 5.501910083889799e-06,
+      "loss": 2.1318,
+      "step": 6108
+    },
+    {
+      "epoch": 2.6353245632952342,
+      "grad_norm": 0.20525705814361572,
+      "learning_rate": 5.489086977251564e-06,
+      "loss": 2.1458,
+      "step": 6109
+    },
+    {
+      "epoch": 2.635755876644382,
+      "grad_norm": 0.20039616525173187,
+      "learning_rate": 5.476278263608025e-06,
+      "loss": 2.3108,
+      "step": 6110
+    },
+    {
+      "epoch": 2.6361871899935303,
+      "grad_norm": 0.19344262778759003,
+      "learning_rate": 5.463483945611377e-06,
+      "loss": 2.1225,
+      "step": 6111
+    },
+    {
+      "epoch": 2.6366185033426786,
+      "grad_norm": 0.19685900211334229,
+      "learning_rate": 5.450704025910793e-06,
+      "loss": 2.1575,
+      "step": 6112
+    },
+    {
+      "epoch": 2.6370498166918264,
+      "grad_norm": 0.20429310202598572,
+      "learning_rate": 5.437938507152556e-06,
+      "loss": 2.0551,
+      "step": 6113
+    },
+    {
+      "epoch": 2.6374811300409746,
+      "grad_norm": 0.195399209856987,
+      "learning_rate": 5.425187391979852e-06,
+      "loss": 2.1962,
+      "step": 6114
+    },
+    {
+      "epoch": 2.637912443390123,
+      "grad_norm": 0.21501316130161285,
+      "learning_rate": 5.4124506830329345e-06,
+      "loss": 2.1833,
+      "step": 6115
+    },
+    {
+      "epoch": 2.638343756739271,
+      "grad_norm": 0.27056029438972473,
+      "learning_rate": 5.39972838294912e-06,
+      "loss": 2.1472,
+      "step": 6116
+    },
+    {
+      "epoch": 2.6387750700884194,
+      "grad_norm": 0.2011452466249466,
+      "learning_rate": 5.3870204943626845e-06,
+      "loss": 2.167,
+      "step": 6117
+    },
+    {
+      "epoch": 2.6392063834375676,
+      "grad_norm": 0.22378626465797424,
+      "learning_rate": 5.374327019904931e-06,
+      "loss": 2.1392,
+      "step": 6118
+    },
+    {
+      "epoch": 2.6396376967867154,
+      "grad_norm": 0.23554293811321259,
+      "learning_rate": 5.361647962204166e-06,
+      "loss": 2.0103,
+      "step": 6119
+    },
+    {
+      "epoch": 2.6400690101358637,
+      "grad_norm": 0.205733984708786,
+      "learning_rate": 5.348983323885756e-06,
+      "loss": 2.1259,
+      "step": 6120
+    },
+    {
+      "epoch": 2.640500323485012,
+      "grad_norm": 0.2434549480676651,
+      "learning_rate": 5.336333107572028e-06,
+      "loss": 2.1261,
+      "step": 6121
+    },
+    {
+      "epoch": 2.6409316368341598,
+      "grad_norm": 0.2025012969970703,
+      "learning_rate": 5.3236973158823695e-06,
+      "loss": 2.1764,
+      "step": 6122
+    },
+    {
+      "epoch": 2.641362950183308,
+      "grad_norm": 0.195558562874794,
+      "learning_rate": 5.311075951433147e-06,
+      "loss": 1.8611,
+      "step": 6123
+    },
+    {
+      "epoch": 2.6417942635324563,
+      "grad_norm": 0.19787761569023132,
+      "learning_rate": 5.298469016837753e-06,
+      "loss": 2.2368,
+      "step": 6124
+    },
+    {
+      "epoch": 2.6422255768816045,
+      "grad_norm": 0.19688139855861664,
+      "learning_rate": 5.285876514706591e-06,
+      "loss": 2.3121,
+      "step": 6125
+    },
+    {
+      "epoch": 2.6422255768816045,
+      "eval_loss": 2.087151527404785,
+      "eval_runtime": 204.0015,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 6125
+    },
+    {
+      "epoch": 2.642656890230753,
+      "grad_norm": 0.2052869200706482,
+      "learning_rate": 5.273298447647067e-06,
+      "loss": 2.0919,
+      "step": 6126
+    },
+    {
+      "epoch": 2.643088203579901,
+      "grad_norm": 0.21629759669303894,
+      "learning_rate": 5.26073481826364e-06,
+      "loss": 2.054,
+      "step": 6127
+    },
+    {
+      "epoch": 2.643519516929049,
+      "grad_norm": 0.17514556646347046,
+      "learning_rate": 5.248185629157711e-06,
+      "loss": 1.8445,
+      "step": 6128
+    },
+    {
+      "epoch": 2.643950830278197,
+      "grad_norm": 0.21411702036857605,
+      "learning_rate": 5.235650882927744e-06,
+      "loss": 1.6976,
+      "step": 6129
+    },
+    {
+      "epoch": 2.6443821436273454,
+      "grad_norm": 0.194255530834198,
+      "learning_rate": 5.22313058216918e-06,
+      "loss": 2.1619,
+      "step": 6130
+    },
+    {
+      "epoch": 2.644813456976493,
+      "grad_norm": 0.1894819736480713,
+      "learning_rate": 5.210624729474503e-06,
+      "loss": 2.2686,
+      "step": 6131
+    },
+    {
+      "epoch": 2.6452447703256414,
+      "grad_norm": 0.21412064135074615,
+      "learning_rate": 5.198133327433199e-06,
+      "loss": 2.2112,
+      "step": 6132
+    },
+    {
+      "epoch": 2.6456760836747897,
+      "grad_norm": 0.20222002267837524,
+      "learning_rate": 5.1856563786317e-06,
+      "loss": 2.128,
+      "step": 6133
+    },
+    {
+      "epoch": 2.646107397023938,
+      "grad_norm": 0.18645766377449036,
+      "learning_rate": 5.17319388565354e-06,
+      "loss": 2.1181,
+      "step": 6134
+    },
+    {
+      "epoch": 2.646538710373086,
+      "grad_norm": 0.19865809381008148,
+      "learning_rate": 5.160745851079196e-06,
+      "loss": 2.1027,
+      "step": 6135
+    },
+    {
+      "epoch": 2.6469700237222344,
+      "grad_norm": 0.8640784025192261,
+      "learning_rate": 5.148312277486172e-06,
+      "loss": 1.9384,
+      "step": 6136
+    },
+    {
+      "epoch": 2.6474013370713823,
+      "grad_norm": 0.22291713953018188,
+      "learning_rate": 5.135893167448976e-06,
+      "loss": 2.3333,
+      "step": 6137
+    },
+    {
+      "epoch": 2.6478326504205305,
+      "grad_norm": 0.22736655175685883,
+      "learning_rate": 5.1234885235391264e-06,
+      "loss": 2.0239,
+      "step": 6138
+    },
+    {
+      "epoch": 2.6482639637696788,
+      "grad_norm": 0.20177805423736572,
+      "learning_rate": 5.111098348325126e-06,
+      "loss": 2.3397,
+      "step": 6139
+    },
+    {
+      "epoch": 2.6486952771188266,
+      "grad_norm": 0.2055068165063858,
+      "learning_rate": 5.098722644372516e-06,
+      "loss": 2.1454,
+      "step": 6140
+    },
+    {
+      "epoch": 2.649126590467975,
+      "grad_norm": 0.20407767593860626,
+      "learning_rate": 5.086361414243803e-06,
+      "loss": 2.2067,
+      "step": 6141
+    },
+    {
+      "epoch": 2.649557903817123,
+      "grad_norm": 0.20581407845020294,
+      "learning_rate": 5.074014660498524e-06,
+      "loss": 2.107,
+      "step": 6142
+    },
+    {
+      "epoch": 2.6499892171662713,
+      "grad_norm": 0.2167045921087265,
+      "learning_rate": 5.0616823856932e-06,
+      "loss": 2.2216,
+      "step": 6143
+    },
+    {
+      "epoch": 2.6504205305154196,
+      "grad_norm": 0.21492063999176025,
+      "learning_rate": 5.049364592381364e-06,
+      "loss": 2.2001,
+      "step": 6144
+    },
+    {
+      "epoch": 2.650851843864568,
+      "grad_norm": 0.19565856456756592,
+      "learning_rate": 5.037061283113583e-06,
+      "loss": 2.0096,
+      "step": 6145
+    },
+    {
+      "epoch": 2.6512831572137157,
+      "grad_norm": 0.19109144806861877,
+      "learning_rate": 5.024772460437345e-06,
+      "loss": 2.0656,
+      "step": 6146
+    },
+    {
+      "epoch": 2.651714470562864,
+      "grad_norm": 0.2673475742340088,
+      "learning_rate": 5.012498126897199e-06,
+      "loss": 2.132,
+      "step": 6147
+    },
+    {
+      "epoch": 2.652145783912012,
+      "grad_norm": 0.19297978281974792,
+      "learning_rate": 5.000238285034702e-06,
+      "loss": 2.0913,
+      "step": 6148
+    },
+    {
+      "epoch": 2.65257709726116,
+      "grad_norm": 0.21174772083759308,
+      "learning_rate": 4.987992937388366e-06,
+      "loss": 2.1064,
+      "step": 6149
+    },
+    {
+      "epoch": 2.6530084106103082,
+      "grad_norm": 0.19300834834575653,
+      "learning_rate": 4.9757620864937454e-06,
+      "loss": 2.1357,
+      "step": 6150
+    },
+    {
+      "epoch": 2.6530084106103082,
+      "eval_loss": 2.087246894836426,
+      "eval_runtime": 204.0893,
+      "eval_samples_per_second": 0.157,
+      "eval_steps_per_second": 0.157,
+      "step": 6150
+    },
+    {
+      "epoch": 2.6534397239594565,
+      "grad_norm": 0.18600746989250183,
+      "learning_rate": 4.963545734883331e-06,
+      "loss": 2.1089,
+      "step": 6151
+    },
+    {
+      "epoch": 2.6538710373086047,
+      "grad_norm": 0.1895558387041092,
+      "learning_rate": 4.951343885086692e-06,
+      "loss": 2.1419,
+      "step": 6152
+    },
+    {
+      "epoch": 2.654302350657753,
+      "grad_norm": 0.21031838655471802,
+      "learning_rate": 4.93915653963034e-06,
+      "loss": 2.2392,
+      "step": 6153
+    },
+    {
+      "epoch": 2.6547336640069013,
+      "grad_norm": 0.2003740668296814,
+      "learning_rate": 4.926983701037807e-06,
+      "loss": 2.0949,
+      "step": 6154
+    },
+    {
+      "epoch": 2.655164977356049,
+      "grad_norm": 0.19085167348384857,
+      "learning_rate": 4.914825371829592e-06,
+      "loss": 2.3022,
+      "step": 6155
+    },
+    {
+      "epoch": 2.6555962907051973,
+      "grad_norm": 0.1863519549369812,
+      "learning_rate": 4.902681554523224e-06,
+      "loss": 2.0893,
+      "step": 6156
+    },
+    {
+      "epoch": 2.6560276040543456,
+      "grad_norm": 0.21060968935489655,
+      "learning_rate": 4.890552251633217e-06,
+      "loss": 2.0933,
+      "step": 6157
+    },
+    {
+      "epoch": 2.6564589174034934,
+      "grad_norm": 0.18363919854164124,
+      "learning_rate": 4.878437465671054e-06,
+      "loss": 2.1239,
+      "step": 6158
+    },
+    {
+      "epoch": 2.6568902307526416,
+      "grad_norm": 0.17458274960517883,
+      "learning_rate": 4.866337199145276e-06,
+      "loss": 1.9382,
+      "step": 6159
+    },
+    {
+      "epoch": 2.65732154410179,
+      "grad_norm": 0.1946101188659668,
+      "learning_rate": 4.8542514545613395e-06,
+      "loss": 2.0561,
+      "step": 6160
+    },
+    {
+      "epoch": 2.657752857450938,
+      "grad_norm": 0.2080899327993393,
+      "learning_rate": 4.842180234421725e-06,
+      "loss": 2.1089,
+      "step": 6161
+    },
+    {
+      "epoch": 2.6581841708000864,
+      "grad_norm": 0.18995599448680878,
+      "learning_rate": 4.830123541225941e-06,
+      "loss": 2.0242,
+      "step": 6162
+    },
+    {
+      "epoch": 2.6586154841492347,
+      "grad_norm": 0.20470154285430908,
+      "learning_rate": 4.818081377470442e-06,
+      "loss": 2.0086,
+      "step": 6163
+    },
+    {
+      "epoch": 2.6590467974983825,
+      "grad_norm": 0.20341189205646515,
+      "learning_rate": 4.8060537456487e-06,
+      "loss": 2.1157,
+      "step": 6164
+    },
+    {
+      "epoch": 2.6594781108475307,
+      "grad_norm": 0.18980471789836884,
+      "learning_rate": 4.794040648251129e-06,
+      "loss": 2.0815,
+      "step": 6165
+    },
+    {
+      "epoch": 2.659909424196679,
+      "grad_norm": 0.2044554501771927,
+      "learning_rate": 4.782042087765226e-06,
+      "loss": 2.1729,
+      "step": 6166
+    },
+    {
+      "epoch": 2.660340737545827,
+      "grad_norm": 0.19274021685123444,
+      "learning_rate": 4.7700580666753915e-06,
+      "loss": 2.1044,
+      "step": 6167
+    },
+    {
+      "epoch": 2.660772050894975,
+      "grad_norm": 0.19175758957862854,
+      "learning_rate": 4.758088587463052e-06,
+      "loss": 2.0502,
+      "step": 6168
+    },
+    {
+      "epoch": 2.6612033642441233,
+      "grad_norm": 0.20993196964263916,
+      "learning_rate": 4.746133652606632e-06,
+      "loss": 2.2449,
+      "step": 6169
+    },
+    {
+      "epoch": 2.6616346775932715,
+      "grad_norm": 0.19075722992420197,
+      "learning_rate": 4.734193264581518e-06,
+      "loss": 2.0345,
+      "step": 6170
+    },
+    {
+      "epoch": 2.66206599094242,
+      "grad_norm": 0.20163051784038544,
+      "learning_rate": 4.722267425860099e-06,
+      "loss": 2.1201,
+      "step": 6171
+    },
+    {
+      "epoch": 2.662497304291568,
+      "grad_norm": 0.20374572277069092,
+      "learning_rate": 4.710356138911759e-06,
+      "loss": 2.2216,
+      "step": 6172
+    },
+    {
+      "epoch": 2.662928617640716,
+      "grad_norm": 0.20015719532966614,
+      "learning_rate": 4.698459406202856e-06,
+      "loss": 2.1439,
+      "step": 6173
+    },
+    {
+      "epoch": 2.663359930989864,
+      "grad_norm": 0.19572551548480988,
+      "learning_rate": 4.686577230196731e-06,
+      "loss": 2.2754,
+      "step": 6174
+    },
+    {
+      "epoch": 2.6637912443390124,
+      "grad_norm": 0.19161763787269592,
+      "learning_rate": 4.674709613353731e-06,
+      "loss": 1.8648,
+      "step": 6175
+    },
+    {
+      "epoch": 2.6637912443390124,
+      "eval_loss": 2.0872437953948975,
+      "eval_runtime": 204.8946,
+      "eval_samples_per_second": 0.156,
+      "eval_steps_per_second": 0.156,
+      "step": 6175
+    },
+    {
+      "epoch": 2.66422255768816,
+      "grad_norm": 0.21940241754055023,
+      "learning_rate": 4.662856558131156e-06,
+      "loss": 2.047,
+      "step": 6176
+    },
+    {
+      "epoch": 2.6646538710373084,
+      "grad_norm": 0.20560306310653687,
+      "learning_rate": 4.651018066983353e-06,
+      "loss": 2.0876,
+      "step": 6177
+    },
+    {
+      "epoch": 2.6650851843864567,
+      "grad_norm": 0.2093990594148636,
+      "learning_rate": 4.63919414236156e-06,
+      "loss": 2.385,
+      "step": 6178
+    },
+    {
+      "epoch": 2.665516497735605,
+      "grad_norm": 0.2323451191186905,
+      "learning_rate": 4.6273847867140696e-06,
+      "loss": 2.0265,
+      "step": 6179
+    },
+    {
+      "epoch": 2.665947811084753,
+      "grad_norm": 0.18504031002521515,
+      "learning_rate": 4.61559000248615e-06,
+      "loss": 2.0652,
+      "step": 6180
+    },
+    {
+      "epoch": 2.6663791244339015,
+      "grad_norm": 0.1730000376701355,
+      "learning_rate": 4.60380979212003e-06,
+      "loss": 1.9666,
+      "step": 6181
+    },
+    {
+      "epoch": 2.6668104377830493,
+      "grad_norm": 0.1883518546819687,
+      "learning_rate": 4.592044158054953e-06,
+      "loss": 2.0666,
+      "step": 6182
+    },
+    {
+      "epoch": 2.6672417511321975,
+      "grad_norm": 0.21099349856376648,
+      "learning_rate": 4.5802931027270674e-06,
+      "loss": 2.1527,
+      "step": 6183
+    },
+    {
+      "epoch": 2.667673064481346,
+      "grad_norm": 0.21935242414474487,
+      "learning_rate": 4.568556628569603e-06,
+      "loss": 2.1865,
+      "step": 6184
+    },
+    {
+      "epoch": 2.6681043778304936,
+      "grad_norm": 0.17997373640537262,
+      "learning_rate": 4.556834738012724e-06,
+      "loss": 2.0517,
+      "step": 6185
+    },
+    {
+      "epoch": 2.668535691179642,
+      "grad_norm": 0.19894824922084808,
+      "learning_rate": 4.545127433483556e-06,
+      "loss": 2.0752,
+      "step": 6186
+    },
+    {
+      "epoch": 2.66896700452879,
+      "grad_norm": 0.24281533062458038,
+      "learning_rate": 4.533434717406234e-06,
+      "loss": 1.9051,
+      "step": 6187
+    },
+    {
+      "epoch": 2.6693983178779384,
+      "grad_norm": 0.19154304265975952,
+      "learning_rate": 4.521756592201872e-06,
+      "loss": 2.1348,
+      "step": 6188
+    },
+    {
+      "epoch": 2.6698296312270866,
+      "grad_norm": 0.20902879536151886,
+      "learning_rate": 4.510093060288536e-06,
+      "loss": 2.1539,
+      "step": 6189
+    },
+    {
+      "epoch": 2.670260944576235,
+      "grad_norm": 0.19530873000621796,
+      "learning_rate": 4.4984441240812935e-06,
+      "loss": 2.0056,
+      "step": 6190
+    },
+    {
+      "epoch": 2.6706922579253827,
+      "grad_norm": 0.19316688179969788,
+      "learning_rate": 4.486809785992207e-06,
+      "loss": 2.1989,
+      "step": 6191
+    },
+    {
+      "epoch": 2.671123571274531,
+      "grad_norm": 0.19766834378242493,
+      "learning_rate": 4.4751900484302574e-06,
+      "loss": 2.2485,
+      "step": 6192
+    },
+    {
+      "epoch": 2.671554884623679,
+      "grad_norm": 0.23230670392513275,
+      "learning_rate": 4.463584913801446e-06,
+      "loss": 1.7903,
+      "step": 6193
+    },
+    {
+      "epoch": 2.671986197972827,
+      "grad_norm": 0.1969883143901825,
+      "learning_rate": 4.451994384508758e-06,
+      "loss": 2.2623,
+      "step": 6194
+    },
+    {
+      "epoch": 2.6724175113219752,
+      "grad_norm": 0.1922171413898468,
+      "learning_rate": 4.440418462952141e-06,
+      "loss": 2.2303,
+      "step": 6195
+    },
+    {
+      "epoch": 2.6728488246711235,
+      "grad_norm": 0.20559678971767426,
+      "learning_rate": 4.42885715152852e-06,
+      "loss": 2.2323,
+      "step": 6196
+    },
+    {
+      "epoch": 2.6732801380202718,
+      "grad_norm": 0.18253296613693237,
+      "learning_rate": 4.4173104526317445e-06,
+      "loss": 2.0705,
+      "step": 6197
+    },
+    {
+      "epoch": 2.67371145136942,
+      "grad_norm": 0.20601263642311096,
+      "learning_rate": 4.405778368652738e-06,
+      "loss": 1.9288,
+      "step": 6198
+    },
+    {
+      "epoch": 2.6741427647185683,
+      "grad_norm": 0.2276991456747055,
+      "learning_rate": 4.394260901979329e-06,
+      "loss": 2.2312,
+      "step": 6199
+    },
+    {
+      "epoch": 2.674574078067716,
+      "grad_norm": 0.18565404415130615,
+      "learning_rate": 4.38275805499633e-06,
+      "loss": 1.9813,
+      "step": 6200
+    },
+    {
+      "epoch": 2.674574078067716,
+      "eval_loss": 2.0872318744659424,
+      "eval_runtime": 201.4551,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 6200
+    },
+    {
+      "epoch": 2.6750053914168643,
+      "grad_norm": 0.18837133049964905,
+      "learning_rate": 4.371269830085525e-06,
+      "loss": 1.9287,
+      "step": 6201
+    },
+    {
+      "epoch": 2.6754367047660126,
+      "grad_norm": 0.20320770144462585,
+      "learning_rate": 4.359796229625695e-06,
+      "loss": 1.988,
+      "step": 6202
+    },
+    {
+      "epoch": 2.6758680181151604,
+      "grad_norm": 0.19712461531162262,
+      "learning_rate": 4.348337255992556e-06,
+      "loss": 2.0474,
+      "step": 6203
+    },
+    {
+      "epoch": 2.6762993314643087,
+      "grad_norm": 0.21028605103492737,
+      "learning_rate": 4.3368929115588115e-06,
+      "loss": 2.1066,
+      "step": 6204
+    },
+    {
+      "epoch": 2.676730644813457,
+      "grad_norm": 0.18788939714431763,
+      "learning_rate": 4.325463198694173e-06,
+      "loss": 2.1886,
+      "step": 6205
+    },
+    {
+      "epoch": 2.677161958162605,
+      "grad_norm": 0.20289573073387146,
+      "learning_rate": 4.314048119765248e-06,
+      "loss": 2.0151,
+      "step": 6206
+    },
+    {
+      "epoch": 2.6775932715117534,
+      "grad_norm": 0.2069668173789978,
+      "learning_rate": 4.302647677135662e-06,
+      "loss": 2.3451,
+      "step": 6207
+    },
+    {
+      "epoch": 2.6780245848609017,
+      "grad_norm": 0.19542808830738068,
+      "learning_rate": 4.2912618731660184e-06,
+      "loss": 2.0026,
+      "step": 6208
+    },
+    {
+      "epoch": 2.6784558982100495,
+      "grad_norm": 0.20399220287799835,
+      "learning_rate": 4.279890710213857e-06,
+      "loss": 2.2483,
+      "step": 6209
+    },
+    {
+      "epoch": 2.6788872115591977,
+      "grad_norm": 0.20325656235218048,
+      "learning_rate": 4.268534190633727e-06,
+      "loss": 2.1639,
+      "step": 6210
+    },
+    {
+      "epoch": 2.679318524908346,
+      "grad_norm": 0.18592952191829681,
+      "learning_rate": 4.257192316777064e-06,
+      "loss": 1.9605,
+      "step": 6211
+    },
+    {
+      "epoch": 2.679749838257494,
+      "grad_norm": 0.19579529762268066,
+      "learning_rate": 4.245865090992382e-06,
+      "loss": 2.1996,
+      "step": 6212
+    },
+    {
+      "epoch": 2.680181151606642,
+      "grad_norm": 0.2062699943780899,
+      "learning_rate": 4.2345525156250865e-06,
+      "loss": 2.1259,
+      "step": 6213
+    },
+    {
+      "epoch": 2.6806124649557903,
+      "grad_norm": 0.19597366452217102,
+      "learning_rate": 4.223254593017564e-06,
