diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11132 @@
+{
+  "best_global_step": 1550,
+  "best_metric": 0.2745862603187561,
+  "best_model_checkpoint": "outputs/checkpoint-1550",
+  "epoch": 1.7337807606263982,
+  "eval_steps": 50,
+  "global_step": 1550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0011185682326621924,
+      "grad_norm": 0.697359561920166,
+      "learning_rate": 0.0,
+      "loss": 1.0228,
+      "step": 1
+    },
+    {
+      "epoch": 0.0022371364653243847,
+      "grad_norm": 0.8241124153137207,
+      "learning_rate": 1.858736059479554e-08,
+      "loss": 1.0078,
+      "step": 2
+    },
+    {
+      "epoch": 0.003355704697986577,
+      "grad_norm": 0.8992052674293518,
+      "learning_rate": 3.717472118959108e-08,
+      "loss": 1.1629,
+      "step": 3
+    },
+    {
+      "epoch": 0.0044742729306487695,
+      "grad_norm": 5.81605863571167,
+      "learning_rate": 5.576208178438662e-08,
+      "loss": 1.1078,
+      "step": 4
+    },
+    {
+      "epoch": 0.005592841163310962,
+      "grad_norm": 0.9619576930999756,
+      "learning_rate": 7.434944237918216e-08,
+      "loss": 1.1955,
+      "step": 5
+    },
+    {
+      "epoch": 0.006711409395973154,
+      "grad_norm": 1.0324803590774536,
+      "learning_rate": 9.293680297397771e-08,
+      "loss": 1.2107,
+      "step": 6
+    },
+    {
+      "epoch": 0.007829977628635347,
+      "grad_norm": 0.7469141483306885,
+      "learning_rate": 1.1152416356877324e-07,
+      "loss": 1.002,
+      "step": 7
+    },
+    {
+      "epoch": 0.008948545861297539,
+      "grad_norm": 1.2946940660476685,
+      "learning_rate": 1.3011152416356877e-07,
+      "loss": 1.0853,
+      "step": 8
+    },
+    {
+      "epoch": 0.010067114093959731,
+      "grad_norm": 0.9244082570075989,
+      "learning_rate": 1.4869888475836432e-07,
+      "loss": 1.1893,
+      "step": 9
+    },
+    {
+      "epoch": 0.011185682326621925,
+      "grad_norm": 0.7072787284851074,
+      "learning_rate": 1.6728624535315985e-07,
+      "loss": 1.0675,
+      "step": 10
+    },
+    {
+      "epoch": 0.012304250559284116,
+      "grad_norm": 0.8728421330451965,
+      "learning_rate": 1.8587360594795542e-07,
+      "loss": 1.1278,
+      "step": 11
+    },
+    {
+      "epoch": 0.013422818791946308,
+      "grad_norm": 0.9901434183120728,
+      "learning_rate": 2.0446096654275095e-07,
+      "loss": 1.0553,
+      "step": 12
+    },
+    {
+      "epoch": 0.0145413870246085,
+      "grad_norm": 0.9405534267425537,
+      "learning_rate": 2.2304832713754647e-07,
+      "loss": 1.1127,
+      "step": 13
+    },
+    {
+      "epoch": 0.015659955257270694,
+      "grad_norm": 0.9357879161834717,
+      "learning_rate": 2.4163568773234205e-07,
+      "loss": 1.2092,
+      "step": 14
+    },
+    {
+      "epoch": 0.016778523489932886,
+      "grad_norm": 1.0680477619171143,
+      "learning_rate": 2.6022304832713755e-07,
+      "loss": 1.1922,
+      "step": 15
+    },
+    {
+      "epoch": 0.017897091722595078,
+      "grad_norm": 2.275601863861084,
+      "learning_rate": 2.788104089219331e-07,
+      "loss": 1.0646,
+      "step": 16
+    },
+    {
+      "epoch": 0.01901565995525727,
+      "grad_norm": 0.8927004337310791,
+      "learning_rate": 2.9739776951672865e-07,
+      "loss": 1.1961,
+      "step": 17
+    },
+    {
+      "epoch": 0.020134228187919462,
+      "grad_norm": 0.8354470133781433,
+      "learning_rate": 3.1598513011152414e-07,
+      "loss": 1.0342,
+      "step": 18
+    },
+    {
+      "epoch": 0.021252796420581657,
+      "grad_norm": 0.8616951704025269,
+      "learning_rate": 3.345724907063197e-07,
+      "loss": 1.0944,
+      "step": 19
+    },
+    {
+      "epoch": 0.02237136465324385,
+      "grad_norm": 0.8386662006378174,
+      "learning_rate": 3.531598513011153e-07,
+      "loss": 1.2301,
+      "step": 20
+    },
+    {
+      "epoch": 0.02348993288590604,
+      "grad_norm": 0.843434751033783,
+      "learning_rate": 3.7174721189591085e-07,
+      "loss": 1.1133,
+      "step": 21
+    },
+    {
+      "epoch": 0.024608501118568233,
+      "grad_norm": 0.7803885340690613,
+      "learning_rate": 3.9033457249070635e-07,
+      "loss": 1.1066,
+      "step": 22
+    },
+    {
+      "epoch": 0.025727069351230425,
+      "grad_norm": 0.8580071330070496,
+      "learning_rate": 4.089219330855019e-07,
+      "loss": 1.1159,
+      "step": 23
+    },
+    {
+      "epoch": 0.026845637583892617,
+      "grad_norm": 0.8866803050041199,
+      "learning_rate": 4.2750929368029745e-07,
+      "loss": 1.166,
+      "step": 24
+    },
+    {
+      "epoch": 0.02796420581655481,
+      "grad_norm": 0.8940161466598511,
+      "learning_rate": 4.4609665427509294e-07,
+      "loss": 0.9731,
+      "step": 25
+    },
+    {
+      "epoch": 0.029082774049217,
+      "grad_norm": 1.0363571643829346,
+      "learning_rate": 4.646840148698885e-07,
+      "loss": 1.2274,
+      "step": 26
+    },
+    {
+      "epoch": 0.030201342281879196,
+      "grad_norm": 8.598701477050781,
+      "learning_rate": 4.832713754646841e-07,
+      "loss": 1.1079,
+      "step": 27
+    },
+    {
+      "epoch": 0.03131991051454139,
+      "grad_norm": 0.8069610595703125,
+      "learning_rate": 5.018587360594796e-07,
+      "loss": 1.0165,
+      "step": 28
+    },
+    {
+      "epoch": 0.03243847874720358,
+      "grad_norm": 0.7143051624298096,
+      "learning_rate": 5.204460966542751e-07,
+      "loss": 0.9828,
+      "step": 29
+    },
+    {
+      "epoch": 0.03355704697986577,
+      "grad_norm": 0.8553173542022705,
+      "learning_rate": 5.390334572490706e-07,
+      "loss": 1.0852,
+      "step": 30
+    },
+    {
+      "epoch": 0.03467561521252797,
+      "grad_norm": 0.9192616939544678,
+      "learning_rate": 5.576208178438662e-07,
+      "loss": 1.1716,
+      "step": 31
+    },
+    {
+      "epoch": 0.035794183445190156,
+      "grad_norm": 0.8169634938240051,
+      "learning_rate": 5.762081784386617e-07,
+      "loss": 1.0572,
+      "step": 32
+    },
+    {
+      "epoch": 0.03691275167785235,
+      "grad_norm": 0.9388527274131775,
+      "learning_rate": 5.947955390334573e-07,
+      "loss": 1.2699,
+      "step": 33
+    },
+    {
+      "epoch": 0.03803131991051454,
+      "grad_norm": 0.8063773512840271,
+      "learning_rate": 6.133828996282528e-07,
+      "loss": 1.0923,
+      "step": 34
+    },
+    {
+      "epoch": 0.039149888143176735,
+      "grad_norm": 1.6729189157485962,
+      "learning_rate": 6.319702602230483e-07,
+      "loss": 1.1288,
+      "step": 35
+    },
+    {
+      "epoch": 0.040268456375838924,
+      "grad_norm": 0.6592870354652405,
+      "learning_rate": 6.505576208178439e-07,
+      "loss": 1.0333,
+      "step": 36
+    },
+    {
+      "epoch": 0.04138702460850112,
+      "grad_norm": 1.0659452676773071,
+      "learning_rate": 6.691449814126394e-07,
+      "loss": 1.1166,
+      "step": 37
+    },
+    {
+      "epoch": 0.042505592841163314,
+      "grad_norm": 0.9113427996635437,
+      "learning_rate": 6.87732342007435e-07,
+      "loss": 1.1662,
+      "step": 38
+    },
+    {
+      "epoch": 0.0436241610738255,
+      "grad_norm": 0.9717181324958801,
+      "learning_rate": 7.063197026022306e-07,
+      "loss": 1.0379,
+      "step": 39
+    },
+    {
+      "epoch": 0.0447427293064877,
+      "grad_norm": 0.7156904339790344,
+      "learning_rate": 7.24907063197026e-07,
+      "loss": 0.9135,
+      "step": 40
+    },
+    {
+      "epoch": 0.04586129753914989,
+      "grad_norm": 0.74992436170578,
+      "learning_rate": 7.434944237918217e-07,
+      "loss": 0.9566,
+      "step": 41
+    },
+    {
+      "epoch": 0.04697986577181208,
+      "grad_norm": 1.3575512170791626,
+      "learning_rate": 7.620817843866171e-07,
+      "loss": 1.0297,
+      "step": 42
+    },
+    {
+      "epoch": 0.04809843400447427,
+      "grad_norm": 0.7977489829063416,
+      "learning_rate": 7.806691449814127e-07,
+      "loss": 1.0686,
+      "step": 43
+    },
+    {
+      "epoch": 0.049217002237136466,
+      "grad_norm": 0.8264037370681763,
+      "learning_rate": 7.992565055762081e-07,
+      "loss": 1.0625,
+      "step": 44
+    },
+    {
+      "epoch": 0.050335570469798654,
+      "grad_norm": 0.8543824553489685,
+      "learning_rate": 8.178438661710038e-07,
+      "loss": 1.0599,
+      "step": 45
+    },
+    {
+      "epoch": 0.05145413870246085,
+      "grad_norm": 0.952981173992157,
+      "learning_rate": 8.364312267657993e-07,
+      "loss": 1.1883,
+      "step": 46
+    },
+    {
+      "epoch": 0.052572706935123045,
+      "grad_norm": 0.8212621212005615,
+      "learning_rate": 8.550185873605949e-07,
+      "loss": 1.0422,
+      "step": 47
+    },
+    {
+      "epoch": 0.053691275167785234,
+      "grad_norm": 0.7974889874458313,
+      "learning_rate": 8.736059479553904e-07,
+      "loss": 1.0582,
+      "step": 48
+    },
+    {
+      "epoch": 0.05480984340044743,
+      "grad_norm": 0.779604434967041,
+      "learning_rate": 8.921933085501859e-07,
+      "loss": 1.0396,
+      "step": 49
+    },
+    {
+      "epoch": 0.05592841163310962,
+      "grad_norm": 1.1233726739883423,
+      "learning_rate": 9.107806691449815e-07,
+      "loss": 1.1547,
+      "step": 50
+    },
+    {
+      "epoch": 0.05592841163310962,
+      "eval_loss": 1.138222575187683,
+      "eval_runtime": 375.3049,
+      "eval_samples_per_second": 2.118,
+      "eval_steps_per_second": 2.118,
+      "step": 50
+    },
+    {
+      "epoch": 0.05704697986577181,
+      "grad_norm": 0.9030459523200989,
+      "learning_rate": 9.29368029739777e-07,
+      "loss": 1.1371,
+      "step": 51
+    },
+    {
+      "epoch": 0.058165548098434,
+      "grad_norm": 0.8469235301017761,
+      "learning_rate": 9.479553903345725e-07,
+      "loss": 1.0534,
+      "step": 52
+    },
+    {
+      "epoch": 0.0592841163310962,
+      "grad_norm": 0.8827993273735046,
+      "learning_rate": 9.665427509293682e-07,
+      "loss": 1.185,
+      "step": 53
+    },
+    {
+      "epoch": 0.06040268456375839,
+      "grad_norm": 0.7730310559272766,
+      "learning_rate": 9.851301115241636e-07,
+      "loss": 1.041,
+      "step": 54
+    },
+    {
+      "epoch": 0.06152125279642058,
+      "grad_norm": 0.9019567966461182,
+      "learning_rate": 1.0037174721189593e-06,
+      "loss": 1.0858,
+      "step": 55
+    },
+    {
+      "epoch": 0.06263982102908278,
+      "grad_norm": 0.7867745757102966,
+      "learning_rate": 1.0223048327137547e-06,
+      "loss": 1.0293,
+      "step": 56
+    },
+    {
+      "epoch": 0.06375838926174497,
+      "grad_norm": 0.8245961666107178,
+      "learning_rate": 1.0408921933085502e-06,
+      "loss": 1.0678,
+      "step": 57
+    },
+    {
+      "epoch": 0.06487695749440715,
+      "grad_norm": 0.9173546433448792,
+      "learning_rate": 1.0594795539033458e-06,
+      "loss": 1.1314,
+      "step": 58
+    },
+    {
+      "epoch": 0.06599552572706935,
+      "grad_norm": 0.751348614692688,
+      "learning_rate": 1.0780669144981413e-06,
+      "loss": 0.9804,
+      "step": 59
+    },
+    {
+      "epoch": 0.06711409395973154,
+      "grad_norm": 0.796683132648468,
+      "learning_rate": 1.096654275092937e-06,
+      "loss": 1.0674,
+      "step": 60
+    },
+    {
+      "epoch": 0.06823266219239374,
+      "grad_norm": 4.512759208679199,
+      "learning_rate": 1.1152416356877324e-06,
+      "loss": 1.0074,
+      "step": 61
+    },
+    {
+      "epoch": 0.06935123042505593,
+      "grad_norm": 0.739653468132019,
+      "learning_rate": 1.133828996282528e-06,
+      "loss": 1.0961,
+      "step": 62
+    },
+    {
+      "epoch": 0.07046979865771812,
+      "grad_norm": 0.8302459716796875,
+      "learning_rate": 1.1524163568773235e-06,
+      "loss": 1.1087,
+      "step": 63
+    },
+    {
+      "epoch": 0.07158836689038031,
+      "grad_norm": 0.8228954076766968,
+      "learning_rate": 1.1710037174721191e-06,
+      "loss": 1.0968,
+      "step": 64
+    },
+    {
+      "epoch": 0.07270693512304251,
+      "grad_norm": 0.7977240085601807,
+      "learning_rate": 1.1895910780669146e-06,
+      "loss": 1.1246,
+      "step": 65
+    },
+    {
+      "epoch": 0.0738255033557047,
+      "grad_norm": 0.8019763231277466,
+      "learning_rate": 1.20817843866171e-06,
+      "loss": 1.0602,
+      "step": 66
+    },
+    {
+      "epoch": 0.07494407158836688,
+      "grad_norm": 0.6944935917854309,
+      "learning_rate": 1.2267657992565057e-06,
+      "loss": 0.9552,
+      "step": 67
+    },
+    {
+      "epoch": 0.07606263982102908,
+      "grad_norm": 0.9132448434829712,
+      "learning_rate": 1.2453531598513011e-06,
+      "loss": 1.14,
+      "step": 68
+    },
+    {
+      "epoch": 0.07718120805369127,
+      "grad_norm": 0.8454166650772095,
+      "learning_rate": 1.2639405204460966e-06,
+      "loss": 1.0735,
+      "step": 69
+    },
+    {
+      "epoch": 0.07829977628635347,
+      "grad_norm": 0.8806889057159424,
+      "learning_rate": 1.2825278810408922e-06,
+      "loss": 1.0534,
+      "step": 70
+    },
+    {
+      "epoch": 0.07941834451901567,
+      "grad_norm": 0.8355393409729004,
+      "learning_rate": 1.3011152416356879e-06,
+      "loss": 1.0796,
+      "step": 71
+    },
+    {
+      "epoch": 0.08053691275167785,
+      "grad_norm": 0.7422959804534912,
+      "learning_rate": 1.3197026022304835e-06,
+      "loss": 0.9629,
+      "step": 72
+    },
+    {
+      "epoch": 0.08165548098434004,
+      "grad_norm": 0.7394403219223022,
+      "learning_rate": 1.3382899628252788e-06,
+      "loss": 0.9845,
+      "step": 73
+    },
+    {
+      "epoch": 0.08277404921700224,
+      "grad_norm": 0.9226001501083374,
+      "learning_rate": 1.3568773234200744e-06,
+      "loss": 1.0786,
+      "step": 74
+    },
+    {
+      "epoch": 0.08389261744966443,
+      "grad_norm": 0.7306943535804749,
+      "learning_rate": 1.37546468401487e-06,
+      "loss": 1.0586,
+      "step": 75
+    },
+    {
+      "epoch": 0.08501118568232663,
+      "grad_norm": 0.7850666046142578,
+      "learning_rate": 1.3940520446096655e-06,
+      "loss": 1.0024,
+      "step": 76
+    },
+    {
+      "epoch": 0.08612975391498881,
+      "grad_norm": 0.8453303575515747,
+      "learning_rate": 1.4126394052044612e-06,
+      "loss": 0.9648,
+      "step": 77
+    },
+    {
+      "epoch": 0.087248322147651,
+      "grad_norm": 0.6994377374649048,
+      "learning_rate": 1.4312267657992564e-06,
+      "loss": 1.0503,
+      "step": 78
+    },
+    {
+      "epoch": 0.0883668903803132,
+      "grad_norm": 0.8966418504714966,
+      "learning_rate": 1.449814126394052e-06,
+      "loss": 0.9281,
+      "step": 79
+    },
+    {
+      "epoch": 0.0894854586129754,
+      "grad_norm": 0.7137410640716553,
+      "learning_rate": 1.4684014869888477e-06,
+      "loss": 0.9106,
+      "step": 80
+    },
+    {
+      "epoch": 0.09060402684563758,
+      "grad_norm": 0.8406743407249451,
+      "learning_rate": 1.4869888475836434e-06,
+      "loss": 1.0357,
+      "step": 81
+    },
+    {
+      "epoch": 0.09172259507829977,
+      "grad_norm": 0.9425244331359863,
+      "learning_rate": 1.5055762081784388e-06,
+      "loss": 1.0755,
+      "step": 82
+    },
+    {
+      "epoch": 0.09284116331096197,
+      "grad_norm": 0.9145747423171997,
+      "learning_rate": 1.5241635687732343e-06,
+      "loss": 1.1066,
+      "step": 83
+    },
+    {
+      "epoch": 0.09395973154362416,
+      "grad_norm": 0.7609133124351501,
+      "learning_rate": 1.54275092936803e-06,
+      "loss": 0.9507,
+      "step": 84
+    },
+    {
+      "epoch": 0.09507829977628636,
+      "grad_norm": 0.8113803267478943,
+      "learning_rate": 1.5613382899628254e-06,
+      "loss": 0.9648,
+      "step": 85
+    },
+    {
+      "epoch": 0.09619686800894854,
+      "grad_norm": 0.929417073726654,
+      "learning_rate": 1.579925650557621e-06,
+      "loss": 1.0954,
+      "step": 86
+    },
+    {
+      "epoch": 0.09731543624161074,
+      "grad_norm": 0.7626249194145203,
+      "learning_rate": 1.5985130111524163e-06,
+      "loss": 0.9776,
+      "step": 87
+    },
+    {
+      "epoch": 0.09843400447427293,
+      "grad_norm": 0.7826419472694397,
+      "learning_rate": 1.617100371747212e-06,
+      "loss": 1.0075,
+      "step": 88
+    },
+    {
+      "epoch": 0.09955257270693513,
+      "grad_norm": 0.7303605079650879,
+      "learning_rate": 1.6356877323420076e-06,
+      "loss": 0.9475,
+      "step": 89
+    },
+    {
+      "epoch": 0.10067114093959731,
+      "grad_norm": 0.8236870169639587,
+      "learning_rate": 1.6542750929368032e-06,
+      "loss": 0.9645,
+      "step": 90
+    },
+    {
+      "epoch": 0.1017897091722595,
+      "grad_norm": 1.6685837507247925,
+      "learning_rate": 1.6728624535315987e-06,
+      "loss": 1.0288,
+      "step": 91
+    },
+    {
+      "epoch": 0.1029082774049217,
+      "grad_norm": 0.8708052039146423,
+      "learning_rate": 1.6914498141263941e-06,
+      "loss": 1.0235,
+      "step": 92
+    },
+    {
+      "epoch": 0.1040268456375839,
+      "grad_norm": 0.8592197895050049,
+      "learning_rate": 1.7100371747211898e-06,
+      "loss": 1.0153,
+      "step": 93
+    },
+    {
+      "epoch": 0.10514541387024609,
+      "grad_norm": 1.0110843181610107,
+      "learning_rate": 1.7286245353159852e-06,
+      "loss": 0.8667,
+      "step": 94
+    },
+    {
+      "epoch": 0.10626398210290827,
+      "grad_norm": 0.8149330019950867,
+      "learning_rate": 1.7472118959107809e-06,
+      "loss": 0.9292,
+      "step": 95
+    },
+    {
+      "epoch": 0.10738255033557047,
+      "grad_norm": 1.8980721235275269,
+      "learning_rate": 1.7657992565055765e-06,
+      "loss": 0.9528,
+      "step": 96
+    },
+    {
+      "epoch": 0.10850111856823266,
+      "grad_norm": 1.9700517654418945,
+      "learning_rate": 1.7843866171003718e-06,
+      "loss": 0.8754,
+      "step": 97
+    },
+    {
+      "epoch": 0.10961968680089486,
+      "grad_norm": 5.660907745361328,
+      "learning_rate": 1.8029739776951674e-06,
+      "loss": 0.8988,
+      "step": 98
+    },
+    {
+      "epoch": 0.11073825503355705,
+      "grad_norm": 2.2622532844543457,
+      "learning_rate": 1.821561338289963e-06,
+      "loss": 0.9347,
+      "step": 99
+    },
+    {
+      "epoch": 0.11185682326621924,
+      "grad_norm": 0.7684885859489441,
+      "learning_rate": 1.8401486988847585e-06,
+      "loss": 0.9158,
+      "step": 100
+    },
+    {
+      "epoch": 0.11185682326621924,
+      "eval_loss": 0.9625731110572815,
+      "eval_runtime": 376.9515,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 2.109,
+      "step": 100
+    },
+    {
+      "epoch": 0.11297539149888143,
+      "grad_norm": 0.7726151943206787,
+      "learning_rate": 1.858736059479554e-06,
+      "loss": 0.964,
+      "step": 101
+    },
+    {
+      "epoch": 0.11409395973154363,
+      "grad_norm": 1.0598620176315308,
+      "learning_rate": 1.8773234200743496e-06,
+      "loss": 0.9,
+      "step": 102
+    },
+    {
+      "epoch": 0.11521252796420582,
+      "grad_norm": 0.8776567578315735,
+      "learning_rate": 1.895910780669145e-06,
+      "loss": 0.9152,
+      "step": 103
+    },
+    {
+      "epoch": 0.116331096196868,
+      "grad_norm": 1.009832501411438,
+      "learning_rate": 1.9144981412639405e-06,
+      "loss": 0.8405,
+      "step": 104
+    },
+    {
+      "epoch": 0.1174496644295302,
+      "grad_norm": 0.9680449962615967,
+      "learning_rate": 1.9330855018587364e-06,
+      "loss": 0.9312,
+      "step": 105
+    },
+    {
+      "epoch": 0.1185682326621924,
+      "grad_norm": 0.8206327557563782,
+      "learning_rate": 1.951672862453532e-06,
+      "loss": 0.8297,
+      "step": 106
+    },
+    {
+      "epoch": 0.11968680089485459,
+      "grad_norm": 0.9549261927604675,
+      "learning_rate": 1.9702602230483273e-06,
+      "loss": 0.8435,
+      "step": 107
+    },
+    {
+      "epoch": 0.12080536912751678,
+      "grad_norm": 1.0195021629333496,