+      "loss": 2.066,
+      "step": 6214
+    },
+    {
+      "epoch": 2.6810437783049386,
+      "grad_norm": 0.22171823680400848,
+      "learning_rate": 4.211971325509183e-06,
+      "loss": 1.9581,
+      "step": 6215
+    },
+    {
+      "epoch": 2.681475091654087,
+      "grad_norm": 0.18330083787441254,
+      "learning_rate": 4.200702715436269e-06,
+      "loss": 1.9926,
+      "step": 6216
+    },
+    {
+      "epoch": 2.681906405003235,
+      "grad_norm": 0.20268268883228302,
+      "learning_rate": 4.189448765132103e-06,
+      "loss": 2.1025,
+      "step": 6217
+    },
+    {
+      "epoch": 2.682337718352383,
+      "grad_norm": 0.18285948038101196,
+      "learning_rate": 4.178209476926939e-06,
+      "loss": 1.8616,
+      "step": 6218
+    },
+    {
+      "epoch": 2.682769031701531,
+      "grad_norm": 0.2060435712337494,
+      "learning_rate": 4.166984853147998e-06,
+      "loss": 2.2504,
+      "step": 6219
+    },
+    {
+      "epoch": 2.6832003450506794,
+      "grad_norm": 0.19653423130512238,
+      "learning_rate": 4.155774896119463e-06,
+      "loss": 2.0712,
+      "step": 6220
+    },
+    {
+      "epoch": 2.683631658399827,
+      "grad_norm": 0.19370831549167633,
+      "learning_rate": 4.144579608162468e-06,
+      "loss": 2.1085,
+      "step": 6221
+    },
+    {
+      "epoch": 2.6840629717489755,
+      "grad_norm": 0.18302783370018005,
+      "learning_rate": 4.133398991595119e-06,
+      "loss": 1.9831,
+      "step": 6222
+    },
+    {
+      "epoch": 2.6844942850981237,
+      "grad_norm": 0.19806934893131256,
+      "learning_rate": 4.122233048732512e-06,
+      "loss": 2.2637,
+      "step": 6223
+    },
+    {
+      "epoch": 2.684925598447272,
+      "grad_norm": 0.19383008778095245,
+      "learning_rate": 4.111081781886657e-06,
+      "loss": 2.1842,
+      "step": 6224
+    },
+    {
+      "epoch": 2.68535691179642,
+      "grad_norm": 0.20208725333213806,
+      "learning_rate": 4.099945193366524e-06,
+      "loss": 2.0627,
+      "step": 6225
+    },
+    {
+      "epoch": 2.68535691179642,
+      "eval_loss": 2.087400436401367,
+      "eval_runtime": 197.412,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 6225
+    },
+    {
+      "epoch": 2.6857882251455685,
+      "grad_norm": 0.20198409259319305,
+      "learning_rate": 4.08882328547811e-06,
+      "loss": 2.0314,
+      "step": 6226
+    },
+    {
+      "epoch": 2.6862195384947163,
+      "grad_norm": 0.19249609112739563,
+      "learning_rate": 4.077716060524297e-06,
+      "loss": 2.004,
+      "step": 6227
+    },
+    {
+      "epoch": 2.6866508518438645,
+      "grad_norm": 0.20324400067329407,
+      "learning_rate": 4.066623520804987e-06,
+      "loss": 2.1872,
+      "step": 6228
+    },
+    {
+      "epoch": 2.687082165193013,
+      "grad_norm": 0.20244289934635162,
+      "learning_rate": 4.055545668616961e-06,
+      "loss": 2.1976,
+      "step": 6229
+    },
+    {
+      "epoch": 2.6875134785421606,
+      "grad_norm": 0.20366866886615753,
+      "learning_rate": 4.044482506254057e-06,
+      "loss": 2.1723,
+      "step": 6230
+    },
+    {
+      "epoch": 2.687944791891309,
+      "grad_norm": 0.21891331672668457,
+      "learning_rate": 4.033434036007002e-06,
+      "loss": 2.2852,
+      "step": 6231
+    },
+    {
+      "epoch": 2.688376105240457,
+      "grad_norm": 0.19543354213237762,
+      "learning_rate": 4.022400260163508e-06,
+      "loss": 2.0835,
+      "step": 6232
+    },
+    {
+      "epoch": 2.6888074185896054,
+      "grad_norm": 0.19975729286670685,
+      "learning_rate": 4.011381181008247e-06,
+      "loss": 2.1318,
+      "step": 6233
+    },
+    {
+      "epoch": 2.6892387319387536,
+      "grad_norm": 0.20842617750167847,
+      "learning_rate": 4.000376800822818e-06,
+      "loss": 2.1383,
+      "step": 6234
+    },
+    {
+      "epoch": 2.689670045287902,
+      "grad_norm": 0.19123625755310059,
+      "learning_rate": 3.989387121885817e-06,
+      "loss": 2.0994,
+      "step": 6235
+    },
+    {
+      "epoch": 2.6901013586370497,
+      "grad_norm": 0.213456928730011,
+      "learning_rate": 3.978412146472757e-06,
+      "loss": 2.2063,
+      "step": 6236
+    },
+    {
+      "epoch": 2.690532671986198,
+      "grad_norm": 0.19223642349243164,
+      "learning_rate": 3.967451876856167e-06,
+      "loss": 1.9427,
+      "step": 6237
+    },
+    {
+      "epoch": 2.690963985335346,
+      "grad_norm": 0.285829097032547,
+      "learning_rate": 3.956506315305452e-06,
+      "loss": 2.2721,
+      "step": 6238
+    },
+    {
+      "epoch": 2.691395298684494,
+      "grad_norm": 0.20903241634368896,
+      "learning_rate": 3.945575464087014e-06,
+      "loss": 2.014,
+      "step": 6239
+    },
+    {
+      "epoch": 2.6918266120336423,
+      "grad_norm": 0.19532136619091034,
+      "learning_rate": 3.934659325464226e-06,
+      "loss": 2.1684,
+      "step": 6240
+    },
+    {
+      "epoch": 2.6922579253827905,
+      "grad_norm": 0.20843470096588135,
+      "learning_rate": 3.92375790169738e-06,
+      "loss": 2.1268,
+      "step": 6241
+    },
+    {
+      "epoch": 2.6926892387319388,
+      "grad_norm": 0.18874512612819672,
+      "learning_rate": 3.912871195043746e-06,
+      "loss": 2.2548,
+      "step": 6242
+    },
+    {
+      "epoch": 2.693120552081087,
+      "grad_norm": 0.19821836054325104,
+      "learning_rate": 3.901999207757503e-06,
+      "loss": 2.1225,
+      "step": 6243
+    },
+    {
+      "epoch": 2.6935518654302353,
+      "grad_norm": 0.19665993750095367,
+      "learning_rate": 3.891141942089851e-06,
+      "loss": 2.1058,
+      "step": 6244
+    },
+    {
+      "epoch": 2.693983178779383,
+      "grad_norm": 0.1842608004808426,
+      "learning_rate": 3.8802994002888995e-06,
+      "loss": 1.9469,
+      "step": 6245
+    },
+    {
+      "epoch": 2.6944144921285313,
+      "grad_norm": 0.2139851152896881,
+      "learning_rate": 3.869471584599704e-06,
+      "loss": 2.3312,
+      "step": 6246
+    },
+    {
+      "epoch": 2.6948458054776796,
+      "grad_norm": 0.19002576172351837,
+      "learning_rate": 3.858658497264286e-06,
+      "loss": 2.0497,
+      "step": 6247
+    },
+    {
+      "epoch": 2.695277118826828,
+      "grad_norm": 0.2080351859331131,
+      "learning_rate": 3.847860140521622e-06,
+      "loss": 2.1032,
+      "step": 6248
+    },
+    {
+      "epoch": 2.6957084321759757,
+      "grad_norm": 0.2042865753173828,
+      "learning_rate": 3.837076516607615e-06,
+      "loss": 2.0746,
+      "step": 6249
+    },
+    {
+      "epoch": 2.696139745525124,
+      "grad_norm": 0.19151441752910614,
+      "learning_rate": 3.826307627755129e-06,
+      "loss": 1.9515,
+      "step": 6250
+    },
+    {
+      "epoch": 2.696139745525124,
+      "eval_loss": 2.087350368499756,
+      "eval_runtime": 196.982,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 6250
+    },
+    {
+      "epoch": 2.696571058874272,
+      "grad_norm": 0.18477578461170197,
+      "learning_rate": 3.8155534761940135e-06,
+      "loss": 2.129,
+      "step": 6251
+    },
+    {
+      "epoch": 2.6970023722234204,
+      "grad_norm": 0.1978139728307724,
+      "learning_rate": 3.8048140641510045e-06,
+      "loss": 2.3592,
+      "step": 6252
+    },
+    {
+      "epoch": 2.6974336855725687,
+      "grad_norm": 0.21185347437858582,
+      "learning_rate": 3.794089393849806e-06,
+      "loss": 2.2006,
+      "step": 6253
+    },
+    {
+      "epoch": 2.6978649989217165,
+      "grad_norm": 0.18322397768497467,
+      "learning_rate": 3.783379467511108e-06,
+      "loss": 2.0021,
+      "step": 6254
+    },
+    {
+      "epoch": 2.6982963122708647,
+      "grad_norm": 0.19997923076152802,
+      "learning_rate": 3.7726842873524955e-06,
+      "loss": 2.0924,
+      "step": 6255
+    },
+    {
+      "epoch": 2.698727625620013,
+      "grad_norm": 0.20193538069725037,
+      "learning_rate": 3.7620038555885467e-06,
+      "loss": 2.2167,
+      "step": 6256
+    },
+    {
+      "epoch": 2.6991589389691613,
+      "grad_norm": 0.20680344104766846,
+      "learning_rate": 3.751338174430718e-06,
+      "loss": 2.0264,
+      "step": 6257
+    },
+    {
+      "epoch": 2.699590252318309,
+      "grad_norm": 0.196910560131073,
+      "learning_rate": 3.7406872460875008e-06,
+      "loss": 2.0407,
+      "step": 6258
+    },
+    {
+      "epoch": 2.7000215656674573,
+      "grad_norm": 0.2135513722896576,
+      "learning_rate": 3.7300510727642647e-06,
+      "loss": 2.181,
+      "step": 6259
+    },
+    {
+      "epoch": 2.7004528790166056,
+      "grad_norm": 0.2150072604417801,
+      "learning_rate": 3.7194296566633474e-06,
+      "loss": 2.1288,
+      "step": 6260
+    },
+    {
+      "epoch": 2.700884192365754,
+      "grad_norm": 0.18341900408267975,
+      "learning_rate": 3.7088229999840317e-06,
+      "loss": 2.1744,
+      "step": 6261
+    },
+    {
+      "epoch": 2.701315505714902,
+      "grad_norm": 0.19466567039489746,
+      "learning_rate": 3.6982311049225444e-06,
+      "loss": 2.2336,
+      "step": 6262
+    },
+    {
+      "epoch": 2.70174681906405,
+      "grad_norm": 0.19956213235855103,
+      "learning_rate": 3.6876539736720475e-06,
+      "loss": 2.2707,
+      "step": 6263
+    },
+    {
+      "epoch": 2.702178132413198,
+      "grad_norm": 0.20101287961006165,
+      "learning_rate": 3.6770916084226554e-06,
+      "loss": 2.1547,
+      "step": 6264
+    },
+    {
+      "epoch": 2.7026094457623464,
+      "grad_norm": 0.22500237822532654,
+      "learning_rate": 3.666544011361411e-06,
+      "loss": 2.0867,
+      "step": 6265
+    },
+    {
+      "epoch": 2.7030407591114947,
+      "grad_norm": 0.20740890502929688,
+      "learning_rate": 3.656011184672325e-06,
+      "loss": 2.1929,
+      "step": 6266
+    },
+    {
+      "epoch": 2.7034720724606425,
+      "grad_norm": 0.1789781153202057,
+      "learning_rate": 3.6454931305363285e-06,
+      "loss": 1.876,
+      "step": 6267
+    },
+    {
+      "epoch": 2.7039033858097907,
+      "grad_norm": 0.2012297362089157,
+      "learning_rate": 3.6349898511312953e-06,
+      "loss": 2.1847,
+      "step": 6268
+    },
+    {
+      "epoch": 2.704334699158939,
+      "grad_norm": 0.19550712406635284,
+      "learning_rate": 3.624501348632061e-06,
+      "loss": 2.1368,
+      "step": 6269
+    },
+    {
+      "epoch": 2.7047660125080872,
+      "grad_norm": 0.1741098314523697,
+      "learning_rate": 3.6140276252103557e-06,
+      "loss": 1.8827,
+      "step": 6270
+    },
+    {
+      "epoch": 2.7051973258572355,
+      "grad_norm": 0.20835484564304352,
+      "learning_rate": 3.6035686830348936e-06,
+      "loss": 2.0917,
+      "step": 6271
+    },
+    {
+      "epoch": 2.7056286392063833,
+      "grad_norm": 0.19523583352565765,
+      "learning_rate": 3.593124524271318e-06,
+      "loss": 2.2537,
+      "step": 6272
+    },
+    {
+      "epoch": 2.7060599525555316,
+      "grad_norm": 0.19004569947719574,
+      "learning_rate": 3.582695151082207e-06,
+      "loss": 2.0502,
+      "step": 6273
+    },
+    {
+      "epoch": 2.70649126590468,
+      "grad_norm": 0.20238938927650452,
+      "learning_rate": 3.5722805656270837e-06,
+      "loss": 2.1448,
+      "step": 6274
+    },
+    {
+      "epoch": 2.706922579253828,
+      "grad_norm": 0.18806307017803192,
+      "learning_rate": 3.561880770062364e-06,
+      "loss": 2.2018,
+      "step": 6275
+    },
+    {
+      "epoch": 2.706922579253828,
+      "eval_loss": 2.0872480869293213,
+      "eval_runtime": 197.1462,
+      "eval_samples_per_second": 0.162,
+      "eval_steps_per_second": 0.162,
+      "step": 6275
+    },
+    {
+      "epoch": 2.707353892602976,
+      "grad_norm": 0.20623880624771118,
+      "learning_rate": 3.5514957665414763e-06,
+      "loss": 2.2654,
+      "step": 6276
+    },
+    {
+      "epoch": 2.707785205952124,
+      "grad_norm": 0.20118732750415802,
+      "learning_rate": 3.541125557214741e-06,
+      "loss": 2.0149,
+      "step": 6277
+    },
+    {
+      "epoch": 2.7082165193012724,
+      "grad_norm": 0.22785833477973938,
+      "learning_rate": 3.5307701442294166e-06,
+      "loss": 2.0617,
+      "step": 6278
+    },
+    {
+      "epoch": 2.7086478326504206,
+      "grad_norm": 0.20555000007152557,
+      "learning_rate": 3.520429529729704e-06,
+      "loss": 2.3689,
+      "step": 6279
+    },
+    {
+      "epoch": 2.709079145999569,
+      "grad_norm": 0.2299307882785797,
+      "learning_rate": 3.510103715856749e-06,
+      "loss": 2.2022,
+      "step": 6280
+    },
+    {
+      "epoch": 2.7095104593487167,
+      "grad_norm": 0.1937788426876068,
+      "learning_rate": 3.499792704748625e-06,
+      "loss": 2.2859,
+      "step": 6281
+    },
+    {
+      "epoch": 2.709941772697865,
+      "grad_norm": 0.21659395098686218,
+      "learning_rate": 3.4894964985403147e-06,
+      "loss": 2.3204,
+      "step": 6282
+    },
+    {
+      "epoch": 2.710373086047013,
+      "grad_norm": 0.1889360249042511,
+      "learning_rate": 3.479215099363805e-06,
+      "loss": 2.3064,
+      "step": 6283
+    },
+    {
+      "epoch": 2.7108043993961615,
+      "grad_norm": 0.18120795488357544,
+      "learning_rate": 3.4689485093479353e-06,
+      "loss": 1.8809,
+      "step": 6284
+    },
+    {
+      "epoch": 2.7112357127453093,
+      "grad_norm": 0.19754794239997864,
+      "learning_rate": 3.458696730618521e-06,
+      "loss": 2.2219,
+      "step": 6285
+    },
+    {
+      "epoch": 2.7116670260944575,
+      "grad_norm": 0.20219269394874573,
+      "learning_rate": 3.4484597652983226e-06,
+      "loss": 2.2278,
+      "step": 6286
+    },
+    {
+      "epoch": 2.712098339443606,
+      "grad_norm": 0.2099461704492569,
+      "learning_rate": 3.4382376155069947e-06,
+      "loss": 2.0175,
+      "step": 6287
+    },
+    {
+      "epoch": 2.712529652792754,
+      "grad_norm": 0.19380366802215576,
+      "learning_rate": 3.428030283361169e-06,
+      "loss": 2.0483,
+      "step": 6288
+    },
+    {
+      "epoch": 2.7129609661419023,
+      "grad_norm": 0.21564538776874542,
+      "learning_rate": 3.417837770974355e-06,
+      "loss": 2.0766,
+      "step": 6289
+    },
+    {
+      "epoch": 2.71339227949105,
+      "grad_norm": 0.17680005729198456,
+      "learning_rate": 3.4076600804570476e-06,
+      "loss": 1.7898,
+      "step": 6290
+    },
+    {
+      "epoch": 2.7138235928401984,
+      "grad_norm": 0.21097064018249512,
+      "learning_rate": 3.3974972139166447e-06,
+      "loss": 2.2372,
+      "step": 6291
+    },
+    {
+      "epoch": 2.7142549061893466,
+      "grad_norm": 0.18284453451633453,
+      "learning_rate": 3.3873491734574708e-06,
+      "loss": 1.9787,
+      "step": 6292
+    },
+    {
+      "epoch": 2.714686219538495,
+      "grad_norm": 0.18313708901405334,
+      "learning_rate": 3.3772159611807876e-06,
+      "loss": 2.2077,
+      "step": 6293
+    },
+    {
+      "epoch": 2.7151175328876427,
+      "grad_norm": 0.20901811122894287,
+      "learning_rate": 3.3670975791847994e-06,
+      "loss": 2.1341,
+      "step": 6294
+    },
+    {
+      "epoch": 2.715548846236791,
+      "grad_norm": 0.18402762711048126,
+      "learning_rate": 3.356994029564622e-06,
+      "loss": 2.1648,
+      "step": 6295
+    },
+    {
+      "epoch": 2.715980159585939,
+      "grad_norm": 0.20775403082370758,
+      "learning_rate": 3.346905314412307e-06,
+      "loss": 2.2146,
+      "step": 6296
+    },
+    {
+      "epoch": 2.7164114729350874,
+      "grad_norm": 0.22491486370563507,
+      "learning_rate": 3.3368314358168334e-06,
+      "loss": 2.0796,
+      "step": 6297
+    },
+    {
+      "epoch": 2.7168427862842357,
+      "grad_norm": 0.2053094506263733,
+      "learning_rate": 3.3267723958641e-06,
+      "loss": 2.0496,
+      "step": 6298
+    },
+    {
+      "epoch": 2.7172740996333835,
+      "grad_norm": 0.1953420639038086,
+      "learning_rate": 3.3167281966369393e-06,
+      "loss": 2.1895,
+      "step": 6299
+    },
+    {
+      "epoch": 2.7177054129825318,
+      "grad_norm": 0.18459086120128632,
+      "learning_rate": 3.30669884021513e-06,
+      "loss": 2.0651,
+      "step": 6300
+    },
+    {
+      "epoch": 2.7177054129825318,
+      "eval_loss": 2.087197780609131,
+      "eval_runtime": 196.2873,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 6300
+    },
+    {
+      "epoch": 2.71813672633168,
+      "grad_norm": 0.17420656979084015,
+      "learning_rate": 3.2966843286753614e-06,
+      "loss": 1.886,
+      "step": 6301
+    },
+    {
+      "epoch": 2.7185680396808283,
+      "grad_norm": 0.19562606513500214,
+      "learning_rate": 3.2866846640912327e-06,
+      "loss": 2.012,
+      "step": 6302
+    },
+    {
+      "epoch": 2.718999353029976,
+      "grad_norm": 0.19525456428527832,
+      "learning_rate": 3.2766998485332705e-06,
+      "loss": 2.222,
+      "step": 6303
+    },
+    {
+      "epoch": 2.7194306663791243,
+      "grad_norm": 0.19238971173763275,
+      "learning_rate": 3.2667298840689637e-06,
+      "loss": 2.0992,
+      "step": 6304
+    },
+    {
+      "epoch": 2.7198619797282726,
+      "grad_norm": 0.19270627200603485,
+      "learning_rate": 3.2567747727627018e-06,
+      "loss": 2.1833,
+      "step": 6305
+    },
+    {
+      "epoch": 2.720293293077421,
+      "grad_norm": 0.20034782588481903,
+      "learning_rate": 3.246834516675795e-06,
+      "loss": 2.2458,
+      "step": 6306
+    },
+    {
+      "epoch": 2.720724606426569,
+      "grad_norm": 0.3612179160118103,
+      "learning_rate": 3.236909117866479e-06,
+      "loss": 2.0146,
+      "step": 6307
+    },
+    {
+      "epoch": 2.721155919775717,
+      "grad_norm": 0.18545496463775635,
+      "learning_rate": 3.2269985783899192e-06,
+      "loss": 2.2781,
+      "step": 6308
+    },
+    {
+      "epoch": 2.721587233124865,
+      "grad_norm": 0.21780475974082947,
+      "learning_rate": 3.2171029002981992e-06,
+      "loss": 2.1374,
+      "step": 6309
+    },
+    {
+      "epoch": 2.7220185464740134,
+      "grad_norm": 0.19212014973163605,
+      "learning_rate": 3.2072220856403293e-06,
+      "loss": 2.3301,
+      "step": 6310
+    },
+    {
+      "epoch": 2.7224498598231617,
+      "grad_norm": 0.1920885145664215,
+      "learning_rate": 3.1973561364622396e-06,
+      "loss": 2.1969,
+      "step": 6311
+    },
+    {
+      "epoch": 2.7228811731723095,
+      "grad_norm": 0.6626337766647339,
+      "learning_rate": 3.1875050548067797e-06,
+      "loss": 2.0958,
+      "step": 6312
+    },
+    {
+      "epoch": 2.7233124865214577,
+      "grad_norm": 0.19787397980690002,
+      "learning_rate": 3.1776688427137264e-06,
+      "loss": 2.2973,
+      "step": 6313
+    },
+    {
+      "epoch": 2.723743799870606,
+      "grad_norm": 0.20773246884346008,
+      "learning_rate": 3.167847502219767e-06,
+      "loss": 2.174,
+      "step": 6314
+    },
+    {
+      "epoch": 2.7241751132197543,
+      "grad_norm": 0.20719939470291138,
+      "learning_rate": 3.158041035358541e-06,
+      "loss": 2.5476,
+      "step": 6315
+    },
+    {
+      "epoch": 2.7246064265689025,
+      "grad_norm": 0.21184757351875305,
+      "learning_rate": 3.148249444160558e-06,
+      "loss": 2.136,
+      "step": 6316
+    },
+    {
+      "epoch": 2.7250377399180503,
+      "grad_norm": 0.20608209073543549,
+      "learning_rate": 3.1384727306532797e-06,
+      "loss": 2.1159,
+      "step": 6317
+    },
+    {
+      "epoch": 2.7254690532671986,
+      "grad_norm": 0.21290935575962067,
+      "learning_rate": 3.128710896861103e-06,
+      "loss": 2.0429,
+      "step": 6318
+    },
+    {
+      "epoch": 2.725900366616347,
+      "grad_norm": 0.1945108324289322,
+      "learning_rate": 3.1189639448052943e-06,
+      "loss": 2.0928,
+      "step": 6319
+    },
+    {
+      "epoch": 2.726331679965495,
+      "grad_norm": 0.2086068093776703,
+      "learning_rate": 3.109231876504098e-06,
+      "loss": 2.3127,
+      "step": 6320
+    },
+    {
+      "epoch": 2.726762993314643,
+      "grad_norm": 0.19717541337013245,
+      "learning_rate": 3.0995146939726017e-06,
+      "loss": 2.205,
+      "step": 6321
+    },
+    {
+      "epoch": 2.727194306663791,
+      "grad_norm": 0.19919461011886597,
+      "learning_rate": 3.089812399222888e-06,
+      "loss": 2.184,
+      "step": 6322
+    },
+    {
+      "epoch": 2.7276256200129394,
+      "grad_norm": 0.19284598529338837,
+      "learning_rate": 3.080124994263916e-06,
+      "loss": 2.2423,
+      "step": 6323
+    },
+    {
+      "epoch": 2.7280569333620877,
+      "grad_norm": 0.2111853063106537,
+      "learning_rate": 3.070452481101565e-06,
+      "loss": 1.9166,
+      "step": 6324
+    },
+    {
+      "epoch": 2.728488246711236,
+      "grad_norm": 0.19298049807548523,
+      "learning_rate": 3.060794861738633e-06,
+      "loss": 2.0194,
+      "step": 6325
+    },
+    {
+      "epoch": 2.728488246711236,
+      "eval_loss": 2.087181806564331,
+      "eval_runtime": 196.6185,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 6325
+    },
+    {
+      "epoch": 2.7289195600603837,
+      "grad_norm": 0.1980437934398651,
+      "learning_rate": 3.0511521381748364e-06,
+      "loss": 2.0237,
+      "step": 6326
+    },
+    {
+      "epoch": 2.729350873409532,
+      "grad_norm": 0.18850843608379364,
+      "learning_rate": 3.0415243124068207e-06,
+      "loss": 2.2705,
+      "step": 6327
+    },
+    {
+      "epoch": 2.7297821867586802,
+      "grad_norm": 0.2052440196275711,
+      "learning_rate": 3.0319113864280987e-06,
+      "loss": 2.2331,
+      "step": 6328
+    },
+    {
+      "epoch": 2.7302135001078285,
+      "grad_norm": 0.21586772799491882,
+      "learning_rate": 3.0223133622291786e-06,
+      "loss": 2.1638,
+      "step": 6329
+    },
+    {
+      "epoch": 2.7306448134569763,
+      "grad_norm": 0.22283998131752014,
+      "learning_rate": 3.0127302417974035e-06,
+      "loss": 2.0749,
+      "step": 6330
+    },
+    {
+      "epoch": 2.7310761268061245,
+      "grad_norm": 0.20513780415058136,
+      "learning_rate": 3.003162027117062e-06,
+      "loss": 2.3214,
+      "step": 6331
+    },
+    {
+      "epoch": 2.731507440155273,
+      "grad_norm": 0.21246421337127686,
+      "learning_rate": 2.9936087201693688e-06,
+      "loss": 2.2214,
+      "step": 6332
+    },
+    {
+      "epoch": 2.731938753504421,
+      "grad_norm": 0.1875782608985901,
+      "learning_rate": 2.98407032293245e-06,
+      "loss": 2.1809,
+      "step": 6333
+    },
+    {
+      "epoch": 2.7323700668535693,
+      "grad_norm": 0.20177769660949707,
+      "learning_rate": 2.9745468373813254e-06,
+      "loss": 1.87,
+      "step": 6334
+    },
+    {
+      "epoch": 2.732801380202717,
+      "grad_norm": 0.19652503728866577,
+      "learning_rate": 2.965038265487918e-06,
+      "loss": 2.1856,
+      "step": 6335
+    },
+    {
+      "epoch": 2.7332326935518654,
+      "grad_norm": 0.20007416605949402,
+      "learning_rate": 2.955544609221111e-06,
+      "loss": 2.1514,
+      "step": 6336
+    },
+    {
+      "epoch": 2.7336640069010136,
+      "grad_norm": 0.21538324654102325,
+      "learning_rate": 2.946065870546649e-06,
+      "loss": 2.1655,
+      "step": 6337
+    },
+    {
+      "epoch": 2.734095320250162,
+      "grad_norm": 0.27241501212120056,
+      "learning_rate": 2.936602051427228e-06,
+      "loss": 2.0861,
+      "step": 6338
+    },
+    {
+      "epoch": 2.7345266335993097,
+      "grad_norm": 0.18923966586589813,
+      "learning_rate": 2.9271531538224147e-06,
+      "loss": 2.1178,
+      "step": 6339
+    },
+    {
+      "epoch": 2.734957946948458,
+      "grad_norm": 0.19602225720882416,
+      "learning_rate": 2.9177191796887266e-06,
+      "loss": 2.1503,
+      "step": 6340
+    },
+    {
+      "epoch": 2.735389260297606,
+      "grad_norm": 0.2139609009027481,
+      "learning_rate": 2.9083001309795517e-06,
+      "loss": 2.3083,
+      "step": 6341
+    },
+    {
+      "epoch": 2.7358205736467545,
+      "grad_norm": 0.18180742859840393,
+      "learning_rate": 2.898896009645221e-06,
+      "loss": 1.9963,
+      "step": 6342
+    },
+    {
+      "epoch": 2.7362518869959027,
+      "grad_norm": 0.20043374598026276,
+      "learning_rate": 2.8895068176329527e-06,
+      "loss": 2.2321,
+      "step": 6343
+    },
+    {
+      "epoch": 2.7366832003450505,
+      "grad_norm": 0.2054060846567154,
+      "learning_rate": 2.8801325568868826e-06,
+      "loss": 2.1354,
+      "step": 6344
+    },
+    {
+      "epoch": 2.737114513694199,
+      "grad_norm": 0.20694616436958313,
+      "learning_rate": 2.8707732293480584e-06,
+      "loss": 1.9688,
+      "step": 6345
+    },
+    {
+      "epoch": 2.737545827043347,
+      "grad_norm": 0.2166181057691574,
+      "learning_rate": 2.8614288369544298e-06,
+      "loss": 2.1904,
+      "step": 6346
+    },
+    {
+      "epoch": 2.7379771403924953,
+      "grad_norm": 0.1976022571325302,
+      "learning_rate": 2.8520993816408655e-06,
+      "loss": 2.1796,
+      "step": 6347
+    },
+    {
+      "epoch": 2.738408453741643,
+      "grad_norm": 0.2170005589723587,
+      "learning_rate": 2.842784865339104e-06,
+      "loss": 2.1227,
+      "step": 6348
+    },
+    {
+      "epoch": 2.7388397670907914,
+      "grad_norm": 0.23662927746772766,
+      "learning_rate": 2.8334852899778355e-06,
+      "loss": 2.0782,
+      "step": 6349
+    },
+    {
+      "epoch": 2.7392710804399396,
+      "grad_norm": 0.17678862810134888,
+      "learning_rate": 2.8242006574826366e-06,
+      "loss": 2.0842,
+      "step": 6350
+    },
+    {
+      "epoch": 2.7392710804399396,
+      "eval_loss": 2.087163209915161,
+      "eval_runtime": 196.5639,
+      "eval_samples_per_second": 0.163,
+      "eval_steps_per_second": 0.163,
+      "step": 6350
+    },
+    {
+      "epoch": 2.739702393789088,
+      "grad_norm": 0.20020273327827454,
+      "learning_rate": 2.8149309697760024e-06,
+      "loss": 2.1296,
+      "step": 6351
+    },
+    {
+      "epoch": 2.740133707138236,
+      "grad_norm": 0.20694181323051453,
+      "learning_rate": 2.8056762287773067e-06,
+      "loss": 2.0802,
+      "step": 6352
+    },
+    {
+      "epoch": 2.740565020487384,
+      "grad_norm": 0.19143207371234894,
+      "learning_rate": 2.7964364364028574e-06,
+      "loss": 2.0352,
+      "step": 6353
+    },
+    {
+      "epoch": 2.740996333836532,
+      "grad_norm": 0.20685730874538422,
+      "learning_rate": 2.7872115945658414e-06,
+      "loss": 2.1375,
+      "step": 6354
+    },
+    {
+      "epoch": 2.7414276471856804,
+      "grad_norm": 0.2112044095993042,
+      "learning_rate": 2.7780017051763803e-06,
+      "loss": 2.1864,
+      "step": 6355
+    },
+    {
+      "epoch": 2.7418589605348287,
+      "grad_norm": 0.18936973810195923,
+      "learning_rate": 2.768806770141466e-06,
+      "loss": 2.1065,
+      "step": 6356
+    },
+    {
+      "epoch": 2.7422902738839765,
+      "grad_norm": 0.19992630183696747,
+      "learning_rate": 2.7596267913650163e-06,
+      "loss": 2.1422,
+      "step": 6357
+    },
+    {
+      "epoch": 2.7427215872331248,
+      "grad_norm": 0.19272871315479279,
+      "learning_rate": 2.7504617707478443e-06,
+      "loss": 2.2792,
+      "step": 6358
+    },
+    {
+      "epoch": 2.743152900582273,
+      "grad_norm": 0.19747158885002136,
+      "learning_rate": 2.741311710187674e-06,
+      "loss": 2.2129,
+      "step": 6359
+    },
+    {
+      "epoch": 2.7435842139314213,
+      "grad_norm": 0.1911972463130951,
+      "learning_rate": 2.732176611579115e-06,
+      "loss": 2.3908,
+      "step": 6360
+    },
+    {
+      "epoch": 2.7440155272805695,
+      "grad_norm": 0.20867711305618286,
+      "learning_rate": 2.723056476813712e-06,
+      "loss": 1.9727,
+      "step": 6361
+    },
+    {
+      "epoch": 2.7444468406297173,
+      "grad_norm": 0.22647875547409058,
+      "learning_rate": 2.713951307779863e-06,
+      "loss": 1.9784,
+      "step": 6362
+    },
+    {
+      "epoch": 2.7448781539788656,
+      "grad_norm": 0.18755938112735748,
+      "learning_rate": 2.704861106362902e-06,
+      "loss": 2.0905,
+      "step": 6363
+    },
+    {
+      "epoch": 2.745309467328014,
+      "grad_norm": 0.18958084285259247,
+      "learning_rate": 2.6957858744450556e-06,
+      "loss": 2.0371,
+      "step": 6364
+    },
+    {
+      "epoch": 2.745740780677162,
+      "grad_norm": 0.19676895439624786,
+      "learning_rate": 2.6867256139054466e-06,
+      "loss": 2.1324,
+      "step": 6365
+    },
+    {
+      "epoch": 2.74617209402631,
+      "grad_norm": 0.2035304605960846,
+      "learning_rate": 2.6776803266201156e-06,
+      "loss": 2.1633,
+      "step": 6366
+    },
+    {
+      "epoch": 2.746603407375458,
+      "grad_norm": 0.18587583303451538,
+      "learning_rate": 2.668650014461965e-06,
+      "loss": 2.2838,
+      "step": 6367
+    },
+    {
+      "epoch": 2.7470347207246064,
+      "grad_norm": 0.20424732565879822,
+      "learning_rate": 2.659634679300832e-06,
+      "loss": 2.1103,
+      "step": 6368
+    },
+    {
+      "epoch": 2.7474660340737547,
+      "grad_norm": 0.19742344319820404,
+      "learning_rate": 2.6506343230034396e-06,
+      "loss": 2.2929,
+      "step": 6369
+    },
+    {
+      "epoch": 2.747897347422903,
+      "grad_norm": 0.19142282009124756,
+      "learning_rate": 2.641648947433414e-06,
+      "loss": 1.8528,
+      "step": 6370
+    },
+    {
+      "epoch": 2.7483286607720507,
+      "grad_norm": 0.1873573362827301,
+      "learning_rate": 2.6326785544512675e-06,
+      "loss": 2.0168,
+      "step": 6371
+    },
+    {
+      "epoch": 2.748759974121199,
+      "grad_norm": 1.0676277875900269,
+      "learning_rate": 2.6237231459144305e-06,
+      "loss": 2.025,
+      "step": 6372
+    },
+    {
+      "epoch": 2.7491912874703472,
+      "grad_norm": 0.19708436727523804,
+      "learning_rate": 2.614782723677203e-06,
+      "loss": 2.3798,
+      "step": 6373
+    },
+    {
+      "epoch": 2.7496226008194955,
+      "grad_norm": 0.1987622082233429,
+      "learning_rate": 2.605857289590796e-06,
+      "loss": 2.1454,
+      "step": 6374
+    },
+    {
+      "epoch": 2.7500539141686433,
+      "grad_norm": 0.22205474972724915,
+      "learning_rate": 2.5969468455033392e-06,
+      "loss": 2.1309,
+      "step": 6375
+    },
+    {
+      "epoch": 2.7500539141686433,
+      "eval_loss": 2.0871455669403076,
+      "eval_runtime": 200.0309,
+      "eval_samples_per_second": 0.16,
+      "eval_steps_per_second": 0.16,
+      "step": 6375
+    },
+    {
+      "epoch": 2.7504852275177916,
+      "grad_norm": 0.20050524175167084,
+      "learning_rate": 2.588051393259824e-06,
+      "loss": 2.1983,
+      "step": 6376
+    },
+    {
+      "epoch": 2.75091654086694,
+      "grad_norm": 0.2021673619747162,
+      "learning_rate": 2.579170934702135e-06,
+      "loss": 2.3204,
+      "step": 6377
+    },
+    {
+      "epoch": 2.751347854216088,
+      "grad_norm": 0.1954621821641922,
+      "learning_rate": 2.570305471669101e-06,
+      "loss": 2.2105,
+      "step": 6378
+    },
+    {
+      "epoch": 2.7517791675652363,
+      "grad_norm": 0.19034738838672638,
+      "learning_rate": 2.561455005996388e-06,
+      "loss": 2.1093,
+      "step": 6379
+    },
+    {
+      "epoch": 2.752210480914384,
+      "grad_norm": 0.19725388288497925,
+      "learning_rate": 2.552619539516604e-06,
+      "loss": 2.0607,
+      "step": 6380
+    },
+    {
+      "epoch": 2.7526417942635324,
+      "grad_norm": 0.19629184901714325,
+      "learning_rate": 2.5437990740592024e-06,
+      "loss": 2.0065,
+      "step": 6381
+    },
+    {
+      "epoch": 2.7530731076126806,
+      "grad_norm": 0.18590472638607025,
+      "learning_rate": 2.5349936114505725e-06,
+      "loss": 1.932,
+      "step": 6382
+    },
+    {
+      "epoch": 2.753504420961829,
+      "grad_norm": 0.20266211032867432,
+      "learning_rate": 2.526203153513981e-06,
+      "loss": 2.1198,
+      "step": 6383
+    },
+    {
+      "epoch": 2.7539357343109767,
+      "grad_norm": 0.2148665189743042,
+      "learning_rate": 2.517427702069588e-06,
+      "loss": 2.2842,
+      "step": 6384
+    },
+    {
+      "epoch": 2.754367047660125,
+      "grad_norm": 0.25880327820777893,
+      "learning_rate": 2.508667258934449e-06,
+      "loss": 2.1491,
+      "step": 6385
+    },
+    {
+      "epoch": 2.7547983610092732,
+      "grad_norm": 0.19644123315811157,
+      "learning_rate": 2.4999218259224962e-06,
+      "loss": 2.189,
+      "step": 6386
+    },
+    {
+      "epoch": 2.7552296743584215,
+      "grad_norm": 0.2401173710823059,
+      "learning_rate": 2.491191404844581e-06,
+      "loss": 1.8995,
+      "step": 6387
+    },
+    {
+      "epoch": 2.7556609877075697,
+      "grad_norm": 0.19858931005001068,
+      "learning_rate": 2.482475997508432e-06,
+      "loss": 2.1006,
+      "step": 6388
+    },
+    {
+      "epoch": 2.7560923010567175,
+      "grad_norm": 0.1920706033706665,
+      "learning_rate": 2.473775605718664e-06,
+      "loss": 2.0843,
+      "step": 6389
+    },
+    {
+      "epoch": 2.756523614405866,
+      "grad_norm": 0.21617066860198975,
+      "learning_rate": 2.465090231276787e-06,
+      "loss": 2.2643,
+      "step": 6390
+    },
+    {
+      "epoch": 2.756954927755014,
+      "grad_norm": 0.192400261759758,
+      "learning_rate": 2.4564198759812026e-06,
+      "loss": 2.0117,
+      "step": 6391
+    },
+    {
+      "epoch": 2.7573862411041623,
+      "grad_norm": 0.19906947016716003,
+      "learning_rate": 2.447764541627209e-06,
+      "loss": 1.9897,
+      "step": 6392
+    },
+    {
+      "epoch": 2.75781755445331,
+      "grad_norm": 0.21517327427864075,
+      "learning_rate": 2.4391242300069974e-06,
+      "loss": 2.1997,
+      "step": 6393
+    },
+    {
+      "epoch": 2.7582488678024584,
+      "grad_norm": 0.20024853944778442,
+      "learning_rate": 2.43049894290962e-06,
+      "loss": 2.059,
+      "step": 6394
+    },
+    {
+      "epoch": 2.7586801811516066,
+      "grad_norm": 0.20617882907390594,
+      "learning_rate": 2.4218886821210316e-06,
+      "loss": 2.0869,
+      "step": 6395
+    },
+    {
+      "epoch": 2.759111494500755,
+      "grad_norm": 0.1851520985364914,
+      "learning_rate": 2.4132934494240972e-06,
+      "loss": 1.9781,
+      "step": 6396
+    },
+    {
+      "epoch": 2.759542807849903,
+      "grad_norm": 0.20461012423038483,
+      "learning_rate": 2.4047132465985525e-06,
+      "loss": 2.1605,
+      "step": 6397
+    },
+    {
+      "epoch": 2.759974121199051,
+      "grad_norm": 0.19638298451900482,
+      "learning_rate": 2.396148075421017e-06,
+      "loss": 1.9916,
+      "step": 6398
+    },
+    {
+      "epoch": 2.760405434548199,
+      "grad_norm": 0.21877293288707733,
+      "learning_rate": 2.3875979376650063e-06,
+      "loss": 2.2603,
+      "step": 6399
+    },
+    {
+      "epoch": 2.7608367478973475,
+      "grad_norm": 0.19478262960910797,
+      "learning_rate": 2.37906283510092e-06,
+      "loss": 2.1077,
+      "step": 6400
+    },
+    {
+      "epoch": 2.7608367478973475,
+      "eval_loss": 2.0871729850769043,
+      "eval_runtime": 194.9904,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6400
+    },
+    {
+      "epoch": 2.7612680612464957,
+      "grad_norm": 0.21150808036327362,
+      "learning_rate": 2.370542769496045e-06,
+      "loss": 2.1381,
+      "step": 6401
+    },
+    {
+      "epoch": 2.7616993745956435,
+      "grad_norm": 0.20274010300636292,
+      "learning_rate": 2.3620377426145448e-06,
+      "loss": 2.2028,
+      "step": 6402
+    },
+    {
+      "epoch": 2.7621306879447918,
+      "grad_norm": 0.17827123403549194,
+      "learning_rate": 2.353547756217486e-06,
+      "loss": 2.1247,
+      "step": 6403
+    },
+    {
+      "epoch": 2.76256200129394,
+      "grad_norm": 0.20930832624435425,
+      "learning_rate": 2.3450728120628205e-06,
+      "loss": 1.9401,
+      "step": 6404
+    },
+    {
+      "epoch": 2.7629933146430883,
+      "grad_norm": 0.2053506076335907,
+      "learning_rate": 2.336612911905361e-06,
+      "loss": 2.0497,
+      "step": 6405
+    },
+    {
+      "epoch": 2.7634246279922365,
+      "grad_norm": 0.2011050581932068,
+      "learning_rate": 2.328168057496832e-06,
+      "loss": 2.1845,
+      "step": 6406
+    },
+    {
+      "epoch": 2.763855941341385,
+      "grad_norm": 0.19295421242713928,
+      "learning_rate": 2.3197382505858423e-06,
+      "loss": 2.0928,
+      "step": 6407
+    },
+    {
+      "epoch": 2.7642872546905326,
+      "grad_norm": 0.19697582721710205,
+      "learning_rate": 2.311323492917855e-06,
+      "loss": 2.2089,
+      "step": 6408
+    },
+    {
+      "epoch": 2.764718568039681,
+      "grad_norm": 0.20078019797801971,
+      "learning_rate": 2.302923786235242e-06,
+      "loss": 2.2633,
+      "step": 6409
+    },
+    {
+      "epoch": 2.765149881388829,
+      "grad_norm": 0.2065783143043518,
+      "learning_rate": 2.2945391322772786e-06,
+      "loss": 2.2333,
+      "step": 6410
+    },
+    {
+      "epoch": 2.765581194737977,
+      "grad_norm": 0.20100045204162598,
+      "learning_rate": 2.286169532780069e-06,
+      "loss": 1.8474,
+      "step": 6411
+    },
+    {
+      "epoch": 2.766012508087125,
+      "grad_norm": 0.1967354416847229,
+      "learning_rate": 2.277814989476659e-06,
+      "loss": 2.1904,
+      "step": 6412
+    },
+    {
+      "epoch": 2.7664438214362734,
+      "grad_norm": 0.20408812165260315,
+      "learning_rate": 2.269475504096907e-06,
+      "loss": 1.9991,
+      "step": 6413
+    },
+    {
+      "epoch": 2.7668751347854217,
+      "grad_norm": 0.1991935819387436,
+      "learning_rate": 2.26115107836764e-06,
+      "loss": 2.0678,
+      "step": 6414
+    },
+    {
+      "epoch": 2.76730644813457,
+      "grad_norm": 0.20208342373371124,
+      "learning_rate": 2.2528417140124876e-06,
+      "loss": 2.2605,
+      "step": 6415
+    },
+    {
+      "epoch": 2.767737761483718,
+      "grad_norm": 0.19435718655586243,
+      "learning_rate": 2.2445474127520152e-06,
+      "loss": 2.1604,
+      "step": 6416
+    },
+    {
+      "epoch": 2.768169074832866,
+      "grad_norm": 0.2047269344329834,
+      "learning_rate": 2.2362681763036404e-06,
+      "loss": 2.1334,
+      "step": 6417
+    },
+    {
+      "epoch": 2.7686003881820143,
+      "grad_norm": 0.19622421264648438,
+      "learning_rate": 2.228004006381667e-06,
+      "loss": 2.2027,
+      "step": 6418
+    },
+    {
+      "epoch": 2.7690317015311625,
+      "grad_norm": 0.1936480700969696,
+      "learning_rate": 2.2197549046972934e-06,
+      "loss": 2.1398,
+      "step": 6419
+    },
+    {
+      "epoch": 2.7694630148803103,
+      "grad_norm": 0.21298177540302277,
+      "learning_rate": 2.2115208729585692e-06,
+      "loss": 1.9736,
+      "step": 6420
+    },
+    {
+      "epoch": 2.7698943282294586,
+      "grad_norm": 0.17468075454235077,
+      "learning_rate": 2.203301912870456e-06,
+      "loss": 1.9918,
+      "step": 6421
+    },
+    {
+      "epoch": 2.770325641578607,
+      "grad_norm": 0.19439345598220825,
+      "learning_rate": 2.1950980261347674e-06,
+      "loss": 2.0712,
+      "step": 6422
+    },
+    {
+      "epoch": 2.770756954927755,
+      "grad_norm": 0.19806329905986786,
+      "learning_rate": 2.186909214450211e-06,
+      "loss": 2.0701,
+      "step": 6423
+    },
+    {
+      "epoch": 2.7711882682769033,
+      "grad_norm": 0.1960648149251938,
+      "learning_rate": 2.1787354795123724e-06,
+      "loss": 2.0607,
+      "step": 6424
+    },
+    {
+      "epoch": 2.7716195816260516,
+      "grad_norm": 0.20692449808120728,
+      "learning_rate": 2.170576823013731e-06,
+      "loss": 2.1282,
+      "step": 6425
+    },
+    {
+      "epoch": 2.7716195816260516,
+      "eval_loss": 2.0871548652648926,
+      "eval_runtime": 194.7721,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6425