+      "learning_rate": 1.9888475836431227e-06,
+      "loss": 0.9342,
+      "step": 108
+    },
+    {
+      "epoch": 0.12192393736017897,
+      "grad_norm": 0.9555221199989319,
+      "learning_rate": 2.0074349442379186e-06,
+      "loss": 0.8428,
+      "step": 109
+    },
+    {
+      "epoch": 0.12304250559284116,
+      "grad_norm": 0.8478293418884277,
+      "learning_rate": 2.0260223048327136e-06,
+      "loss": 0.8582,
+      "step": 110
+    },
+    {
+      "epoch": 0.12416107382550336,
+      "grad_norm": 0.8309235572814941,
+      "learning_rate": 2.0446096654275095e-06,
+      "loss": 0.8005,
+      "step": 111
+    },
+    {
+      "epoch": 0.12527964205816555,
+      "grad_norm": 0.9847236275672913,
+      "learning_rate": 2.063197026022305e-06,
+      "loss": 0.9191,
+      "step": 112
+    },
+    {
+      "epoch": 0.12639821029082773,
+      "grad_norm": 1.016616702079773,
+      "learning_rate": 2.0817843866171004e-06,
+      "loss": 0.857,
+      "step": 113
+    },
+    {
+      "epoch": 0.12751677852348994,
+      "grad_norm": 1.0105868577957153,
+      "learning_rate": 2.1003717472118962e-06,
+      "loss": 0.8997,
+      "step": 114
+    },
+    {
+      "epoch": 0.12863534675615212,
+      "grad_norm": 0.9318265914916992,
+      "learning_rate": 2.1189591078066917e-06,
+      "loss": 0.8289,
+      "step": 115
+    },
+    {
+      "epoch": 0.1297539149888143,
+      "grad_norm": 0.9799047708511353,
+      "learning_rate": 2.137546468401487e-06,
+      "loss": 0.8192,
+      "step": 116
+    },
+    {
+      "epoch": 0.13087248322147652,
+      "grad_norm": 0.8267964720726013,
+      "learning_rate": 2.1561338289962826e-06,
+      "loss": 0.7895,
+      "step": 117
+    },
+    {
+      "epoch": 0.1319910514541387,
+      "grad_norm": 0.8401578664779663,
+      "learning_rate": 2.1747211895910784e-06,
+      "loss": 0.8308,
+      "step": 118
+    },
+    {
+      "epoch": 0.1331096196868009,
+      "grad_norm": 1.4501429796218872,
+      "learning_rate": 2.193308550185874e-06,
+      "loss": 0.7761,
+      "step": 119
+    },
+    {
+      "epoch": 0.1342281879194631,
+      "grad_norm": 1.0903233289718628,
+      "learning_rate": 2.2118959107806693e-06,
+      "loss": 0.7694,
+      "step": 120
+    },
+    {
+      "epoch": 0.13534675615212527,
+      "grad_norm": 0.9950303435325623,
+      "learning_rate": 2.2304832713754648e-06,
+      "loss": 0.7387,
+      "step": 121
+    },
+    {
+      "epoch": 0.13646532438478748,
+      "grad_norm": 0.950897753238678,
+      "learning_rate": 2.2490706319702602e-06,
+      "loss": 0.7648,
+      "step": 122
+    },
+    {
+      "epoch": 0.13758389261744966,
+      "grad_norm": 1.0191601514816284,
+      "learning_rate": 2.267657992565056e-06,
+      "loss": 0.7433,
+      "step": 123
+    },
+    {
+      "epoch": 0.13870246085011187,
+      "grad_norm": 0.8940446376800537,
+      "learning_rate": 2.2862453531598515e-06,
+      "loss": 0.7671,
+      "step": 124
+    },
+    {
+      "epoch": 0.13982102908277405,
+      "grad_norm": 1.025028944015503,
+      "learning_rate": 2.304832713754647e-06,
+      "loss": 0.7688,
+      "step": 125
+    },
+    {
+      "epoch": 0.14093959731543623,
+      "grad_norm": 0.8699458241462708,
+      "learning_rate": 2.3234200743494424e-06,
+      "loss": 0.6479,
+      "step": 126
+    },
+    {
+      "epoch": 0.14205816554809844,
+      "grad_norm": 1.6527787446975708,
+      "learning_rate": 2.3420074349442383e-06,
+      "loss": 0.7275,
+      "step": 127
+    },
+    {
+      "epoch": 0.14317673378076062,
+      "grad_norm": 0.8746642470359802,
+      "learning_rate": 2.3605947955390337e-06,
+      "loss": 0.7161,
+      "step": 128
+    },
+    {
+      "epoch": 0.14429530201342283,
+      "grad_norm": 0.8040481805801392,
+      "learning_rate": 2.379182156133829e-06,
+      "loss": 0.6186,
+      "step": 129
+    },
+    {
+      "epoch": 0.14541387024608501,
+      "grad_norm": 0.9275310635566711,
+      "learning_rate": 2.3977695167286246e-06,
+      "loss": 0.6387,
+      "step": 130
+    },
+    {
+      "epoch": 0.1465324384787472,
+      "grad_norm": 0.7265205979347229,
+      "learning_rate": 2.41635687732342e-06,
+      "loss": 0.6432,
+      "step": 131
+    },
+    {
+      "epoch": 0.1476510067114094,
+      "grad_norm": 1.032467007637024,
+      "learning_rate": 2.434944237918216e-06,
+      "loss": 0.608,
+      "step": 132
+    },
+    {
+      "epoch": 0.1487695749440716,
+      "grad_norm": 0.8901663422584534,
+      "learning_rate": 2.4535315985130114e-06,
+      "loss": 0.6688,
+      "step": 133
+    },
+    {
+      "epoch": 0.14988814317673377,
+      "grad_norm": 0.7955690622329712,
+      "learning_rate": 2.472118959107807e-06,
+      "loss": 0.5902,
+      "step": 134
+    },
+    {
+      "epoch": 0.15100671140939598,
+      "grad_norm": 0.6967551708221436,
+      "learning_rate": 2.4907063197026023e-06,
+      "loss": 0.5483,
+      "step": 135
+    },
+    {
+      "epoch": 0.15212527964205816,
+      "grad_norm": 0.6825600862503052,
+      "learning_rate": 2.5092936802973977e-06,
+      "loss": 0.5912,
+      "step": 136
+    },
+    {
+      "epoch": 0.15324384787472037,
+      "grad_norm": 0.8622741103172302,
+      "learning_rate": 2.527881040892193e-06,
+      "loss": 0.5764,
+      "step": 137
+    },
+    {
+      "epoch": 0.15436241610738255,
+      "grad_norm": 0.7212573289871216,
+      "learning_rate": 2.546468401486989e-06,
+      "loss": 0.5788,
+      "step": 138
+    },
+    {
+      "epoch": 0.15548098434004473,
+      "grad_norm": 0.6434993743896484,
+      "learning_rate": 2.5650557620817845e-06,
+      "loss": 0.5684,
+      "step": 139
+    },
+    {
+      "epoch": 0.15659955257270694,
+      "grad_norm": 0.6838394999504089,
+      "learning_rate": 2.5836431226765803e-06,
+      "loss": 0.58,
+      "step": 140
+    },
+    {
+      "epoch": 0.15771812080536912,
+      "grad_norm": 0.5968404412269592,
+      "learning_rate": 2.6022304832713758e-06,
+      "loss": 0.6487,
+      "step": 141
+    },
+    {
+      "epoch": 0.15883668903803133,
+      "grad_norm": 1.2263623476028442,
+      "learning_rate": 2.6208178438661712e-06,
+      "loss": 0.5063,
+      "step": 142
+    },
+    {
+      "epoch": 0.1599552572706935,
+      "grad_norm": 0.8275245428085327,
+      "learning_rate": 2.639405204460967e-06,
+      "loss": 0.5821,
+      "step": 143
+    },
+    {
+      "epoch": 0.1610738255033557,
+      "grad_norm": 0.6955662369728088,
+      "learning_rate": 2.6579925650557625e-06,
+      "loss": 0.5675,
+      "step": 144
+    },
+    {
+      "epoch": 0.1621923937360179,
+      "grad_norm": 0.6327409148216248,
+      "learning_rate": 2.6765799256505576e-06,
+      "loss": 0.5509,
+      "step": 145
+    },
+    {
+      "epoch": 0.16331096196868009,
+      "grad_norm": 0.8698219060897827,
+      "learning_rate": 2.695167286245353e-06,
+      "loss": 0.5811,
+      "step": 146
+    },
+    {
+      "epoch": 0.1644295302013423,
+      "grad_norm": 0.5564067959785461,
+      "learning_rate": 2.713754646840149e-06,
+      "loss": 0.627,
+      "step": 147
+    },
+    {
+      "epoch": 0.16554809843400448,
+      "grad_norm": 0.6057626605033875,
+      "learning_rate": 2.7323420074349443e-06,
+      "loss": 0.5444,
+      "step": 148
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.5734674334526062,
+      "learning_rate": 2.75092936802974e-06,
+      "loss": 0.4719,
+      "step": 149
+    },
+    {
+      "epoch": 0.16778523489932887,
+      "grad_norm": 0.6801913976669312,
+      "learning_rate": 2.7695167286245356e-06,
+      "loss": 0.5713,
+      "step": 150
+    },
+    {
+      "epoch": 0.16778523489932887,
+      "eval_loss": 0.5491015911102295,
+      "eval_runtime": 377.0287,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 2.109,
+      "step": 150
+    },
+    {
+      "epoch": 0.16890380313199105,
+      "grad_norm": 0.5245065093040466,
+      "learning_rate": 2.788104089219331e-06,
+      "loss": 0.4768,
+      "step": 151
+    },
+    {
+      "epoch": 0.17002237136465326,
+      "grad_norm": 0.49822941422462463,
+      "learning_rate": 2.806691449814127e-06,
+      "loss": 0.6005,
+      "step": 152
+    },
+    {
+      "epoch": 0.17114093959731544,
+      "grad_norm": 0.5362038016319275,
+      "learning_rate": 2.8252788104089224e-06,
+      "loss": 0.4879,
+      "step": 153
+    },
+    {
+      "epoch": 0.17225950782997762,
+      "grad_norm": 0.500764787197113,
+      "learning_rate": 2.843866171003718e-06,
+      "loss": 0.4906,
+      "step": 154
+    },
+    {
+      "epoch": 0.17337807606263983,
+      "grad_norm": 0.4894807040691376,
+      "learning_rate": 2.862453531598513e-06,
+      "loss": 0.6338,
+      "step": 155
+    },
+    {
+      "epoch": 0.174496644295302,
+      "grad_norm": 0.4735981523990631,
+      "learning_rate": 2.8810408921933087e-06,
+      "loss": 0.5166,
+      "step": 156
+    },
+    {
+      "epoch": 0.1756152125279642,
+      "grad_norm": 0.5308732390403748,
+      "learning_rate": 2.899628252788104e-06,
+      "loss": 0.4998,
+      "step": 157
+    },
+    {
+      "epoch": 0.1767337807606264,
+      "grad_norm": 0.46382462978363037,
+      "learning_rate": 2.9182156133829e-06,
+      "loss": 0.4968,
+      "step": 158
+    },
+    {
+      "epoch": 0.17785234899328858,
+      "grad_norm": 0.4579423666000366,
+      "learning_rate": 2.9368029739776955e-06,
+      "loss": 0.4666,
+      "step": 159
+    },
+    {
+      "epoch": 0.1789709172259508,
+      "grad_norm": 0.414998859167099,
+      "learning_rate": 2.955390334572491e-06,
+      "loss": 0.5298,
+      "step": 160
+    },
+    {
+      "epoch": 0.18008948545861297,
+      "grad_norm": 0.4376446008682251,
+      "learning_rate": 2.973977695167287e-06,
+      "loss": 0.534,
+      "step": 161
+    },
+    {
+      "epoch": 0.18120805369127516,
+      "grad_norm": 0.4209648370742798,
+      "learning_rate": 2.9925650557620822e-06,
+      "loss": 0.5765,
+      "step": 162
+    },
+    {
+      "epoch": 0.18232662192393737,
+      "grad_norm": 0.6292575597763062,
+      "learning_rate": 3.0111524163568777e-06,
+      "loss": 0.4897,
+      "step": 163
+    },
+    {
+      "epoch": 0.18344519015659955,
+      "grad_norm": 0.4352070689201355,
+      "learning_rate": 3.0297397769516727e-06,
+      "loss": 0.5151,
+      "step": 164
+    },
+    {
+      "epoch": 0.18456375838926176,
+      "grad_norm": 0.4680917263031006,
+      "learning_rate": 3.0483271375464686e-06,
+      "loss": 0.514,
+      "step": 165
+    },
+    {
+      "epoch": 0.18568232662192394,
+      "grad_norm": 0.3949781060218811,
+      "learning_rate": 3.066914498141264e-06,
+      "loss": 0.5045,
+      "step": 166
+    },
+    {
+      "epoch": 0.18680089485458612,
+      "grad_norm": 0.557427704334259,
+      "learning_rate": 3.08550185873606e-06,
+      "loss": 0.439,
+      "step": 167
+    },
+    {
+      "epoch": 0.18791946308724833,
+      "grad_norm": 0.38726532459259033,
+      "learning_rate": 3.1040892193308553e-06,
+      "loss": 0.4472,
+      "step": 168
+    },
+    {
+      "epoch": 0.1890380313199105,
+      "grad_norm": 0.39277103543281555,
+      "learning_rate": 3.1226765799256508e-06,
+      "loss": 0.5121,
+      "step": 169
+    },
+    {
+      "epoch": 0.19015659955257272,
+      "grad_norm": 4.005094528198242,
+      "learning_rate": 3.1412639405204466e-06,
+      "loss": 0.5303,
+      "step": 170
+    },
+    {
+      "epoch": 0.1912751677852349,
+      "grad_norm": 0.41803544759750366,
+      "learning_rate": 3.159851301115242e-06,
+      "loss": 0.4559,
+      "step": 171
+    },
+    {
+      "epoch": 0.19239373601789708,
+      "grad_norm": 0.4186052083969116,
+      "learning_rate": 3.1784386617100375e-06,
+      "loss": 0.4638,
+      "step": 172
+    },
+    {
+      "epoch": 0.1935123042505593,
+      "grad_norm": 0.35806962847709656,
+      "learning_rate": 3.1970260223048326e-06,
+      "loss": 0.5106,
+      "step": 173
+    },
+    {
+      "epoch": 0.19463087248322147,
+      "grad_norm": 0.3786235749721527,
+      "learning_rate": 3.2156133828996284e-06,
+      "loss": 0.5428,
+      "step": 174
+    },
+    {
+      "epoch": 0.19574944071588368,
+      "grad_norm": 0.35347774624824524,
+      "learning_rate": 3.234200743494424e-06,
+      "loss": 0.5238,
+      "step": 175
+    },
+    {
+      "epoch": 0.19686800894854586,
+      "grad_norm": 0.3494052588939667,
+      "learning_rate": 3.2527881040892197e-06,
+      "loss": 0.4969,
+      "step": 176
+    },
+    {
+      "epoch": 0.19798657718120805,
+      "grad_norm": 0.37865009903907776,
+      "learning_rate": 3.271375464684015e-06,
+      "loss": 0.5233,
+      "step": 177
+    },
+    {
+      "epoch": 0.19910514541387025,
+      "grad_norm": 0.36743587255477905,
+      "learning_rate": 3.2899628252788106e-06,
+      "loss": 0.4309,
+      "step": 178
+    },
+    {
+      "epoch": 0.20022371364653244,
+      "grad_norm": 0.35772132873535156,
+      "learning_rate": 3.3085501858736065e-06,
+      "loss": 0.4845,
+      "step": 179
+    },
+    {
+      "epoch": 0.20134228187919462,
+      "grad_norm": 0.35337793827056885,
+      "learning_rate": 3.327137546468402e-06,
+      "loss": 0.5278,
+      "step": 180
+    },
+    {
+      "epoch": 0.20246085011185683,
+      "grad_norm": 0.3413907587528229,
+      "learning_rate": 3.3457249070631974e-06,
+      "loss": 0.5051,
+      "step": 181
+    },
+    {
+      "epoch": 0.203579418344519,
+      "grad_norm": 0.3378649353981018,
+      "learning_rate": 3.3643122676579924e-06,
+      "loss": 0.4627,
+      "step": 182
+    },
+    {
+      "epoch": 0.20469798657718122,
+      "grad_norm": 0.3554287850856781,
+      "learning_rate": 3.3828996282527883e-06,
+      "loss": 0.5132,
+      "step": 183
+    },
+    {
+      "epoch": 0.2058165548098434,
+      "grad_norm": 0.339643657207489,
+      "learning_rate": 3.4014869888475837e-06,
+      "loss": 0.485,
+      "step": 184
+    },
+    {
+      "epoch": 0.20693512304250558,
+      "grad_norm": 0.7494234442710876,
+      "learning_rate": 3.4200743494423796e-06,
+      "loss": 0.5497,
+      "step": 185
+    },
+    {
+      "epoch": 0.2080536912751678,
+      "grad_norm": 0.3561193645000458,
+      "learning_rate": 3.438661710037175e-06,
+      "loss": 0.5524,
+      "step": 186
+    },
+    {
+      "epoch": 0.20917225950782997,
+      "grad_norm": 0.38029277324676514,
+      "learning_rate": 3.4572490706319705e-06,
+      "loss": 0.4592,
+      "step": 187
+    },
+    {
+      "epoch": 0.21029082774049218,
+      "grad_norm": 0.6649860143661499,
+      "learning_rate": 3.4758364312267663e-06,
+      "loss": 0.5482,
+      "step": 188
+    },
+    {
+      "epoch": 0.21140939597315436,
+      "grad_norm": 0.3561515212059021,
+      "learning_rate": 3.4944237918215618e-06,
+      "loss": 0.4041,
+      "step": 189
+    },
+    {
+      "epoch": 0.21252796420581654,
+      "grad_norm": 0.392444908618927,
+      "learning_rate": 3.5130111524163572e-06,
+      "loss": 0.4632,
+      "step": 190
+    },
+    {
+      "epoch": 0.21364653243847875,
+      "grad_norm": 0.36593955755233765,
+      "learning_rate": 3.531598513011153e-06,
+      "loss": 0.4717,
+      "step": 191
+    },
+    {
+      "epoch": 0.21476510067114093,
+      "grad_norm": 0.8896877765655518,
+      "learning_rate": 3.550185873605948e-06,
+      "loss": 0.4966,
+      "step": 192
+    },
+    {
+      "epoch": 0.21588366890380314,
+      "grad_norm": 0.38146454095840454,
+      "learning_rate": 3.5687732342007436e-06,
+      "loss": 0.4788,
+      "step": 193
+    },
+    {
+      "epoch": 0.21700223713646533,
+      "grad_norm": 0.3630574941635132,
+      "learning_rate": 3.5873605947955394e-06,
+      "loss": 0.4598,
+      "step": 194
+    },
+    {
+      "epoch": 0.2181208053691275,
+      "grad_norm": 0.3017019033432007,
+      "learning_rate": 3.605947955390335e-06,
+      "loss": 0.4141,
+      "step": 195
+    },
+    {
+      "epoch": 0.21923937360178972,
+      "grad_norm": 0.3096812665462494,
+      "learning_rate": 3.6245353159851303e-06,
+      "loss": 0.3976,
+      "step": 196
+    },
+    {
+      "epoch": 0.2203579418344519,
+      "grad_norm": 0.3209673762321472,
+      "learning_rate": 3.643122676579926e-06,
+      "loss": 0.4609,
+      "step": 197
+    },
+    {
+      "epoch": 0.2214765100671141,
+      "grad_norm": 0.3192496597766876,
+      "learning_rate": 3.6617100371747216e-06,
+      "loss": 0.4714,
+      "step": 198
+    },
+    {
+      "epoch": 0.2225950782997763,
+      "grad_norm": 0.4107726514339447,
+      "learning_rate": 3.680297397769517e-06,
+      "loss": 0.5456,
+      "step": 199
+    },
+    {
+      "epoch": 0.22371364653243847,
+      "grad_norm": 0.7011065483093262,
+      "learning_rate": 3.698884758364313e-06,
+      "loss": 0.4095,
+      "step": 200
+    },
+    {
+      "epoch": 0.22371364653243847,
+      "eval_loss": 0.45907658338546753,
+      "eval_runtime": 376.8292,
+      "eval_samples_per_second": 2.11,
+      "eval_steps_per_second": 2.11,
+      "step": 200
+    },
+    {
+      "epoch": 0.22483221476510068,
+      "grad_norm": 0.31661006808280945,
+      "learning_rate": 3.717472118959108e-06,
+      "loss": 0.443,
+      "step": 201
+    },
+    {
+      "epoch": 0.22595078299776286,
+      "grad_norm": 0.3416866064071655,
+      "learning_rate": 3.7360594795539034e-06,
+      "loss": 0.4225,
+      "step": 202
+    },
+    {
+      "epoch": 0.22706935123042504,
+      "grad_norm": 0.333741694688797,
+      "learning_rate": 3.7546468401486993e-06,
+      "loss": 0.4585,
+      "step": 203
+    },
+    {
+      "epoch": 0.22818791946308725,
+      "grad_norm": 0.3095785677433014,
+      "learning_rate": 3.7732342007434947e-06,
+      "loss": 0.4169,
+      "step": 204
+    },
+    {
+      "epoch": 0.22930648769574943,
+      "grad_norm": 0.34538066387176514,
+      "learning_rate": 3.79182156133829e-06,
+      "loss": 0.5144,
+      "step": 205
+    },
+    {
+      "epoch": 0.23042505592841164,
+      "grad_norm": 0.3058728873729706,
+      "learning_rate": 3.810408921933086e-06,
+      "loss": 0.4855,
+      "step": 206
+    },
+    {
+      "epoch": 0.23154362416107382,
+      "grad_norm": 0.31272053718566895,
+      "learning_rate": 3.828996282527881e-06,
+      "loss": 0.4437,
+      "step": 207
+    },
+    {
+      "epoch": 0.232662192393736,
+      "grad_norm": 0.3328510820865631,
+      "learning_rate": 3.847583643122677e-06,
+      "loss": 0.411,
+      "step": 208
+    },
+    {
+      "epoch": 0.23378076062639822,
+      "grad_norm": 0.3082751929759979,
+      "learning_rate": 3.866171003717473e-06,
+      "loss": 0.4448,
+      "step": 209
+    },
+    {
+      "epoch": 0.2348993288590604,
+      "grad_norm": 0.305494487285614,
+      "learning_rate": 3.884758364312268e-06,
+      "loss": 0.4481,
+      "step": 210
+    },
+    {
+      "epoch": 0.2360178970917226,
+      "grad_norm": 0.2996240258216858,
+      "learning_rate": 3.903345724907064e-06,
+      "loss": 0.4494,
+      "step": 211
+    },
+    {
+      "epoch": 0.2371364653243848,
+      "grad_norm": 0.2894361913204193,
+      "learning_rate": 3.921933085501859e-06,
+      "loss": 0.4325,
+      "step": 212
+    },
+    {
+      "epoch": 0.23825503355704697,
+      "grad_norm": 0.3032226860523224,
+      "learning_rate": 3.9405204460966546e-06,
+      "loss": 0.444,
+      "step": 213
+    },
+    {
+      "epoch": 0.23937360178970918,
+      "grad_norm": 0.3289664089679718,
+      "learning_rate": 3.9591078066914504e-06,
+      "loss": 0.469,
+      "step": 214
+    },
+    {
+      "epoch": 0.24049217002237136,
+      "grad_norm": 0.322125107049942,
+      "learning_rate": 3.9776951672862455e-06,
+      "loss": 0.5105,
+      "step": 215
+    },
+    {
+      "epoch": 0.24161073825503357,
+      "grad_norm": 0.34320828318595886,
+      "learning_rate": 3.996282527881041e-06,
+      "loss": 0.4929,
+      "step": 216
+    },