+    },
+    {
+      "epoch": 2.7720508949751994,
+      "grad_norm": 0.20055831968784332,
+      "learning_rate": 2.1624332466435938e-06,
+      "loss": 2.1578,
+      "step": 6426
+    },
+    {
+      "epoch": 2.7724822083243477,
+      "grad_norm": 0.1872549206018448,
+      "learning_rate": 2.154304752088179e-06,
+      "loss": 2.2042,
+      "step": 6427
+    },
+    {
+      "epoch": 2.772913521673496,
+      "grad_norm": 0.19771480560302734,
+      "learning_rate": 2.1461913410306064e-06,
+      "loss": 2.1327,
+      "step": 6428
+    },
+    {
+      "epoch": 2.7733448350226437,
+      "grad_norm": 0.23074021935462952,
+      "learning_rate": 2.1380930151508325e-06,
+      "loss": 2.1386,
+      "step": 6429
+    },
+    {
+      "epoch": 2.773776148371792,
+      "grad_norm": 0.20994645357131958,
+      "learning_rate": 2.130009776125699e-06,
+      "loss": 2.1653,
+      "step": 6430
+    },
+    {
+      "epoch": 2.7742074617209402,
+      "grad_norm": 0.19817151129245758,
+      "learning_rate": 2.1219416256289424e-06,
+      "loss": 2.185,
+      "step": 6431
+    },
+    {
+      "epoch": 2.7746387750700885,
+      "grad_norm": 0.20503103733062744,
+      "learning_rate": 2.1138885653311426e-06,
+      "loss": 2.1917,
+      "step": 6432
+    },
+    {
+      "epoch": 2.7750700884192367,
+      "grad_norm": 0.18863680958747864,
+      "learning_rate": 2.105850596899791e-06,
+      "loss": 2.1454,
+      "step": 6433
+    },
+    {
+      "epoch": 2.775501401768385,
+      "grad_norm": 0.1986525058746338,
+      "learning_rate": 2.0978277219992318e-06,
+      "loss": 2.2787,
+      "step": 6434
+    },
+    {
+      "epoch": 2.775932715117533,
+      "grad_norm": 0.20332397520542145,
+      "learning_rate": 2.089819942290685e-06,
+      "loss": 1.9675,
+      "step": 6435
+    },
+    {
+      "epoch": 2.776364028466681,
+      "grad_norm": 0.18802419304847717,
+      "learning_rate": 2.0818272594322507e-06,
+      "loss": 2.1166,
+      "step": 6436
+    },
+    {
+      "epoch": 2.7767953418158293,
+      "grad_norm": 0.2151930332183838,
+      "learning_rate": 2.073849675078895e-06,
+      "loss": 2.1658,
+      "step": 6437
+    },
+    {
+      "epoch": 2.777226655164977,
+      "grad_norm": 0.18824675679206848,
+      "learning_rate": 2.0658871908824816e-06,
+      "loss": 2.1774,
+      "step": 6438
+    },
+    {
+      "epoch": 2.7776579685141254,
+      "grad_norm": 0.20375822484493256,
+      "learning_rate": 2.0579398084917236e-06,
+      "loss": 2.1393,
+      "step": 6439
+    },
+    {
+      "epoch": 2.7780892818632736,
+      "grad_norm": 0.21004673838615417,
+      "learning_rate": 2.0500075295521973e-06,
+      "loss": 1.9528,
+      "step": 6440
+    },
+    {
+      "epoch": 2.778520595212422,
+      "grad_norm": 0.24513062834739685,
+      "learning_rate": 2.04209035570638e-06,
+      "loss": 1.7822,
+      "step": 6441
+    },
+    {
+      "epoch": 2.77895190856157,
+      "grad_norm": 0.19536660611629486,
+      "learning_rate": 2.0341882885936107e-06,
+      "loss": 2.0755,
+      "step": 6442
+    },
+    {
+      "epoch": 2.7793832219107184,
+      "grad_norm": 0.25875434279441833,
+      "learning_rate": 2.0263013298500967e-06,
+      "loss": 2.2622,
+      "step": 6443
+    },
+    {
+      "epoch": 2.779814535259866,
+      "grad_norm": 0.24299484491348267,
+      "learning_rate": 2.018429481108924e-06,
+      "loss": 2.2031,
+      "step": 6444
+    },
+    {
+      "epoch": 2.7802458486090145,
+      "grad_norm": 0.1964147537946701,
+      "learning_rate": 2.0105727440000386e-06,
+      "loss": 2.1252,
+      "step": 6445
+    },
+    {
+      "epoch": 2.7806771619581627,
+      "grad_norm": 0.19906258583068848,
+      "learning_rate": 2.0027311201502638e-06,
+      "loss": 2.1964,
+      "step": 6446
+    },
+    {
+      "epoch": 2.7811084753073105,
+      "grad_norm": 0.19478124380111694,
+      "learning_rate": 1.994904611183293e-06,
+      "loss": 1.9829,
+      "step": 6447
+    },
+    {
+      "epoch": 2.781539788656459,
+      "grad_norm": 0.1987592726945877,
+      "learning_rate": 1.9870932187196957e-06,
+      "loss": 2.1168,
+      "step": 6448
+    },
+    {
+      "epoch": 2.781971102005607,
+      "grad_norm": 0.19542710483074188,
+      "learning_rate": 1.979296944376904e-06,
+      "loss": 2.3019,
+      "step": 6449
+    },
+    {
+      "epoch": 2.7824024153547553,
+      "grad_norm": 0.2147682160139084,
+      "learning_rate": 1.971515789769218e-06,
+      "loss": 2.2137,
+      "step": 6450
+    },
+    {
+      "epoch": 2.7824024153547553,
+      "eval_loss": 2.0871737003326416,
+      "eval_runtime": 194.617,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6450
+    },
+    {
+      "epoch": 2.7828337287039036,
+      "grad_norm": 0.19122406840324402,
+      "learning_rate": 1.963749756507807e-06,
+      "loss": 2.143,
+      "step": 6451
+    },
+    {
+      "epoch": 2.783265042053052,
+      "grad_norm": 0.20361442863941193,
+      "learning_rate": 1.9559988462007348e-06,
+      "loss": 2.109,
+      "step": 6452
+    },
+    {
+      "epoch": 2.7836963554021996,
+      "grad_norm": 0.19597600400447845,
+      "learning_rate": 1.9482630604528927e-06,
+      "loss": 2.2253,
+      "step": 6453
+    },
+    {
+      "epoch": 2.784127668751348,
+      "grad_norm": 0.19568900763988495,
+      "learning_rate": 1.9405424008660657e-06,
+      "loss": 2.2234,
+      "step": 6454
+    },
+    {
+      "epoch": 2.784558982100496,
+      "grad_norm": 0.20625774562358856,
+      "learning_rate": 1.9328368690388925e-06,
+      "loss": 2.0696,
+      "step": 6455
+    },
+    {
+      "epoch": 2.784990295449644,
+      "grad_norm": 0.19813793897628784,
+      "learning_rate": 1.925146466566896e-06,
+      "loss": 2.2189,
+      "step": 6456
+    },
+    {
+      "epoch": 2.785421608798792,
+      "grad_norm": 0.21412412822246552,
+      "learning_rate": 1.917471195042461e-06,
+      "loss": 2.1537,
+      "step": 6457
+    },
+    {
+      "epoch": 2.7858529221479404,
+      "grad_norm": 0.20112594962120056,
+      "learning_rate": 1.909811056054833e-06,
+      "loss": 2.3006,
+      "step": 6458
+    },
+    {
+      "epoch": 2.7862842354970887,
+      "grad_norm": 0.1943495273590088,
+      "learning_rate": 1.9021660511901182e-06,
+      "loss": 1.949,
+      "step": 6459
+    },
+    {
+      "epoch": 2.786715548846237,
+      "grad_norm": 0.21642576158046722,
+      "learning_rate": 1.8945361820313087e-06,
+      "loss": 2.3344,
+      "step": 6460
+    },
+    {
+      "epoch": 2.787146862195385,
+      "grad_norm": 0.20003938674926758,
+      "learning_rate": 1.8869214501582575e-06,
+      "loss": 2.3522,
+      "step": 6461
+    },
+    {
+      "epoch": 2.787578175544533,
+      "grad_norm": 0.19744734466075897,
+      "learning_rate": 1.8793218571476697e-06,
+      "loss": 2.2744,
+      "step": 6462
+    },
+    {
+      "epoch": 2.7880094888936813,
+      "grad_norm": 0.18666155636310577,
+      "learning_rate": 1.8717374045731204e-06,
+      "loss": 1.9739,
+      "step": 6463
+    },
+    {
+      "epoch": 2.7884408022428295,
+      "grad_norm": 0.19375626742839813,
+      "learning_rate": 1.8641680940050612e-06,
+      "loss": 2.0815,
+      "step": 6464
+    },
+    {
+      "epoch": 2.7888721155919773,
+      "grad_norm": 0.19252249598503113,
+      "learning_rate": 1.856613927010797e-06,
+      "loss": 1.968,
+      "step": 6465
+    },
+    {
+      "epoch": 2.7893034289411256,
+      "grad_norm": 0.21271540224552155,
+      "learning_rate": 1.8490749051545017e-06,
+      "loss": 2.2135,
+      "step": 6466
+    },
+    {
+      "epoch": 2.789734742290274,
+      "grad_norm": 0.23579350113868713,
+      "learning_rate": 1.841551029997218e-06,
+      "loss": 1.9886,
+      "step": 6467
+    },
+    {
+      "epoch": 2.790166055639422,
+      "grad_norm": 0.19472530484199524,
+      "learning_rate": 1.8340423030968332e-06,
+      "loss": 2.2144,
+      "step": 6468
+    },
+    {
+      "epoch": 2.7905973689885704,
+      "grad_norm": 0.21968962252140045,
+      "learning_rate": 1.8265487260081202e-06,
+      "loss": 2.2461,
+      "step": 6469
+    },
+    {
+      "epoch": 2.7910286823377186,
+      "grad_norm": 0.19856375455856323,
+      "learning_rate": 1.8190703002827046e-06,
+      "loss": 2.1164,
+      "step": 6470
+    },
+    {
+      "epoch": 2.7914599956868664,
+      "grad_norm": 0.19614583253860474,
+      "learning_rate": 1.811607027469089e-06,
+      "loss": 2.0143,
+      "step": 6471
+    },
+    {
+      "epoch": 2.7918913090360147,
+      "grad_norm": 0.4438173174858093,
+      "learning_rate": 1.8041589091126047e-06,
+      "loss": 2.1001,
+      "step": 6472
+    },
+    {
+      "epoch": 2.792322622385163,
+      "grad_norm": 0.21552805602550507,
+      "learning_rate": 1.796725946755459e-06,
+      "loss": 2.2342,
+      "step": 6473
+    },
+    {
+      "epoch": 2.7927539357343107,
+      "grad_norm": 0.18399421870708466,
+      "learning_rate": 1.7893081419367627e-06,
+      "loss": 2.0788,
+      "step": 6474
+    },
+    {
+      "epoch": 2.793185249083459,
+      "grad_norm": 0.18250444531440735,
+      "learning_rate": 1.781905496192429e-06,
+      "loss": 1.8374,
+      "step": 6475
+    },
+    {
+      "epoch": 2.793185249083459,
+      "eval_loss": 2.0871214866638184,
+      "eval_runtime": 194.9282,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6475
+    },
+    {
+      "epoch": 2.7936165624326073,
+      "grad_norm": 0.1958213448524475,
+      "learning_rate": 1.7745180110552565e-06,
+      "loss": 2.3356,
+      "step": 6476
+    },
+    {
+      "epoch": 2.7940478757817555,
+      "grad_norm": 0.2061747908592224,
+      "learning_rate": 1.7671456880549217e-06,
+      "loss": 2.1613,
+      "step": 6477
+    },
+    {
+      "epoch": 2.7944791891309038,
+      "grad_norm": 0.20832446217536926,
+      "learning_rate": 1.7597885287179203e-06,
+      "loss": 2.0902,
+      "step": 6478
+    },
+    {
+      "epoch": 2.794910502480052,
+      "grad_norm": 0.192495658993721,
+      "learning_rate": 1.7524465345676503e-06,
+      "loss": 1.8938,
+      "step": 6479
+    },
+    {
+      "epoch": 2.7953418158292,
+      "grad_norm": 0.21228113770484924,
+      "learning_rate": 1.7451197071243538e-06,
+      "loss": 1.9777,
+      "step": 6480
+    },
+    {
+      "epoch": 2.795773129178348,
+      "grad_norm": 0.19520573318004608,
+      "learning_rate": 1.7378080479051176e-06,
+      "loss": 2.0726,
+      "step": 6481
+    },
+    {
+      "epoch": 2.7962044425274963,
+      "grad_norm": 0.19283892214298248,
+      "learning_rate": 1.7305115584239055e-06,
+      "loss": 2.2036,
+      "step": 6482
+    },
+    {
+      "epoch": 2.796635755876644,
+      "grad_norm": 0.1813032180070877,
+      "learning_rate": 1.7232302401915337e-06,
+      "loss": 2.4088,
+      "step": 6483
+    },
+    {
+      "epoch": 2.7970670692257924,
+      "grad_norm": 0.2084456980228424,
+      "learning_rate": 1.7159640947156884e-06,
+      "loss": 2.1702,
+      "step": 6484
+    },
+    {
+      "epoch": 2.7974983825749407,
+      "grad_norm": 0.21089473366737366,
+      "learning_rate": 1.7087131235009072e-06,
+      "loss": 2.5504,
+      "step": 6485
+    },
+    {
+      "epoch": 2.797929695924089,
+      "grad_norm": 0.19794204831123352,
+      "learning_rate": 1.701477328048556e-06,
+      "loss": 2.179,
+      "step": 6486
+    },
+    {
+      "epoch": 2.798361009273237,
+      "grad_norm": 0.22298972308635712,
+      "learning_rate": 1.6942567098569032e-06,
+      "loss": 2.0364,
+      "step": 6487
+    },
+    {
+      "epoch": 2.7987923226223854,
+      "grad_norm": 0.21681968867778778,
+      "learning_rate": 1.6870512704210691e-06,
+      "loss": 2.2683,
+      "step": 6488
+    },
+    {
+      "epoch": 2.7992236359715332,
+      "grad_norm": 0.2068440020084381,
+      "learning_rate": 1.6798610112330024e-06,
+      "loss": 2.2263,
+      "step": 6489
+    },
+    {
+      "epoch": 2.7996549493206815,
+      "grad_norm": 0.20238614082336426,
+      "learning_rate": 1.6726859337815285e-06,
+      "loss": 2.0219,
+      "step": 6490
+    },
+    {
+      "epoch": 2.8000862626698297,
+      "grad_norm": 0.20311656594276428,
+      "learning_rate": 1.6655260395523257e-06,
+      "loss": 2.1749,
+      "step": 6491
+    },
+    {
+      "epoch": 2.8005175760189776,
+      "grad_norm": 0.2581019699573517,
+      "learning_rate": 1.6583813300279247e-06,
+      "loss": 2.0499,
+      "step": 6492
+    },
+    {
+      "epoch": 2.800948889368126,
+      "grad_norm": 0.20405562222003937,
+      "learning_rate": 1.6512518066877256e-06,
+      "loss": 2.1406,
+      "step": 6493
+    },
+    {
+      "epoch": 2.801380202717274,
+      "grad_norm": 0.20483388006687164,
+      "learning_rate": 1.6441374710079642e-06,
+      "loss": 2.1057,
+      "step": 6494
+    },
+    {
+      "epoch": 2.8018115160664223,
+      "grad_norm": 0.19205987453460693,
+      "learning_rate": 1.6370383244617452e-06,
+      "loss": 2.0372,
+      "step": 6495
+    },
+    {
+      "epoch": 2.8022428294155706,
+      "grad_norm": 0.20968760550022125,
+      "learning_rate": 1.6299543685190263e-06,
+      "loss": 2.0498,
+      "step": 6496
+    },
+    {
+      "epoch": 2.802674142764719,
+      "grad_norm": 0.22050540149211884,
+      "learning_rate": 1.6228856046466093e-06,
+      "loss": 2.1664,
+      "step": 6497
+    },
+    {
+      "epoch": 2.8031054561138666,
+      "grad_norm": 0.1926727592945099,
+      "learning_rate": 1.6158320343081732e-06,
+      "loss": 2.3444,
+      "step": 6498
+    },
+    {
+      "epoch": 2.803536769463015,
+      "grad_norm": 0.2012040913105011,
+      "learning_rate": 1.6087936589642414e-06,
+      "loss": 2.2633,
+      "step": 6499
+    },
+    {
+      "epoch": 2.803968082812163,
+      "grad_norm": 0.21898652613162994,
+      "learning_rate": 1.6017704800721648e-06,
+      "loss": 2.1332,
+      "step": 6500
+    },
+    {
+      "epoch": 2.803968082812163,
+      "eval_loss": 2.0871434211730957,
+      "eval_runtime": 194.753,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6500
+    },
+    {
+      "epoch": 2.804399396161311,
+      "grad_norm": 0.1935500055551529,
+      "learning_rate": 1.5947624990861718e-06,
+      "loss": 2.0575,
+      "step": 6501
+    },
+    {
+      "epoch": 2.804830709510459,
+      "grad_norm": 0.1991499662399292,
+      "learning_rate": 1.5877697174573601e-06,
+      "loss": 2.0898,
+      "step": 6502
+    },
+    {
+      "epoch": 2.8052620228596075,
+      "grad_norm": 0.18370495736598969,
+      "learning_rate": 1.5807921366336463e-06,
+      "loss": 2.1516,
+      "step": 6503
+    },
+    {
+      "epoch": 2.8056933362087557,
+      "grad_norm": 0.2187313735485077,
+      "learning_rate": 1.573829758059833e-06,
+      "loss": 1.7816,
+      "step": 6504
+    },
+    {
+      "epoch": 2.806124649557904,
+      "grad_norm": 0.21289658546447754,
+      "learning_rate": 1.5668825831775172e-06,
+      "loss": 2.2105,
+      "step": 6505
+    },
+    {
+      "epoch": 2.8065559629070522,
+      "grad_norm": 0.20149406790733337,
+      "learning_rate": 1.5599506134252315e-06,
+      "loss": 2.2682,
+      "step": 6506
+    },
+    {
+      "epoch": 2.8069872762562,
+      "grad_norm": 0.2140059471130371,
+      "learning_rate": 1.5530338502382945e-06,
+      "loss": 2.0684,
+      "step": 6507
+    },
+    {
+      "epoch": 2.8074185896053483,
+      "grad_norm": 0.21615104377269745,
+      "learning_rate": 1.5461322950489018e-06,
+      "loss": 2.1763,
+      "step": 6508
+    },
+    {
+      "epoch": 2.8078499029544965,
+      "grad_norm": 0.20524483919143677,
+      "learning_rate": 1.5392459492860942e-06,
+      "loss": 2.1927,
+      "step": 6509
+    },
+    {
+      "epoch": 2.8082812163036444,
+      "grad_norm": 0.1816680133342743,
+      "learning_rate": 1.5323748143757725e-06,
+      "loss": 2.072,
+      "step": 6510
+    },
+    {
+      "epoch": 2.8087125296527926,
+      "grad_norm": 0.1891278773546219,
+      "learning_rate": 1.5255188917406652e-06,
+      "loss": 2.2138,
+      "step": 6511
+    },
+    {
+      "epoch": 2.809143843001941,
+      "grad_norm": 0.20423439145088196,
+      "learning_rate": 1.5186781828003786e-06,
+      "loss": 2.0443,
+      "step": 6512
+    },
+    {
+      "epoch": 2.809575156351089,
+      "grad_norm": 0.20651452243328094,
+      "learning_rate": 1.5118526889713462e-06,
+      "loss": 2.1039,
+      "step": 6513
+    },
+    {
+      "epoch": 2.8100064697002374,
+      "grad_norm": 0.8881313800811768,
+      "learning_rate": 1.5050424116668792e-06,
+      "loss": 2.1473,
+      "step": 6514
+    },
+    {
+      "epoch": 2.8104377830493856,
+      "grad_norm": 0.19139255583286285,
+      "learning_rate": 1.4982473522970912e-06,
+      "loss": 2.0588,
+      "step": 6515
+    },
+    {
+      "epoch": 2.8108690963985334,
+      "grad_norm": 0.1904522329568863,
+      "learning_rate": 1.4914675122689983e-06,
+      "loss": 2.1976,
+      "step": 6516
+    },
+    {
+      "epoch": 2.8113004097476817,
+      "grad_norm": 0.22774598002433777,
+      "learning_rate": 1.4847028929864447e-06,
+      "loss": 1.7157,
+      "step": 6517
+    },
+    {
+      "epoch": 2.81173172309683,
+      "grad_norm": 0.201594278216362,
+      "learning_rate": 1.477953495850101e-06,
+      "loss": 2.1143,
+      "step": 6518
+    },
+    {
+      "epoch": 2.8121630364459778,
+      "grad_norm": 0.20529593527317047,
+      "learning_rate": 1.4712193222574997e-06,
+      "loss": 2.1312,
+      "step": 6519
+    },
+    {
+      "epoch": 2.812594349795126,
+      "grad_norm": 0.19255559146404266,
+      "learning_rate": 1.464500373603042e-06,
+      "loss": 2.1235,
+      "step": 6520
+    },
+    {
+      "epoch": 2.8130256631442743,
+      "grad_norm": 0.2061937004327774,
+      "learning_rate": 1.4577966512779482e-06,
+      "loss": 2.1598,
+      "step": 6521
+    },
+    {
+      "epoch": 2.8134569764934225,
+      "grad_norm": 0.2003118246793747,
+      "learning_rate": 1.4511081566703081e-06,
+      "loss": 2.2089,
+      "step": 6522
+    },
+    {
+      "epoch": 2.8138882898425708,
+      "grad_norm": 0.20519734919071198,
+      "learning_rate": 1.4444348911650388e-06,
+      "loss": 2.2433,
+      "step": 6523
+    },
+    {
+      "epoch": 2.814319603191719,
+      "grad_norm": 0.19626106321811676,
+      "learning_rate": 1.4377768561439102e-06,
+      "loss": 2.2808,
+      "step": 6524
+    },
+    {
+      "epoch": 2.814750916540867,
+      "grad_norm": 0.20269042253494263,
+      "learning_rate": 1.4311340529855525e-06,
+      "loss": 2.0386,
+      "step": 6525
+    },
+    {
+      "epoch": 2.814750916540867,
+      "eval_loss": 2.0870800018310547,
+      "eval_runtime": 194.8169,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6525
+    },
+    {
+      "epoch": 2.815182229890015,
+      "grad_norm": 0.2238195538520813,
+      "learning_rate": 1.424506483065424e-06,
+      "loss": 2.2912,
+      "step": 6526
+    },
+    {
+      "epoch": 2.8156135432391634,
+      "grad_norm": 0.19504454731941223,
+      "learning_rate": 1.4178941477558353e-06,
+      "loss": 2.0852,
+      "step": 6527
+    },
+    {
+      "epoch": 2.816044856588311,
+      "grad_norm": 0.20155642926692963,
+      "learning_rate": 1.4112970484259411e-06,
+      "loss": 1.9959,
+      "step": 6528
+    },
+    {
+      "epoch": 2.8164761699374594,
+      "grad_norm": 0.1982235610485077,
+      "learning_rate": 1.4047151864417405e-06,
+      "loss": 2.0988,
+      "step": 6529
+    },
+    {
+      "epoch": 2.8169074832866077,
+      "grad_norm": 0.20831064879894257,
+      "learning_rate": 1.3981485631660933e-06,
+      "loss": 2.1964,
+      "step": 6530
+    },
+    {
+      "epoch": 2.817338796635756,
+      "grad_norm": 0.2066282033920288,
+      "learning_rate": 1.3915971799586867e-06,
+      "loss": 2.1122,
+      "step": 6531
+    },
+    {