+    {
+      "epoch": 0.24272930648769575,
+      "grad_norm": 0.284434050321579,
+      "learning_rate": 4.014869888475837e-06,
+      "loss": 0.3635,
+      "step": 217
+    },
+    {
+      "epoch": 0.24384787472035793,
+      "grad_norm": 0.3490959405899048,
+      "learning_rate": 4.033457249070632e-06,
+      "loss": 0.5105,
+      "step": 218
+    },
+    {
+      "epoch": 0.24496644295302014,
+      "grad_norm": 0.4406474828720093,
+      "learning_rate": 4.052044609665427e-06,
+      "loss": 0.5779,
+      "step": 219
+    },
+    {
+      "epoch": 0.24608501118568232,
+      "grad_norm": 0.3179807960987091,
+      "learning_rate": 4.070631970260223e-06,
+      "loss": 0.3967,
+      "step": 220
+    },
+    {
+      "epoch": 0.24720357941834453,
+      "grad_norm": 0.3659047782421112,
+      "learning_rate": 4.089219330855019e-06,
+      "loss": 0.3817,
+      "step": 221
+    },
+    {
+      "epoch": 0.2483221476510067,
+      "grad_norm": 0.37222808599472046,
+      "learning_rate": 4.107806691449814e-06,
+      "loss": 0.4326,
+      "step": 222
+    },
+    {
+      "epoch": 0.2494407158836689,
+      "grad_norm": 0.3316914737224579,
+      "learning_rate": 4.12639405204461e-06,
+      "loss": 0.409,
+      "step": 223
+    },
+    {
+      "epoch": 0.2505592841163311,
+      "grad_norm": 0.3071722984313965,
+      "learning_rate": 4.144981412639406e-06,
+      "loss": 0.4349,
+      "step": 224
+    },
+    {
+      "epoch": 0.2516778523489933,
+      "grad_norm": 0.2802477777004242,
+      "learning_rate": 4.163568773234201e-06,
+      "loss": 0.3572,
+      "step": 225
+    },
+    {
+      "epoch": 0.25279642058165547,
+      "grad_norm": 0.3275560140609741,
+      "learning_rate": 4.182156133828997e-06,
+      "loss": 0.4759,
+      "step": 226
+    },
+    {
+      "epoch": 0.2539149888143177,
+      "grad_norm": 0.38033831119537354,
+      "learning_rate": 4.2007434944237925e-06,
+      "loss": 0.4101,
+      "step": 227
+    },
+    {
+      "epoch": 0.2550335570469799,
+      "grad_norm": 0.32060644030570984,
+      "learning_rate": 4.219330855018588e-06,
+      "loss": 0.4361,
+      "step": 228
+    },
+    {
+      "epoch": 0.25615212527964204,
+      "grad_norm": 0.5606611967086792,
+      "learning_rate": 4.237918215613383e-06,
+      "loss": 0.4363,
+      "step": 229
+    },
+    {
+      "epoch": 0.25727069351230425,
+      "grad_norm": 0.31278616189956665,
+      "learning_rate": 4.256505576208178e-06,
+      "loss": 0.4427,
+      "step": 230
+    },
+    {
+      "epoch": 0.25838926174496646,
+      "grad_norm": 0.2902446389198303,
+      "learning_rate": 4.275092936802974e-06,
+      "loss": 0.3848,
+      "step": 231
+    },
+    {
+      "epoch": 0.2595078299776286,
+      "grad_norm": 0.33259814977645874,
+      "learning_rate": 4.29368029739777e-06,
+      "loss": 0.4595,
+      "step": 232
+    },
+    {
+      "epoch": 0.2606263982102908,
+      "grad_norm": 0.30889761447906494,
+      "learning_rate": 4.312267657992565e-06,
+      "loss": 0.4208,
+      "step": 233
+    },
+    {
+      "epoch": 0.26174496644295303,
+      "grad_norm": 0.3423914909362793,
+      "learning_rate": 4.330855018587361e-06,
+      "loss": 0.4652,
+      "step": 234
+    },
+    {
+      "epoch": 0.26286353467561524,
+      "grad_norm": 0.3276422619819641,
+      "learning_rate": 4.349442379182157e-06,
+      "loss": 0.4503,
+      "step": 235
+    },
+    {
+      "epoch": 0.2639821029082774,
+      "grad_norm": 0.3135182857513428,
+      "learning_rate": 4.368029739776952e-06,
+      "loss": 0.4387,
+      "step": 236
+    },
+    {
+      "epoch": 0.2651006711409396,
+      "grad_norm": 0.2975255846977234,
+      "learning_rate": 4.386617100371748e-06,
+      "loss": 0.3928,
+      "step": 237
+    },
+    {
+      "epoch": 0.2662192393736018,
+      "grad_norm": 0.353934109210968,
+      "learning_rate": 4.405204460966543e-06,
+      "loss": 0.4785,
+      "step": 238
+    },
+    {
+      "epoch": 0.26733780760626397,
+      "grad_norm": 0.31682029366493225,
+      "learning_rate": 4.423791821561339e-06,
+      "loss": 0.4355,
+      "step": 239
+    },
+    {
+      "epoch": 0.2684563758389262,
+      "grad_norm": 0.30169084668159485,
+      "learning_rate": 4.442379182156134e-06,
+      "loss": 0.4226,
+      "step": 240
+    },
+    {
+      "epoch": 0.2695749440715884,
+      "grad_norm": 0.35651126503944397,
+      "learning_rate": 4.4609665427509296e-06,
+      "loss": 0.4171,
+      "step": 241
+    },
+    {
+      "epoch": 0.27069351230425054,
+      "grad_norm": 0.313177227973938,
+      "learning_rate": 4.479553903345725e-06,
+      "loss": 0.3637,
+      "step": 242
+    },
+    {
+      "epoch": 0.27181208053691275,
+      "grad_norm": 0.29632848501205444,
+      "learning_rate": 4.4981412639405204e-06,
+      "loss": 0.4398,
+      "step": 243
+    },
+    {
+      "epoch": 0.27293064876957496,
+      "grad_norm": 0.3289046883583069,
+      "learning_rate": 4.516728624535316e-06,
+      "loss": 0.3457,
+      "step": 244
+    },
+    {
+      "epoch": 0.2740492170022371,
+      "grad_norm": 0.31114718317985535,
+      "learning_rate": 4.535315985130112e-06,
+      "loss": 0.4324,
+      "step": 245
+    },
+    {
+      "epoch": 0.2751677852348993,
+      "grad_norm": 0.2849334478378296,
+      "learning_rate": 4.553903345724908e-06,
+      "loss": 0.3488,
+      "step": 246
+    },
+    {
+      "epoch": 0.27628635346756153,
+      "grad_norm": 0.32387015223503113,
+      "learning_rate": 4.572490706319703e-06,
+      "loss": 0.4026,
+      "step": 247
+    },
+    {
+      "epoch": 0.27740492170022374,
+      "grad_norm": 0.30889856815338135,
+      "learning_rate": 4.591078066914498e-06,
+      "loss": 0.4513,
+      "step": 248
+    },
+    {
+      "epoch": 0.2785234899328859,
+      "grad_norm": 0.33541280031204224,
+      "learning_rate": 4.609665427509294e-06,
+      "loss": 0.4549,
+      "step": 249
+    },
+    {
+      "epoch": 0.2796420581655481,
+      "grad_norm": 0.3286095857620239,
+      "learning_rate": 4.62825278810409e-06,
+      "loss": 0.4373,
+      "step": 250
+    },
+    {
+      "epoch": 0.2796420581655481,
+      "eval_loss": 0.4131084084510803,
+      "eval_runtime": 376.7437,
+      "eval_samples_per_second": 2.11,
+      "eval_steps_per_second": 2.11,
+      "step": 250
+    },
+    {
+      "epoch": 0.2807606263982103,
+      "grad_norm": 0.26990702748298645,
+      "learning_rate": 4.646840148698885e-06,
+      "loss": 0.3192,
+      "step": 251
+    },
+    {
+      "epoch": 0.28187919463087246,
+      "grad_norm": 0.28313425183296204,
+      "learning_rate": 4.665427509293681e-06,
+      "loss": 0.38,
+      "step": 252
+    },
+    {
+      "epoch": 0.2829977628635347,
+      "grad_norm": 0.42295464873313904,
+      "learning_rate": 4.684014869888477e-06,
+      "loss": 0.4542,
+      "step": 253
+    },
+    {
+      "epoch": 0.2841163310961969,
+      "grad_norm": 0.30162960290908813,
+      "learning_rate": 4.702602230483272e-06,
+      "loss": 0.3455,
+      "step": 254
+    },
+    {
+      "epoch": 0.28523489932885904,
+      "grad_norm": 0.27259892225265503,
+      "learning_rate": 4.7211895910780675e-06,
+      "loss": 0.3299,
+      "step": 255
+    },
+    {
+      "epoch": 0.28635346756152125,
+      "grad_norm": 0.37468963861465454,
+      "learning_rate": 4.739776951672863e-06,
+      "loss": 0.458,
+      "step": 256
+    },
+    {
+      "epoch": 0.28747203579418346,
+      "grad_norm": 0.41411465406417847,
+      "learning_rate": 4.758364312267658e-06,
+      "loss": 0.4466,
+      "step": 257
+    },
+    {
+      "epoch": 0.28859060402684567,
+      "grad_norm": 0.35305294394493103,
+      "learning_rate": 4.776951672862453e-06,
+      "loss": 0.3834,
+      "step": 258
+    },
+    {
+      "epoch": 0.2897091722595078,
+      "grad_norm": 0.29956525564193726,
+      "learning_rate": 4.795539033457249e-06,
+      "loss": 0.3762,
+      "step": 259
+    },
+    {
+      "epoch": 0.29082774049217003,
+      "grad_norm": 0.39069366455078125,
+      "learning_rate": 4.814126394052045e-06,
+      "loss": 0.4058,
+      "step": 260
+    },
+    {
+      "epoch": 0.29194630872483224,
+      "grad_norm": 0.3356025516986847,
+      "learning_rate": 4.83271375464684e-06,
+      "loss": 0.3725,
+      "step": 261
+    },
+    {
+      "epoch": 0.2930648769574944,
+      "grad_norm": 0.45072317123413086,
+      "learning_rate": 4.851301115241636e-06,
+      "loss": 0.3914,
+      "step": 262
+    },
+    {
+      "epoch": 0.2941834451901566,
+      "grad_norm": 0.44041502475738525,
+      "learning_rate": 4.869888475836432e-06,
+      "loss": 0.4268,
+      "step": 263
+    },
+    {
+      "epoch": 0.2953020134228188,
+      "grad_norm": 0.30348747968673706,
+      "learning_rate": 4.888475836431228e-06,
+      "loss": 0.3625,
+      "step": 264
+    },
+    {
+      "epoch": 0.29642058165548096,
+      "grad_norm": 0.3397924304008484,
+      "learning_rate": 4.907063197026023e-06,
+      "loss": 0.385,
+      "step": 265
+    },
+    {
+      "epoch": 0.2975391498881432,
+      "grad_norm": 0.29160842299461365,
+      "learning_rate": 4.925650557620818e-06,
+      "loss": 0.3514,
+      "step": 266
+    },
+    {
+      "epoch": 0.2986577181208054,
+      "grad_norm": 0.3206447660923004,
+      "learning_rate": 4.944237918215614e-06,
+      "loss": 0.4451,
+      "step": 267
+    },
+    {
+      "epoch": 0.29977628635346754,
+      "grad_norm": 0.39959484338760376,
+      "learning_rate": 4.9628252788104095e-06,
+      "loss": 0.5073,
+      "step": 268
+    },
+    {
+      "epoch": 0.30089485458612975,
+      "grad_norm": 0.28956446051597595,
+      "learning_rate": 4.9814126394052045e-06,
+      "loss": 0.3624,
+      "step": 269
+    },
+    {
+      "epoch": 0.30201342281879195,
+      "grad_norm": 0.38245102763175964,
+      "learning_rate": 5e-06,
+      "loss": 0.3833,
+      "step": 270
+    },
+    {
+      "epoch": 0.30313199105145416,
+      "grad_norm": 0.3376619517803192,
+      "learning_rate": 4.999997881175173e-06,
+      "loss": 0.3934,
+      "step": 271
+    },
+    {
+      "epoch": 0.3042505592841163,
+      "grad_norm": 0.3451228737831116,
+      "learning_rate": 4.9999915247042805e-06,
+      "loss": 0.4057,
+      "step": 272
+    },
+    {
+      "epoch": 0.3053691275167785,
+      "grad_norm": 0.2706328332424164,
+      "learning_rate": 4.9999809305981e-06,
+      "loss": 0.3723,
+      "step": 273
+    },
+    {
+      "epoch": 0.30648769574944074,
+      "grad_norm": 0.45232075452804565,
+      "learning_rate": 4.999966098874587e-06,
+      "loss": 0.5705,
+      "step": 274
+    },
+    {
+      "epoch": 0.3076062639821029,
+      "grad_norm": 0.32139483094215393,
+      "learning_rate": 4.999947029558883e-06,
+      "loss": 0.4073,
+      "step": 275
+    },
+    {
+      "epoch": 0.3087248322147651,
+      "grad_norm": 0.30884405970573425,
+      "learning_rate": 4.9999237226833104e-06,
+      "loss": 0.3773,
+      "step": 276
+    },
+    {
+      "epoch": 0.3098434004474273,
+      "grad_norm": 0.32114043831825256,
+      "learning_rate": 4.999896178287378e-06,
+      "loss": 0.3567,
+      "step": 277
+    },
+    {
+      "epoch": 0.31096196868008946,
+      "grad_norm": 0.3145090341567993,
+      "learning_rate": 4.999864396417774e-06,
+      "loss": 0.3853,
+      "step": 278
+    },
+    {
+      "epoch": 0.31208053691275167,
+      "grad_norm": 0.3029519021511078,
+      "learning_rate": 4.99982837712837e-06,
+      "loss": 0.3655,
+      "step": 279
+    },
+    {
+      "epoch": 0.3131991051454139,
+      "grad_norm": 0.39528998732566833,
+      "learning_rate": 4.999788120480222e-06,
+      "loss": 0.4127,
+      "step": 280
+    },
+    {
+      "epoch": 0.3143176733780761,
+      "grad_norm": 0.30222469568252563,
+      "learning_rate": 4.999743626541567e-06,
+      "loss": 0.4113,
+      "step": 281
+    },
+    {
+      "epoch": 0.31543624161073824,
+      "grad_norm": 0.28925079107284546,
+      "learning_rate": 4.999694895387825e-06,
+      "loss": 0.3442,
+      "step": 282
+    },
+    {
+      "epoch": 0.31655480984340045,
+      "grad_norm": 0.3931190073490143,
+      "learning_rate": 4.999641927101597e-06,
+      "loss": 0.368,
+      "step": 283
+    },
+    {
+      "epoch": 0.31767337807606266,
+      "grad_norm": 0.3024509847164154,
+      "learning_rate": 4.999584721772669e-06,
+      "loss": 0.4465,
+      "step": 284
+    },
+    {
+      "epoch": 0.3187919463087248,
+      "grad_norm": 0.3034071922302246,
+      "learning_rate": 4.999523279498007e-06,
+      "loss": 0.4664,
+      "step": 285
+    },
+    {
+      "epoch": 0.319910514541387,
+      "grad_norm": 0.32978177070617676,
+      "learning_rate": 4.999457600381758e-06,
+      "loss": 0.4191,
+      "step": 286
+    },
+    {
+      "epoch": 0.32102908277404923,
+      "grad_norm": 0.284646600484848,
+      "learning_rate": 4.999387684535255e-06,
+      "loss": 0.322,
+      "step": 287
+    },
+    {
+      "epoch": 0.3221476510067114,
+      "grad_norm": 0.3090347945690155,
+      "learning_rate": 4.999313532077007e-06,
+      "loss": 0.3882,
+      "step": 288
+    },
+    {
+      "epoch": 0.3232662192393736,
+      "grad_norm": 0.34074026346206665,
+      "learning_rate": 4.999235143132708e-06,
+      "loss": 0.4021,
+      "step": 289
+    },
+    {
+      "epoch": 0.3243847874720358,
+      "grad_norm": 0.3228881061077118,
+      "learning_rate": 4.999152517835232e-06,
+      "loss": 0.3467,
+      "step": 290
+    },
+    {
+      "epoch": 0.32550335570469796,
+      "grad_norm": 0.2910652458667755,
+      "learning_rate": 4.999065656324632e-06,
+      "loss": 0.3551,
+      "step": 291
+    },
+    {
+      "epoch": 0.32662192393736017,
+      "grad_norm": 0.38834840059280396,
+      "learning_rate": 4.998974558748146e-06,
+      "loss": 0.4833,
+      "step": 292
+    },
+    {
+      "epoch": 0.3277404921700224,
+      "grad_norm": 0.27184849977493286,
+      "learning_rate": 4.998879225260189e-06,
+      "loss": 0.3509,
+      "step": 293
+    },
+    {
+      "epoch": 0.3288590604026846,
+      "grad_norm": 0.29583096504211426,
+      "learning_rate": 4.998779656022356e-06,
+      "loss": 0.3283,
+      "step": 294
+    },
+    {
+      "epoch": 0.32997762863534674,
+      "grad_norm": 0.30960991978645325,
+      "learning_rate": 4.998675851203423e-06,
+      "loss": 0.3533,
+      "step": 295
+    },
+    {
+      "epoch": 0.33109619686800895,
+      "grad_norm": 0.2893279194831848,
+      "learning_rate": 4.998567810979347e-06,
+      "loss": 0.3227,
+      "step": 296
+    },
+    {
+      "epoch": 0.33221476510067116,
+      "grad_norm": 0.30004778504371643,
+      "learning_rate": 4.998455535533261e-06,
+      "loss": 0.3823,
+      "step": 297
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.3450540602207184,
+      "learning_rate": 4.9983390250554795e-06,
+      "loss": 0.3911,
+      "step": 298
+    },
+    {
+      "epoch": 0.3344519015659955,
+      "grad_norm": 0.28002044558525085,
+      "learning_rate": 4.998218279743494e-06,
+      "loss": 0.3474,
+      "step": 299
+    },
+    {
+      "epoch": 0.33557046979865773,
+      "grad_norm": 0.28742578625679016,
+      "learning_rate": 4.998093299801976e-06,
+      "loss": 0.3502,
+      "step": 300
+    },
+    {
+      "epoch": 0.33557046979865773,
+      "eval_loss": 0.38576123118400574,
+      "eval_runtime": 376.6193,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 300
+    },
+    {
+      "epoch": 0.3366890380313199,
+      "grad_norm": 0.33350133895874023,
+      "learning_rate": 4.997964085442772e-06,
+      "loss": 0.4133,
+      "step": 301
+    },
+    {
+      "epoch": 0.3378076062639821,
+      "grad_norm": 0.2661343216896057,
+      "learning_rate": 4.997830636884911e-06,
+      "loss": 0.3021,
+      "step": 302
+    },
+    {
+      "epoch": 0.3389261744966443,
+      "grad_norm": 0.34488674998283386,
+      "learning_rate": 4.997692954354594e-06,
+      "loss": 0.4405,
+      "step": 303
+    },
+    {
+      "epoch": 0.3400447427293065,
+      "grad_norm": 0.30503594875335693,
+      "learning_rate": 4.997551038085203e-06,
+      "loss": 0.3489,
+      "step": 304
+    },
+    {
+      "epoch": 0.34116331096196867,
+      "grad_norm": 0.32220813632011414,
+      "learning_rate": 4.997404888317292e-06,
+      "loss": 0.3798,
+      "step": 305
+    },
+    {
+      "epoch": 0.3422818791946309,
+      "grad_norm": 0.26920878887176514,
+      "learning_rate": 4.997254505298596e-06,
+      "loss": 0.313,
+      "step": 306
+    },
+    {
+      "epoch": 0.3434004474272931,
+      "grad_norm": 0.3296228349208832,
+      "learning_rate": 4.997099889284022e-06,
+      "loss": 0.3938,
+      "step": 307
+    },
+    {
+      "epoch": 0.34451901565995524,
+      "grad_norm": 0.30485832691192627,
+      "learning_rate": 4.996941040535654e-06,
+      "loss": 0.3955,
+      "step": 308
+    },
+    {
+      "epoch": 0.34563758389261745,
+      "grad_norm": 0.2854394018650055,
+      "learning_rate": 4.9967779593227496e-06,
+      "loss": 0.3372,
+      "step": 309
+    },
+    {
+      "epoch": 0.34675615212527966,
+      "grad_norm": 0.30855122208595276,
+      "learning_rate": 4.996610645921741e-06,
+      "loss": 0.4171,
+      "step": 310
+    },
+    {
+      "epoch": 0.3478747203579418,
+      "grad_norm": 0.36724746227264404,
+      "learning_rate": 4.996439100616235e-06,
+      "loss": 0.3319,
+      "step": 311
+    },
+    {
+      "epoch": 0.348993288590604,
+      "grad_norm": 0.3428019881248474,
+      "learning_rate": 4.9962633236970115e-06,
+      "loss": 0.3661,
+      "step": 312
+    },
+    {
+      "epoch": 0.35011185682326623,
+      "grad_norm": 0.32487553358078003,
+      "learning_rate": 4.996083315462023e-06,
+      "loss": 0.4127,
+      "step": 313
+    },
+    {
+      "epoch": 0.3512304250559284,
+      "grad_norm": 0.2544281482696533,
+      "learning_rate": 4.995899076216392e-06,
+      "loss": 0.3048,
+      "step": 314
+    },
+    {
+      "epoch": 0.3523489932885906,
+      "grad_norm": 0.36887869238853455,
+      "learning_rate": 4.9957106062724185e-06,
+      "loss": 0.4714,
+      "step": 315
+    },
+    {
+      "epoch": 0.3534675615212528,
+      "grad_norm": 0.34506508708000183,
+      "learning_rate": 4.995517905949567e-06,
+      "loss": 0.4655,
+      "step": 316
+    },
+    {
+      "epoch": 0.354586129753915,
+      "grad_norm": 0.3289477229118347,
+      "learning_rate": 4.995320975574479e-06,
+      "loss": 0.3223,
+      "step": 317
+    },
+    {
+      "epoch": 0.35570469798657717,
+      "grad_norm": 0.28838202357292175,
+      "learning_rate": 4.995119815480961e-06,
+      "loss": 0.3769,
+      "step": 318
+    },
+    {
+      "epoch": 0.3568232662192394,
+      "grad_norm": 0.35621005296707153,
+      "learning_rate": 4.994914426009992e-06,
+      "loss": 0.3668,
+      "step": 319
+    },
+    {
+      "epoch": 0.3579418344519016,
+      "grad_norm": 0.5147815346717834,
+      "learning_rate": 4.994704807509721e-06,
+      "loss": 0.4145,
+      "step": 320
+    },
+    {
+      "epoch": 0.35906040268456374,
+      "grad_norm": 0.34419554471969604,
+      "learning_rate": 4.994490960335462e-06,
+      "loss": 0.3969,
+      "step": 321
+    },
+    {
+      "epoch": 0.36017897091722595,
+      "grad_norm": 0.3444998264312744,
+      "learning_rate": 4.9942728848496995e-06,
+      "loss": 0.3635,
+      "step": 322
+    },
+    {
+      "epoch": 0.36129753914988816,
+      "grad_norm": 0.3281770944595337,
+      "learning_rate": 4.9940505814220845e-06,
+      "loss": 0.355,
+      "step": 323
+    },
+    {
+      "epoch": 0.3624161073825503,
+      "grad_norm": 0.28537437319755554,
+      "learning_rate": 4.993824050429434e-06,
+      "loss": 0.3255,
+      "step": 324
+    },
+    {
+      "epoch": 0.3635346756152125,
+      "grad_norm": 0.27903199195861816,
+      "learning_rate": 4.993593292255733e-06,
+      "loss": 0.2951,
+      "step": 325
+    },
+    {
+      "epoch": 0.36465324384787473,
+      "grad_norm": 0.30100008845329285,
+      "learning_rate": 4.99335830729213e-06,
+      "loss": 0.3651,
+      "step": 326
+    },
+    {
+      "epoch": 0.36577181208053694,
+      "grad_norm": 0.2706114649772644,
+      "learning_rate": 4.993119095936937e-06,
+      "loss": 0.3046,
+      "step": 327
+    },
+    {
+      "epoch": 0.3668903803131991,
+      "grad_norm": 0.3363129198551178,
+      "learning_rate": 4.992875658595633e-06,
+      "loss": 0.4092,
+      "step": 328
+    },
+    {
+      "epoch": 0.3680089485458613,
+      "grad_norm": 0.33284562826156616,
+      "learning_rate": 4.992627995680859e-06,
+      "loss": 0.3601,
+      "step": 329
+    },
+    {
+      "epoch": 0.3691275167785235,
+      "grad_norm": 0.2990688383579254,
+      "learning_rate": 4.992376107612418e-06,
+      "loss": 0.3716,
+      "step": 330
+    },
+    {
+      "epoch": 0.37024608501118567,
+      "grad_norm": 0.31069859862327576,
+      "learning_rate": 4.992119994817275e-06,
+      "loss": 0.3718,
+      "step": 331
+    },
+    {
+      "epoch": 0.3713646532438479,
+      "grad_norm": 0.3040110468864441,
+      "learning_rate": 4.991859657729557e-06,
+      "loss": 0.3414,
+      "step": 332
+    },
+    {
+      "epoch": 0.3724832214765101,
+      "grad_norm": 0.2946312129497528,
+      "learning_rate": 4.991595096790552e-06,
+      "loss": 0.3422,
+      "step": 333
+    },
+    {
+      "epoch": 0.37360178970917224,
+      "grad_norm": 0.3156837522983551,
+      "learning_rate": 4.9913263124487035e-06,
+      "loss": 0.3663,
+      "step": 334
+    },
+    {
+      "epoch": 0.37472035794183445,
+      "grad_norm": 0.32687637209892273,
+      "learning_rate": 4.99105330515962e-06,
+      "loss": 0.3748,
+      "step": 335
+    },
+    {
+      "epoch": 0.37583892617449666,
+      "grad_norm": 0.3898880183696747,
+      "learning_rate": 4.990776075386064e-06,
+      "loss": 0.3862,
+      "step": 336
+    },
+    {
+      "epoch": 0.3769574944071588,
+      "grad_norm": 0.2722826302051544,
+      "learning_rate": 4.990494623597958e-06,
+      "loss": 0.3093,
+      "step": 337
+    },
+    {
+      "epoch": 0.378076062639821,
+      "grad_norm": 0.36950165033340454,
+      "learning_rate": 4.990208950272378e-06,
+      "loss": 0.4385,
+      "step": 338
+    },
+    {
+      "epoch": 0.37919463087248323,
+      "grad_norm": 0.297207236289978,
+      "learning_rate": 4.989919055893557e-06,
+      "loss": 0.3591,
+      "step": 339
+    },
+    {
+      "epoch": 0.38031319910514544,
+      "grad_norm": 0.4501153826713562,
+      "learning_rate": 4.989624940952884e-06,
+      "loss": 0.321,
+      "step": 340
+    },
+    {
+      "epoch": 0.3814317673378076,
+      "grad_norm": 0.30277740955352783,
+      "learning_rate": 4.989326605948902e-06,
+      "loss": 0.3242,
+      "step": 341
+    },
+    {
+      "epoch": 0.3825503355704698,
+      "grad_norm": 0.3023722171783447,
+      "learning_rate": 4.989024051387307e-06,
+      "loss": 0.351,
+      "step": 342
+    },
+    {
+      "epoch": 0.383668903803132,
+      "grad_norm": 0.35803765058517456,
+      "learning_rate": 4.988717277780946e-06,
+      "loss": 0.3742,
+      "step": 343
+    },
+    {
+      "epoch": 0.38478747203579416,
+      "grad_norm": 0.2736894488334656,
+      "learning_rate": 4.9884062856498185e-06,
+      "loss": 0.3276,
+      "step": 344
+    },
+    {
+      "epoch": 0.3859060402684564,
+      "grad_norm": 0.31314927339553833,
+      "learning_rate": 4.988091075521076e-06,
+      "loss": 0.3729,
+      "step": 345
+    },
+    {
+      "epoch": 0.3870246085011186,
+      "grad_norm": 0.30895665287971497,
+      "learning_rate": 4.987771647929017e-06,
+      "loss": 0.3914,
+      "step": 346
+    },
+    {
+      "epoch": 0.38814317673378074,
+      "grad_norm": 0.3819127082824707,
+      "learning_rate": 4.9874480034150915e-06,
+      "loss": 0.438,
+      "step": 347
+    },
+    {
+      "epoch": 0.38926174496644295,
+      "grad_norm": 0.31712761521339417,
+      "learning_rate": 4.987120142527896e-06,
+      "loss": 0.4094,
+      "step": 348
+    },
+    {
+      "epoch": 0.39038031319910516,
+      "grad_norm": 0.3410664498806,
+      "learning_rate": 4.986788065823175e-06,
+      "loss": 0.4708,
+      "step": 349
+    },
+    {
+      "epoch": 0.39149888143176736,
+      "grad_norm": 0.2916277050971985,
+      "learning_rate": 4.986451773863817e-06,
+      "loss": 0.3627,
+      "step": 350
+    },
+    {
+      "epoch": 0.39149888143176736,
+      "eval_loss": 0.3686007857322693,
+      "eval_runtime": 376.6949,
+      "eval_samples_per_second": 2.11,
+      "eval_steps_per_second": 2.11,
+      "step": 350
+    },
+    {
+      "epoch": 0.3926174496644295,
+      "grad_norm": 0.326291024684906,
+      "learning_rate": 4.9861112672198575e-06,
+      "loss": 0.3473,
+      "step": 351
+    },
+    {
+      "epoch": 0.39373601789709173,
+      "grad_norm": 0.32254719734191895,
+      "learning_rate": 4.985766546468477e-06,
+      "loss": 0.4182,
+      "step": 352
+    },
+    {
+      "epoch": 0.39485458612975394,
+      "grad_norm": 0.4112759232521057,
+      "learning_rate": 4.985417612193996e-06,
+      "loss": 0.3725,
+      "step": 353
+    },
+    {
+      "epoch": 0.3959731543624161,
+      "grad_norm": 0.3436565697193146,
+      "learning_rate": 4.98506446498788e-06,
+      "loss": 0.3168,
+      "step": 354
+    },
+    {
+      "epoch": 0.3970917225950783,
+      "grad_norm": 0.3482731282711029,
+      "learning_rate": 4.984707105448734e-06,
+      "loss": 0.3574,
+      "step": 355
+    },
+    {
+      "epoch": 0.3982102908277405,
+      "grad_norm": 0.29381173849105835,
+      "learning_rate": 4.9843455341823044e-06,
+      "loss": 0.327,
+      "step": 356
+    },
+    {
+      "epoch": 0.39932885906040266,
+      "grad_norm": 0.3394116461277008,
+      "learning_rate": 4.983979751801475e-06,
+      "loss": 0.4046,
+      "step": 357
+    },
+    {
+      "epoch": 0.4004474272930649,
+      "grad_norm": 0.2818320393562317,
+      "learning_rate": 4.983609758926271e-06,
+      "loss": 0.3565,
+      "step": 358
+    },
+    {
+      "epoch": 0.4015659955257271,
+      "grad_norm": 0.3093859851360321,
+      "learning_rate": 4.983235556183851e-06,
+      "loss": 0.3704,
+      "step": 359
+    },
+    {
+      "epoch": 0.40268456375838924,
+      "grad_norm": 0.3526606261730194,
+      "learning_rate": 4.9828571442085104e-06,
+      "loss": 0.3563,
+      "step": 360
+    },
+    {
+      "epoch": 0.40380313199105144,
+      "grad_norm": 0.2993238866329193,
+      "learning_rate": 4.982474523641682e-06,
+      "loss": 0.379,
+      "step": 361
+    },
+    {
+      "epoch": 0.40492170022371365,
+      "grad_norm": 0.34483104944229126,
+      "learning_rate": 4.982087695131929e-06,
+      "loss": 0.3996,
+      "step": 362
+    },
+    {
+      "epoch": 0.40604026845637586,
+      "grad_norm": 0.4885406792163849,
+      "learning_rate": 4.981696659334949e-06,
+      "loss": 0.4449,
+      "step": 363
+    },
+    {
+      "epoch": 0.407158836689038,
+      "grad_norm": 0.28771209716796875,
+      "learning_rate": 4.981301416913573e-06,
+      "loss": 0.3441,
+      "step": 364
+    },
+    {
+      "epoch": 0.4082774049217002,
+      "grad_norm": 0.3830564320087433,
+      "learning_rate": 4.980901968537758e-06,
+      "loss": 0.3599,
+      "step": 365
+    },
+    {
+      "epoch": 0.40939597315436244,
+      "grad_norm": 0.26629677414894104,
+      "learning_rate": 4.980498314884594e-06,
+      "loss": 0.3372,
+      "step": 366
+    },
+    {
+      "epoch": 0.4105145413870246,
+      "grad_norm": 0.34145382046699524,
+      "learning_rate": 4.980090456638299e-06,
+      "loss": 0.4427,
+      "step": 367
+    },
+    {
+      "epoch": 0.4116331096196868,
+      "grad_norm": 0.3672374188899994,
+      "learning_rate": 4.979678394490215e-06,
+      "loss": 0.4151,
+      "step": 368
+    },
+    {
+      "epoch": 0.412751677852349,
+      "grad_norm": 0.3465013802051544,
+      "learning_rate": 4.979262129138814e-06,
+      "loss": 0.429,
+      "step": 369
+    },
+    {
+      "epoch": 0.41387024608501116,
+      "grad_norm": 0.26761046051979065,
+      "learning_rate": 4.978841661289689e-06,
+      "loss": 0.3124,
+      "step": 370
+    },
+    {
+      "epoch": 0.41498881431767337,
+      "grad_norm": 0.41746488213539124,
+      "learning_rate": 4.978416991655561e-06,
+      "loss": 0.3544,
+      "step": 371
+    },
+    {
+      "epoch": 0.4161073825503356,
+      "grad_norm": 0.30543252825737,
+      "learning_rate": 4.977988120956268e-06,
+      "loss": 0.3815,
+      "step": 372
+    },
+    {
+      "epoch": 0.4172259507829978,
+      "grad_norm": 0.29315754771232605,
+      "learning_rate": 4.977555049918771e-06,
+      "loss": 0.2687,
+      "step": 373
+    },
+    {
+      "epoch": 0.41834451901565994,
+      "grad_norm": 0.3436398506164551,
+      "learning_rate": 4.977117779277153e-06,
+      "loss": 0.431,
+      "step": 374
+    },
+    {
+      "epoch": 0.41946308724832215,
+      "grad_norm": 0.31415262818336487,
+      "learning_rate": 4.976676309772613e-06,
+      "loss": 0.3513,
+      "step": 375
+    },
+    {
+      "epoch": 0.42058165548098436,
+      "grad_norm": 0.3410176932811737,
+      "learning_rate": 4.9762306421534695e-06,
+      "loss": 0.3502,
+      "step": 376
+    },
+    {
+      "epoch": 0.4217002237136465,
+      "grad_norm": 0.30478906631469727,
+      "learning_rate": 4.9757807771751545e-06,
+      "loss": 0.3324,
+      "step": 377
+    },
+    {
+      "epoch": 0.4228187919463087,
+      "grad_norm": 0.3103587031364441,
+      "learning_rate": 4.975326715600216e-06,
+      "loss": 0.3529,
+      "step": 378
+    },
+    {
+      "epoch": 0.42393736017897093,
+      "grad_norm": 0.3906938433647156,
+      "learning_rate": 4.974868458198317e-06,
+      "loss": 0.4729,
+      "step": 379
+    },
+    {
+      "epoch": 0.4250559284116331,
+      "grad_norm": 0.3342723846435547,
+      "learning_rate": 4.974406005746228e-06,
+      "loss": 0.4115,
+      "step": 380
+    },
+    {
+      "epoch": 0.4261744966442953,
+      "grad_norm": 0.2926039397716522,
+      "learning_rate": 4.973939359027838e-06,
+      "loss": 0.3647,
+      "step": 381
+    },
+    {
+      "epoch": 0.4272930648769575,
+      "grad_norm": 0.32736265659332275,
+      "learning_rate": 4.973468518834138e-06,
+      "loss": 0.4128,
+      "step": 382
+    },
+    {
+      "epoch": 0.42841163310961966,
+      "grad_norm": 0.2867015600204468,
+      "learning_rate": 4.972993485963231e-06,
+      "loss": 0.352,
+      "step": 383
+    },
+    {
+      "epoch": 0.42953020134228187,
+      "grad_norm": 0.28875496983528137,
+      "learning_rate": 4.972514261220327e-06,
+      "loss": 0.3636,
+      "step": 384
+    },
+    {
+      "epoch": 0.4306487695749441,
+      "grad_norm": 0.3173429071903229,
+      "learning_rate": 4.9720308454177405e-06,
+      "loss": 0.3314,
+      "step": 385
+    },
+    {
+      "epoch": 0.4317673378076063,
+      "grad_norm": 0.29224690794944763,
+      "learning_rate": 4.971543239374889e-06,
+      "loss": 0.2705,
+      "step": 386
+    },
+    {
+      "epoch": 0.43288590604026844,
+      "grad_norm": 0.33834773302078247,
+      "learning_rate": 4.9710514439182945e-06,
+      "loss": 0.3637,
+      "step": 387
+    },
+    {
+      "epoch": 0.43400447427293065,
+      "grad_norm": 0.3793725371360779,
+      "learning_rate": 4.97055545988158e-06,
+      "loss": 0.3267,
+      "step": 388
+    },
+    {
+      "epoch": 0.43512304250559286,
+      "grad_norm": 0.3243982791900635,
+      "learning_rate": 4.970055288105468e-06,
+      "loss": 0.3482,
+      "step": 389
+    },
+    {
+      "epoch": 0.436241610738255,
+      "grad_norm": 0.3603496849536896,
+      "learning_rate": 4.96955092943778e-06,
+      "loss": 0.3929,
+      "step": 390
+    },
+    {
+      "epoch": 0.4373601789709172,
+      "grad_norm": 0.3242970108985901,
+      "learning_rate": 4.969042384733434e-06,
+      "loss": 0.3696,
+      "step": 391
+    },
+    {
+      "epoch": 0.43847874720357943,
+      "grad_norm": 0.3667403757572174,
+      "learning_rate": 4.968529654854443e-06,
+      "loss": 0.3843,
+      "step": 392
+    },
+    {
+      "epoch": 0.4395973154362416,
+      "grad_norm": 0.299001544713974,
+      "learning_rate": 4.9680127406699156e-06,
+      "loss": 0.3374,
+      "step": 393
+    },
+    {
+      "epoch": 0.4407158836689038,
+      "grad_norm": 0.29333847761154175,
+      "learning_rate": 4.967491643056051e-06,
+      "loss": 0.3174,
+      "step": 394
+    },
+    {
+      "epoch": 0.441834451901566,
+      "grad_norm": 0.3308899402618408,
+      "learning_rate": 4.966966362896143e-06,
+      "loss": 0.3489,
+      "step": 395
+    },
+    {
+      "epoch": 0.4429530201342282,
+      "grad_norm": 0.3108149766921997,
+      "learning_rate": 4.966436901080571e-06,
+      "loss": 0.3199,
+      "step": 396
+    },
+    {
+      "epoch": 0.44407158836689037,
+      "grad_norm": 0.29311317205429077,
+      "learning_rate": 4.965903258506806e-06,
+      "loss": 0.2703,
+      "step": 397
+    },
+    {
+      "epoch": 0.4451901565995526,
+      "grad_norm": 0.32560640573501587,
+      "learning_rate": 4.965365436079404e-06,
+      "loss": 0.3201,
+      "step": 398
+    },
+    {
+      "epoch": 0.4463087248322148,
+      "grad_norm": 0.28091391921043396,
+      "learning_rate": 4.964823434710005e-06,
+      "loss": 0.3026,
+      "step": 399
+    },
+    {
+      "epoch": 0.44742729306487694,
+      "grad_norm": 0.34165579080581665,
+      "learning_rate": 4.9642772553173335e-06,
+      "loss": 0.3384,
+      "step": 400
+    },
+    {
+      "epoch": 0.44742729306487694,
+      "eval_loss": 0.3573058843612671,
+      "eval_runtime": 376.8802,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 2.109,
+      "step": 400
+    },
+    {
+      "epoch": 0.44854586129753915,
+      "grad_norm": 0.31163346767425537,
+      "learning_rate": 4.9637268988271975e-06,
+      "loss": 0.3285,
+      "step": 401
+    },
+    {
+      "epoch": 0.44966442953020136,
+      "grad_norm": 0.31296831369400024,
+      "learning_rate": 4.963172366172484e-06,
+      "loss": 0.3588,
+      "step": 402
+    },
+    {
+      "epoch": 0.4507829977628635,
+      "grad_norm": 0.3521609902381897,
+      "learning_rate": 4.962613658293158e-06,
+      "loss": 0.352,
+      "step": 403
+    },
+    {
+      "epoch": 0.4519015659955257,
+      "grad_norm": 0.31171315908432007,
+      "learning_rate": 4.962050776136265e-06,
+      "loss": 0.3619,
+      "step": 404
+    },
+    {
+      "epoch": 0.45302013422818793,
+      "grad_norm": 0.3633255362510681,
+      "learning_rate": 4.961483720655922e-06,
+      "loss": 0.404,
+      "step": 405
+    },
+    {
+      "epoch": 0.4541387024608501,
+      "grad_norm": 0.4151226580142975,
+      "learning_rate": 4.960912492813323e-06,
+      "loss": 0.4436,
+      "step": 406
+    },
+    {
+      "epoch": 0.4552572706935123,
+      "grad_norm": 0.33188238739967346,
+      "learning_rate": 4.960337093576732e-06,
+      "loss": 0.4034,
+      "step": 407
+    },
+    {
+      "epoch": 0.4563758389261745,
+      "grad_norm": 0.3268672823905945,
+      "learning_rate": 4.959757523921488e-06,
+      "loss": 0.3912,
+      "step": 408
+    },
+    {
+      "epoch": 0.4574944071588367,
+      "grad_norm": 0.3294200897216797,
+      "learning_rate": 4.959173784829993e-06,
+      "loss": 0.3231,
+      "step": 409
+    },
+    {
+      "epoch": 0.45861297539149887,
+      "grad_norm": 0.3666338622570038,
+      "learning_rate": 4.9585858772917204e-06,
+      "loss": 0.4279,
+      "step": 410
+    },
+    {
+      "epoch": 0.4597315436241611,
+      "grad_norm": 0.3587000370025635,
+      "learning_rate": 4.957993802303211e-06,
+      "loss": 0.3389,
+      "step": 411
+    },
+    {
+      "epoch": 0.4608501118568233,
+      "grad_norm": 0.3554668128490448,
+      "learning_rate": 4.9573975608680646e-06,
+      "loss": 0.3822,
+      "step": 412
+    },
+    {
+      "epoch": 0.46196868008948544,
+      "grad_norm": 0.33347633481025696,
+      "learning_rate": 4.9567971539969475e-06,
+      "loss": 0.3477,
+      "step": 413
+    },
+    {
+      "epoch": 0.46308724832214765,
+      "grad_norm": 0.332907110452652,
+      "learning_rate": 4.9561925827075856e-06,
+      "loss": 0.3262,
+      "step": 414
+    },
+    {
+      "epoch": 0.46420581655480986,
+      "grad_norm": 0.2875790297985077,
+      "learning_rate": 4.955583848024762e-06,
+      "loss": 0.3188,
+      "step": 415
+    },
+    {
+      "epoch": 0.465324384787472,
+      "grad_norm": 0.3714223802089691,
+      "learning_rate": 4.954970950980319e-06,
+      "loss": 0.4167,
+      "step": 416
+    },
+    {
+      "epoch": 0.4664429530201342,
+      "grad_norm": 0.38172417879104614,
+      "learning_rate": 4.954353892613155e-06,
+      "loss": 0.3922,
+      "step": 417
+    },
+    {
+      "epoch": 0.46756152125279643,
+      "grad_norm": 0.33700963854789734,
+      "learning_rate": 4.953732673969219e-06,
+      "loss": 0.3246,
+      "step": 418
+    },
+    {
+      "epoch": 0.46868008948545864,
+      "grad_norm": 0.33790937066078186,
+      "learning_rate": 4.9531072961015155e-06,
+      "loss": 0.3576,
+      "step": 419
+    },
+    {
+      "epoch": 0.4697986577181208,
+      "grad_norm": 0.4202900230884552,
+      "learning_rate": 4.9524777600700964e-06,
+      "loss": 0.3272,
+      "step": 420
+    },
+    {
+      "epoch": 0.470917225950783,
+      "grad_norm": 0.3359755277633667,
+      "learning_rate": 4.951844066942063e-06,
+      "loss": 0.3695,
+      "step": 421
+    },
+    {
+      "epoch": 0.4720357941834452,
+      "grad_norm": 0.37388232350349426,
+      "learning_rate": 4.9512062177915634e-06,
+      "loss": 0.3432,
+      "step": 422
+    },
+    {
+      "epoch": 0.47315436241610737,
+      "grad_norm": 0.30955401062965393,
+      "learning_rate": 4.95056421369979e-06,
+      "loss": 0.3416,
+      "step": 423
+    },
+    {
+      "epoch": 0.4742729306487696,
+      "grad_norm": 0.3202754557132721,
+      "learning_rate": 4.949918055754979e-06,
+      "loss": 0.3013,
+      "step": 424
+    },
+    {
+      "epoch": 0.4753914988814318,
+      "grad_norm": 0.31085923314094543,
+      "learning_rate": 4.9492677450524055e-06,
+      "loss": 0.315,
+      "step": 425
+    },
+    {
+      "epoch": 0.47651006711409394,
+      "grad_norm": 0.3000643849372864,
+      "learning_rate": 4.948613282694384e-06,
+      "loss": 0.3098,
+      "step": 426
+    },
+    {
+      "epoch": 0.47762863534675615,
+      "grad_norm": 0.5085323452949524,
+      "learning_rate": 4.94795466979027e-06,
+      "loss": 0.323,
+      "step": 427
+    },
+    {
+      "epoch": 0.47874720357941836,
+      "grad_norm": 0.32867252826690674,
+      "learning_rate": 4.947291907456451e-06,
+      "loss": 0.3718,
+      "step": 428
+    },
+    {
+      "epoch": 0.4798657718120805,
+      "grad_norm": 0.30551809072494507,
+      "learning_rate": 4.946624996816348e-06,
+      "loss": 0.3889,
+      "step": 429
+    },
+    {
+      "epoch": 0.4809843400447427,
+      "grad_norm": 0.292198121547699,
+      "learning_rate": 4.9459539390004155e-06,
+      "loss": 0.2933,
+      "step": 430
+    },
+    {
+      "epoch": 0.48210290827740493,
+      "grad_norm": 0.3698483109474182,
+      "learning_rate": 4.945278735146136e-06,
+      "loss": 0.3111,
+      "step": 431
+    },
+    {
+      "epoch": 0.48322147651006714,
+      "grad_norm": 0.3626561164855957,
+      "learning_rate": 4.94459938639802e-06,
+      "loss": 0.4021,
+      "step": 432
+    },
+    {
+      "epoch": 0.4843400447427293,
+      "grad_norm": 0.34384214878082275,
+      "learning_rate": 4.943915893907606e-06,
+      "loss": 0.3616,
+      "step": 433
+    },
+    {
+      "epoch": 0.4854586129753915,
+      "grad_norm": 0.36774563789367676,