+      "epoch": 2.817770109984904,
+      "grad_norm": 0.19970938563346863,
+      "learning_rate": 1.3850610381760524e-06,
+      "loss": 2.1798,
+      "step": 6532
+    },
+    {
+      "epoch": 2.8182014233340524,
+      "grad_norm": 0.20368598401546478,
+      "learning_rate": 1.3785401391715579e-06,
+      "loss": 2.0832,
+      "step": 6533
+    },
+    {
+      "epoch": 2.8186327366832002,
+      "grad_norm": 0.2219233214855194,
+      "learning_rate": 1.3720344842954395e-06,
+      "loss": 2.2272,
+      "step": 6534
+    },
+    {
+      "epoch": 2.8190640500323485,
+      "grad_norm": 0.189238041639328,
+      "learning_rate": 1.36554407489477e-06,
+      "loss": 2.1196,
+      "step": 6535
+    },
+    {
+      "epoch": 2.8194953633814968,
+      "grad_norm": 0.19806627929210663,
+      "learning_rate": 1.3590689123134491e-06,
+      "loss": 2.158,
+      "step": 6536
+    },
+    {
+      "epoch": 2.8199266767306446,
+      "grad_norm": 0.20515137910842896,
+      "learning_rate": 1.3526089978922377e-06,
+      "loss": 2.2419,
+      "step": 6537
+    },
+    {
+      "epoch": 2.820357990079793,
+      "grad_norm": 0.22096523642539978,
+      "learning_rate": 1.3461643329687244e-06,
+      "loss": 2.18,
+      "step": 6538
+    },
+    {
+      "epoch": 2.820789303428941,
+      "grad_norm": 0.18739445507526398,
+      "learning_rate": 1.3397349188773498e-06,
+      "loss": 1.9742,
+      "step": 6539
+    },
+    {
+      "epoch": 2.8212206167780893,
+      "grad_norm": 0.19757986068725586,
+      "learning_rate": 1.3333207569493991e-06,
+      "loss": 2.2201,
+      "step": 6540
+    },
+    {
+      "epoch": 2.8216519301272376,
+      "grad_norm": 0.2100953906774521,
+      "learning_rate": 1.3269218485129852e-06,
+      "loss": 2.1429,
+      "step": 6541
+    },
+    {
+      "epoch": 2.822083243476386,
+      "grad_norm": 0.19657377898693085,
+      "learning_rate": 1.320538194893081e-06,
+      "loss": 2.0707,
+      "step": 6542
+    },
+    {
+      "epoch": 2.8225145568255336,
+      "grad_norm": 0.19098903238773346,
+      "learning_rate": 1.3141697974114796e-06,
+      "loss": 1.9887,
+      "step": 6543
+    },
+    {
+      "epoch": 2.822945870174682,
+      "grad_norm": 0.210743710398674,
+      "learning_rate": 1.307816657386851e-06,
+      "loss": 2.1582,
+      "step": 6544
+    },
+    {
+      "epoch": 2.82337718352383,
+      "grad_norm": 0.18364079296588898,
+      "learning_rate": 1.3014787761346596e-06,
+      "loss": 1.8924,
+      "step": 6545
+    },
+    {
+      "epoch": 2.823808496872978,
+      "grad_norm": 0.2031092643737793,
+      "learning_rate": 1.2951561549672473e-06,
+      "loss": 2.2685,
+      "step": 6546
+    },
+    {
+      "epoch": 2.8242398102221262,
+      "grad_norm": 0.21850912272930145,
+      "learning_rate": 1.288848795193767e-06,
+      "loss": 2.1427,
+      "step": 6547
+    },
+    {
+      "epoch": 2.8246711235712745,
+      "grad_norm": 0.19463056325912476,
+      "learning_rate": 1.2825566981202324e-06,
+      "loss": 2.1355,
+      "step": 6548
+    },
+    {
+      "epoch": 2.8251024369204227,
+      "grad_norm": 0.21621888875961304,
+      "learning_rate": 1.2762798650495094e-06,
+      "loss": 2.0943,
+      "step": 6549
+    },
+    {
+      "epoch": 2.825533750269571,
+      "grad_norm": 0.17011678218841553,
+      "learning_rate": 1.2700182972812585e-06,
+      "loss": 1.9588,
+      "step": 6550
+    },
+    {
+      "epoch": 2.825533750269571,
+      "eval_loss": 2.0870361328125,
+      "eval_runtime": 195.0598,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6550
+    },
+    {
+      "epoch": 2.8259650636187192,
+      "grad_norm": 0.1891787052154541,
+      "learning_rate": 1.2637719961120175e-06,
+      "loss": 2.1756,
+      "step": 6551
+    },
+    {
+      "epoch": 2.826396376967867,
+      "grad_norm": 0.17923136055469513,
+      "learning_rate": 1.2575409628351518e-06,
+      "loss": 2.2323,
+      "step": 6552
+    },
+    {
+      "epoch": 2.8268276903170153,
+      "grad_norm": 0.20607276260852814,
+      "learning_rate": 1.251325198740863e-06,
+      "loss": 2.1205,
+      "step": 6553
+    },
+    {
+      "epoch": 2.8272590036661636,
+      "grad_norm": 0.19592034816741943,
+      "learning_rate": 1.2451247051161878e-06,
+      "loss": 2.0756,
+      "step": 6554
+    },
+    {
+      "epoch": 2.8276903170153114,
+      "grad_norm": 0.20986421406269073,
+      "learning_rate": 1.2389394832450245e-06,
+      "loss": 2.0254,
+      "step": 6555
+    },
+    {
+      "epoch": 2.8281216303644596,
+      "grad_norm": 0.18774859607219696,
+      "learning_rate": 1.2327695344080735e-06,
+      "loss": 2.0611,
+      "step": 6556
+    },
+    {
+      "epoch": 2.828552943713608,
+      "grad_norm": 0.19214344024658203,
+      "learning_rate": 1.226614859882888e-06,
+      "loss": 2.1418,
+      "step": 6557
+    },
+    {
+      "epoch": 2.828984257062756,
+      "grad_norm": 0.18636095523834229,
+      "learning_rate": 1.2204754609438733e-06,
+      "loss": 2.0401,
+      "step": 6558
+    },
+    {
+      "epoch": 2.8294155704119044,
+      "grad_norm": 0.20176027715206146,
+      "learning_rate": 1.2143513388622544e-06,
+      "loss": 2.0793,
+      "step": 6559
+    },
+    {
+      "epoch": 2.8298468837610526,
+      "grad_norm": 0.18498626351356506,
+      "learning_rate": 1.2082424949061003e-06,
+      "loss": 2.0077,
+      "step": 6560
+    },
+    {
+      "epoch": 2.8302781971102005,
+      "grad_norm": 0.1844421774148941,
+      "learning_rate": 1.2021489303402992e-06,
+      "loss": 1.9651,
+      "step": 6561
+    },
+    {
+      "epoch": 2.8307095104593487,
+      "grad_norm": 0.20158296823501587,
+      "learning_rate": 1.1960706464266085e-06,
+      "loss": 1.9676,
+      "step": 6562
+    },
+    {
+      "epoch": 2.831140823808497,
+      "grad_norm": 0.20933398604393005,
+      "learning_rate": 1.1900076444236046e-06,
+      "loss": 2.1597,
+      "step": 6563
+    },
+    {
+      "epoch": 2.8315721371576448,
+      "grad_norm": 0.38234084844589233,
+      "learning_rate": 1.1839599255866832e-06,
+      "loss": 2.1744,
+      "step": 6564
+    },
+    {
+      "epoch": 2.832003450506793,
+      "grad_norm": 0.19370846450328827,
+      "learning_rate": 1.1779274911680931e-06,
+      "loss": 1.9428,
+      "step": 6565
+    },
+    {
+      "epoch": 2.8324347638559413,
+      "grad_norm": 0.1982627660036087,
+      "learning_rate": 1.1719103424169263e-06,
+      "loss": 2.1265,
+      "step": 6566
+    },
+    {
+      "epoch": 2.8328660772050895,
+      "grad_norm": 0.23792126774787903,
+      "learning_rate": 1.1659084805790946e-06,
+      "loss": 1.9542,
+      "step": 6567
+    },
+    {
+      "epoch": 2.833297390554238,
+      "grad_norm": 0.21161462366580963,
+      "learning_rate": 1.1599219068973458e-06,
+      "loss": 2.149,
+      "step": 6568
+    },
+    {
+      "epoch": 2.833728703903386,
+      "grad_norm": 0.2071421891450882,
+      "learning_rate": 1.1539506226112634e-06,
+      "loss": 2.1924,
+      "step": 6569
+    },
+    {
+      "epoch": 2.834160017252534,
+      "grad_norm": 0.2017904669046402,
+      "learning_rate": 1.1479946289572828e-06,
+      "loss": 2.1854,
+      "step": 6570
+    },
+    {
+      "epoch": 2.834591330601682,
+      "grad_norm": 0.18799950182437897,
+      "learning_rate": 1.1420539271686346e-06,
+      "loss": 2.188,
+      "step": 6571
+    },
+    {
+      "epoch": 2.8350226439508304,
+      "grad_norm": 0.1960608810186386,
+      "learning_rate": 1.1361285184754266e-06,
+      "loss": 2.1766,
+      "step": 6572
+    },
+    {
+      "epoch": 2.835453957299978,
+      "grad_norm": 0.7240349054336548,
+      "learning_rate": 1.1302184041045603e-06,
+      "loss": 2.1145,
+      "step": 6573
+    },
+    {
+      "epoch": 2.8358852706491264,
+      "grad_norm": 0.20830345153808594,
+      "learning_rate": 1.124323585279807e-06,
+      "loss": 2.101,
+      "step": 6574
+    },
+    {
+      "epoch": 2.8363165839982747,
+      "grad_norm": 0.1981867402791977,
+      "learning_rate": 1.1184440632217405e-06,
+      "loss": 2.1011,
+      "step": 6575
+    },
+    {
+      "epoch": 2.8363165839982747,
+      "eval_loss": 2.087026834487915,
+      "eval_runtime": 199.1904,
+      "eval_samples_per_second": 0.161,
+      "eval_steps_per_second": 0.161,
+      "step": 6575
+    },
+    {
+      "epoch": 2.836747897347423,
+      "grad_norm": 0.2012898027896881,
+      "learning_rate": 1.1125798391477869e-06,
+      "loss": 2.1873,
+      "step": 6576
+    },
+    {
+      "epoch": 2.837179210696571,
+      "grad_norm": 0.2115420699119568,
+      "learning_rate": 1.106730914272208e-06,
+      "loss": 2.0934,
+      "step": 6577
+    },
+    {
+      "epoch": 2.8376105240457195,
+      "grad_norm": 0.19753608107566833,
+      "learning_rate": 1.1008972898060687e-06,
+      "loss": 2.066,
+      "step": 6578
+    },
+    {
+      "epoch": 2.8380418373948673,
+      "grad_norm": 0.20166726410388947,
+      "learning_rate": 1.0950789669572858e-06,
+      "loss": 2.2392,
+      "step": 6579
+    },
+    {
+      "epoch": 2.8384731507440155,
+      "grad_norm": 0.855771005153656,
+      "learning_rate": 1.0892759469306206e-06,
+      "loss": 2.285,
+      "step": 6580
+    },
+    {
+      "epoch": 2.8389044640931638,
+      "grad_norm": 0.20052401721477509,
+      "learning_rate": 1.0834882309276455e-06,
+      "loss": 2.0705,
+      "step": 6581
+    },
+    {
+      "epoch": 2.8393357774423116,
+      "grad_norm": 0.19966664910316467,
+      "learning_rate": 1.07771582014676e-06,
+      "loss": 2.0723,
+      "step": 6582
+    },
+    {
+      "epoch": 2.83976709079146,
+      "grad_norm": 0.18459691107273102,
+      "learning_rate": 1.0719587157832248e-06,
+      "loss": 2.116,
+      "step": 6583
+    },
+    {
+      "epoch": 2.840198404140608,
+      "grad_norm": 0.19035299122333527,
+      "learning_rate": 1.0662169190290949e-06,
+      "loss": 2.109,
+      "step": 6584
+    },
+    {
+      "epoch": 2.8406297174897563,
+      "grad_norm": 0.21567097306251526,
+      "learning_rate": 1.0604904310732853e-06,
+      "loss": 2.2523,
+      "step": 6585
+    },
+    {
+      "epoch": 2.8410610308389046,
+      "grad_norm": 0.2520385682582855,
+      "learning_rate": 1.0547792531015148e-06,
+      "loss": 2.1898,
+      "step": 6586
+    },
+    {
+      "epoch": 2.841492344188053,
+      "grad_norm": 0.19074073433876038,
+      "learning_rate": 1.0490833862963539e-06,
+      "loss": 2.1059,
+      "step": 6587
+    },
+    {
+      "epoch": 2.8419236575372007,
+      "grad_norm": 0.20969784259796143,
+      "learning_rate": 1.0434028318371924e-06,
+      "loss": 2.1702,
+      "step": 6588
+    },
+    {
+      "epoch": 2.842354970886349,
+      "grad_norm": 0.22954483330249786,
+      "learning_rate": 1.0377375909002393e-06,
+      "loss": 2.1248,
+      "step": 6589
+    },
+    {
+      "epoch": 2.842786284235497,
+      "grad_norm": 0.18923214077949524,
+      "learning_rate": 1.0320876646585736e-06,
+      "loss": 2.0614,
+      "step": 6590
+    },
+    {
+      "epoch": 2.843217597584645,
+      "grad_norm": 0.20723962783813477,
+      "learning_rate": 1.026453054282042e-06,
+      "loss": 2.1693,
+      "step": 6591
+    },
+    {
+      "epoch": 2.8436489109337932,
+      "grad_norm": 0.19316589832305908,
+      "learning_rate": 1.0208337609373784e-06,
+      "loss": 2.1487,
+      "step": 6592
+    },
+    {
+      "epoch": 2.8440802242829415,
+      "grad_norm": 0.19485367834568024,
+      "learning_rate": 1.0152297857880936e-06,
+      "loss": 2.3646,
+      "step": 6593
+    },
+    {
+      "epoch": 2.8445115376320897,
+      "grad_norm": 0.1999531090259552,
+      "learning_rate": 1.0096411299945677e-06,
+      "loss": 2.114,
+      "step": 6594
+    },
+    {
+      "epoch": 2.844942850981238,
+      "grad_norm": 0.2081172615289688,
+      "learning_rate": 1.0040677947140002e-06,
+      "loss": 2.3284,
+      "step": 6595
+    },
+    {
+      "epoch": 2.8453741643303863,
+      "grad_norm": 0.21178525686264038,
+      "learning_rate": 9.985097811004012e-07,
+      "loss": 2.2318,
+      "step": 6596
+    },
+    {
+      "epoch": 2.845805477679534,
+      "grad_norm": 0.2165527045726776,
+      "learning_rate": 9.929670903046084e-07,
+      "loss": 2.1899,
+      "step": 6597
+    },
+    {
+      "epoch": 2.8462367910286823,
+      "grad_norm": 0.20413519442081451,
+      "learning_rate": 9.87439723474312e-07,
+      "loss": 2.2067,
+      "step": 6598
+    },
+    {
+      "epoch": 2.8466681043778306,
+      "grad_norm": 0.1823328584432602,
+      "learning_rate": 9.81927681754005e-07,
+      "loss": 2.1177,
+      "step": 6599
+    },
+    {
+      "epoch": 2.8470994177269784,
+      "grad_norm": 0.2507426142692566,
+      "learning_rate": 9.764309662850244e-07,
+      "loss": 2.2117,
+      "step": 6600
+    },
+    {
+      "epoch": 2.8470994177269784,
+      "eval_loss": 2.087045907974243,
+      "eval_runtime": 195.3532,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6600
+    },
+    {
+      "epoch": 2.8475307310761266,
+      "grad_norm": 0.19653716683387756,
+      "learning_rate": 9.709495782055266e-07,
+      "loss": 2.452,
+      "step": 6601
+    },
+    {
+      "epoch": 2.847962044425275,
+      "grad_norm": 1.8744388818740845,
+      "learning_rate": 9.654835186504784e-07,
+      "loss": 2.0288,
+      "step": 6602
+    },
+    {
+      "epoch": 2.848393357774423,
+      "grad_norm": 0.2228662669658661,
+      "learning_rate": 9.600327887516912e-07,
+      "loss": 2.2299,
+      "step": 6603
+    },
+    {
+      "epoch": 2.8488246711235714,
+      "grad_norm": 0.2115723341703415,
+      "learning_rate": 9.545973896378123e-07,
+      "loss": 2.2138,
+      "step": 6604
+    },
+    {
+      "epoch": 2.8492559844727197,
+      "grad_norm": 0.20439818501472473,
+      "learning_rate": 9.491773224342831e-07,
+      "loss": 1.9981,
+      "step": 6605
+    },
+    {
+      "epoch": 2.8496872978218675,
+      "grad_norm": 0.20456363260746002,
+      "learning_rate": 9.437725882633973e-07,
+      "loss": 2.1978,
+      "step": 6606
+    },
+    {
+      "epoch": 2.8501186111710157,
+      "grad_norm": 0.19252854585647583,
+      "learning_rate": 9.383831882442599e-07,
+      "loss": 2.1711,
+      "step": 6607
+    },
+    {
+      "epoch": 2.850549924520164,
+      "grad_norm": 0.21124014258384705,
+      "learning_rate": 9.330091234928117e-07,
+      "loss": 2.0787,
+      "step": 6608
+    },
+    {
+      "epoch": 2.850981237869312,
+      "grad_norm": 0.19842417538166046,
+      "learning_rate": 9.276503951218123e-07,
+      "loss": 2.0478,
+      "step": 6609
+    },
+    {
+      "epoch": 2.85141255121846,
+      "grad_norm": 0.20528315007686615,
+      "learning_rate": 9.223070042408326e-07,
+      "loss": 2.1122,
+      "step": 6610
+    },
+    {
+      "epoch": 2.8518438645676083,
+      "grad_norm": 0.2027859389781952,
+      "learning_rate": 9.169789519562875e-07,
+      "loss": 2.1932,
+      "step": 6611
+    },
+    {
+      "epoch": 2.8522751779167566,
+      "grad_norm": 0.20300999283790588,
+      "learning_rate": 9.116662393714114e-07,
+      "loss": 2.0943,
+      "step": 6612
+    },
+    {
+      "epoch": 2.852706491265905,
+      "grad_norm": 0.18875077366828918,
+      "learning_rate": 9.063688675862574e-07,
+      "loss": 2.117,
+      "step": 6613
+    },
+    {
+      "epoch": 2.853137804615053,
+      "grad_norm": 0.19429738819599152,
+      "learning_rate": 9.010868376976982e-07,
+      "loss": 2.1098,
+      "step": 6614
+    },
+    {
+      "epoch": 2.853569117964201,
+      "grad_norm": 0.1997184306383133,
+      "learning_rate": 8.958201507994506e-07,
+      "loss": 2.1689,
+      "step": 6615
+    },
+    {
+      "epoch": 2.854000431313349,
+      "grad_norm": 0.693905234336853,
+      "learning_rate": 8.905688079820256e-07,
+      "loss": 2.1667,
+      "step": 6616
+    },
+    {
+      "epoch": 2.8544317446624974,
+      "grad_norm": 0.20525632798671722,
+      "learning_rate": 8.8533281033277e-07,
+      "loss": 2.0395,
+      "step": 6617
+    },
+    {
+      "epoch": 2.854863058011645,
+      "grad_norm": 0.2016616314649582,
+      "learning_rate": 8.801121589358667e-07,
+      "loss": 2.1236,
+      "step": 6618
+    },
+    {
+      "epoch": 2.8552943713607934,
+      "grad_norm": 0.19586916267871857,
+      "learning_rate": 8.74906854872301e-07,
+      "loss": 2.0706,
+      "step": 6619
+    },
+    {
+      "epoch": 2.8557256847099417,
+      "grad_norm": 0.20806916058063507,
+      "learning_rate": 8.697168992198939e-07,
+      "loss": 2.2386,
+      "step": 6620
+    },
+    {
+      "epoch": 2.85615699805909,
+      "grad_norm": 0.19900496304035187,
+      "learning_rate": 8.645422930532692e-07,
+      "loss": 2.2531,
+      "step": 6621
+    },
+    {
+      "epoch": 2.856588311408238,
+      "grad_norm": 0.1811569631099701,
+      "learning_rate": 8.593830374438948e-07,
+      "loss": 2.1097,
+      "step": 6622
+    },
+    {
+      "epoch": 2.8570196247573865,
+      "grad_norm": 0.2104291319847107,
+      "learning_rate": 8.542391334600579e-07,
+      "loss": 1.9623,
+      "step": 6623
+    },
+    {
+      "epoch": 2.8574509381065343,
+      "grad_norm": 0.21303077042102814,
+      "learning_rate": 8.491105821668565e-07,
+      "loss": 2.2264,
+      "step": 6624
+    },
+    {
+      "epoch": 2.8578822514556825,
+      "grad_norm": 0.2043152004480362,
+      "learning_rate": 8.43997384626191e-07,
+      "loss": 2.1682,
+      "step": 6625
+    },
+    {
+      "epoch": 2.8578822514556825,
+      "eval_loss": 2.0864996910095215,
+      "eval_runtime": 195.4386,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6625
+    },
+    {
+      "epoch": 2.858313564804831,
+      "grad_norm": 0.19965513050556183,
+      "learning_rate": 8.388995418968397e-07,
+      "loss": 2.224,
+      "step": 6626
+    },
+    {
+      "epoch": 2.8587448781539786,
+      "grad_norm": 0.20936112105846405,
+      "learning_rate": 8.338170550343415e-07,
+      "loss": 2.256,
+      "step": 6627
+    },
+    {
+      "epoch": 2.859176191503127,
+      "grad_norm": 0.20981188118457794,
+      "learning_rate": 8.287499250910961e-07,
+      "loss": 2.2108,
+      "step": 6628
+    },
+    {
+      "epoch": 2.859607504852275,
+      "grad_norm": 0.19933246076107025,
+      "learning_rate": 8.236981531162979e-07,
+      "loss": 2.1298,
+      "step": 6629
+    },
+    {
+      "epoch": 2.8600388182014234,
+      "grad_norm": 0.18843577802181244,
+      "learning_rate": 8.186617401559764e-07,
+      "loss": 2.1118,
+      "step": 6630
+    },
+    {
+      "epoch": 2.8604701315505716,
+      "grad_norm": 0.19854816794395447,
+      "learning_rate": 8.13640687252981e-07,
+      "loss": 2.2842,
+      "step": 6631
+    },
+    {
+      "epoch": 2.86090144489972,
+      "grad_norm": 0.1979244500398636,
+      "learning_rate": 8.086349954469634e-07,
+      "loss": 2.0514,
+      "step": 6632
+    },
+    {
+      "epoch": 2.8613327582488677,
+      "grad_norm": 0.2001362293958664,
+      "learning_rate": 8.036446657744111e-07,
+      "loss": 1.6759,
+      "step": 6633
+    },
+    {
+      "epoch": 2.861764071598016,
+      "grad_norm": 0.19848394393920898,
+      "learning_rate": 7.986696992686308e-07,
+      "loss": 2.0379,
+      "step": 6634
+    },
+    {
+      "epoch": 2.862195384947164,
+      "grad_norm": 0.20530439913272858,
+      "learning_rate": 7.937100969597404e-07,
+      "loss": 2.1757,
+      "step": 6635
+    },
+    {
+      "epoch": 2.862626698296312,
+      "grad_norm": 0.20105339586734772,
+      "learning_rate": 7.887658598746849e-07,
+      "loss": 2.2595,
+      "step": 6636
+    },
+    {
+      "epoch": 2.8630580116454603,
+      "grad_norm": 0.19826927781105042,
+      "learning_rate": 7.83836989037212e-07,
+      "loss": 2.1847,
+      "step": 6637
+    },
+    {
+      "epoch": 2.8634893249946085,
+      "grad_norm": 0.20068801939487457,
+      "learning_rate": 7.789234854679055e-07,
+      "loss": 2.1122,
+      "step": 6638
+    },
+    {
+      "epoch": 2.8639206383437568,
+      "grad_norm": 0.18873481452465057,
+      "learning_rate": 7.740253501841598e-07,
+      "loss": 2.0476,
+      "step": 6639
+    },
+    {
+      "epoch": 2.864351951692905,
+      "grad_norm": 0.17728032171726227,
+      "learning_rate": 7.691425842001886e-07,
+      "loss": 1.7599,
+      "step": 6640
+    },
+    {
+      "epoch": 2.8647832650420533,
+      "grad_norm": 0.1999674141407013,
+      "learning_rate": 7.642751885270249e-07,
+      "loss": 2.1667,
+      "step": 6641
+    },
+    {
+      "epoch": 2.865214578391201,
+      "grad_norm": 0.18546587228775024,
+      "learning_rate": 7.59423164172504e-07,
+      "loss": 2.1342,
+      "step": 6642
+    },
+    {
+      "epoch": 2.8656458917403493,
+      "grad_norm": 0.19256991147994995,
+      "learning_rate": 7.545865121413059e-07,
+      "loss": 2.0104,
+      "step": 6643
+    },
+    {
+      "epoch": 2.8660772050894976,
+      "grad_norm": 0.22530266642570496,
+      "learning_rate": 7.497652334348958e-07,
+      "loss": 2.3658,
+      "step": 6644
+    },
+    {
+      "epoch": 2.8665085184386454,
+      "grad_norm": 0.680363118648529,
+      "learning_rate": 7.449593290515837e-07,
+      "loss": 2.2008,
+      "step": 6645
+    },
+    {
+      "epoch": 2.8669398317877937,
+      "grad_norm": 0.20634908974170685,
+      "learning_rate": 7.401687999864903e-07,
+      "loss": 2.0761,
+      "step": 6646
+    },
+    {
+      "epoch": 2.867371145136942,
+      "grad_norm": 0.2039497047662735,
+      "learning_rate": 7.353936472315302e-07,
+      "loss": 2.2705,
+      "step": 6647
+    },
+    {
+      "epoch": 2.86780245848609,
+      "grad_norm": 0.1839836686849594,
+      "learning_rate": 7.306338717754629e-07,
+      "loss": 2.1424,
+      "step": 6648
+    },
+    {
+      "epoch": 2.8682337718352384,
+      "grad_norm": 0.19734084606170654,
+      "learning_rate": 7.258894746038413e-07,
+      "loss": 2.0179,
+      "step": 6649
+    },
+    {
+      "epoch": 2.8686650851843867,
+      "grad_norm": 0.20046739280223846,
+      "learning_rate": 7.211604566990631e-07,
+      "loss": 2.2146,
+      "step": 6650
+    },
+    {
+      "epoch": 2.8686650851843867,
+      "eval_loss": 2.086951732635498,
+      "eval_runtime": 200.8992,
+      "eval_samples_per_second": 0.159,
+      "eval_steps_per_second": 0.159,
+      "step": 6650
+    },
+    {
+      "epoch": 2.8690963985335345,
+      "grad_norm": 0.17358741164207458,
+      "learning_rate": 7.164468190403034e-07,
+      "loss": 2.0335,
+      "step": 6651
+    },
+    {
+      "epoch": 2.8695277118826827,
+      "grad_norm": 0.20116187632083893,
+      "learning_rate": 7.117485626035813e-07,
+      "loss": 2.0944,
+      "step": 6652
+    },
+    {
+      "epoch": 2.869959025231831,
+      "grad_norm": 0.20895633101463318,
+      "learning_rate": 7.070656883617188e-07,
+      "loss": 2.2085,
+      "step": 6653
+    },
+    {
+      "epoch": 2.870390338580979,
+      "grad_norm": 0.21144898235797882,
+      "learning_rate": 7.023981972843651e-07,
+      "loss": 1.9656,
+      "step": 6654
+    },
+    {
+      "epoch": 2.870821651930127,
+      "grad_norm": 0.19829332828521729,
+      "learning_rate": 6.977460903379806e-07,
+      "loss": 2.195,
+      "step": 6655
+    },
+    {
+      "epoch": 2.8712529652792753,
+      "grad_norm": 0.2046317160129547,
+      "learning_rate": 6.931093684858114e-07,
+      "loss": 2.1055,
+      "step": 6656
+    },
+    {
+      "epoch": 2.8716842786284236,
+      "grad_norm": 0.20307636260986328,
+      "learning_rate": 6.884880326879477e-07,
+      "loss": 2.214,
+      "step": 6657
+    },
+    {
+      "epoch": 2.872115591977572,
+      "grad_norm": 0.20070429146289825,
+      "learning_rate": 6.838820839012993e-07,
+      "loss": 2.2231,
+      "step": 6658
+    },
+    {
+      "epoch": 2.87254690532672,
+      "grad_norm": 0.21028077602386475,
+      "learning_rate": 6.792915230795782e-07,
+      "loss": 2.1243,
+      "step": 6659
+    },
+    {
+      "epoch": 2.872978218675868,
+      "grad_norm": 0.215521901845932,
+      "learning_rate": 6.747163511732989e-07,
+      "loss": 2.2529,
+      "step": 6660
+    },
+    {
+      "epoch": 2.873409532025016,
+      "grad_norm": 0.20972667634487152,
+      "learning_rate": 6.701565691298122e-07,
+      "loss": 2.159,
+      "step": 6661
+    },
+    {
+      "epoch": 2.8738408453741644,
+      "grad_norm": 0.20830823481082916,
+      "learning_rate": 6.656121778932627e-07,
+      "loss": 2.221,
+      "step": 6662
+    },
+    {
+      "epoch": 2.874272158723312,
+      "grad_norm": 0.2061716765165329,
+      "learning_rate": 6.610831784046145e-07,
+      "loss": 2.2175,
+      "step": 6663
+    },
+    {
+      "epoch": 2.8747034720724605,
+      "grad_norm": 0.19651354849338531,
+      "learning_rate": 6.565695716016589e-07,
+      "loss": 2.1074,
+      "step": 6664
+    },
+    {
+      "epoch": 2.8751347854216087,
+      "grad_norm": 0.2101098895072937,
+      "learning_rate": 6.520713584189736e-07,
+      "loss": 2.121,
+      "step": 6665
+    },
+    {
+      "epoch": 2.875566098770757,
+      "grad_norm": 0.22959531843662262,
+      "learning_rate": 6.475885397879715e-07,
+      "loss": 2.1796,
+      "step": 6666
+    },
+    {
+      "epoch": 2.8759974121199052,
+      "grad_norm": 0.24616841971874237,
+      "learning_rate": 6.431211166368605e-07,
+      "loss": 2.0884,
+      "step": 6667
+    },
+    {
+      "epoch": 2.8764287254690535,
+      "grad_norm": 0.19734734296798706,
+      "learning_rate": 6.386690898906837e-07,
+      "loss": 1.9628,
+      "step": 6668
+    },
+    {
+      "epoch": 2.8768600388182013,
+      "grad_norm": 0.19654397666454315,
+      "learning_rate": 6.342324604712706e-07,
+      "loss": 2.1327,
+      "step": 6669
+    },
+    {
+      "epoch": 2.8772913521673495,
+      "grad_norm": 0.19043001532554626,
+      "learning_rate": 6.298112292972779e-07,
+      "loss": 2.0345,
+      "step": 6670
+    },
+    {
+      "epoch": 2.877722665516498,
+      "grad_norm": 0.2039814293384552,
+      "learning_rate": 6.254053972841566e-07,
+      "loss": 2.2965,
+      "step": 6671
+    },
+    {
+      "epoch": 2.878153978865646,
+      "grad_norm": 0.19430115818977356,
+      "learning_rate": 6.210149653442104e-07,
+      "loss": 2.0156,
+      "step": 6672
+    },
+    {
+      "epoch": 2.878585292214794,
+      "grad_norm": 0.19798149168491364,
+      "learning_rate": 6.166399343865036e-07,
+      "loss": 2.1365,
+      "step": 6673
+    },
+    {
+      "epoch": 2.879016605563942,
+      "grad_norm": 0.1976615935564041,
+      "learning_rate": 6.122803053169451e-07,
+      "loss": 2.1069,
+      "step": 6674
+    },
+    {
+      "epoch": 2.8794479189130904,
+      "grad_norm": 0.19123438000679016,
+      "learning_rate": 6.079360790382375e-07,
+      "loss": 2.1346,
+      "step": 6675
+    },
+    {
+      "epoch": 2.8794479189130904,
+      "eval_loss": 2.087125778198242,
+      "eval_runtime": 195.1716,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6675
+    },
+    {
+      "epoch": 2.8798792322622386,
+      "grad_norm": 0.2105484902858734,
+      "learning_rate": 6.036072564498951e-07,
+      "loss": 2.0971,
+      "step": 6676
+    },
+    {
+      "epoch": 2.880310545611387,
+      "grad_norm": 0.2146306186914444,
+      "learning_rate": 5.992938384482671e-07,
+      "loss": 2.1788,
+      "step": 6677
+    },
+    {
+      "epoch": 2.8807418589605347,
+      "grad_norm": 0.18713943660259247,
+      "learning_rate": 5.949958259264808e-07,
+      "loss": 2.1241,
+      "step": 6678
+    },
+    {
+      "epoch": 2.881173172309683,
+      "grad_norm": 0.20983433723449707,
+      "learning_rate": 5.907132197744829e-07,
+      "loss": 2.1917,
+      "step": 6679
+    },
+    {
+      "epoch": 2.881604485658831,
+      "grad_norm": 0.1955743134021759,
+      "learning_rate": 5.864460208790389e-07,
+      "loss": 2.1598,
+      "step": 6680
+    },
+    {
+      "epoch": 2.8820357990079795,
+      "grad_norm": 0.23252135515213013,
+      "learning_rate": 5.82194230123717e-07,
+      "loss": 1.7775,
+      "step": 6681
+    },
+    {
+      "epoch": 2.8824671123571273,
+      "grad_norm": 0.21306568384170532,
+      "learning_rate": 5.779578483889131e-07,
+      "loss": 2.0399,
+      "step": 6682
+    },
+    {
+      "epoch": 2.8828984257062755,
+      "grad_norm": 0.35586538910865784,
+      "learning_rate": 5.737368765517919e-07,
+      "loss": 2.0184,
+      "step": 6683
+    },
+    {
+      "epoch": 2.883329739055424,
+      "grad_norm": 0.22086109220981598,
+      "learning_rate": 5.695313154863629e-07,
+      "loss": 2.2016,
+      "step": 6684
+    },
+    {
+      "epoch": 2.883761052404572,
+      "grad_norm": 0.20301884412765503,
+      "learning_rate": 5.653411660634294e-07,
+      "loss": 2.1293,
+      "step": 6685
+    },
+    {
+      "epoch": 2.8841923657537203,
+      "grad_norm": 0.20924760401248932,
+      "learning_rate": 5.61166429150614e-07,
+      "loss": 2.3138,
+      "step": 6686
+    },
+    {
+      "epoch": 2.884623679102868,
+      "grad_norm": 0.20393526554107666,
+      "learning_rate": 5.570071056123421e-07,
+      "loss": 2.1015,
+      "step": 6687
+    },
+    {
+      "epoch": 2.8850549924520164,
+      "grad_norm": 0.2079361379146576,
+      "learning_rate": 5.528631963098412e-07,
+      "loss": 2.1904,
+      "step": 6688
+    },
+    {
+      "epoch": 2.8854863058011646,
+      "grad_norm": 0.21120835840702057,
+      "learning_rate": 5.487347021011502e-07,
+      "loss": 2.2447,
+      "step": 6689
+    },
+    {
+      "epoch": 2.885917619150313,
+      "grad_norm": 0.19092771410942078,
+      "learning_rate": 5.44621623841135e-07,
+      "loss": 2.0738,
+      "step": 6690
+    },
+    {
+      "epoch": 2.8863489324994607,
+      "grad_norm": 0.2014131247997284,
+      "learning_rate": 5.405239623814394e-07,
+      "loss": 2.217,
+      "step": 6691
+    },
+    {
+      "epoch": 2.886780245848609,
+      "grad_norm": 0.2046353667974472,
+      "learning_rate": 5.364417185705267e-07,
+      "loss": 2.1468,
+      "step": 6692
+    },
+    {
+      "epoch": 2.887211559197757,
+      "grad_norm": 0.19931462407112122,
+      "learning_rate": 5.323748932536787e-07,
+      "loss": 2.1508,
+      "step": 6693
+    },
+    {
+      "epoch": 2.8876428725469054,
+      "grad_norm": 0.1995890736579895,
+      "learning_rate": 5.283234872729802e-07,
+      "loss": 1.9074,
+      "step": 6694
+    },
+    {
+      "epoch": 2.8880741858960537,
+      "grad_norm": 0.1915419101715088,
+      "learning_rate": 5.24287501467302e-07,
+      "loss": 2.1993,
+      "step": 6695
+    },
+    {
+      "epoch": 2.8885054992452015,
+      "grad_norm": 0.1968814730644226,
+      "learning_rate": 5.202669366723588e-07,
+      "loss": 2.1575,
+      "step": 6696
+    },
+    {
+      "epoch": 2.8889368125943498,
+      "grad_norm": 0.19840653240680695,
+      "learning_rate": 5.162617937206348e-07,
+      "loss": 2.0741,
+      "step": 6697
+    },
+    {
+      "epoch": 2.889368125943498,
+      "grad_norm": 0.19516044855117798,
+      "learning_rate": 5.122720734414498e-07,
+      "loss": 2.3114,
+      "step": 6698
+    },
+    {
+      "epoch": 2.8897994392926463,
+      "grad_norm": 0.19493445754051208,
+      "learning_rate": 5.082977766609098e-07,
+      "loss": 2.0649,
+      "step": 6699
+    },
+    {
+      "epoch": 2.890230752641794,
+      "grad_norm": 0.20990261435508728,
+      "learning_rate": 5.04338904201948e-07,
+      "loss": 2.2653,
+      "step": 6700
+    },
+    {
+      "epoch": 2.890230752641794,
+      "eval_loss": 2.087057113647461,
+      "eval_runtime": 195.522,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6700
+    },
+    {
+      "epoch": 2.8906620659909423,
+      "grad_norm": 0.19304528832435608,
+      "learning_rate": 5.003954568842922e-07,
+      "loss": 2.1359,
+      "step": 6701
+    },
+    {
+      "epoch": 2.8910933793400906,
+      "grad_norm": 0.1956472247838974,
+      "learning_rate": 4.964674355244724e-07,
+      "loss": 1.9652,
+      "step": 6702
+    },
+    {
+      "epoch": 2.891524692689239,
+      "grad_norm": 0.20953501760959625,
+      "learning_rate": 4.925548409358133e-07,
+      "loss": 2.1561,
+      "step": 6703
+    },
+    {
+      "epoch": 2.891956006038387,
+      "grad_norm": 0.19960734248161316,
+      "learning_rate": 4.886576739284831e-07,
+      "loss": 2.2626,
+      "step": 6704
+    },
+    {
+      "epoch": 2.892387319387535,
+      "grad_norm": 0.19962123036384583,
+      "learning_rate": 4.847759353094199e-07,
+      "loss": 2.2388,
+      "step": 6705
+    },
+    {
+      "epoch": 2.892818632736683,
+      "grad_norm": 0.2108890563249588,
+      "learning_rate": 4.809096258823892e-07,
+      "loss": 2.1739,
+      "step": 6706
+    },
+    {
+      "epoch": 2.8932499460858314,
+      "grad_norm": 0.20290786027908325,
+      "learning_rate": 4.770587464479503e-07,
+      "loss": 1.9976,
+      "step": 6707
+    },
+    {
+      "epoch": 2.8936812594349797,
+      "grad_norm": 0.1943017840385437,
+      "learning_rate": 4.732232978034572e-07,
+      "loss": 2.044,
+      "step": 6708
+    },
+    {
+      "epoch": 2.8941125727841275,
+      "grad_norm": 0.1958266943693161,
+      "learning_rate": 4.694032807430997e-07,
+      "loss": 1.9093,
+      "step": 6709
+    },
+    {
+      "epoch": 2.8945438861332757,
+      "grad_norm": 0.22454294562339783,
+      "learning_rate": 4.6559869605783675e-07,
+      "loss": 2.2015,
+      "step": 6710
+    },
+    {
+      "epoch": 2.894975199482424,
+      "grad_norm": 0.20868833363056183,
+      "learning_rate": 4.61809544535463e-07,
+      "loss": 2.0411,
+      "step": 6711
+    },
+    {
+      "epoch": 2.8954065128315722,
+      "grad_norm": 2.48274302482605,
+      "learning_rate": 4.580358269605594e-07,
+      "loss": 2.0262,
+      "step": 6712
+    },
+    {
+      "epoch": 2.8958378261807205,
+      "grad_norm": 0.19596897065639496,
+      "learning_rate": 4.542775441145174e-07,
+      "loss": 2.0699,
+      "step": 6713
+    },
+    {
+      "epoch": 2.8962691395298683,
+      "grad_norm": 0.8381218314170837,
+      "learning_rate": 4.50534696775523e-07,
+      "loss": 2.2512,
+      "step": 6714
+    },
+    {
+      "epoch": 2.8967004528790166,
+      "grad_norm": 0.22735945880413055,
+      "learning_rate": 4.4680728571858116e-07,
+      "loss": 2.2737,
+      "step": 6715
+    },
+    {
+      "epoch": 2.897131766228165,
+      "grad_norm": 0.20122380554676056,
+      "learning_rate": 4.430953117154912e-07,
+      "loss": 2.1732,
+      "step": 6716
+    },
+    {
+      "epoch": 2.897563079577313,
+      "grad_norm": 0.17533114552497864,
+      "learning_rate": 4.393987755348549e-07,
+      "loss": 2.1612,
+      "step": 6717
+    },
+    {
+      "epoch": 2.897994392926461,
+      "grad_norm": 0.18631507456302643,
+      "learning_rate": 4.357176779420851e-07,
+      "loss": 2.2448,
+      "step": 6718
+    },
+    {
+      "epoch": 2.898425706275609,
+      "grad_norm": 0.19202157855033875,
+      "learning_rate": 4.320520196993971e-07,
+      "loss": 2.0164,
+      "step": 6719
+    },
+    {
+      "epoch": 2.8988570196247574,
+      "grad_norm": 0.20462560653686523,
+      "learning_rate": 4.28401801565792e-07,
+      "loss": 2.0823,
+      "step": 6720
+    },
+    {
+      "epoch": 2.8992883329739056,
+      "grad_norm": 0.20073947310447693,
+      "learning_rate": 4.247670242970985e-07,
+      "loss": 2.049,
+      "step": 6721
+    },
+    {
+      "epoch": 2.899719646323054,
+      "grad_norm": 0.2194797843694687,
+      "learning_rate": 4.211476886459314e-07,
+      "loss": 2.2606,
+      "step": 6722
+    },
+    {
+      "epoch": 2.9001509596722017,
+      "grad_norm": 0.19137798249721527,
+      "learning_rate": 4.175437953617161e-07,
+      "loss": 2.2349,
+      "step": 6723
+    },
+    {
+      "epoch": 2.90058227302135,
+      "grad_norm": 0.19921918213367462,
+      "learning_rate": 4.1395534519068076e-07,
+      "loss": 2.4475,
+      "step": 6724
+    },
+    {
+      "epoch": 2.901013586370498,
+      "grad_norm": 0.21443259716033936,
+      "learning_rate": 4.103823388758476e-07,
+      "loss": 2.3751,
+      "step": 6725
+    },
+    {
+      "epoch": 2.901013586370498,
+      "eval_loss": 2.0870168209075928,
+      "eval_runtime": 195.6161,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6725
+    },
+    {
+      "epoch": 2.9014448997196465,
+      "grad_norm": 0.19477905333042145,
+      "learning_rate": 4.0682477715704986e-07,
+      "loss": 1.9528,
+      "step": 6726
+    },
+    {
+      "epoch": 2.9018762130687943,
+      "grad_norm": 0.19327762722969055,
+      "learning_rate": 4.03282660770915e-07,
+      "loss": 2.1517,
+      "step": 6727
+    },
+    {
+      "epoch": 2.9023075264179425,
+      "grad_norm": 0.6009685397148132,
+      "learning_rate": 3.9975599045088955e-07,
+      "loss": 2.0693,
+      "step": 6728
+    },
+    {
+      "epoch": 2.902738839767091,
+      "grad_norm": 0.1924435794353485,
+      "learning_rate": 3.962447669271979e-07,
+      "loss": 2.2527,
+      "step": 6729
+    },
+    {
+      "epoch": 2.903170153116239,
+      "grad_norm": 0.20312826335430145,
+      "learning_rate": 3.9274899092687497e-07,
+      "loss": 2.267,
+      "step": 6730
+    },
+    {
+      "epoch": 2.9036014664653873,
+      "grad_norm": 0.1943555325269699,
+      "learning_rate": 3.892686631737668e-07,
+      "loss": 2.1596,
+      "step": 6731
+    },
+    {
+      "epoch": 2.904032779814535,
+      "grad_norm": 0.1989336460828781,
+      "learning_rate": 3.8580378438851367e-07,
+      "loss": 2.2043,
+      "step": 6732
+    },
+    {
+      "epoch": 2.9044640931636834,
+      "grad_norm": 0.19924335181713104,
+      "learning_rate": 3.823543552885583e-07,
+      "loss": 2.0053,
+      "step": 6733
+    },
+    {
+      "epoch": 2.9048954065128316,
+      "grad_norm": 0.20633479952812195,
+      "learning_rate": 3.7892037658812933e-07,
+      "loss": 2.2411,
+      "step": 6734
+    },
+    {
+      "epoch": 2.90532671986198,
+      "grad_norm": 0.1956966370344162,
+      "learning_rate": 3.7550184899828304e-07,
+      "loss": 2.1115,
+      "step": 6735
+    },
+    {
+      "epoch": 2.9057580332111277,
+      "grad_norm": 0.19222424924373627,
+      "learning_rate": 3.720987732268532e-07,
+      "loss": 2.1966,
+      "step": 6736
+    },
+    {
+      "epoch": 2.906189346560276,
+      "grad_norm": 0.20548641681671143,
+      "learning_rate": 3.6871114997850115e-07,
+      "loss": 2.3462,
+      "step": 6737
+    },
+    {
+      "epoch": 2.906620659909424,
+      "grad_norm": 0.21557635068893433,
+      "learning_rate": 3.6533897995464913e-07,
+      "loss": 1.9703,
+      "step": 6738
+    },
+    {
+      "epoch": 2.9070519732585725,
+      "grad_norm": 0.217384472489357,
+      "learning_rate": 3.619822638535552e-07,
+      "loss": 2.2944,
+      "step": 6739
+    },
+    {
+      "epoch": 2.9074832866077207,
+      "grad_norm": 0.20118100941181183,
+      "learning_rate": 3.586410023702635e-07,
+      "loss": 1.9737,
+      "step": 6740
+    },
+    {
+      "epoch": 2.9079145999568685,
+      "grad_norm": 0.19280241429805756,
+      "learning_rate": 3.553151961966122e-07,
+      "loss": 2.1149,
+      "step": 6741
+    },
+    {
+      "epoch": 2.9083459133060168,
+      "grad_norm": 0.18102608621120453,
+      "learning_rate": 3.5200484602125045e-07,
+      "loss": 2.0382,
+      "step": 6742
+    },
+    {
+      "epoch": 2.908777226655165,
+      "grad_norm": 0.19837021827697754,
+      "learning_rate": 3.4870995252962994e-07,
+      "loss": 2.2212,
+      "step": 6743
+    },
+    {
+      "epoch": 2.9092085400043133,
+      "grad_norm": 0.20592765510082245,
+      "learning_rate": 3.4543051640398e-07,
+      "loss": 2.28,