+      "learning_rate": 4.943228258833453e-06,
+      "loss": 0.4211,
+      "step": 434
+    },
+    {
+      "epoch": 0.4865771812080537,
+      "grad_norm": 0.3687521815299988,
+      "learning_rate": 4.942536482341144e-06,
+      "loss": 0.3537,
+      "step": 435
+    },
+    {
+      "epoch": 0.48769574944071586,
+      "grad_norm": 0.3728216290473938,
+      "learning_rate": 4.941840565603282e-06,
+      "loss": 0.4552,
+      "step": 436
+    },
+    {
+      "epoch": 0.4888143176733781,
+      "grad_norm": 0.31738924980163574,
+      "learning_rate": 4.941140509799487e-06,
+      "loss": 0.3107,
+      "step": 437
+    },
+    {
+      "epoch": 0.4899328859060403,
+      "grad_norm": 0.3757306635379791,
+      "learning_rate": 4.940436316116395e-06,
+      "loss": 0.3345,
+      "step": 438
+    },
+    {
+      "epoch": 0.49105145413870244,
+      "grad_norm": 0.3138746917247772,
+      "learning_rate": 4.939727985747659e-06,
+      "loss": 0.2864,
+      "step": 439
+    },
+    {
+      "epoch": 0.49217002237136465,
+      "grad_norm": 0.33581066131591797,
+      "learning_rate": 4.939015519893938e-06,
+      "loss": 0.3426,
+      "step": 440
+    },
+    {
+      "epoch": 0.49328859060402686,
+      "grad_norm": 0.3957127630710602,
+      "learning_rate": 4.938298919762907e-06,
+      "loss": 0.4665,
+      "step": 441
+    },
+    {
+      "epoch": 0.49440715883668906,
+      "grad_norm": 0.3494655191898346,
+      "learning_rate": 4.9375781865692445e-06,
+      "loss": 0.3754,
+      "step": 442
+    },
+    {
+      "epoch": 0.4955257270693512,
+      "grad_norm": 0.3607597053050995,
+      "learning_rate": 4.936853321534637e-06,
+      "loss": 0.3488,
+      "step": 443
+    },
+    {
+      "epoch": 0.4966442953020134,
+      "grad_norm": 0.290842205286026,
+      "learning_rate": 4.9361243258877745e-06,
+      "loss": 0.2672,
+      "step": 444
+    },
+    {
+      "epoch": 0.49776286353467564,
+      "grad_norm": 0.3222283124923706,
+      "learning_rate": 4.935391200864347e-06,
+      "loss": 0.3058,
+      "step": 445
+    },
+    {
+      "epoch": 0.4988814317673378,
+      "grad_norm": 0.3919990360736847,
+      "learning_rate": 4.934653947707047e-06,
+      "loss": 0.3667,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.319731205701828,
+      "learning_rate": 4.933912567665561e-06,
+      "loss": 0.3336,
+      "step": 447
+    },
+    {
+      "epoch": 0.5011185682326622,
+      "grad_norm": 0.39007750153541565,
+      "learning_rate": 4.9331670619965735e-06,
+      "loss": 0.4083,
+      "step": 448
+    },
+    {
+      "epoch": 0.5022371364653244,
+      "grad_norm": 0.35970282554626465,
+      "learning_rate": 4.932417431963761e-06,
+      "loss": 0.353,
+      "step": 449
+    },
+    {
+      "epoch": 0.5033557046979866,
+      "grad_norm": 0.3694131672382355,
+      "learning_rate": 4.931663678837791e-06,
+      "loss": 0.375,
+      "step": 450
+    },
+    {
+      "epoch": 0.5033557046979866,
+      "eval_loss": 0.34162551164627075,
+      "eval_runtime": 374.9491,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 2.12,
+      "step": 450
+    },
+    {
+      "epoch": 0.5044742729306487,
+      "grad_norm": 0.3365922272205353,
+      "learning_rate": 4.930905803896322e-06,
+      "loss": 0.3321,
+      "step": 451
+    },
+    {
+      "epoch": 0.5055928411633109,
+      "grad_norm": 0.36184102296829224,
+      "learning_rate": 4.930143808423995e-06,
+      "loss": 0.3432,
+      "step": 452
+    },
+    {
+      "epoch": 0.5067114093959731,
+      "grad_norm": 0.3486451208591461,
+      "learning_rate": 4.929377693712439e-06,
+      "loss": 0.3152,
+      "step": 453
+    },
+    {
+      "epoch": 0.5078299776286354,
+      "grad_norm": 0.3770626485347748,
+      "learning_rate": 4.928607461060264e-06,
+      "loss": 0.3313,
+      "step": 454
+    },
+    {
+      "epoch": 0.5089485458612976,
+      "grad_norm": 0.3081826865673065,
+      "learning_rate": 4.927833111773061e-06,
+      "loss": 0.3778,
+      "step": 455
+    },
+    {
+      "epoch": 0.5100671140939598,
+      "grad_norm": 0.324587881565094,
+      "learning_rate": 4.927054647163398e-06,
+      "loss": 0.2801,
+      "step": 456
+    },
+    {
+      "epoch": 0.5111856823266219,
+      "grad_norm": 0.386663556098938,
+      "learning_rate": 4.926272068550819e-06,
+      "loss": 0.3481,
+      "step": 457
+    },
+    {
+      "epoch": 0.5123042505592841,
+      "grad_norm": 0.3981550335884094,
+      "learning_rate": 4.925485377261842e-06,
+      "loss": 0.3421,
+      "step": 458
+    },
+    {
+      "epoch": 0.5134228187919463,
+      "grad_norm": 0.41805386543273926,
+      "learning_rate": 4.924694574629955e-06,
+      "loss": 0.3654,
+      "step": 459
+    },
+    {
+      "epoch": 0.5145413870246085,
+      "grad_norm": 0.32252582907676697,
+      "learning_rate": 4.923899661995617e-06,
+      "loss": 0.2812,
+      "step": 460
+    },
+    {
+      "epoch": 0.5156599552572707,
+      "grad_norm": 0.43598851561546326,
+      "learning_rate": 4.923100640706251e-06,
+      "loss": 0.2856,
+      "step": 461
+    },
+    {
+      "epoch": 0.5167785234899329,
+      "grad_norm": 0.36268118023872375,
+      "learning_rate": 4.922297512116248e-06,
+      "loss": 0.3757,
+      "step": 462
+    },
+    {
+      "epoch": 0.5178970917225951,
+      "grad_norm": 0.30699658393859863,
+      "learning_rate": 4.921490277586957e-06,
+      "loss": 0.3104,
+      "step": 463
+    },
+    {
+      "epoch": 0.5190156599552572,
+      "grad_norm": 0.4442712664604187,
+      "learning_rate": 4.9206789384866905e-06,
+      "loss": 0.3255,
+      "step": 464
+    },
+    {
+      "epoch": 0.5201342281879194,
+      "grad_norm": 0.37133488059043884,
+      "learning_rate": 4.919863496190715e-06,
+      "loss": 0.3589,
+      "step": 465
+    },
+    {
+      "epoch": 0.5212527964205816,
+      "grad_norm": 0.36069542169570923,
+      "learning_rate": 4.919043952081256e-06,
+      "loss": 0.3172,
+      "step": 466
+    },
+    {
+      "epoch": 0.5223713646532439,
+      "grad_norm": 0.3288801610469818,
+      "learning_rate": 4.918220307547488e-06,
+      "loss": 0.2677,
+      "step": 467
+    },
+    {
+      "epoch": 0.5234899328859061,
+      "grad_norm": 0.364067018032074,
+      "learning_rate": 4.917392563985539e-06,
+      "loss": 0.3426,
+      "step": 468
+    },
+    {
+      "epoch": 0.5246085011185683,
+      "grad_norm": 0.35532283782958984,
+      "learning_rate": 4.916560722798484e-06,
+      "loss": 0.3129,
+      "step": 469
+    },
+    {
+      "epoch": 0.5257270693512305,
+      "grad_norm": 0.38478919863700867,
+      "learning_rate": 4.915724785396343e-06,
+      "loss": 0.3126,
+      "step": 470
+    },
+    {
+      "epoch": 0.5268456375838926,
+      "grad_norm": 0.39041411876678467,
+      "learning_rate": 4.91488475319608e-06,
+      "loss": 0.3825,
+      "step": 471
+    },
+    {
+      "epoch": 0.5279642058165548,
+      "grad_norm": 0.36006051301956177,
+      "learning_rate": 4.914040627621599e-06,
+      "loss": 0.2781,
+      "step": 472
+    },
+    {
+      "epoch": 0.529082774049217,
+      "grad_norm": 0.38814491033554077,
+      "learning_rate": 4.9131924101037455e-06,
+      "loss": 0.3888,
+      "step": 473
+    },
+    {
+      "epoch": 0.5302013422818792,
+      "grad_norm": 0.3350260555744171,
+      "learning_rate": 4.912340102080297e-06,
+      "loss": 0.263,
+      "step": 474
+    },
+    {
+      "epoch": 0.5313199105145414,
+      "grad_norm": 0.39259982109069824,
+      "learning_rate": 4.911483704995968e-06,
+      "loss": 0.3587,
+      "step": 475
+    },
+    {
+      "epoch": 0.5324384787472036,
+      "grad_norm": 0.39165207743644714,
+      "learning_rate": 4.910623220302402e-06,
+      "loss": 0.4614,
+      "step": 476
+    },
+    {
+      "epoch": 0.5335570469798657,
+      "grad_norm": 0.38493409752845764,
+      "learning_rate": 4.909758649458172e-06,
+      "loss": 0.3791,
+      "step": 477
+    },
+    {
+      "epoch": 0.5346756152125279,
+      "grad_norm": 0.41047391295433044,
+      "learning_rate": 4.9088899939287785e-06,
+      "loss": 0.4756,
+      "step": 478
+    },
+    {
+      "epoch": 0.5357941834451901,
+      "grad_norm": 0.38805124163627625,
+      "learning_rate": 4.908017255186644e-06,
+      "loss": 0.424,
+      "step": 479
+    },
+    {
+      "epoch": 0.5369127516778524,
+      "grad_norm": 0.4114658236503601,
+      "learning_rate": 4.90714043471111e-06,
+      "loss": 0.4448,
+      "step": 480
+    },
+    {
+      "epoch": 0.5380313199105146,
+      "grad_norm": 0.35568225383758545,
+      "learning_rate": 4.9062595339884455e-06,
+      "loss": 0.3168,
+      "step": 481
+    },
+    {
+      "epoch": 0.5391498881431768,
+      "grad_norm": 0.3408364951610565,
+      "learning_rate": 4.905374554511825e-06,
+      "loss": 0.2688,
+      "step": 482
+    },
+    {
+      "epoch": 0.540268456375839,
+      "grad_norm": 0.3925073742866516,
+      "learning_rate": 4.904485497781345e-06,
+      "loss": 0.3761,
+      "step": 483
+    },
+    {
+      "epoch": 0.5413870246085011,
+      "grad_norm": 0.4615781307220459,
+      "learning_rate": 4.903592365304007e-06,
+      "loss": 0.2519,
+      "step": 484
+    },
+    {
+      "epoch": 0.5425055928411633,
+      "grad_norm": 0.3747728765010834,
+      "learning_rate": 4.902695158593727e-06,
+      "loss": 0.258,
+      "step": 485
+    },
+    {
+      "epoch": 0.5436241610738255,
+      "grad_norm": 0.43629682064056396,
+      "learning_rate": 4.901793879171321e-06,
+      "loss": 0.3415,
+      "step": 486
+    },
+    {
+      "epoch": 0.5447427293064877,
+      "grad_norm": 0.370054692029953,
+      "learning_rate": 4.900888528564513e-06,
+      "loss": 0.2745,
+      "step": 487
+    },
+    {
+      "epoch": 0.5458612975391499,
+      "grad_norm": 0.3784375786781311,
+      "learning_rate": 4.8999791083079275e-06,
+      "loss": 0.3461,
+      "step": 488
+    },
+    {
+      "epoch": 0.5469798657718121,
+      "grad_norm": 0.37398561835289,
+      "learning_rate": 4.899065619943085e-06,
+      "loss": 0.3678,
+      "step": 489
+    },
+    {
+      "epoch": 0.5480984340044742,
+      "grad_norm": 0.410810649394989,
+      "learning_rate": 4.898148065018403e-06,
+      "loss": 0.3748,
+      "step": 490
+    },
+    {
+      "epoch": 0.5492170022371364,
+      "grad_norm": 0.3225620985031128,
+      "learning_rate": 4.897226445089193e-06,
+      "loss": 0.3077,
+      "step": 491
+    },
+    {
+      "epoch": 0.5503355704697986,
+      "grad_norm": 0.3191516399383545,
+      "learning_rate": 4.896300761717654e-06,
+      "loss": 0.2825,
+      "step": 492
+    },
+    {
+      "epoch": 0.5514541387024608,
+      "grad_norm": 0.3571834862232208,
+      "learning_rate": 4.895371016472877e-06,
+      "loss": 0.2557,
+      "step": 493
+    },
+    {
+      "epoch": 0.5525727069351231,
+      "grad_norm": 0.3537055253982544,
+      "learning_rate": 4.894437210930834e-06,
+      "loss": 0.262,
+      "step": 494
+    },
+    {
+      "epoch": 0.5536912751677853,
+      "grad_norm": 0.35530024766921997,
+      "learning_rate": 4.893499346674383e-06,
+      "loss": 0.2748,
+      "step": 495
+    },
+    {
+      "epoch": 0.5548098434004475,
+      "grad_norm": 0.33276456594467163,
+      "learning_rate": 4.892557425293259e-06,
+      "loss": 0.2717,
+      "step": 496
+    },
+    {
+      "epoch": 0.5559284116331096,
+      "grad_norm": 0.3380317687988281,
+      "learning_rate": 4.891611448384075e-06,
+      "loss": 0.3635,
+      "step": 497
+    },
+    {
+      "epoch": 0.5570469798657718,
+      "grad_norm": 0.35890451073646545,
+      "learning_rate": 4.8906614175503196e-06,
+      "loss": 0.2696,
+      "step": 498
+    },
+    {
+      "epoch": 0.558165548098434,
+      "grad_norm": 0.34443095326423645,
+      "learning_rate": 4.8897073344023505e-06,
+      "loss": 0.2892,
+      "step": 499
+    },
+    {
+      "epoch": 0.5592841163310962,
+      "grad_norm": 0.3407563269138336,
+      "learning_rate": 4.888749200557396e-06,
+      "loss": 0.3215,
+      "step": 500
+    },
+    {
+      "epoch": 0.5592841163310962,
+      "eval_loss": 0.3103519380092621,
+      "eval_runtime": 375.5247,
+      "eval_samples_per_second": 2.117,
+      "eval_steps_per_second": 2.117,
+      "step": 500
+    },
+    {
+      "epoch": 0.5604026845637584,
+      "grad_norm": 0.37557560205459595,
+      "learning_rate": 4.887787017639551e-06,
+      "loss": 0.2832,
+      "step": 501
+    },
+    {
+      "epoch": 0.5615212527964206,
+      "grad_norm": 0.3939456045627594,
+      "learning_rate": 4.886820787279773e-06,
+      "loss": 0.3601,
+      "step": 502
+    },
+    {
+      "epoch": 0.5626398210290827,
+      "grad_norm": 0.3418801426887512,
+      "learning_rate": 4.885850511115881e-06,
+      "loss": 0.2868,
+      "step": 503
+    },
+    {
+      "epoch": 0.5637583892617449,
+      "grad_norm": 0.33409515023231506,
+      "learning_rate": 4.884876190792548e-06,
+      "loss": 0.2853,
+      "step": 504
+    },
+    {
+      "epoch": 0.5648769574944071,
+      "grad_norm": 0.4141145646572113,
+      "learning_rate": 4.883897827961309e-06,
+      "loss": 0.3075,
+      "step": 505
+    },
+    {
+      "epoch": 0.5659955257270693,
+      "grad_norm": 0.3346996307373047,
+      "learning_rate": 4.882915424280546e-06,
+      "loss": 0.2845,
+      "step": 506
+    },
+    {
+      "epoch": 0.5671140939597316,
+      "grad_norm": 0.3863886296749115,
+      "learning_rate": 4.881928981415493e-06,
+      "loss": 0.2922,
+      "step": 507
+    },
+    {
+      "epoch": 0.5682326621923938,
+      "grad_norm": 0.3816783130168915,
+      "learning_rate": 4.880938501038228e-06,
+      "loss": 0.4305,
+      "step": 508
+    },
+    {
+      "epoch": 0.569351230425056,
+      "grad_norm": 0.3404105007648468,
+      "learning_rate": 4.8799439848276765e-06,
+      "loss": 0.2579,
+      "step": 509
+    },
+    {
+      "epoch": 0.5704697986577181,
+      "grad_norm": 0.36595046520233154,
+      "learning_rate": 4.8789454344696e-06,
+      "loss": 0.3559,
+      "step": 510
+    },
+    {
+      "epoch": 0.5715883668903803,
+      "grad_norm": 0.410990834236145,
+      "learning_rate": 4.8779428516566056e-06,
+      "loss": 0.3898,
+      "step": 511
+    },
+    {
+      "epoch": 0.5727069351230425,
+      "grad_norm": 0.31936854124069214,
+      "learning_rate": 4.876936238088127e-06,
+      "loss": 0.3242,
+      "step": 512
+    },
+    {
+      "epoch": 0.5738255033557047,
+      "grad_norm": 0.37220728397369385,
+      "learning_rate": 4.875925595470436e-06,
+      "loss": 0.2998,
+      "step": 513
+    },
+    {
+      "epoch": 0.5749440715883669,
+      "grad_norm": 0.6426952481269836,
+      "learning_rate": 4.874910925516634e-06,
+      "loss": 0.2962,
+      "step": 514
+    },
+    {
+      "epoch": 0.5760626398210291,
+      "grad_norm": 0.32255858182907104,
+      "learning_rate": 4.8738922299466446e-06,
+      "loss": 0.3576,
+      "step": 515
+    },
+    {
+      "epoch": 0.5771812080536913,
+      "grad_norm": 0.3229198455810547,
+      "learning_rate": 4.872869510487219e-06,
+      "loss": 0.2526,
+      "step": 516
+    },
+    {
+      "epoch": 0.5782997762863534,
+      "grad_norm": 0.40679529309272766,
+      "learning_rate": 4.871842768871928e-06,
+      "loss": 0.2371,
+      "step": 517
+    },
+    {
+      "epoch": 0.5794183445190156,
+      "grad_norm": 0.35106325149536133,
+      "learning_rate": 4.87081200684116e-06,
+      "loss": 0.2813,
+      "step": 518
+    },
+    {
+      "epoch": 0.5805369127516778,
+      "grad_norm": 0.33908048272132874,
+      "learning_rate": 4.869777226142118e-06,
+      "loss": 0.2526,
+      "step": 519
+    },
+    {
+      "epoch": 0.5816554809843401,
+      "grad_norm": 0.32178765535354614,
+      "learning_rate": 4.8687384285288185e-06,
+      "loss": 0.2872,
+      "step": 520
+    },
+    {
+      "epoch": 0.5827740492170023,
+      "grad_norm": 0.3607175350189209,
+      "learning_rate": 4.867695615762084e-06,
+      "loss": 0.3648,
+      "step": 521
+    },
+    {
+      "epoch": 0.5838926174496645,
+      "grad_norm": 0.3652980625629425,
+      "learning_rate": 4.866648789609545e-06,
+      "loss": 0.3449,
+      "step": 522
+    },
+    {
+      "epoch": 0.5850111856823266,
+      "grad_norm": 0.7061839699745178,
+      "learning_rate": 4.865597951845636e-06,
+      "loss": 0.2983,
+      "step": 523
+    },
+    {
+      "epoch": 0.5861297539149888,
+      "grad_norm": 0.3158840239048004,
+      "learning_rate": 4.864543104251587e-06,
+      "loss": 0.3356,
+      "step": 524
+    },
+    {
+      "epoch": 0.587248322147651,
+      "grad_norm": 0.3423825204372406,
+      "learning_rate": 4.86348424861543e-06,
+      "loss": 0.2918,
+      "step": 525
+    },
+    {
+      "epoch": 0.5883668903803132,
+      "grad_norm": 0.3407121002674103,
+      "learning_rate": 4.862421386731989e-06,
+      "loss": 0.3427,
+      "step": 526
+    },
+    {
+      "epoch": 0.5894854586129754,
+      "grad_norm": 0.36023014783859253,
+      "learning_rate": 4.861354520402877e-06,
+      "loss": 0.3002,
+      "step": 527
+    },
+    {
+      "epoch": 0.5906040268456376,
+      "grad_norm": 0.30209678411483765,
+      "learning_rate": 4.860283651436496e-06,
+      "loss": 0.2951,
+      "step": 528
+    },
+    {
+      "epoch": 0.5917225950782998,
+      "grad_norm": 0.3096085786819458,
+      "learning_rate": 4.859208781648036e-06,
+      "loss": 0.2747,
+      "step": 529
+    },
+    {
+      "epoch": 0.5928411633109619,
+      "grad_norm": 0.353807270526886,
+      "learning_rate": 4.8581299128594625e-06,
+      "loss": 0.2487,
+      "step": 530
+    },
+    {
+      "epoch": 0.5939597315436241,
+      "grad_norm": 0.3648567199707031,
+      "learning_rate": 4.857047046899524e-06,
+      "loss": 0.3151,
+      "step": 531
+    },
+    {
+      "epoch": 0.5950782997762863,
+      "grad_norm": 0.3880610167980194,
+      "learning_rate": 4.855960185603742e-06,
+      "loss": 0.3301,
+      "step": 532
+    },
+    {
+      "epoch": 0.5961968680089486,
+      "grad_norm": 0.36094367504119873,
+      "learning_rate": 4.854869330814413e-06,
+      "loss": 0.2662,
+      "step": 533
+    },
+    {
+      "epoch": 0.5973154362416108,
+      "grad_norm": 0.32027992606163025,
+      "learning_rate": 4.853774484380601e-06,
+      "loss": 0.2479,
+      "step": 534
+    },
+    {
+      "epoch": 0.598434004474273,
+      "grad_norm": 0.293278306722641,
+      "learning_rate": 4.852675648158136e-06,
+      "loss": 0.229,
+      "step": 535
+    },
+    {
+      "epoch": 0.5995525727069351,
+      "grad_norm": 0.3266526460647583,
+      "learning_rate": 4.85157282400961e-06,
+      "loss": 0.2592,
+      "step": 536
+    },
+    {
+      "epoch": 0.6006711409395973,
+      "grad_norm": 0.3274593949317932,
+      "learning_rate": 4.850466013804377e-06,
+      "loss": 0.2634,
+      "step": 537
+    },
+    {
+      "epoch": 0.6017897091722595,
+      "grad_norm": 0.35161513090133667,
+      "learning_rate": 4.849355219418547e-06,
+      "loss": 0.2913,
+      "step": 538
+    },
+    {
+      "epoch": 0.6029082774049217,
+      "grad_norm": 0.3486526906490326,
+      "learning_rate": 4.848240442734983e-06,
+      "loss": 0.2959,
+      "step": 539
+    },
+    {
+      "epoch": 0.6040268456375839,
+      "grad_norm": 0.39304664731025696,
+      "learning_rate": 4.8471216856432965e-06,
+      "loss": 0.3706,
+      "step": 540
+    },
+    {
+      "epoch": 0.6051454138702461,
+      "grad_norm": 0.31824731826782227,
+      "learning_rate": 4.84599895003985e-06,
+      "loss": 0.2644,
+      "step": 541
+    },
+    {
+      "epoch": 0.6062639821029083,
+      "grad_norm": 0.34036195278167725,
+      "learning_rate": 4.844872237827745e-06,
+      "loss": 0.3077,
+      "step": 542
+    },
+    {
+      "epoch": 0.6073825503355704,
+      "grad_norm": 0.3363955020904541,
+      "learning_rate": 4.8437415509168285e-06,