+      "step": 6744
+    },
+    {
+      "epoch": 2.909639853353461,
+      "grad_norm": 0.21061819791793823,
+      "learning_rate": 3.4216653832334904e-07,
+      "loss": 2.0615,
+      "step": 6745
+    },
+    {
+      "epoch": 2.9100711667026093,
+      "grad_norm": 0.19395215809345245,
+      "learning_rate": 3.3891801896357984e-07,
+      "loss": 2.0634,
+      "step": 6746
+    },
+    {
+      "epoch": 2.9105024800517576,
+      "grad_norm": 0.20471517741680145,
+      "learning_rate": 3.3568495899731765e-07,
+      "loss": 2.2366,
+      "step": 6747
+    },
+    {
+      "epoch": 2.910933793400906,
+      "grad_norm": 0.9463796615600586,
+      "learning_rate": 3.3246735909400193e-07,
+      "loss": 2.1938,
+      "step": 6748
+    },
+    {
+      "epoch": 2.911365106750054,
+      "grad_norm": 0.19731606543064117,
+      "learning_rate": 3.2926521991986646e-07,
+      "loss": 2.0529,
+      "step": 6749
+    },
+    {
+      "epoch": 2.911796420099202,
+      "grad_norm": 0.21093003451824188,
+      "learning_rate": 3.260785421379475e-07,
+      "loss": 2.1669,
+      "step": 6750
+    },
+    {
+      "epoch": 2.911796420099202,
+      "eval_loss": 2.087052345275879,
+      "eval_runtime": 195.3611,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6750
+    },
+    {
+      "epoch": 2.91222773344835,
+      "grad_norm": 0.2138030230998993,
+      "learning_rate": 3.2290732640809223e-07,
+      "loss": 2.3074,
+      "step": 6751
+    },
+    {
+      "epoch": 2.9126590467974984,
+      "grad_norm": 0.20158720016479492,
+      "learning_rate": 3.197515733869338e-07,
+      "loss": 2.1314,
+      "step": 6752
+    },
+    {
+      "epoch": 2.9130903601466467,
+      "grad_norm": 0.20197947323322296,
+      "learning_rate": 3.166112837278995e-07,
+      "loss": 2.0844,
+      "step": 6753
+    },
+    {
+      "epoch": 2.9135216734957945,
+      "grad_norm": 0.20278272032737732,
+      "learning_rate": 3.134864580812191e-07,
+      "loss": 2.1534,
+      "step": 6754
+    },
+    {
+      "epoch": 2.9139529868449427,
+      "grad_norm": 0.21093043684959412,
+      "learning_rate": 3.103770970939251e-07,
+      "loss": 2.2496,
+      "step": 6755
+    },
+    {
+      "epoch": 2.914384300194091,
+      "grad_norm": 0.1912979632616043,
+      "learning_rate": 3.072832014098442e-07,
+      "loss": 2.195,
+      "step": 6756
+    },
+    {
+      "epoch": 2.9148156135432393,
+      "grad_norm": 0.2099740356206894,
+      "learning_rate": 3.0420477166960557e-07,
+      "loss": 2.1556,
+      "step": 6757
+    },
+    {
+      "epoch": 2.9152469268923875,
+      "grad_norm": 0.20231591165065765,
+      "learning_rate": 3.0114180851063265e-07,
+      "loss": 2.2292,
+      "step": 6758
+    },
+    {
+      "epoch": 2.9156782402415353,
+      "grad_norm": 0.19204868376255035,
+      "learning_rate": 2.980943125671348e-07,
+      "loss": 2.0469,
+      "step": 6759
+    },
+    {
+      "epoch": 2.9161095535906836,
+      "grad_norm": 0.212842658162117,
+      "learning_rate": 2.950622844701406e-07,
+      "loss": 2.2559,
+      "step": 6760
+    },
+    {
+      "epoch": 2.916540866939832,
+      "grad_norm": 0.20621280372142792,
+      "learning_rate": 2.920457248474561e-07,
+      "loss": 2.2678,
+      "step": 6761
+    },
+    {
+      "epoch": 2.91697218028898,
+      "grad_norm": 0.21888796985149384,
+      "learning_rate": 2.890446343236902e-07,
+      "loss": 2.1982,
+      "step": 6762
+    },
+    {
+      "epoch": 2.917403493638128,
+      "grad_norm": 0.19971078634262085,
+      "learning_rate": 2.860590135202623e-07,
+      "loss": 2.1474,
+      "step": 6763
+    },
+    {
+      "epoch": 2.917834806987276,
+      "grad_norm": 0.21392220258712769,
+      "learning_rate": 2.8308886305537794e-07,
+      "loss": 2.2514,
+      "step": 6764
+    },
+    {
+      "epoch": 2.9182661203364244,
+      "grad_norm": 0.1981843262910843,
+      "learning_rate": 2.801341835440368e-07,
+      "loss": 2.1369,
+      "step": 6765
+    },
+    {
+      "epoch": 2.9186974336855727,
+      "grad_norm": 0.21266666054725647,
+      "learning_rate": 2.7719497559803295e-07,
+      "loss": 2.072,
+      "step": 6766
+    },
+    {
+      "epoch": 2.919128747034721,
+      "grad_norm": 0.20510557293891907,
+      "learning_rate": 2.7427123982596277e-07,
+      "loss": 2.1128,
+      "step": 6767
+    },
+    {
+      "epoch": 2.9195600603838687,
+      "grad_norm": 0.18966813385486603,
+      "learning_rate": 2.7136297683321706e-07,
+      "loss": 2.3035,
+      "step": 6768
+    },
+    {
+      "epoch": 2.919991373733017,
+      "grad_norm": 0.20073847472667694,
+      "learning_rate": 2.684701872219891e-07,
+      "loss": 2.1059,
+      "step": 6769
+    },
+    {
+      "epoch": 2.9204226870821652,
+      "grad_norm": 0.20553511381149292,
+      "learning_rate": 2.655928715912581e-07,
+      "loss": 2.1751,
+      "step": 6770
+    },
+    {
+      "epoch": 2.9208540004313135,
+      "grad_norm": 0.20025672018527985,
+      "learning_rate": 2.6273103053680576e-07,
+      "loss": 2.1073,
+      "step": 6771
+    },
+    {
+      "epoch": 2.9212853137804613,
+      "grad_norm": 0.18325988948345184,
+      "learning_rate": 2.5988466465119983e-07,
+      "loss": 2.0485,
+      "step": 6772
+    },
+    {
+      "epoch": 2.9217166271296096,
+      "grad_norm": 0.19736038148403168,
+      "learning_rate": 2.570537745238188e-07,
+      "loss": 2.127,
+      "step": 6773
+    },
+    {
+      "epoch": 2.922147940478758,
+      "grad_norm": 0.213426873087883,
+      "learning_rate": 2.5423836074083546e-07,
+      "loss": 2.4094,
+      "step": 6774
+    },
+    {
+      "epoch": 2.922579253827906,
+      "grad_norm": 0.2321801632642746,
+      "learning_rate": 2.5143842388520854e-07,
+      "loss": 2.1095,
+      "step": 6775
+    },
+    {
+      "epoch": 2.922579253827906,
+      "eval_loss": 2.0869569778442383,
+      "eval_runtime": 195.3991,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6775
+    },
+    {
+      "epoch": 2.9230105671770543,
+      "grad_norm": 0.20658846199512482,
+      "learning_rate": 2.4865396453668254e-07,
+      "loss": 2.0625,
+      "step": 6776
+    },
+    {
+      "epoch": 2.923441880526202,
+      "grad_norm": 0.20817962288856506,
+      "learning_rate": 2.458849832718213e-07,
+      "loss": 2.1999,
+      "step": 6777
+    },
+    {
+      "epoch": 2.9238731938753504,
+      "grad_norm": 0.20461788773536682,
+      "learning_rate": 2.431314806639828e-07,
+      "loss": 2.0633,
+      "step": 6778
+    },
+    {
+      "epoch": 2.9243045072244986,
+      "grad_norm": 0.19786451756954193,
+      "learning_rate": 2.403934572832861e-07,
+      "loss": 2.0404,
+      "step": 6779
+    },
+    {
+      "epoch": 2.924735820573647,
+      "grad_norm": 0.213621124625206,
+      "learning_rate": 2.3767091369668588e-07,
+      "loss": 2.2852,
+      "step": 6780
+    },
+    {
+      "epoch": 2.9251671339227947,
+      "grad_norm": 0.1997172236442566,
+      "learning_rate": 2.3496385046790634e-07,
+      "loss": 2.0288,
+      "step": 6781
+    },
+    {
+      "epoch": 2.925598447271943,
+      "grad_norm": 0.1925685852766037,
+      "learning_rate": 2.3227226815748235e-07,
+      "loss": 2.1788,
+      "step": 6782
+    },
+    {
+      "epoch": 2.926029760621091,
+      "grad_norm": 0.2174510508775711,
+      "learning_rate": 2.2959616732273478e-07,
+      "loss": 2.1952,
+      "step": 6783
+    },
+    {
+      "epoch": 2.9264610739702395,
+      "grad_norm": 0.2124984711408615,
+      "learning_rate": 2.269355485177704e-07,
+      "loss": 2.1769,
+      "step": 6784
+    },
+    {
+      "epoch": 2.9268923873193877,
+      "grad_norm": 0.17805612087249756,
+      "learning_rate": 2.2429041229350686e-07,
+      "loss": 2.1134,
+      "step": 6785
+    },
+    {
+      "epoch": 2.9273237006685355,
+      "grad_norm": 0.19361688196659088,
+      "learning_rate": 2.216607591976477e-07,
+      "loss": 2.1236,
+      "step": 6786
+    },
+    {
+      "epoch": 2.927755014017684,
+      "grad_norm": 0.17942456901073456,
+      "learning_rate": 2.190465897746907e-07,
+      "loss": 1.9414,
+      "step": 6787
+    },
+    {
+      "epoch": 2.928186327366832,
+      "grad_norm": 0.21018028259277344,
+      "learning_rate": 2.1644790456592786e-07,
+      "loss": 1.942,
+      "step": 6788
+    },
+    {
+      "epoch": 2.9286176407159803,
+      "grad_norm": 0.21197490394115448,
+      "learning_rate": 2.1386470410943713e-07,
+      "loss": 2.0872,
+      "step": 6789
+    },
+    {
+      "epoch": 2.929048954065128,
+      "grad_norm": 0.1928757131099701,
+      "learning_rate": 2.112969889401156e-07,
+      "loss": 2.1669,
+      "step": 6790
+    },
+    {
+      "epoch": 2.9294802674142764,
+      "grad_norm": 0.1942148655653,
+      "learning_rate": 2.0874475958962967e-07,
+      "loss": 2.1107,
+      "step": 6791
+    },
+    {
+      "epoch": 2.9299115807634246,
+      "grad_norm": 0.18924149870872498,
+      "learning_rate": 2.0620801658643992e-07,
+      "loss": 1.922,
+      "step": 6792
+    },
+    {
+      "epoch": 2.930342894112573,
+      "grad_norm": 0.22006838023662567,
+      "learning_rate": 2.0368676045580957e-07,
+      "loss": 2.1127,
+      "step": 6793
+    },
+    {
+      "epoch": 2.930774207461721,
+      "grad_norm": 0.1920243501663208,
+      "learning_rate": 2.0118099171979595e-07,
+      "loss": 2.1766,
+      "step": 6794
+    },
+    {
+      "epoch": 2.931205520810869,
+      "grad_norm": 0.19455263018608093,
+      "learning_rate": 1.9869071089725075e-07,
+      "loss": 2.0103,
+      "step": 6795
+    },
+    {
+      "epoch": 2.931636834160017,
+      "grad_norm": 0.18723560869693756,
+      "learning_rate": 1.9621591850379482e-07,
+      "loss": 1.9857,
+      "step": 6796
+    },
+    {
+      "epoch": 2.9320681475091654,
+      "grad_norm": 0.20894886553287506,
+      "learning_rate": 1.9375661505188488e-07,
+      "loss": 1.9011,
+      "step": 6797
+    },
+    {
+      "epoch": 2.9324994608583137,
+      "grad_norm": 0.20964136719703674,
+      "learning_rate": 1.9131280105073032e-07,
+      "loss": 2.1019,
+      "step": 6798
+    },
+    {
+      "epoch": 2.9329307742074615,
+      "grad_norm": 0.18412867188453674,
+      "learning_rate": 1.8888447700635134e-07,
+      "loss": 1.8822,
+      "step": 6799
+    },
+    {
+      "epoch": 2.9333620875566098,
+      "grad_norm": 0.20818638801574707,
+      "learning_rate": 1.864716434215624e-07,
+      "loss": 2.2422,
+      "step": 6800
+    },
+    {
+      "epoch": 2.9333620875566098,
+      "eval_loss": 2.0870261192321777,
+      "eval_runtime": 195.3813,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6800
+    },
+    {
+      "epoch": 2.933793400905758,
+      "grad_norm": 0.18927675485610962,
+      "learning_rate": 1.8407430079597217e-07,
+      "loss": 2.1812,
+      "step": 6801
+    },
+    {
+      "epoch": 2.9342247142549063,
+      "grad_norm": 0.27150091528892517,
+      "learning_rate": 1.8169244962596695e-07,
+      "loss": 2.0754,
+      "step": 6802
+    },
+    {
+      "epoch": 2.9346560276040545,
+      "grad_norm": 0.7071654200553894,
+      "learning_rate": 1.7932609040474388e-07,
+      "loss": 2.107,
+      "step": 6803
+    },
+    {
+      "epoch": 2.9350873409532023,
+      "grad_norm": 0.20735763013362885,
+      "learning_rate": 1.7697522362226934e-07,
+      "loss": 2.2393,
+      "step": 6804
+    },
+    {
+      "epoch": 2.9355186543023506,
+      "grad_norm": 0.22723041474819183,
+      "learning_rate": 1.7463984976532896e-07,
+      "loss": 2.2172,
+      "step": 6805
+    },
+    {
+      "epoch": 2.935949967651499,
+      "grad_norm": 0.20030942559242249,
+      "learning_rate": 1.7231996931749425e-07,
+      "loss": 2.2804,
+      "step": 6806
+    },
+    {
+      "epoch": 2.936381281000647,
+      "grad_norm": 0.9719837307929993,
+      "learning_rate": 1.7001558275909766e-07,
+      "loss": 2.1877,
+      "step": 6807
+    },
+    {
+      "epoch": 2.936812594349795,
+      "grad_norm": 0.20374545454978943,
+      "learning_rate": 1.677266905673158e-07,
+      "loss": 2.0767,
+      "step": 6808
+    },
+    {
+      "epoch": 2.937243907698943,
+      "grad_norm": 0.2079235166311264,
+      "learning_rate": 1.6545329321606127e-07,
+      "loss": 2.2619,
+      "step": 6809
+    },
+    {
+      "epoch": 2.9376752210480914,
+      "grad_norm": 0.21248282492160797,
+      "learning_rate": 1.631953911760825e-07,
+      "loss": 2.1212,
+      "step": 6810
+    },
+    {
+      "epoch": 2.9381065343972397,
+      "grad_norm": 0.19418877363204956,
+      "learning_rate": 1.6095298491490548e-07,
+      "loss": 2.2536,
+      "step": 6811
+    },
+    {
+      "epoch": 2.938537847746388,
+      "grad_norm": 0.1790664941072464,
+      "learning_rate": 1.5872607489683387e-07,
+      "loss": 2.1337,
+      "step": 6812
+    },
+    {
+      "epoch": 2.9389691610955357,
+      "grad_norm": 0.19266265630722046,
+      "learning_rate": 1.5651466158297377e-07,
+      "loss": 2.1115,
+      "step": 6813
+    },
+    {
+      "epoch": 2.939400474444684,
+      "grad_norm": 0.19078189134597778,
+      "learning_rate": 1.543187454312339e-07,
+      "loss": 2.3414,
+      "step": 6814
+    },
+    {
+      "epoch": 2.9398317877938323,
+      "grad_norm": 0.1999526023864746,
+      "learning_rate": 1.5213832689630057e-07,
+      "loss": 2.0213,
+      "step": 6815
+    },
+    {
+      "epoch": 2.9402631011429805,
+      "grad_norm": 0.19977650046348572,
+      "learning_rate": 1.4997340642963762e-07,
+      "loss": 2.0637,
+      "step": 6816
+    },
+    {
+      "epoch": 2.9406944144921283,
+      "grad_norm": 0.19836311042308807,
+      "learning_rate": 1.4782398447953646e-07,
+      "loss": 2.2932,
+      "step": 6817
+    },
+    {
+      "epoch": 2.9411257278412766,
+      "grad_norm": 0.19646064937114716,
+      "learning_rate": 1.4569006149104944e-07,
+      "loss": 2.0503,
+      "step": 6818
+    },
+    {
+      "epoch": 2.941557041190425,
+      "grad_norm": 0.1919003129005432,
+      "learning_rate": 1.4357163790602312e-07,
+      "loss": 2.2208,
+      "step": 6819
+    },
+    {
+      "epoch": 2.941988354539573,
+      "grad_norm": 0.1872098445892334,
+      "learning_rate": 1.4146871416310657e-07,
+      "loss": 2.069,
+      "step": 6820
+    },
+    {
+      "epoch": 2.9424196678887213,
+      "grad_norm": 0.21568883955478668,
+      "learning_rate": 1.3938129069772652e-07,
+      "loss": 2.1917,
+      "step": 6821
+    },
+    {
+      "epoch": 2.942850981237869,
+      "grad_norm": 0.2069755494594574,
+      "learning_rate": 1.3730936794212056e-07,
+      "loss": 2.0024,
+      "step": 6822
+    },
+    {
+      "epoch": 2.9432822945870174,
+      "grad_norm": 0.19497476518154144,
+      "learning_rate": 1.3525294632528716e-07,
+      "loss": 2.1144,
+      "step": 6823
+    },
+    {
+      "epoch": 2.9437136079361657,
+      "grad_norm": 0.20551230013370514,
+      "learning_rate": 1.3321202627304407e-07,
+      "loss": 2.1161,
+      "step": 6824
+    },
+    {
+      "epoch": 2.944144921285314,
+      "grad_norm": 0.19996224343776703,
+      "learning_rate": 1.3118660820797822e-07,
+      "loss": 2.2457,
+      "step": 6825
+    },
+    {
+      "epoch": 2.944144921285314,
+      "eval_loss": 2.0869040489196777,
+      "eval_runtime": 195.4597,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6825
+    },
+    {
+      "epoch": 2.9445762346344617,
+      "grad_norm": 3.4457085132598877,
+      "learning_rate": 1.2917669254947915e-07,
+      "loss": 2.2843,
+      "step": 6826
+    },
+    {
+      "epoch": 2.94500754798361,
+      "grad_norm": 0.19179941713809967,
+      "learning_rate": 1.2718227971371397e-07,
+      "loss": 2.1372,
+      "step": 6827
+    },
+    {
+      "epoch": 2.9454388613327582,
+      "grad_norm": 0.20089589059352875,
+      "learning_rate": 1.2520337011366066e-07,
+      "loss": 2.069,
+      "step": 6828
+    },
+    {
+      "epoch": 2.9458701746819065,
+      "grad_norm": 0.19916307926177979,
+      "learning_rate": 1.2323996415905814e-07,
+      "loss": 2.3106,
+      "step": 6829
+    },
+    {
+      "epoch": 2.9463014880310547,
+      "grad_norm": 0.26394152641296387,
+      "learning_rate": 1.212920622564645e-07,
+      "loss": 2.0409,
+      "step": 6830
+    },
+    {
+      "epoch": 2.946732801380203,
+      "grad_norm": 0.1933462917804718,
+      "learning_rate": 1.193596648092071e-07,
+      "loss": 2.2213,
+      "step": 6831
+    },
+    {
+      "epoch": 2.947164114729351,
+      "grad_norm": 0.19561220705509186,
+      "learning_rate": 1.1744277221741594e-07,
+      "loss": 2.1113,
+      "step": 6832
+    },
+    {
+      "epoch": 2.947595428078499,
+      "grad_norm": 0.2019442617893219,
+      "learning_rate": 1.1554138487799847e-07,
+      "loss": 2.0393,
+      "step": 6833
+    },
+    {
+      "epoch": 2.9480267414276473,
+      "grad_norm": 0.20612679421901703,
+      "learning_rate": 1.1365550318465644e-07,
+      "loss": 2.1104,
+      "step": 6834
+    },
+    {
+      "epoch": 2.948458054776795,
+      "grad_norm": 0.2055290788412094,
+      "learning_rate": 1.1178512752788582e-07,
+      "loss": 2.1996,
+      "step": 6835
+    },
+    {
+      "epoch": 2.9488893681259434,
+      "grad_norm": 0.2121615707874298,
+      "learning_rate": 1.0993025829497682e-07,
+      "loss": 2.3635,
+      "step": 6836
+    },
+    {
+      "epoch": 2.9493206814750916,
+      "grad_norm": 0.19621258974075317,
+      "learning_rate": 1.0809089586998054e-07,
+      "loss": 2.1606,
+      "step": 6837
+    },
+    {
+      "epoch": 2.94975199482424,
+      "grad_norm": 0.2023998647928238,
+      "learning_rate": 1.0626704063377567e-07,
+      "loss": 2.3725,
+      "step": 6838
+    },
+    {
+      "epoch": 2.950183308173388,
+      "grad_norm": 0.20256933569908142,
+      "learning_rate": 1.0445869296400178e-07,
+      "loss": 2.1603,
+      "step": 6839
+    },
+    {
+      "epoch": 2.9506146215225364,
+      "grad_norm": 0.1928139477968216,
+      "learning_rate": 1.0266585323510102e-07,
+      "loss": 2.2596,
+      "step": 6840
+    },
+    {
+      "epoch": 2.951045934871684,
+      "grad_norm": 0.18831153213977814,
+      "learning_rate": 1.008885218182931e-07,
+      "loss": 2.1917,
+      "step": 6841
+    },
+    {
+      "epoch": 2.9514772482208325,
+      "grad_norm": 0.21388104557991028,
+      "learning_rate": 9.912669908160031e-08,
+      "loss": 2.1683,
+      "step": 6842
+    },
+    {
+      "epoch": 2.9519085615699807,
+      "grad_norm": 0.18549636006355286,
+      "learning_rate": 9.738038538983084e-08,
+      "loss": 2.1971,
+      "step": 6843
+    },
+    {
+      "epoch": 2.9523398749191285,
+      "grad_norm": 0.20190787315368652,
+      "learning_rate": 9.564958110457877e-08,
+      "loss": 2.184,
+      "step": 6844
+    },
+    {
+      "epoch": 2.952771188268277,
+      "grad_norm": 0.20602178573608398,
+      "learning_rate": 9.393428658420743e-08,
+      "loss": 2.1097,
+      "step": 6845
+    },
+    {
+      "epoch": 2.953202501617425,
+      "grad_norm": 0.21606449782848358,
+      "learning_rate": 9.223450218391604e-08,
+      "loss": 2.136,
+      "step": 6846
+    },
+    {
+      "epoch": 2.9536338149665733,
+      "grad_norm": 0.21144703030586243,
+      "learning_rate": 9.055022825563974e-08,
+      "loss": 2.1398,
+      "step": 6847
+    },
+    {
+      "epoch": 2.9540651283157215,
+      "grad_norm": 0.22148758172988892,
+      "learning_rate": 8.888146514813287e-08,
+      "loss": 1.7514,
+      "step": 6848
+    },
+    {
+      "epoch": 2.95449644166487,
+      "grad_norm": 0.19042165577411652,
+      "learning_rate": 8.722821320693573e-08,
+      "loss": 2.0921,
+      "step": 6849
+    },
+    {
+      "epoch": 2.9549277550140176,
+      "grad_norm": 0.21668869256973267,
+      "learning_rate": 8.559047277436614e-08,
+      "loss": 2.178,
+      "step": 6850
+    },
+    {
+      "epoch": 2.9549277550140176,
+      "eval_loss": 2.086881160736084,
+      "eval_runtime": 195.3724,
+      "eval_samples_per_second": 0.164,
+      "eval_steps_per_second": 0.164,
+      "step": 6850
+    },
+    {
+      "epoch": 2.955359068363166,
+      "grad_norm": 0.2040349543094635,
+      "learning_rate": 8.39682441895445e-08,
+      "loss": 2.2474,
+      "step": 6851
+    },
+    {
+      "epoch": 2.955790381712314,
+      "grad_norm": 0.19920016825199127,
+      "learning_rate": 8.236152778836047e-08,
+      "loss": 1.9796,
+      "step": 6852
+    },
+    {
+      "epoch": 2.956221695061462,
+      "grad_norm": 0.19970017671585083,
+      "learning_rate": 8.077032390351457e-08,
+      "loss": 2.2889,
+      "step": 6853
+    },
+    {
+      "epoch": 2.95665300841061,
+      "grad_norm": 0.20651085674762726,
+      "learning_rate": 7.919463286446825e-08,
+      "loss": 2.1612,
+      "step": 6854
+    },
+    {
+      "epoch": 2.9570843217597584,
+      "grad_norm": 0.19311478734016418,
+      "learning_rate": 7.763445499750221e-08,
+      "loss": 2.2837,
+      "step": 6855
+    },
+    {
+      "epoch": 2.9575156351089067,
+      "grad_norm": 0.20735017955303192,
+      "learning_rate": 7.608979062564968e-08,
+      "loss": 2.3233,
+      "step": 6856
+    },
+    {
+      "epoch": 2.957946948458055,
+      "grad_norm": 0.19792737066745758,
+      "learning_rate": 7.45606400687715e-08,
+      "loss": 2.1628,
+      "step": 6857
+    },
+    {
+      "epoch": 2.958378261807203,
+      "grad_norm": 0.20489484071731567,
+      "learning_rate": 7.304700364347271e-08,
+      "loss": 2.076,
+      "step": 6858
+    },
+    {
+      "epoch": 2.958809575156351,
+      "grad_norm": 0.19102559983730316,
+      "learning_rate": 7.154888166317762e-08,
+      "loss": 1.9899,
+      "step": 6859
+    },
+    {
+      "epoch": 2.9592408885054993,
+      "grad_norm": 0.1942603439092636,
+      "learning_rate": 7.006627443809643e-08,
+      "loss": 2.2985,
+      "step": 6860
+    },
+    {
+      "epoch": 2.9596722018546475,
+      "grad_norm": 0.1981513351202011,
+      "learning_rate": 6.859918227521688e-08,
+      "loss": 2.1052,
+      "step": 6861
+    },
+    {
+      "epoch": 2.9601035152037953,
+      "grad_norm": 0.1908348798751831,
+      "learning_rate": 6.714760547830433e-08,
+      "loss": 2.1002,
+      "step": 6862
+    },
+    {
+      "epoch": 2.9605348285529436,
+      "grad_norm": 0.20092721283435822,
+      "learning_rate": 6.571154434793502e-08,
+      "loss": 2.2455,
+      "step": 6863
+    },
+    {
+      "epoch": 2.960966141902092,
+      "grad_norm": 0.2005387842655182,
+      "learning_rate": 6.429099918146274e-08,
+      "loss": 2.1606,
+      "step": 6864
+    },
+    {
+      "epoch": 2.96139745525124,
+      "grad_norm": 0.2052379548549652,
+      "learning_rate": 6.28859702730189e-08,
+      "loss": 2.1535,
+      "step": 6865
+    },
+    {
+      "epoch": 2.9618287686003884,
+      "grad_norm": 0.21680006384849548,
+      "learning_rate": 6.149645791352908e-08,
+      "loss": 2.1254,
+      "step": 6866
+    },
+    {
+      "epoch": 2.9622600819495366,
+      "grad_norm": 0.19733987748622894,
+      "learning_rate": 6.012246239071317e-08,
+      "loss": 2.2219,
+      "step": 6867
+    },
+    {
+      "epoch": 2.9626913952986844,
+      "grad_norm": 0.21556080877780914,
+      "learning_rate": 5.876398398907689e-08,
+      "loss": 2.3021,
+      "step": 6868
+    },
+    {
+      "epoch": 2.9631227086478327,
+      "grad_norm": 0.17450863122940063,
+      "learning_rate": 5.7421022989886935e-08,
+      "loss": 1.8516,
+      "step": 6869
+    },
+    {
+      "epoch": 2.963554021996981,
+      "grad_norm": 0.20078998804092407,
+      "learning_rate": 5.609357967124584e-08,
+      "loss": 2.298,
+      "step": 6870
+    },
+    {
+      "epoch": 2.9639853353461287,
+      "grad_norm": 0.20523545145988464,
+      "learning_rate": 5.478165430800041e-08,
+      "loss": 2.1605,
+      "step": 6871
+    },
+    {
+      "epoch": 2.964416648695277,
+      "grad_norm": 0.24319957196712494,
+      "learning_rate": 5.3485247171791703e-08,
+      "loss": 2.306,
+      "step": 6872
+    },
+    {
+      "epoch": 2.9648479620444252,
+      "grad_norm": 13.429825782775879,
+      "learning_rate": 5.2204358531071636e-08,
+      "loss": 2.2616,
+      "step": 6873
+    },
+    {
+      "epoch": 2.9652792753935735,
+      "grad_norm": 0.20633433759212494,
+      "learning_rate": 5.093898865105306e-08,
+      "loss": 2.3296,
+      "step": 6874
+    },
+    {
+      "epoch": 2.9657105887427218,
+      "grad_norm": 0.1916525661945343,
+      "learning_rate": 4.9689137793743064e-08,
+      "loss": 2.1803,
+      "step": 6875
+    },
+    {
+      "epoch": 2.9657105887427218,
+      "eval_loss": 2.0869922637939453,
+      "eval_runtime": 212.9874,
+      "eval_samples_per_second": 0.15,
+      "eval_steps_per_second": 0.15,
+      "step": 6875
+    },
+    {
+      "epoch": 2.96614190209187,
+      "grad_norm": 0.1868002563714981,
+      "learning_rate": 4.845480621794295e-08,
+      "loss": 2.2291,
+      "step": 6876
+    },
+    {
+      "epoch": 2.966573215441018,
+      "grad_norm": 0.20137692987918854,
+      "learning_rate": 4.723599417923163e-08,
+      "loss": 2.1532,
+      "step": 6877
+    },
+    {
+      "epoch": 2.967004528790166,
+      "grad_norm": 0.19183698296546936,
+      "learning_rate": 4.6032701929973884e-08,
+      "loss": 2.1687,
+      "step": 6878
+    },
+    {
+      "epoch": 2.9674358421393143,
+      "grad_norm": 0.2428313046693802,
+      "learning_rate": 4.484492971932874e-08,
+      "loss": 2.2349,
+      "step": 6879
+    },
+    {
+      "epoch": 2.967867155488462,
+      "grad_norm": 0.2129974216222763,
+      "learning_rate": 4.367267779324113e-08,
+      "loss": 2.2224,
+      "step": 6880
+    },
+    {
+      "epoch": 2.9682984688376104,
+      "grad_norm": 0.20092099905014038,
+      "learning_rate": 4.251594639442524e-08,
+      "loss": 2.0661,
+      "step": 6881
+    },
+    {
+      "epoch": 2.9687297821867586,
+      "grad_norm": 0.18725912272930145,
+      "learning_rate": 4.137473576241446e-08,
+      "loss": 2.0123,
+      "step": 6882
+    },
+    {
+      "epoch": 2.969161095535907,
+      "grad_norm": 0.20052030682563782,
+      "learning_rate": 4.024904613348645e-08,
+      "loss": 2.3143,
+      "step": 6883
+    },
+    {
+      "epoch": 2.969592408885055,
+      "grad_norm": 0.20144249498844147,
+      "learning_rate": 3.913887774074642e-08,
+      "loss": 2.1777,
+      "step": 6884
+    },
+    {
+      "epoch": 2.9700237222342034,
+      "grad_norm": 0.1862250715494156,
+      "learning_rate": 3.8044230814060496e-08,
+      "loss": 2.0669,
+      "step": 6885
+    },
+    {
+      "epoch": 2.9704550355833512,
+      "grad_norm": 0.19004148244857788,
+      "learning_rate": 3.696510558008903e-08,
+      "loss": 2.1808,
+      "step": 6886
+    },
+    {
+      "epoch": 2.9708863489324995,
+      "grad_norm": 0.19828954339027405,
+      "learning_rate": 3.5901502262269954e-08,
+      "loss": 2.2483,
+      "step": 6887
+    },
+    {
+      "epoch": 2.9713176622816477,
+      "grad_norm": 0.22624792158603668,
+      "learning_rate": 3.4853421080835444e-08,
+      "loss": 2.1074,
+      "step": 6888
+    },
+    {
+      "epoch": 2.9717489756307955,
+      "grad_norm": 0.20023319125175476,
+      "learning_rate": 3.3820862252811884e-08,
+      "loss": 2.1283,
+      "step": 6889
+    },
+    {
+      "epoch": 2.972180288979944,
+      "grad_norm": 0.1807999610900879,
+      "learning_rate": 3.2803825991994934e-08,
+      "loss": 2.0663,
+      "step": 6890
+    },
+    {
+      "epoch": 2.972611602329092,
+      "grad_norm": 0.19238461554050446,
+      "learning_rate": 3.1802312508966146e-08,
+      "loss": 1.9663,
+      "step": 6891
+    },
+    {
+      "epoch": 2.9730429156782403,
+      "grad_norm": 0.1982019543647766,
+      "learning_rate": 3.081632201111794e-08,
+      "loss": 2.084,
+      "step": 6892
+    },
+    {
+      "epoch": 2.9734742290273886,
+      "grad_norm": 0.19389572739601135,
+      "learning_rate": 2.9845854702587043e-08,
+      "loss": 2.1794,
+      "step": 6893
+    },
+    {
+      "epoch": 2.973905542376537,
+      "grad_norm": 0.19226449728012085,
+      "learning_rate": 2.8890910784337694e-08,
+      "loss": 1.9868,
+      "step": 6894
+    },
+    {
+      "epoch": 2.9743368557256846,
+      "grad_norm": 0.20984889566898346,
+      "learning_rate": 2.7951490454095062e-08,
+      "loss": 2.0223,
+      "step": 6895
+    },
+    {
+      "epoch": 2.974768169074833,
+      "grad_norm": 0.42044931650161743,
+      "learning_rate": 2.7027593906378542e-08,
+      "loss": 2.2972,
+      "step": 6896
+    },
+    {
+      "epoch": 2.975199482423981,
+      "grad_norm": 0.2137708216905594,
+      "learning_rate": 2.611922133248512e-08,
+      "loss": 2.254,
+      "step": 6897
+    },
+    {
+      "epoch": 2.975630795773129,
+      "grad_norm": 0.18392716348171234,
+      "learning_rate": 2.5226372920514326e-08,
+      "loss": 2.0367,
+      "step": 6898
+    },
+    {
+      "epoch": 2.976062109122277,
+      "grad_norm": 0.20180554687976837,
+      "learning_rate": 2.4349048855326625e-08,
+      "loss": 2.1274,
+      "step": 6899
+    },
+    {
+      "epoch": 2.9764934224714255,
+      "grad_norm": 0.18473385274410248,
+      "learning_rate": 2.3487249318593358e-08,
+      "loss": 1.9986,
+      "step": 6900
+    },
+    {
+      "epoch": 2.9764934224714255,
+      "eval_loss": 2.086920738220215,
+      "eval_runtime": 207.6476,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 6900
+    },
+    {
+      "epoch": 2.9769247358205737,
+      "grad_norm": 0.20555371046066284,
+      "learning_rate": 2.2640974488746798e-08,
+      "loss": 2.2448,
+      "step": 6901
+    },
+    {
+      "epoch": 2.977356049169722,
+      "grad_norm": 0.18608687818050385,
+      "learning_rate": 2.1810224541030096e-08,
+      "loss": 2.1915,
+      "step": 6902
+    },
+    {
+      "epoch": 2.97778736251887,
+      "grad_norm": 0.19687053561210632,
+      "learning_rate": 2.0994999647455656e-08,
+      "loss": 2.1627,
+      "step": 6903
+    },
+    {
+      "epoch": 2.978218675868018,
+      "grad_norm": 0.20249271392822266,
+      "learning_rate": 2.0195299976821787e-08,
+      "loss": 2.2829,
+      "step": 6904
+    },
+    {
+      "epoch": 2.9786499892171663,
+      "grad_norm": 0.1894199401140213,
+      "learning_rate": 1.9411125694712704e-08,
+      "loss": 2.0305,
+      "step": 6905
+    },
+    {
+      "epoch": 2.9790813025663145,
+      "grad_norm": 0.20701386034488678,
+      "learning_rate": 1.864247696350685e-08,
+      "loss": 2.2719,
+      "step": 6906
+    },
+    {
+      "epoch": 2.9795126159154623,
+      "grad_norm": 0.18859870731830597,
+      "learning_rate": 1.788935394236024e-08,
+      "loss": 2.1539,
+      "step": 6907
+    },
+    {
+      "epoch": 2.9799439292646106,
+      "grad_norm": 0.2163713276386261,
+      "learning_rate": 1.7151756787206483e-08,
+      "loss": 2.0162,
+      "step": 6908
+    },
+    {
+      "epoch": 2.980375242613759,
+      "grad_norm": 0.21868695318698883,
+      "learning_rate": 1.642968565079006e-08,
+      "loss": 1.9967,
+      "step": 6909
+    },
+    {
+      "epoch": 2.980806555962907,
+      "grad_norm": 0.1910344958305359,
+      "learning_rate": 1.572314068260805e-08,
+      "loss": 2.0954,
+      "step": 6910
+    },
+    {
+      "epoch": 2.9812378693120554,
+      "grad_norm": 0.2031833529472351,
+      "learning_rate": 1.503212202896009e-08,
+      "loss": 2.2385,
+      "step": 6911
+    },
+    {
+      "epoch": 2.9816691826612036,
+      "grad_norm": 0.20792736113071442,
+      "learning_rate": 1.435662983294006e-08,
+      "loss": 2.162,
+      "step": 6912
+    },
+    {
+      "epoch": 2.9821004960103514,
+      "grad_norm": 0.22117304801940918,
+      "learning_rate": 1.369666423441107e-08,
+      "loss": 1.8649,
+      "step": 6913
+    },
+    {
+      "epoch": 2.9825318093594997,
+      "grad_norm": 0.2013591080904007,
+      "learning_rate": 1.3052225370013825e-08,
+      "loss": 2.0469,
+      "step": 6914
+    },
+    {
+      "epoch": 2.982963122708648,
+      "grad_norm": 0.3142205774784088,
+      "learning_rate": 1.2423313373199906e-08,
+      "loss": 2.0327,
+      "step": 6915
+    },
+    {
+      "epoch": 2.9833944360577958,
+      "grad_norm": 0.2305009365081787,
+      "learning_rate": 1.1809928374198475e-08,
+      "loss": 2.0315,
+      "step": 6916
+    },
+    {
+      "epoch": 2.983825749406944,
+      "grad_norm": 0.20702128112316132,
+      "learning_rate": 1.1212070499999615e-08,
+      "loss": 2.0369,
+      "step": 6917
+    },
+    {
+      "epoch": 2.9842570627560923,
+      "grad_norm": 0.18710415065288544,
+      "learning_rate": 1.0629739874404297e-08,
+      "loss": 1.9671,
+      "step": 6918
+    },
+    {
+      "epoch": 2.9846883761052405,
+      "grad_norm": 0.20689338445663452,
+      "learning_rate": 1.006293661799107e-08,
+      "loss": 2.1959,
+      "step": 6919
+    },
+    {
+      "epoch": 2.9851196894543888,
+      "grad_norm": 0.19970650970935822,
+      "learning_rate": 9.511660848132707e-09,
+      "loss": 2.1549,
+      "step": 6920
+    },
+    {
+      "epoch": 2.985551002803537,
+      "grad_norm": 0.19728341698646545,
+      "learning_rate": 8.975912678954589e-09,
+      "loss": 1.9665,
+      "step": 6921
+    },
+    {
+      "epoch": 2.985982316152685,
+      "grad_norm": 0.2005823254585266,
+      "learning_rate": 8.455692221409627e-09,
+      "loss": 2.3352,
+      "step": 6922
+    },
+    {
+      "epoch": 2.986413629501833,
+      "grad_norm": 0.20024988055229187,
+      "learning_rate": 7.950999583203333e-09,
+      "loss": 2.3154,
+      "step": 6923
+    },
+    {
+      "epoch": 2.9868449428509813,
+      "grad_norm": 0.19718816876411438,
+      "learning_rate": 7.461834868843775e-09,
+      "loss": 2.1285,
+      "step": 6924
+    },
+    {
+      "epoch": 2.987276256200129,
+      "grad_norm": 0.20390065014362335,
+      "learning_rate": 6.988198179616599e-09,
+      "loss": 2.1669,
+      "step": 6925
+    },
+    {
+      "epoch": 2.987276256200129,
+      "eval_loss": 2.0870180130004883,
+      "eval_runtime": 207.3305,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 6925
+    },
+    {
+      "epoch": 2.9877075695492774,
+      "grad_norm": 0.197320818901062,
+      "learning_rate": 6.530089613593359e-09,
+      "loss": 2.139,
+      "step": 6926
+    },
+    {
+      "epoch": 2.9881388828984257,
+      "grad_norm": 0.19850942492485046,
+      "learning_rate": 6.087509265623181e-09,
+      "loss": 2.1301,
+      "step": 6927
+    },
+    {
+      "epoch": 2.988570196247574,
+      "grad_norm": 0.252987802028656,
+      "learning_rate": 5.660457227357751e-09,
+      "loss": 2.1739,
+      "step": 6928
+    },
+    {
+      "epoch": 2.989001509596722,
+      "grad_norm": 0.20021499693393707,
+      "learning_rate": 5.248933587218007e-09,
+      "loss": 2.0975,
+      "step": 6929
+    },
+    {
+      "epoch": 2.9894328229458704,
+      "grad_norm": 0.18422454595565796,
+      "learning_rate": 4.852938430419118e-09,
+      "loss": 2.1753,
+      "step": 6930
+    },
+    {
+      "epoch": 2.9898641362950182,
+      "grad_norm": 0.19496819376945496,
+      "learning_rate": 4.4724718389538286e-09,
+      "loss": 2.1651,
+      "step": 6931
+    },
+    {
+      "epoch": 2.9902954496441665,
+      "grad_norm": 0.18620194494724274,
+      "learning_rate": 4.107533891592462e-09,
+      "loss": 1.9124,
+      "step": 6932
+    },
+    {
+      "epoch": 2.9907267629933147,
+      "grad_norm": 0.19842314720153809,
+      "learning_rate": 3.758124663916229e-09,
+      "loss": 2.1233,
+      "step": 6933
+    },
+    {
+      "epoch": 2.9911580763424626,
+      "grad_norm": 0.19758078455924988,
+      "learning_rate": 3.4242442282589322e-09,
+      "loss": 2.0477,
+      "step": 6934
+    },
+    {
+      "epoch": 2.991589389691611,
+      "grad_norm": 0.2080450803041458,
+      "learning_rate": 3.105892653765263e-09,
+      "loss": 2.1694,
+      "step": 6935
+    },
+    {
+      "epoch": 2.992020703040759,
+      "grad_norm": 0.20617592334747314,
+      "learning_rate": 2.803070006340835e-09,
+      "loss": 2.2902,
+      "step": 6936
+    },
+    {
+      "epoch": 2.9924520163899073,
+      "grad_norm": 0.20559188723564148,
+      "learning_rate": 2.5157763487021477e-09,
+      "loss": 1.8361,
+      "step": 6937
+    },
+    {
+      "epoch": 2.9928833297390556,
+      "grad_norm": 0.20320501923561096,
+      "learning_rate": 2.2440117403266233e-09,
+      "loss": 1.8617,
+      "step": 6938
+    },
+    {
+      "epoch": 2.993314643088204,
+      "grad_norm": 0.18574926257133484,
+      "learning_rate": 1.987776237494243e-09,
+      "loss": 2.0917,
+      "step": 6939
+    },
+    {
+      "epoch": 2.9937459564373516,
+      "grad_norm": 0.1990656554698944,
+      "learning_rate": 1.747069893262565e-09,
+      "loss": 2.3291,
+      "step": 6940
+    },
+    {
+      "epoch": 2.9941772697865,
+      "grad_norm": 0.20082135498523712,
+      "learning_rate": 1.5218927574583983e-09,
+      "loss": 2.1523,
+      "step": 6941
+    },
+    {
+      "epoch": 2.994608583135648,
+      "grad_norm": 0.2077145278453827,
+      "learning_rate": 1.312244876719437e-09,
+      "loss": 2.3263,
+      "step": 6942
+    },
+    {
+      "epoch": 2.995039896484796,
+      "grad_norm": 0.1954571008682251,
+      "learning_rate": 1.1181262944442993e-09,
+      "loss": 2.1045,
+      "step": 6943
+    },
+    {
+      "epoch": 2.995471209833944,
+      "grad_norm": 0.18653079867362976,
+      "learning_rate": 9.395370508424871e-10,
+      "loss": 1.9513,
+      "step": 6944
+    },
+    {
+      "epoch": 2.9959025231830925,
+      "grad_norm": 0.18665404617786407,
+      "learning_rate": 7.764771828844274e-10,
+      "loss": 2.1955,
+      "step": 6945
+    },
+    {
+      "epoch": 2.9963338365322407,
+      "grad_norm": 0.2045997977256775,
+      "learning_rate": 6.289467243347779e-10,
+      "loss": 2.2903,
+      "step": 6946
+    },
+    {
+      "epoch": 2.996765149881389,
+      "grad_norm": 0.18565250933170319,
+      "learning_rate": 4.969457057441007e-10,
+      "loss": 2.2809,
+      "step": 6947
+    },
+    {
+      "epoch": 2.9971964632305372,
+      "grad_norm": 0.21470415592193604,
+      "learning_rate": 3.8047415443220873e-10,
+      "loss": 2.1994,
+      "step": 6948
+    },
+    {
+      "epoch": 2.997627776579685,
+      "grad_norm": 0.20408982038497925,
+      "learning_rate": 2.795320945297996e-10,
+      "loss": 2.2236,
+      "step": 6949
+    },
+    {
+      "epoch": 2.9980590899288333,
+      "grad_norm": 0.20640811324119568,
+      "learning_rate": 1.9411954693682176e-10,
+      "loss": 2.0928,
+      "step": 6950
+    },
+    {
+      "epoch": 2.9980590899288333,
+      "eval_loss": 2.0869898796081543,
+      "eval_runtime": 207.3656,
+      "eval_samples_per_second": 0.154,
+      "eval_steps_per_second": 0.154,
+      "step": 6950
+    },
+    {
+      "epoch": 2.9984904032779816,
+      "grad_norm": 0.19671790301799774,
+      "learning_rate": 1.242365293308012e-10,
+      "loss": 2.1314,
+      "step": 6951
+    },
+    {
+      "epoch": 2.9989217166271294,
+      "grad_norm": 0.20118442177772522,
+      "learning_rate": 6.988305619182178e-11,
+      "loss": 2.0186,
+      "step": 6952
+    },
+    {
+      "epoch": 2.9993530299762776,
+      "grad_norm": 0.19977568089962006,
+      "learning_rate": 3.105913876089161e-11,
+      "loss": 2.1311,
+      "step": 6953
+    },
+    {
+      "epoch": 2.999784343325426,
+      "grad_norm": 0.2036544382572174,
+      "learning_rate": 7.764785089903192e-12,
+      "loss": 2.289,
+      "step": 6954
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 6954,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.62625932186227e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}