+      "loss": 0.3036,
+      "step": 543
+    },
+    {
+      "epoch": 0.6085011185682326,
+      "grad_norm": 0.3290499448776245,
+      "learning_rate": 4.842606891223682e-06,
+      "loss": 0.2637,
+      "step": 544
+    },
+    {
+      "epoch": 0.6096196868008948,
+      "grad_norm": 0.37139827013015747,
+      "learning_rate": 4.84146826067162e-06,
+      "loss": 0.385,
+      "step": 545
+    },
+    {
+      "epoch": 0.610738255033557,
+      "grad_norm": 0.31628963351249695,
+      "learning_rate": 4.8403256611906915e-06,
+      "loss": 0.2385,
+      "step": 546
+    },
+    {
+      "epoch": 0.6118568232662193,
+      "grad_norm": 0.3218952715396881,
+      "learning_rate": 4.83917909471767e-06,
+      "loss": 0.2412,
+      "step": 547
+    },
+    {
+      "epoch": 0.6129753914988815,
+      "grad_norm": 0.32762303948402405,
+      "learning_rate": 4.838028563196055e-06,
+      "loss": 0.2969,
+      "step": 548
+    },
+    {
+      "epoch": 0.6140939597315436,
+      "grad_norm": 0.3849232494831085,
+      "learning_rate": 4.836874068576065e-06,
+      "loss": 0.4314,
+      "step": 549
+    },
+    {
+      "epoch": 0.6152125279642058,
+      "grad_norm": 0.38277336955070496,
+      "learning_rate": 4.835715612814639e-06,
+      "loss": 0.379,
+      "step": 550
+    },
+    {
+      "epoch": 0.6152125279642058,
+      "eval_loss": 0.3042142987251282,
+      "eval_runtime": 375.2691,
+      "eval_samples_per_second": 2.118,
+      "eval_steps_per_second": 2.118,
+      "step": 550
+    },
+    {
+      "epoch": 0.616331096196868,
+      "grad_norm": 0.4057752788066864,
+      "learning_rate": 4.834553197875427e-06,
+      "loss": 0.2942,
+      "step": 551
+    },
+    {
+      "epoch": 0.6174496644295302,
+      "grad_norm": 0.3637617528438568,
+      "learning_rate": 4.833386825728794e-06,
+      "loss": 0.3591,
+      "step": 552
+    },
+    {
+      "epoch": 0.6185682326621924,
+      "grad_norm": 0.39566802978515625,
+      "learning_rate": 4.83221649835181e-06,
+      "loss": 0.2746,
+      "step": 553
+    },
+    {
+      "epoch": 0.6196868008948546,
+      "grad_norm": 0.40520596504211426,
+      "learning_rate": 4.831042217728248e-06,
+      "loss": 0.3298,
+      "step": 554
+    },
+    {
+      "epoch": 0.6208053691275168,
+      "grad_norm": 0.3547848165035248,
+      "learning_rate": 4.829863985848587e-06,
+      "loss": 0.2757,
+      "step": 555
+    },
+    {
+      "epoch": 0.6219239373601789,
+      "grad_norm": 0.3893035352230072,
+      "learning_rate": 4.828681804709998e-06,
+      "loss": 0.2917,
+      "step": 556
+    },
+    {
+      "epoch": 0.6230425055928411,
+      "grad_norm": 0.7000828981399536,
+      "learning_rate": 4.8274956763163515e-06,
+      "loss": 0.3711,
+      "step": 557
+    },
+    {
+      "epoch": 0.6241610738255033,
+      "grad_norm": 0.4453166127204895,
+      "learning_rate": 4.826305602678204e-06,
+      "loss": 0.2663,
+      "step": 558
+    },
+    {
+      "epoch": 0.6252796420581656,
+      "grad_norm": 0.7207187414169312,
+      "learning_rate": 4.8251115858128016e-06,
+      "loss": 0.2985,
+      "step": 559
+    },
+    {
+      "epoch": 0.6263982102908278,
+      "grad_norm": 0.39802902936935425,
+      "learning_rate": 4.8239136277440745e-06,
+      "loss": 0.3771,
+      "step": 560
+    },
+    {
+      "epoch": 0.62751677852349,
+      "grad_norm": 0.36102500557899475,
+      "learning_rate": 4.822711730502634e-06,
+      "loss": 0.3538,
+      "step": 561
+    },
+    {
+      "epoch": 0.6286353467561522,
+      "grad_norm": 0.4261714518070221,
+      "learning_rate": 4.821505896125769e-06,
+      "loss": 0.4727,
+      "step": 562
+    },
+    {
+      "epoch": 0.6297539149888143,
+      "grad_norm": 0.38941535353660583,
+      "learning_rate": 4.820296126657438e-06,
+      "loss": 0.384,
+      "step": 563
+    },
+    {
+      "epoch": 0.6308724832214765,
+      "grad_norm": 0.3647603392601013,
+      "learning_rate": 4.819082424148275e-06,
+      "loss": 0.3239,
+      "step": 564
+    },
+    {
+      "epoch": 0.6319910514541387,
+      "grad_norm": 0.3924996554851532,
+      "learning_rate": 4.817864790655577e-06,
+      "loss": 0.272,
+      "step": 565
+    },
+    {
+      "epoch": 0.6331096196868009,
+      "grad_norm": 0.4429261386394501,
+      "learning_rate": 4.816643228243307e-06,
+      "loss": 0.3123,
+      "step": 566
+    },
+    {
+      "epoch": 0.6342281879194631,
+      "grad_norm": 0.39329850673675537,
+      "learning_rate": 4.815417738982085e-06,
+      "loss": 0.3435,
+      "step": 567
+    },
+    {
+      "epoch": 0.6353467561521253,
+      "grad_norm": 0.3604203164577484,
+      "learning_rate": 4.814188324949189e-06,
+      "loss": 0.3506,
+      "step": 568
+    },
+    {
+      "epoch": 0.6364653243847874,
+      "grad_norm": 0.3354223668575287,
+      "learning_rate": 4.812954988228551e-06,
+      "loss": 0.2482,
+      "step": 569
+    },
+    {
+      "epoch": 0.6375838926174496,
+      "grad_norm": 0.3591589033603668,
+      "learning_rate": 4.811717730910749e-06,
+      "loss": 0.3461,
+      "step": 570
+    },
+    {
+      "epoch": 0.6387024608501118,
+      "grad_norm": 0.41761314868927,
+      "learning_rate": 4.810476555093007e-06,
+      "loss": 0.4136,
+      "step": 571
+    },
+    {
+      "epoch": 0.639821029082774,
+      "grad_norm": 0.3376199007034302,
+      "learning_rate": 4.809231462879196e-06,
+      "loss": 0.2805,
+      "step": 572
+    },
+    {
+      "epoch": 0.6409395973154363,
+      "grad_norm": 0.3496900200843811,
+      "learning_rate": 4.807982456379819e-06,
+      "loss": 0.3958,
+      "step": 573
+    },
+    {
+      "epoch": 0.6420581655480985,
+      "grad_norm": 0.3497297167778015,
+      "learning_rate": 4.806729537712017e-06,
+      "loss": 0.2408,
+      "step": 574
+    },
+    {
+      "epoch": 0.6431767337807607,
+      "grad_norm": 0.34045857191085815,
+      "learning_rate": 4.805472708999563e-06,
+      "loss": 0.2552,
+      "step": 575
+    },
+    {
+      "epoch": 0.6442953020134228,
+      "grad_norm": 0.3492688536643982,
+      "learning_rate": 4.804211972372857e-06,
+      "loss": 0.2875,
+      "step": 576
+    },
+    {
+      "epoch": 0.645413870246085,
+      "grad_norm": 0.34412750601768494,
+      "learning_rate": 4.802947329968921e-06,
+      "loss": 0.2865,
+      "step": 577
+    },
+    {
+      "epoch": 0.6465324384787472,
+      "grad_norm": 0.33287253975868225,
+      "learning_rate": 4.8016787839314024e-06,
+      "loss": 0.2862,
+      "step": 578
+    },
+    {
+      "epoch": 0.6476510067114094,
+      "grad_norm": 0.3494202494621277,
+      "learning_rate": 4.800406336410561e-06,
+      "loss": 0.2793,
+      "step": 579
+    },
+    {
+      "epoch": 0.6487695749440716,
+      "grad_norm": 0.4269697368144989,
+      "learning_rate": 4.7991299895632724e-06,
+      "loss": 0.3033,
+      "step": 580
+    },
+    {
+      "epoch": 0.6498881431767338,
+      "grad_norm": 0.4095929265022278,
+      "learning_rate": 4.797849745553019e-06,
+      "loss": 0.2803,
+      "step": 581
+    },
+    {
+      "epoch": 0.6510067114093959,
+      "grad_norm": 0.37961679697036743,
+      "learning_rate": 4.7965656065498945e-06,
+      "loss": 0.2723,
+      "step": 582
+    },
+    {
+      "epoch": 0.6521252796420581,
+      "grad_norm": 0.35413551330566406,
+      "learning_rate": 4.795277574730588e-06,
+      "loss": 0.2267,
+      "step": 583
+    },
+    {
+      "epoch": 0.6532438478747203,
+      "grad_norm": 0.3585027754306793,
+      "learning_rate": 4.793985652278392e-06,
+      "loss": 0.3559,
+      "step": 584
+    },
+    {
+      "epoch": 0.6543624161073825,
+      "grad_norm": 0.33826395869255066,
+      "learning_rate": 4.792689841383193e-06,
+      "loss": 0.2521,
+      "step": 585
+    },
+    {
+      "epoch": 0.6554809843400448,
+      "grad_norm": 0.3641104996204376,
+      "learning_rate": 4.7913901442414665e-06,
+      "loss": 0.4002,
+      "step": 586
+    },
+    {
+      "epoch": 0.656599552572707,
+      "grad_norm": 0.38858771324157715,
+      "learning_rate": 4.790086563056278e-06,
+      "loss": 0.2982,
+      "step": 587
+    },
+    {
+      "epoch": 0.6577181208053692,
+      "grad_norm": 0.30337172746658325,
+      "learning_rate": 4.788779100037276e-06,
+      "loss": 0.2777,
+      "step": 588
+    },
+    {
+      "epoch": 0.6588366890380313,
+      "grad_norm": 0.33638280630111694,
+      "learning_rate": 4.787467757400687e-06,
+      "loss": 0.2822,
+      "step": 589
+    },
+    {
+      "epoch": 0.6599552572706935,
+      "grad_norm": 0.4196690022945404,
+      "learning_rate": 4.7861525373693175e-06,
+      "loss": 0.3408,
+      "step": 590
+    },
+    {
+      "epoch": 0.6610738255033557,
+      "grad_norm": 0.33803901076316833,
+      "learning_rate": 4.784833442172543e-06,
+      "loss": 0.3056,
+      "step": 591
+    },
+    {
+      "epoch": 0.6621923937360179,
+      "grad_norm": 0.3208058774471283,
+      "learning_rate": 4.783510474046308e-06,
+      "loss": 0.3096,
+      "step": 592
+    },
+    {
+      "epoch": 0.6633109619686801,
+      "grad_norm": 0.36712682247161865,
+      "learning_rate": 4.7821836352331235e-06,
+      "loss": 0.3167,
+      "step": 593
+    },
+    {
+      "epoch": 0.6644295302013423,
+      "grad_norm": 0.3843868672847748,
+      "learning_rate": 4.780852927982061e-06,
+      "loss": 0.269,
+      "step": 594
+    },
+    {
+      "epoch": 0.6655480984340044,
+      "grad_norm": 0.4317105710506439,
+      "learning_rate": 4.779518354548749e-06,
+      "loss": 0.3149,
+      "step": 595
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.3311421871185303,
+      "learning_rate": 4.778179917195368e-06,
+      "loss": 0.2439,
+      "step": 596
+    },
+    {
+      "epoch": 0.6677852348993288,
+      "grad_norm": 0.3715791702270508,
+      "learning_rate": 4.776837618190651e-06,
+      "loss": 0.3331,
+      "step": 597
+    },
+    {
+      "epoch": 0.668903803131991,
+      "grad_norm": 0.37263649702072144,
+      "learning_rate": 4.775491459809874e-06,
+      "loss": 0.4465,
+      "step": 598
+    },
+    {
+      "epoch": 0.6700223713646533,
+      "grad_norm": 0.4538126289844513,
+      "learning_rate": 4.774141444334858e-06,
+      "loss": 0.3255,
+      "step": 599
+    },
+    {
+      "epoch": 0.6711409395973155,
+      "grad_norm": 0.37712544202804565,
+      "learning_rate": 4.772787574053957e-06,
+      "loss": 0.3447,
+      "step": 600
+    },
+    {
+      "epoch": 0.6711409395973155,
+      "eval_loss": 0.3000986874103546,
+      "eval_runtime": 375.749,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 2.116,
+      "step": 600
+    },
+    {
+      "epoch": 0.6722595078299777,
+      "grad_norm": 0.36250224709510803,
+      "learning_rate": 4.771429851262065e-06,
+      "loss": 0.3003,
+      "step": 601
+    },
+    {
+      "epoch": 0.6733780760626398,
+      "grad_norm": 0.4212695360183716,
+      "learning_rate": 4.770068278260601e-06,
+      "loss": 0.3036,
+      "step": 602
+    },
+    {
+      "epoch": 0.674496644295302,
+      "grad_norm": 0.41591283679008484,
+      "learning_rate": 4.768702857357516e-06,
+      "loss": 0.2781,
+      "step": 603
+    },
+    {
+      "epoch": 0.6756152125279642,
+      "grad_norm": 0.3553246855735779,
+      "learning_rate": 4.767333590867277e-06,
+      "loss": 0.261,
+      "step": 604
+    },
+    {
+      "epoch": 0.6767337807606264,
+      "grad_norm": 0.3625635802745819,
+      "learning_rate": 4.765960481110874e-06,
+      "loss": 0.3651,
+      "step": 605
+    },
+    {
+      "epoch": 0.6778523489932886,
+      "grad_norm": 0.3491491377353668,
+      "learning_rate": 4.76458353041581e-06,
+      "loss": 0.377,
+      "step": 606
+    },
+    {
+      "epoch": 0.6789709172259508,
+      "grad_norm": 0.42099276185035706,
+      "learning_rate": 4.763202741116099e-06,
+      "loss": 0.3013,
+      "step": 607
+    },
+    {
+      "epoch": 0.680089485458613,
+      "grad_norm": 0.5076472759246826,
+      "learning_rate": 4.761818115552262e-06,
+      "loss": 0.3929,
+      "step": 608
+    },
+    {
+      "epoch": 0.6812080536912751,
+      "grad_norm": 0.3877736032009125,
+      "learning_rate": 4.76042965607132e-06,
+      "loss": 0.2744,
+      "step": 609
+    },
+    {
+      "epoch": 0.6823266219239373,
+      "grad_norm": 0.34696388244628906,
+      "learning_rate": 4.759037365026797e-06,
+      "loss": 0.332,
+      "step": 610
+    },
+    {
+      "epoch": 0.6834451901565995,
+      "grad_norm": 0.3803223967552185,
+      "learning_rate": 4.7576412447787105e-06,
+      "loss": 0.282,
+      "step": 611
+    },
+    {
+      "epoch": 0.6845637583892618,
+      "grad_norm": 0.36747977137565613,
+      "learning_rate": 4.7562412976935655e-06,
+      "loss": 0.253,
+      "step": 612
+    },
+    {
+      "epoch": 0.685682326621924,
+      "grad_norm": 0.43595513701438904,
+      "learning_rate": 4.754837526144359e-06,
+      "loss": 0.2988,
+      "step": 613
+    },
+    {
+      "epoch": 0.6868008948545862,
+      "grad_norm": 0.36716851592063904,
+      "learning_rate": 4.753429932510564e-06,
+      "loss": 0.2794,
+      "step": 614
+    },
+    {
+      "epoch": 0.6879194630872483,
+      "grad_norm": 0.37363508343696594,
+      "learning_rate": 4.752018519178139e-06,
+      "loss": 0.355,
+      "step": 615
+    },
+    {
+      "epoch": 0.6890380313199105,
+      "grad_norm": 0.40918856859207153,
+      "learning_rate": 4.750603288539513e-06,
+      "loss": 0.2933,
+      "step": 616
+    },
+    {
+      "epoch": 0.6901565995525727,
+      "grad_norm": 0.3837878108024597,
+      "learning_rate": 4.749184242993588e-06,
+      "loss": 0.3301,
+      "step": 617
+    },
+    {
+      "epoch": 0.6912751677852349,
+      "grad_norm": 0.3654283285140991,
+      "learning_rate": 4.747761384945729e-06,
+      "loss": 0.3006,
+      "step": 618
+    },
+    {
+      "epoch": 0.6923937360178971,
+      "grad_norm": 0.40632256865501404,
+      "learning_rate": 4.746334716807765e-06,
+      "loss": 0.3027,
+      "step": 619
+    },
+    {
+      "epoch": 0.6935123042505593,
+      "grad_norm": 0.3480345904827118,
+      "learning_rate": 4.744904240997987e-06,
+      "loss": 0.2879,
+      "step": 620
+    },
+    {
+      "epoch": 0.6946308724832215,
+      "grad_norm": 0.32933899760246277,
+      "learning_rate": 4.743469959941136e-06,
+      "loss": 0.2497,
+      "step": 621
+    },
+    {
+      "epoch": 0.6957494407158836,
+      "grad_norm": 0.35397520661354065,
+      "learning_rate": 4.742031876068403e-06,
+      "loss": 0.3149,
+      "step": 622
+    },
+    {
+      "epoch": 0.6968680089485458,
+      "grad_norm": 0.33340832591056824,
+      "learning_rate": 4.740589991817427e-06,
+      "loss": 0.3438,
+      "step": 623
+    },
+    {
+      "epoch": 0.697986577181208,
+      "grad_norm": 0.39394107460975647,
+      "learning_rate": 4.739144309632289e-06,
+      "loss": 0.2833,
+      "step": 624
+    },
+    {
+      "epoch": 0.6991051454138703,
+      "grad_norm": 0.4287222921848297,
+      "learning_rate": 4.737694831963505e-06,
+      "loss": 0.2896,
+      "step": 625
+    },
+    {
+      "epoch": 0.7002237136465325,
+      "grad_norm": 0.35358521342277527,
+      "learning_rate": 4.7362415612680274e-06,
+      "loss": 0.3168,
+      "step": 626
+    },
+    {
+      "epoch": 0.7013422818791947,
+      "grad_norm": 0.3645099699497223,
+      "learning_rate": 4.734784500009238e-06,
+      "loss": 0.2538,
+      "step": 627
+    },
+    {
+      "epoch": 0.7024608501118568,
+      "grad_norm": 0.31505364179611206,
+      "learning_rate": 4.73332365065694e-06,
+      "loss": 0.277,
+      "step": 628
+    },
+    {
+      "epoch": 0.703579418344519,
+      "grad_norm": 0.3579901158809662,
+      "learning_rate": 4.731859015687364e-06,
+      "loss": 0.2942,
+      "step": 629
+    },
+    {
+      "epoch": 0.7046979865771812,
+      "grad_norm": 0.3697115182876587,
+      "learning_rate": 4.730390597583152e-06,
+      "loss": 0.3501,
+      "step": 630
+    },
+    {
+      "epoch": 0.7058165548098434,
+      "grad_norm": 0.34573182463645935,
+      "learning_rate": 4.728918398833361e-06,
+      "loss": 0.3684,
+      "step": 631
+    },
+    {
+      "epoch": 0.7069351230425056,
+      "grad_norm": 0.3500659167766571,
+      "learning_rate": 4.727442421933456e-06,
+      "loss": 0.2558,
+      "step": 632
+    },
+    {
+      "epoch": 0.7080536912751678,
+      "grad_norm": 0.3355782926082611,
+      "learning_rate": 4.725962669385307e-06,
+      "loss": 0.2504,
+      "step": 633
+    },
+    {
+      "epoch": 0.70917225950783,
+      "grad_norm": 0.393700510263443,
+      "learning_rate": 4.724479143697181e-06,
+      "loss": 0.2627,
+      "step": 634
+    },
+    {
+      "epoch": 0.7102908277404921,
+      "grad_norm": 0.37306371331214905,
+      "learning_rate": 4.722991847383744e-06,
+      "loss": 0.3402,
+      "step": 635
+    },
+    {
+      "epoch": 0.7114093959731543,
+      "grad_norm": 0.35806167125701904,
+      "learning_rate": 4.721500782966053e-06,
+      "loss": 0.327,
+      "step": 636
+    },
+    {
+      "epoch": 0.7125279642058165,
+      "grad_norm": 0.3309209942817688,
+      "learning_rate": 4.720005952971551e-06,
+      "loss": 0.2586,
+      "step": 637
+    },
+    {
+      "epoch": 0.7136465324384788,
+      "grad_norm": 0.3596161901950836,
+      "learning_rate": 4.718507359934064e-06,
+      "loss": 0.3108,
+      "step": 638
+    },
+    {
+      "epoch": 0.714765100671141,
+      "grad_norm": 0.4232602119445801,
+      "learning_rate": 4.717005006393798e-06,
+      "loss": 0.2909,
+      "step": 639
+    },
+    {
+      "epoch": 0.7158836689038032,
+      "grad_norm": 0.3771987557411194,
+      "learning_rate": 4.71549889489733e-06,
+      "loss": 0.2957,
+      "step": 640
+    },
+    {
+      "epoch": 0.7170022371364653,
+      "grad_norm": 0.3458472788333893,
+      "learning_rate": 4.713989027997611e-06,
+      "loss": 0.27,
+      "step": 641
+    },
+    {
+      "epoch": 0.7181208053691275,
+      "grad_norm": 0.3060201406478882,
+      "learning_rate": 4.712475408253955e-06,
+      "loss": 0.228,
+      "step": 642
+    },
+    {
+      "epoch": 0.7192393736017897,
+      "grad_norm": 0.36420348286628723,
+      "learning_rate": 4.710958038232039e-06,
+      "loss": 0.2666,
+      "step": 643
+    },
+    {
+      "epoch": 0.7203579418344519,
+      "grad_norm": 0.3402693569660187,
+      "learning_rate": 4.709436920503894e-06,
+      "loss": 0.2929,
+      "step": 644
+    },
+    {
+      "epoch": 0.7214765100671141,
+      "grad_norm": 0.3150545060634613,
+      "learning_rate": 4.7079120576479085e-06,
+      "loss": 0.2718,
+      "step": 645
+    },
+    {
+      "epoch": 0.7225950782997763,
+      "grad_norm": 0.35974034667015076,
+      "learning_rate": 4.706383452248814e-06,
+      "loss": 0.2622,
+      "step": 646
+    },
+    {
+      "epoch": 0.7237136465324385,
+      "grad_norm": 0.34490182995796204,
+      "learning_rate": 4.704851106897689e-06,
+      "loss": 0.3222,
+      "step": 647
+    },
+    {
+      "epoch": 0.7248322147651006,
+      "grad_norm": 0.40183958411216736,
+      "learning_rate": 4.70331502419195e-06,
+      "loss": 0.2867,
+      "step": 648
+    },
+    {
+      "epoch": 0.7259507829977628,
+      "grad_norm": 0.3651914894580841,
+      "learning_rate": 4.70177520673535e-06,
+      "loss": 0.3279,
+      "step": 649
+    },
+    {
+      "epoch": 0.727069351230425,
+      "grad_norm": 0.408110111951828,
+      "learning_rate": 4.700231657137971e-06,
+      "loss": 0.2693,
+      "step": 650
+    },
+    {
+      "epoch": 0.727069351230425,
+      "eval_loss": 0.29643547534942627,
+      "eval_runtime": 377.1333,
+      "eval_samples_per_second": 2.108,
+      "eval_steps_per_second": 2.108,
+      "step": 650
+    },
+    {
+      "epoch": 0.7281879194630873,
+      "grad_norm": 0.46308496594429016,
+      "learning_rate": 4.698684378016223e-06,
+      "loss": 0.3118,
+      "step": 651
+    },
+    {
+      "epoch": 0.7293064876957495,
+      "grad_norm": 0.39952579140663147,
+      "learning_rate": 4.697133371992836e-06,
+      "loss": 0.3703,
+      "step": 652
+    },
+    {
+      "epoch": 0.7304250559284117,
+      "grad_norm": 0.3774736523628235,
+      "learning_rate": 4.6955786416968586e-06,
+      "loss": 0.2499,
+      "step": 653
+    },
+    {
+      "epoch": 0.7315436241610739,
+      "grad_norm": 0.44428813457489014,
+      "learning_rate": 4.694020189763651e-06,
+      "loss": 0.3331,
+      "step": 654
+    },
+    {
+      "epoch": 0.732662192393736,
+      "grad_norm": 0.35620760917663574,
+      "learning_rate": 4.692458018834883e-06,
+      "loss": 0.3706,
+      "step": 655
+    },
+    {
+      "epoch": 0.7337807606263982,
+      "grad_norm": 0.35481327772140503,
+      "learning_rate": 4.6908921315585265e-06,
+      "loss": 0.3212,
+      "step": 656
+    },
+    {
+      "epoch": 0.7348993288590604,
+      "grad_norm": 0.34074386954307556,
+      "learning_rate": 4.689322530588856e-06,
+      "loss": 0.3117,
+      "step": 657
+    },
+    {
+      "epoch": 0.7360178970917226,
+      "grad_norm": 0.3884759545326233,
+      "learning_rate": 4.687749218586438e-06,
+      "loss": 0.246,
+      "step": 658
+    },
+    {
+      "epoch": 0.7371364653243848,
+      "grad_norm": 0.3636641800403595,
+      "learning_rate": 4.686172198218132e-06,
+      "loss": 0.2335,
+      "step": 659
+    },
+    {
+      "epoch": 0.738255033557047,
+      "grad_norm": 0.4108135998249054,
+      "learning_rate": 4.68459147215708e-06,
+      "loss": 0.2894,
+      "step": 660
+    },
+    {
+      "epoch": 0.7393736017897091,
+      "grad_norm": 0.3478459119796753,
+      "learning_rate": 4.683007043082709e-06,
+      "loss": 0.329,
+      "step": 661
+    },
+    {
+      "epoch": 0.7404921700223713,
+      "grad_norm": 0.3461150527000427,
+      "learning_rate": 4.68141891368072e-06,
+      "loss": 0.2342,
+      "step": 662
+    },
+    {
+      "epoch": 0.7416107382550335,
+      "grad_norm": 0.4223523437976837,
+      "learning_rate": 4.679827086643088e-06,
+      "loss": 0.3195,
+      "step": 663
+    },
+    {
+      "epoch": 0.7427293064876958,
+      "grad_norm": 0.3769877552986145,
+      "learning_rate": 4.678231564668053e-06,
+      "loss": 0.3113,
+      "step": 664
+    },
+    {
+      "epoch": 0.743847874720358,
+      "grad_norm": 0.3045426309108734,
+      "learning_rate": 4.6766323504601245e-06,
+      "loss": 0.2473,
+      "step": 665
+    },
+    {
+      "epoch": 0.7449664429530202,
+      "grad_norm": 0.41605642437934875,
+      "learning_rate": 4.675029446730062e-06,
+      "loss": 0.2852,
+      "step": 666
+    },
+    {
+      "epoch": 0.7460850111856824,
+      "grad_norm": 0.3897973597049713,
+      "learning_rate": 4.673422856194886e-06,
+      "loss": 0.2876,
+      "step": 667
+    },
+    {
+      "epoch": 0.7472035794183445,
+      "grad_norm": 0.385809063911438,
+      "learning_rate": 4.671812581577863e-06,
+      "loss": 0.3509,
+      "step": 668
+    },
+    {
+      "epoch": 0.7483221476510067,
+      "grad_norm": 0.3703514337539673,
+      "learning_rate": 4.670198625608504e-06,
+      "loss": 0.232,
+      "step": 669
+    },
+    {
+      "epoch": 0.7494407158836689,
+      "grad_norm": 0.38114774227142334,
+      "learning_rate": 4.668580991022563e-06,
+      "loss": 0.3684,
+      "step": 670
+    },
+    {
+      "epoch": 0.7505592841163311,
+      "grad_norm": 0.38675299286842346,
+      "learning_rate": 4.666959680562026e-06,
+      "loss": 0.2834,
+      "step": 671
+    },
+    {
+      "epoch": 0.7516778523489933,
+      "grad_norm": 0.3351726830005646,
+      "learning_rate": 4.66533469697511e-06,
+      "loss": 0.2557,
+      "step": 672
+    },
+    {
+      "epoch": 0.7527964205816555,
+      "grad_norm": 0.30592837929725647,
+      "learning_rate": 4.663706043016262e-06,
+      "loss": 0.2378,
+      "step": 673
+    },
+    {
+      "epoch": 0.7539149888143176,
+      "grad_norm": 0.3512606918811798,
+      "learning_rate": 4.662073721446146e-06,
+      "loss": 0.276,
+      "step": 674
+    },
+    {
+      "epoch": 0.7550335570469798,
+      "grad_norm": 0.3326805830001831,
+      "learning_rate": 4.660437735031646e-06,
+      "loss": 0.289,
+      "step": 675
+    },
+    {
+      "epoch": 0.756152125279642,
+      "grad_norm": 0.372894287109375,
+      "learning_rate": 4.658798086545857e-06,
+      "loss": 0.2555,
+      "step": 676
+    },
+    {
+      "epoch": 0.7572706935123042,
+      "grad_norm": 0.4499683678150177,
+      "learning_rate": 4.65715477876808e-06,
+      "loss": 0.309,
+      "step": 677
+    },
+    {
+      "epoch": 0.7583892617449665,
+      "grad_norm": 0.36224496364593506,
+      "learning_rate": 4.655507814483821e-06,
+      "loss": 0.2918,
+      "step": 678
+    },
+    {
+      "epoch": 0.7595078299776287,
+      "grad_norm": 0.3316250741481781,
+      "learning_rate": 4.653857196484783e-06,
+      "loss": 0.2893,
+      "step": 679
+    },
+    {
+      "epoch": 0.7606263982102909,
+      "grad_norm": 0.38180679082870483,
+      "learning_rate": 4.652202927568863e-06,
+      "loss": 0.3297,
+      "step": 680
+    },
+    {
+      "epoch": 0.761744966442953,
+      "grad_norm": 0.5836800932884216,
+      "learning_rate": 4.650545010540145e-06,
+      "loss": 0.3664,
+      "step": 681
+    },
+    {
+      "epoch": 0.7628635346756152,
+      "grad_norm": 0.3977658450603485,
+      "learning_rate": 4.648883448208897e-06,
+      "loss": 0.3003,
+      "step": 682
+    },
+    {
+      "epoch": 0.7639821029082774,
+      "grad_norm": 0.3326352834701538,
+      "learning_rate": 4.647218243391568e-06,
+      "loss": 0.2257,
+      "step": 683
+    },
+    {
+      "epoch": 0.7651006711409396,
+      "grad_norm": 0.36793944239616394,
+      "learning_rate": 4.645549398910778e-06,
+      "loss": 0.3141,
+      "step": 684
+    },
+    {
+      "epoch": 0.7662192393736018,
+      "grad_norm": 0.4063091278076172,
+      "learning_rate": 4.643876917595321e-06,
+      "loss": 0.3874,
+      "step": 685
+    },
+    {
+      "epoch": 0.767337807606264,
+      "grad_norm": 0.33739620447158813,
+      "learning_rate": 4.64220080228015e-06,
+      "loss": 0.2685,
+      "step": 686
+    },
+    {
+      "epoch": 0.7684563758389261,
+      "grad_norm": 0.47535163164138794,
+      "learning_rate": 4.6405210558063835e-06,
+      "loss": 0.2874,
+      "step": 687
+    },
+    {
+      "epoch": 0.7695749440715883,
+      "grad_norm": 0.3732830882072449,
+      "learning_rate": 4.63883768102129e-06,
+      "loss": 0.2404,
+      "step": 688
+    },
+    {
+      "epoch": 0.7706935123042505,
+      "grad_norm": 0.420602411031723,
+      "learning_rate": 4.637150680778292e-06,
+      "loss": 0.4128,
+      "step": 689
+    },
+    {
+      "epoch": 0.7718120805369127,
+      "grad_norm": 0.3910024166107178,
+      "learning_rate": 4.635460057936956e-06,
+      "loss": 0.2739,
+      "step": 690
+    },
+    {
+      "epoch": 0.772930648769575,
+      "grad_norm": 0.47667405009269714,
+      "learning_rate": 4.633765815362988e-06,
+      "loss": 0.3901,
+      "step": 691
+    },
+    {
+      "epoch": 0.7740492170022372,
+      "grad_norm": 1.0164694786071777,
+      "learning_rate": 4.632067955928231e-06,
+      "loss": 0.346,
+      "step": 692
+    },
+    {
+      "epoch": 0.7751677852348994,
+      "grad_norm": 0.3763993978500366,
+      "learning_rate": 4.630366482510658e-06,
+      "loss": 0.3185,
+      "step": 693
+    },
+    {
+      "epoch": 0.7762863534675615,
+      "grad_norm": 0.3833650052547455,
+      "learning_rate": 4.628661397994369e-06,
+      "loss": 0.2598,
+      "step": 694
+    },
+    {
+      "epoch": 0.7774049217002237,
+      "grad_norm": 0.42143768072128296,
+      "learning_rate": 4.626952705269584e-06,
+      "loss": 0.2992,
+      "step": 695
+    },
+    {
+      "epoch": 0.7785234899328859,
+      "grad_norm": 0.36815062165260315,
+      "learning_rate": 4.625240407232639e-06,
+      "loss": 0.3139,
+      "step": 696
+    },
+    {
+      "epoch": 0.7796420581655481,
+      "grad_norm": 0.4066043198108673,
+      "learning_rate": 4.623524506785983e-06,
+      "loss": 0.3403,
+      "step": 697
+    },
+    {
+      "epoch": 0.7807606263982103,
+      "grad_norm": 0.4246339201927185,
+      "learning_rate": 4.621805006838168e-06,
+      "loss": 0.2979,
+      "step": 698
+    },
+    {
+      "epoch": 0.7818791946308725,
+      "grad_norm": 0.3898549973964691,
+      "learning_rate": 4.620081910303852e-06,
+      "loss": 0.2881,
+      "step": 699
+    },
+    {
+      "epoch": 0.7829977628635347,
+      "grad_norm": 0.4031178653240204,
+      "learning_rate": 4.618355220103784e-06,
+      "loss": 0.3835,
+      "step": 700
+    },
+    {
+      "epoch": 0.7829977628635347,
+      "eval_loss": 0.29387032985687256,
+      "eval_runtime": 376.665,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 700
+    },
+    {
+      "epoch": 0.7841163310961968,
+      "grad_norm": 0.3796100616455078,
+      "learning_rate": 4.616624939164808e-06,
+      "loss": 0.3091,
+      "step": 701
+    },
+    {
+      "epoch": 0.785234899328859,
+      "grad_norm": 0.36631500720977783,
+      "learning_rate": 4.614891070419855e-06,
+      "loss": 0.3799,
+      "step": 702
+    },
+    {
+      "epoch": 0.7863534675615212,
+      "grad_norm": 0.3151400685310364,
+      "learning_rate": 4.613153616807935e-06,
+      "loss": 0.2385,
+      "step": 703
+    },
+    {
+      "epoch": 0.7874720357941835,
+      "grad_norm": 0.4115789830684662,
+      "learning_rate": 4.611412581274137e-06,
+      "loss": 0.2792,
+      "step": 704
+    },
+    {
+      "epoch": 0.7885906040268457,
+      "grad_norm": 0.3879226744174957,
+      "learning_rate": 4.60966796676962e-06,
+      "loss": 0.2603,
+      "step": 705
+    },
+    {
+      "epoch": 0.7897091722595079,
+      "grad_norm": 0.3998829424381256,
+      "learning_rate": 4.60791977625161e-06,
+      "loss": 0.2914,
+      "step": 706
+    },
+    {
+      "epoch": 0.79082774049217,
+      "grad_norm": 0.43842077255249023,
+      "learning_rate": 4.606168012683394e-06,
+      "loss": 0.3534,
+      "step": 707
+    },
+    {
+      "epoch": 0.7919463087248322,
+      "grad_norm": 0.3857001066207886,
+      "learning_rate": 4.604412679034317e-06,
+      "loss": 0.2994,
+      "step": 708
+    },
+    {
+      "epoch": 0.7930648769574944,
+      "grad_norm": 0.47911903262138367,
+      "learning_rate": 4.602653778279774e-06,
+      "loss": 0.3526,
+      "step": 709
+    },
+    {
+      "epoch": 0.7941834451901566,
+      "grad_norm": 0.3647831678390503,
+      "learning_rate": 4.600891313401207e-06,
+      "loss": 0.2733,
+      "step": 710
+    },
+    {
+      "epoch": 0.7953020134228188,
+      "grad_norm": 0.39061617851257324,
+      "learning_rate": 4.5991252873861e-06,
+      "loss": 0.262,
+      "step": 711
+    },
+    {
+      "epoch": 0.796420581655481,
+      "grad_norm": 0.3592703342437744,
+      "learning_rate": 4.597355703227972e-06,
+      "loss": 0.2599,
+      "step": 712
+    },
+    {
+      "epoch": 0.7975391498881432,
+      "grad_norm": 0.359822154045105,
+      "learning_rate": 4.5955825639263754e-06,
+      "loss": 0.3651,
+      "step": 713
+    },
+    {
+      "epoch": 0.7986577181208053,
+      "grad_norm": 0.4158327877521515,
+      "learning_rate": 4.5938058724868865e-06,
+      "loss": 0.3054,
+      "step": 714
+    },
+    {
+      "epoch": 0.7997762863534675,
+      "grad_norm": 0.4156656265258789,
+      "learning_rate": 4.592025631921104e-06,
+      "loss": 0.2982,
+      "step": 715
+    },
+    {
+      "epoch": 0.8008948545861297,
+      "grad_norm": 0.38249292969703674,
+      "learning_rate": 4.590241845246641e-06,
+      "loss": 0.2431,
+      "step": 716
+    },
+    {
+      "epoch": 0.802013422818792,
+      "grad_norm": 0.4058525264263153,
+      "learning_rate": 4.588454515487125e-06,
+      "loss": 0.2528,
+      "step": 717
+    },
+    {
+      "epoch": 0.8031319910514542,
+      "grad_norm": 0.37386444211006165,
+      "learning_rate": 4.586663645672186e-06,
+      "loss": 0.259,
+      "step": 718
+    },
+    {
+      "epoch": 0.8042505592841164,
+      "grad_norm": 0.3539482057094574,
+      "learning_rate": 4.584869238837454e-06,
+      "loss": 0.2857,
+      "step": 719
+    },
+    {
+      "epoch": 0.8053691275167785,
+      "grad_norm": 0.3337392210960388,
+      "learning_rate": 4.583071298024558e-06,
+      "loss": 0.2459,
+      "step": 720
+    },
+    {
+      "epoch": 0.8064876957494407,
+      "grad_norm": 0.407926470041275,
+      "learning_rate": 4.581269826281114e-06,
+      "loss": 0.236,
+      "step": 721
+    },
+    {
+      "epoch": 0.8076062639821029,
+      "grad_norm": 0.36108073592185974,
+      "learning_rate": 4.5794648266607256e-06,
+      "loss": 0.2705,
+      "step": 722
+    },
+    {
+      "epoch": 0.8087248322147651,
+      "grad_norm": 0.39510682225227356,
+      "learning_rate": 4.577656302222975e-06,
+      "loss": 0.2768,
+      "step": 723
+    },
+    {
+      "epoch": 0.8098434004474273,
+      "grad_norm": 0.33281171321868896,
+      "learning_rate": 4.575844256033418e-06,
+      "loss": 0.2748,
+      "step": 724
+    },
+    {
+      "epoch": 0.8109619686800895,
+      "grad_norm": 0.44088882207870483,
+      "learning_rate": 4.574028691163583e-06,
+      "loss": 0.321,
+      "step": 725
+    },
+    {
+      "epoch": 0.8120805369127517,
+      "grad_norm": 0.4108178913593292,
+      "learning_rate": 4.5722096106909595e-06,
+      "loss": 0.3029,
+      "step": 726
+    },
+    {
+      "epoch": 0.8131991051454138,
+      "grad_norm": 0.33357682824134827,
+      "learning_rate": 4.570387017699e-06,
+      "loss": 0.2778,
+      "step": 727
+    },
+    {
+      "epoch": 0.814317673378076,
+      "grad_norm": 0.4473627209663391,
+      "learning_rate": 4.568560915277107e-06,
+      "loss": 0.4199,
+      "step": 728
+    },
+    {
+      "epoch": 0.8154362416107382,
+      "grad_norm": 0.43677130341529846,
+      "learning_rate": 4.566731306520634e-06,
+      "loss": 0.2726,
+      "step": 729
+    },
+    {
+      "epoch": 0.8165548098434005,
+      "grad_norm": 0.4755209982395172,
+      "learning_rate": 4.564898194530878e-06,
+      "loss": 0.3478,
+      "step": 730
+    },
+    {
+      "epoch": 0.8176733780760627,
+      "grad_norm": 0.33642473816871643,
+      "learning_rate": 4.563061582415072e-06,
+      "loss": 0.252,
+      "step": 731
+    },
+    {
+      "epoch": 0.8187919463087249,
+      "grad_norm": 0.34328266978263855,
+      "learning_rate": 4.561221473286384e-06,
+      "loss": 0.277,
+      "step": 732
+    },
+    {
+      "epoch": 0.819910514541387,
+      "grad_norm": 0.38910964131355286,
+      "learning_rate": 4.55937787026391e-06,
+      "loss": 0.2533,
+      "step": 733
+    },
+    {
+      "epoch": 0.8210290827740492,
+      "grad_norm": 0.3730185925960541,
+      "learning_rate": 4.557530776472667e-06,
+      "loss": 0.26,
+      "step": 734
+    },
+    {
+      "epoch": 0.8221476510067114,
+      "grad_norm": 0.39365464448928833,
+      "learning_rate": 4.55568019504359e-06,
+      "loss": 0.3035,
+      "step": 735
+    },
+    {
+      "epoch": 0.8232662192393736,
+      "grad_norm": 0.35591381788253784,
+      "learning_rate": 4.553826129113524e-06,
+      "loss": 0.2565,
+      "step": 736
+    },
+    {
+      "epoch": 0.8243847874720358,
+      "grad_norm": 0.36858460307121277,
+      "learning_rate": 4.551968581825223e-06,
+      "loss": 0.3147,
+      "step": 737
+    },
+    {
+      "epoch": 0.825503355704698,
+      "grad_norm": 0.6574727296829224,
+      "learning_rate": 4.55010755632734e-06,
+      "loss": 0.2467,
+      "step": 738
+    },
+    {
+      "epoch": 0.8266219239373602,
+      "grad_norm": 0.4751858115196228,
+      "learning_rate": 4.548243055774425e-06,
+      "loss": 0.3666,
+      "step": 739
+    },
+    {
+      "epoch": 0.8277404921700223,
+      "grad_norm": 0.3789668083190918,
+      "learning_rate": 4.546375083326918e-06,
+      "loss": 0.2415,
+      "step": 740
+    },
+    {
+      "epoch": 0.8288590604026845,
+      "grad_norm": 0.38406869769096375,
+      "learning_rate": 4.544503642151144e-06,
+      "loss": 0.323,
+      "step": 741
+    },
+    {
+      "epoch": 0.8299776286353467,
+      "grad_norm": 0.36564141511917114,
+      "learning_rate": 4.542628735419309e-06,
+      "loss": 0.2953,
+      "step": 742
+    },
+    {
+      "epoch": 0.831096196868009,
+      "grad_norm": 0.3515996038913727,
+      "learning_rate": 4.540750366309491e-06,
+      "loss": 0.2387,
+      "step": 743
+    },
+    {
+      "epoch": 0.8322147651006712,
+      "grad_norm": 0.34407639503479004,
+      "learning_rate": 4.538868538005638e-06,
+      "loss": 0.239,
+      "step": 744
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.4639814496040344,
+      "learning_rate": 4.536983253697561e-06,
+      "loss": 0.3733,
+      "step": 745
+    },
+    {
+      "epoch": 0.8344519015659956,
+      "grad_norm": 0.3663330674171448,
+      "learning_rate": 4.535094516580932e-06,
+      "loss": 0.2688,
+      "step": 746
+    },
+    {
+      "epoch": 0.8355704697986577,
+      "grad_norm": 0.390386164188385,
+      "learning_rate": 4.533202329857271e-06,
+      "loss": 0.2861,
+      "step": 747
+    },
+    {
+      "epoch": 0.8366890380313199,
+      "grad_norm": 0.3823988139629364,
+      "learning_rate": 4.531306696733949e-06,
+      "loss": 0.2926,
+      "step": 748
+    },
+    {
+      "epoch": 0.8378076062639821,
+      "grad_norm": 0.40171313285827637,
+      "learning_rate": 4.529407620424178e-06,
+      "loss": 0.2547,
+      "step": 749
+    },
+    {
+      "epoch": 0.8389261744966443,
+      "grad_norm": 0.4024547040462494,
+      "learning_rate": 4.527505104147006e-06,
+      "loss": 0.2848,
+      "step": 750
+    },
+    {
+      "epoch": 0.8389261744966443,
+      "eval_loss": 0.2920389771461487,
+      "eval_runtime": 376.6865,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 750
+    },
+    {
+      "epoch": 0.8400447427293065,
+      "grad_norm": 0.3801092207431793,
+      "learning_rate": 4.52559915112731e-06,
+      "loss": 0.3202,
+      "step": 751
+    },
+    {
+      "epoch": 0.8411633109619687,
+      "grad_norm": 0.35585805773735046,
+      "learning_rate": 4.523689764595797e-06,
+      "loss": 0.2682,
+      "step": 752
+    },
+    {
+      "epoch": 0.8422818791946308,
+      "grad_norm": 0.4018070101737976,
+      "learning_rate": 4.521776947788991e-06,
+      "loss": 0.3044,
+      "step": 753
+    },
+    {
+      "epoch": 0.843400447427293,
+      "grad_norm": 0.3804682791233063,
+      "learning_rate": 4.519860703949229e-06,
+      "loss": 0.3092,
+      "step": 754
+    },
+    {
+      "epoch": 0.8445190156599552,
+      "grad_norm": 0.3889797031879425,
+      "learning_rate": 4.5179410363246616e-06,
+      "loss": 0.2914,
+      "step": 755
+    },
+    {
+      "epoch": 0.8456375838926175,
+      "grad_norm": 0.41364794969558716,
+      "learning_rate": 4.516017948169239e-06,
+      "loss": 0.4066,
+      "step": 756
+    },
+    {
+      "epoch": 0.8467561521252797,
+      "grad_norm": 0.39795881509780884,
+      "learning_rate": 4.51409144274271e-06,
+      "loss": 0.3522,
+      "step": 757
+    },
+    {
+      "epoch": 0.8478747203579419,
+      "grad_norm": 0.3604715168476105,
+      "learning_rate": 4.512161523310618e-06,
+      "loss": 0.2494,
+      "step": 758
+    },
+    {
+      "epoch": 0.8489932885906041,
+      "grad_norm": 0.3989139795303345,
+      "learning_rate": 4.510228193144292e-06,
+      "loss": 0.2833,
+      "step": 759
+    },
+    {
+      "epoch": 0.8501118568232662,
+      "grad_norm": 0.46256813406944275,
+      "learning_rate": 4.508291455520841e-06,
+      "loss": 0.3088,
+      "step": 760
+    },
+    {
+      "epoch": 0.8512304250559284,
+      "grad_norm": 0.3795095682144165,
+      "learning_rate": 4.506351313723153e-06,
+      "loss": 0.2536,
+      "step": 761
+    },
+    {
+      "epoch": 0.8523489932885906,
+      "grad_norm": 0.3441731929779053,
+      "learning_rate": 4.504407771039883e-06,
+      "loss": 0.2427,
+      "step": 762
+    },
+    {
+      "epoch": 0.8534675615212528,
+      "grad_norm": 0.4435843527317047,
+      "learning_rate": 4.502460830765453e-06,
+      "loss": 0.3819,
+      "step": 763
+    },
+    {
+      "epoch": 0.854586129753915,
+      "grad_norm": 0.4349744915962219,
+      "learning_rate": 4.500510496200043e-06,
+      "loss": 0.3239,
+      "step": 764
+    },
+    {
+      "epoch": 0.8557046979865772,
+      "grad_norm": 0.4064520597457886,
+      "learning_rate": 4.498556770649587e-06,
+      "loss": 0.3367,
+      "step": 765
+    },
+    {
+      "epoch": 0.8568232662192393,
+      "grad_norm": 0.41930535435676575,
+      "learning_rate": 4.496599657425766e-06,
+      "loss": 0.4612,
+      "step": 766
+    },
+    {
+      "epoch": 0.8579418344519015,
+      "grad_norm": 0.4040359854698181,
+      "learning_rate": 4.494639159846006e-06,
+      "loss": 0.3265,
+      "step": 767
+    },
+    {
+      "epoch": 0.8590604026845637,
+      "grad_norm": 0.3989505469799042,
+      "learning_rate": 4.492675281233466e-06,
+      "loss": 0.2964,
+      "step": 768
+    },
+    {
+      "epoch": 0.860178970917226,
+      "grad_norm": 0.3589783012866974,
+      "learning_rate": 4.490708024917039e-06,
+      "loss": 0.2776,
+      "step": 769
+    },
+    {
+      "epoch": 0.8612975391498882,
+      "grad_norm": 0.43537792563438416,
+      "learning_rate": 4.48873739423134e-06,
+      "loss": 0.293,
+      "step": 770
+    },
+    {
+      "epoch": 0.8624161073825504,
+      "grad_norm": 0.3926440477371216,
+      "learning_rate": 4.486763392516709e-06,
+      "loss": 0.3465,
+      "step": 771
+    },
+    {
+      "epoch": 0.8635346756152126,
+      "grad_norm": 0.37360385060310364,
+      "learning_rate": 4.484786023119194e-06,
+      "loss": 0.3115,
+      "step": 772
+    },
+    {
+      "epoch": 0.8646532438478747,
+      "grad_norm": 0.41157010197639465,
+      "learning_rate": 4.482805289390557e-06,
+      "loss": 0.3122,
+      "step": 773
+    },
+    {
+      "epoch": 0.8657718120805369,
+      "grad_norm": 0.39628759026527405,
+      "learning_rate": 4.48082119468826e-06,
+      "loss": 0.2359,
+      "step": 774
+    },
+    {
+      "epoch": 0.8668903803131991,
+      "grad_norm": 0.3240848183631897,
+      "learning_rate": 4.478833742375461e-06,
+      "loss": 0.2728,
+      "step": 775
+    },
+    {
+      "epoch": 0.8680089485458613,
+      "grad_norm": 0.4541075527667999,
+      "learning_rate": 4.476842935821011e-06,
+      "loss": 0.2731,
+      "step": 776
+    },
+    {
+      "epoch": 0.8691275167785235,
+      "grad_norm": 0.48087626695632935,
+      "learning_rate": 4.474848778399447e-06,
+      "loss": 0.289,
+      "step": 777
+    },
+    {
+      "epoch": 0.8702460850111857,
+      "grad_norm": 0.40191930532455444,
+      "learning_rate": 4.472851273490985e-06,
+      "loss": 0.284,
+      "step": 778
+    },
+    {
+      "epoch": 0.8713646532438478,
+      "grad_norm": 0.3911993205547333,
+      "learning_rate": 4.470850424481514e-06,
+      "loss": 0.2529,
+      "step": 779
+    },
+    {
+      "epoch": 0.87248322147651,
+      "grad_norm": 0.39867544174194336,
+      "learning_rate": 4.468846234762595e-06,
+      "loss": 0.2777,
+      "step": 780
+    },
+    {
+      "epoch": 0.8736017897091722,
+      "grad_norm": 0.3462413251399994,
+      "learning_rate": 4.466838707731448e-06,
+      "loss": 0.2399,
+      "step": 781
+    },
+    {
+      "epoch": 0.8747203579418344,
+      "grad_norm": 0.45927783846855164,
+      "learning_rate": 4.464827846790952e-06,
+      "loss": 0.2703,
+      "step": 782
+    },
+    {
+      "epoch": 0.8758389261744967,
+      "grad_norm": 0.36693263053894043,
+      "learning_rate": 4.462813655349637e-06,
+      "loss": 0.315,
+      "step": 783
+    },
+    {
+      "epoch": 0.8769574944071589,
+      "grad_norm": 0.35959646105766296,
+      "learning_rate": 4.460796136821679e-06,
+      "loss": 0.2729,
+      "step": 784
+    },
+    {
+      "epoch": 0.8780760626398211,
+      "grad_norm": 0.4232349991798401,
+      "learning_rate": 4.45877529462689e-06,
+      "loss": 0.2663,
+      "step": 785
+    },
+    {
+      "epoch": 0.8791946308724832,
+      "grad_norm": 0.5229600667953491,
+      "learning_rate": 4.45675113219072e-06,
+      "loss": 0.2819,
+      "step": 786
+    },
+    {
+      "epoch": 0.8803131991051454,
+      "grad_norm": 0.3806210458278656,
+      "learning_rate": 4.454723652944246e-06,
+      "loss": 0.3358,
+      "step": 787
+    },
+    {
+      "epoch": 0.8814317673378076,
+      "grad_norm": 0.35532307624816895,
+      "learning_rate": 4.452692860324166e-06,
+      "loss": 0.2343,
+      "step": 788
+    },
+    {
+      "epoch": 0.8825503355704698,
+      "grad_norm": 0.3916340470314026,
+      "learning_rate": 4.4506587577727945e-06,
+      "loss": 0.2807,
+      "step": 789
+    },
+    {
+      "epoch": 0.883668903803132,
+      "grad_norm": 0.40005603432655334,
+      "learning_rate": 4.448621348738059e-06,
+      "loss": 0.3033,
+      "step": 790
+    },
+    {
+      "epoch": 0.8847874720357942,
+      "grad_norm": 0.4400363564491272,
+      "learning_rate": 4.446580636673487e-06,
+      "loss": 0.3366,
+      "step": 791
+    },
+    {
+      "epoch": 0.8859060402684564,
+      "grad_norm": 0.4062058627605438,
+      "learning_rate": 4.44453662503821e-06,
+      "loss": 0.3194,
+      "step": 792
+    },
+    {
+      "epoch": 0.8870246085011185,
+      "grad_norm": 0.4038722515106201,
+      "learning_rate": 4.44248931729695e-06,
+      "loss": 0.3076,
+      "step": 793
+    },
+    {
+      "epoch": 0.8881431767337807,
+      "grad_norm": 0.4124155640602112,
+      "learning_rate": 4.440438716920015e-06,
+      "loss": 0.3788,
+      "step": 794
+    },
+    {
+      "epoch": 0.889261744966443,
+      "grad_norm": 0.3759743571281433,
+      "learning_rate": 4.438384827383295e-06,
+      "loss": 0.2347,
+      "step": 795
+    },
+    {
+      "epoch": 0.8903803131991052,
+      "grad_norm": 0.44200465083122253,
+      "learning_rate": 4.436327652168257e-06,
+      "loss": 0.3847,
+      "step": 796
+    },
+    {
+      "epoch": 0.8914988814317674,
+      "grad_norm": 0.4350292980670929,
+      "learning_rate": 4.434267194761936e-06,
+      "loss": 0.3082,
+      "step": 797
+    },
+    {
+      "epoch": 0.8926174496644296,
+      "grad_norm": 0.4255637228488922,
+      "learning_rate": 4.432203458656931e-06,
+      "loss": 0.3374,
+      "step": 798
+    },
+    {
+      "epoch": 0.8937360178970917,
+      "grad_norm": 0.36973047256469727,
+      "learning_rate": 4.430136447351397e-06,
+      "loss": 0.2586,
+      "step": 799
+    },
+    {
+      "epoch": 0.8948545861297539,
+      "grad_norm": 0.3531433045864105,
+      "learning_rate": 4.428066164349042e-06,
+      "loss": 0.24,
+      "step": 800
+    },
+    {
+      "epoch": 0.8948545861297539,
+      "eval_loss": 0.2899681329727173,
+      "eval_runtime": 376.5794,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 800
+    },
+    {
+      "epoch": 0.8959731543624161,
+      "grad_norm": 0.3802293837070465,
+      "learning_rate": 4.42599261315912e-06,
+      "loss": 0.2704,
+      "step": 801
+    },
+    {
+      "epoch": 0.8970917225950783,
+      "grad_norm": 0.41422635316848755,
+      "learning_rate": 4.423915797296426e-06,
+      "loss": 0.3015,
+      "step": 802
+    },
+    {
+      "epoch": 0.8982102908277405,
+      "grad_norm": 0.3657076358795166,
+      "learning_rate": 4.421835720281283e-06,
+      "loss": 0.2735,
+      "step": 803
+    },
+    {
+      "epoch": 0.8993288590604027,
+      "grad_norm": 0.39450007677078247,
+      "learning_rate": 4.4197523856395505e-06,
+      "loss": 0.2576,
+      "step": 804
+    },
+    {
+      "epoch": 0.9004474272930649,
+      "grad_norm": 0.3786505460739136,
+      "learning_rate": 4.4176657969026035e-06,
+      "loss": 0.3504,
+      "step": 805
+    },
+    {
+      "epoch": 0.901565995525727,
+      "grad_norm": 0.41710010170936584,
+      "learning_rate": 4.415575957607335e-06,
+      "loss": 0.2822,
+      "step": 806
+    },
+    {
+      "epoch": 0.9026845637583892,
+      "grad_norm": 0.3772369921207428,
+      "learning_rate": 4.413482871296148e-06,
+      "loss": 0.2809,
+      "step": 807
+    },
+    {
+      "epoch": 0.9038031319910514,
+      "grad_norm": 0.3759446442127228,
+      "learning_rate": 4.4113865415169485e-06,
+      "loss": 0.2816,
+      "step": 808
+    },
+    {
+      "epoch": 0.9049217002237137,
+      "grad_norm": 0.3392682373523712,
+      "learning_rate": 4.409286971823141e-06,
+      "loss": 0.2367,
+      "step": 809
+    },
+    {
+      "epoch": 0.9060402684563759,
+      "grad_norm": 0.42997559905052185,
+      "learning_rate": 4.407184165773622e-06,
+      "loss": 0.2973,
+      "step": 810
+    },
+    {
+      "epoch": 0.9071588366890381,
+      "grad_norm": 0.35543403029441833,
+      "learning_rate": 4.405078126932775e-06,
+      "loss": 0.3123,
+      "step": 811
+    },
+    {
+      "epoch": 0.9082774049217002,
+      "grad_norm": 0.44964316487312317,
+      "learning_rate": 4.40296885887046e-06,
+      "loss": 0.3353,
+      "step": 812
+    },
+    {
+      "epoch": 0.9093959731543624,
+      "grad_norm": 0.4058191776275635,
+      "learning_rate": 4.400856365162012e-06,
+      "loss": 0.2824,
+      "step": 813
+    },
+    {
+      "epoch": 0.9105145413870246,
+      "grad_norm": 0.438174307346344,
+      "learning_rate": 4.398740649388237e-06,
+      "loss": 0.2766,
+      "step": 814
+    },
+    {
+      "epoch": 0.9116331096196868,
+      "grad_norm": 0.3943518102169037,
+      "learning_rate": 4.396621715135397e-06,
+      "loss": 0.3257,
+      "step": 815
+    },
+    {
+      "epoch": 0.912751677852349,
+      "grad_norm": 0.3897485136985779,
+      "learning_rate": 4.394499565995215e-06,
+      "loss": 0.2877,
+      "step": 816
+    },
+    {
+      "epoch": 0.9138702460850112,
+      "grad_norm": 0.385063111782074,
+      "learning_rate": 4.39237420556486e-06,
+      "loss": 0.2521,
+      "step": 817
+    },
+    {
+      "epoch": 0.9149888143176734,
+      "grad_norm": 0.3870301842689514,
+      "learning_rate": 4.390245637446944e-06,
+      "loss": 0.3379,
+      "step": 818
+    },
+    {
+      "epoch": 0.9161073825503355,
+      "grad_norm": 0.361459344625473,
+      "learning_rate": 4.388113865249518e-06,
+      "loss": 0.2909,
+      "step": 819
+    },
+    {
+      "epoch": 0.9172259507829977,
+      "grad_norm": 0.37545695900917053,
+      "learning_rate": 4.385978892586065e-06,
+      "loss": 0.2614,
+      "step": 820
+    },
+    {
+      "epoch": 0.9183445190156599,
+      "grad_norm": 0.43916240334510803,
+      "learning_rate": 4.383840723075488e-06,
+      "loss": 0.3595,
+      "step": 821
+    },
+    {
+      "epoch": 0.9194630872483222,
+      "grad_norm": 0.3690807819366455,
+      "learning_rate": 4.381699360342117e-06,
+      "loss": 0.3478,
+      "step": 822
+    },
+    {
+      "epoch": 0.9205816554809844,
+      "grad_norm": 0.39453035593032837,
+      "learning_rate": 4.379554808015686e-06,
+      "loss": 0.2988,
+      "step": 823
+    },
+    {
+      "epoch": 0.9217002237136466,
+      "grad_norm": 0.3663634657859802,
+      "learning_rate": 4.377407069731342e-06,
+      "loss": 0.2411,
+      "step": 824
+    },
+    {
+      "epoch": 0.9228187919463087,
+      "grad_norm": 0.3578493297100067,
+      "learning_rate": 4.375256149129628e-06,
+      "loss": 0.2337,
+      "step": 825
+    },
+    {
+      "epoch": 0.9239373601789709,
+      "grad_norm": 0.40120798349380493,
+      "learning_rate": 4.373102049856485e-06,
+      "loss": 0.2586,
+      "step": 826
+    },
+    {
+      "epoch": 0.9250559284116331,
+      "grad_norm": 0.4924633502960205,
+      "learning_rate": 4.370944775563239e-06,
+      "loss": 0.3175,
+      "step": 827
+    },
+    {
+      "epoch": 0.9261744966442953,
+      "grad_norm": 0.4092751145362854,
+      "learning_rate": 4.368784329906599e-06,
+      "loss": 0.2443,
+      "step": 828
+    },
+    {
+      "epoch": 0.9272930648769575,
+      "grad_norm": 0.3876032829284668,
+      "learning_rate": 4.366620716548651e-06,
+      "loss": 0.3178,
+      "step": 829
+    },
+    {
+      "epoch": 0.9284116331096197,
+      "grad_norm": 0.34863290190696716,
+      "learning_rate": 4.364453939156849e-06,
+      "loss": 0.2833,
+      "step": 830
+    },
+    {
+      "epoch": 0.9295302013422819,
+      "grad_norm": 0.41042447090148926,
+      "learning_rate": 4.362284001404008e-06,
+      "loss": 0.2817,
+      "step": 831
+    },
+    {
+      "epoch": 0.930648769574944,
+      "grad_norm": 0.3950803279876709,
+      "learning_rate": 4.3601109069683056e-06,
+      "loss": 0.2349,
+      "step": 832
+    },
+    {
+      "epoch": 0.9317673378076062,
+      "grad_norm": 0.45157188177108765,
+      "learning_rate": 4.357934659533265e-06,
+      "loss": 0.2931,
+      "step": 833
+    },
+    {
+      "epoch": 0.9328859060402684,
+      "grad_norm": 0.37973472476005554,
+      "learning_rate": 4.355755262787756e-06,
+      "loss": 0.2628,
+      "step": 834
+    },
+    {
+      "epoch": 0.9340044742729307,
+      "grad_norm": 0.3617250621318817,
+      "learning_rate": 4.353572720425987e-06,
+      "loss": 0.2627,
+      "step": 835
+    },
+    {
+      "epoch": 0.9351230425055929,
+      "grad_norm": 0.41152244806289673,
+      "learning_rate": 4.3513870361474985e-06,
+      "loss": 0.257,
+      "step": 836
+    },
+    {
+      "epoch": 0.9362416107382551,
+      "grad_norm": 0.46206122636795044,
+      "learning_rate": 4.349198213657155e-06,
+      "loss": 0.2733,
+      "step": 837
+    },
+    {
+      "epoch": 0.9373601789709173,
+      "grad_norm": 0.49034422636032104,
+      "learning_rate": 4.347006256665143e-06,
+      "loss": 0.2981,
+      "step": 838
+    },
+    {
+      "epoch": 0.9384787472035794,
+      "grad_norm": 0.8672022223472595,
+      "learning_rate": 4.34481116888696e-06,
+      "loss": 0.3104,
+      "step": 839
+    },
+    {
+      "epoch": 0.9395973154362416,
+      "grad_norm": 0.35544389486312866,
+      "learning_rate": 4.34261295404341e-06,
+      "loss": 0.2271,
+      "step": 840
+    },
+    {
+      "epoch": 0.9407158836689038,
+      "grad_norm": 0.4257761836051941,
+      "learning_rate": 4.340411615860601e-06,
+      "loss": 0.2924,
+      "step": 841
+    },
+    {
+      "epoch": 0.941834451901566,
+      "grad_norm": 0.45104363560676575,
+      "learning_rate": 4.338207158069931e-06,
+      "loss": 0.2772,
+      "step": 842
+    },
+    {
+      "epoch": 0.9429530201342282,
+      "grad_norm": 0.4651937484741211,
+      "learning_rate": 4.33599958440809e-06,
+      "loss": 0.4152,
+      "step": 843
+    },
+    {
+      "epoch": 0.9440715883668904,
+      "grad_norm": 0.39263612031936646,
+      "learning_rate": 4.333788898617046e-06,
+      "loss": 0.324,
+      "step": 844
+    },
+    {
+      "epoch": 0.9451901565995525,
+      "grad_norm": 0.3783542513847351,
+      "learning_rate": 4.331575104444045e-06,
+      "loss": 0.3151,
+      "step": 845
+    },
+    {
+      "epoch": 0.9463087248322147,
+      "grad_norm": 0.4123061001300812,
+      "learning_rate": 4.329358205641598e-06,
+      "loss": 0.373,
+      "step": 846
+    },
+    {
+      "epoch": 0.9474272930648769,
+      "grad_norm": 0.3542521297931671,
+      "learning_rate": 4.327138205967484e-06,
+      "loss": 0.2381,
+      "step": 847
+    },
+    {
+      "epoch": 0.9485458612975392,
+      "grad_norm": 0.44262054562568665,
+      "learning_rate": 4.324915109184734e-06,
+      "loss": 0.2861,
+      "step": 848
+    },
+    {
+      "epoch": 0.9496644295302014,
+      "grad_norm": 0.36385682225227356,
+      "learning_rate": 4.322688919061631e-06,
+      "loss": 0.2571,
+      "step": 849
+    },
+    {
+      "epoch": 0.9507829977628636,
+      "grad_norm": 0.4241195023059845,
+      "learning_rate": 4.3204596393716994e-06,
+      "loss": 0.3664,
+      "step": 850
+    },
+    {
+      "epoch": 0.9507829977628636,
+      "eval_loss": 0.28848111629486084,
+      "eval_runtime": 376.6073,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 850
+    },
+    {
+      "epoch": 0.9519015659955258,
+      "grad_norm": 0.4178703725337982,
+      "learning_rate": 4.3182272738937025e-06,
+      "loss": 0.3462,
+      "step": 851
+    },
+    {
+      "epoch": 0.9530201342281879,
+      "grad_norm": 0.4074319303035736,
+      "learning_rate": 4.315991826411634e-06,
+      "loss": 0.2684,
+      "step": 852
+    },
+    {
+      "epoch": 0.9541387024608501,
+      "grad_norm": 0.4007607400417328,
+      "learning_rate": 4.313753300714709e-06,
+      "loss": 0.3562,
+      "step": 853
+    },
+    {
+      "epoch": 0.9552572706935123,
+      "grad_norm": 0.3619527518749237,
+      "learning_rate": 4.311511700597365e-06,
+      "loss": 0.2882,
+      "step": 854
+    },
+    {
+      "epoch": 0.9563758389261745,
+      "grad_norm": 0.4196773171424866,
+      "learning_rate": 4.3092670298592464e-06,
+      "loss": 0.4282,
+      "step": 855
+    },
+    {
+      "epoch": 0.9574944071588367,
+      "grad_norm": 0.3927515745162964,
+      "learning_rate": 4.307019292305207e-06,
+      "loss": 0.3052,
+      "step": 856
+    },
+    {
+      "epoch": 0.9586129753914989,
+      "grad_norm": 0.3654141426086426,
+      "learning_rate": 4.304768491745293e-06,
+      "loss": 0.2484,
+      "step": 857
+    },
+    {
+      "epoch": 0.959731543624161,
+      "grad_norm": 0.3797949254512787,
+      "learning_rate": 4.30251463199475e-06,
+      "loss": 0.2659,
+      "step": 858
+    },
+    {
+      "epoch": 0.9608501118568232,
+      "grad_norm": 0.40369683504104614,
+      "learning_rate": 4.300257716874001e-06,
+      "loss": 0.3088,
+      "step": 859
+    },
+    {
+      "epoch": 0.9619686800894854,
+      "grad_norm": 0.44334644079208374,
+      "learning_rate": 4.297997750208656e-06,
+      "loss": 0.2987,
+      "step": 860
+    },
+    {
+      "epoch": 0.9630872483221476,
+      "grad_norm": 0.4055168032646179,
+      "learning_rate": 4.295734735829492e-06,
+      "loss": 0.2677,
+      "step": 861
+    },
+    {
+      "epoch": 0.9642058165548099,
+      "grad_norm": 0.3873702883720398,
+      "learning_rate": 4.293468677572454e-06,
+      "loss": 0.2798,
+      "step": 862
+    },
+    {
+      "epoch": 0.9653243847874721,
+      "grad_norm": 0.3605068027973175,
+      "learning_rate": 4.291199579278647e-06,
+      "loss": 0.2373,
+      "step": 863
+    },
+    {
+      "epoch": 0.9664429530201343,
+      "grad_norm": 0.3896702229976654,
+      "learning_rate": 4.288927444794328e-06,
+      "loss": 0.2494,
+      "step": 864
+    },
+    {
+      "epoch": 0.9675615212527964,
+      "grad_norm": 0.5111702084541321,
+      "learning_rate": 4.2866522779709e-06,
+      "loss": 0.3052,
+      "step": 865
+    },
+    {
+      "epoch": 0.9686800894854586,
+      "grad_norm": 0.56321781873703,
+      "learning_rate": 4.284374082664908e-06,
+      "loss": 0.3414,
+      "step": 866
+    },
+    {
+      "epoch": 0.9697986577181208,
+      "grad_norm": 0.44593819975852966,
+      "learning_rate": 4.282092862738029e-06,
+      "loss": 0.3528,
+      "step": 867
+    },
+    {
+      "epoch": 0.970917225950783,
+      "grad_norm": 0.35562431812286377,
+      "learning_rate": 4.279808622057068e-06,
+      "loss": 0.2527,
+      "step": 868
+    },
+    {
+      "epoch": 0.9720357941834452,
+      "grad_norm": 0.43215683102607727,
+      "learning_rate": 4.27752136449395e-06,
+      "loss": 0.2989,
+      "step": 869
+    },
+    {
+      "epoch": 0.9731543624161074,
+      "grad_norm": 0.45298337936401367,
+      "learning_rate": 4.2752310939257115e-06,
+      "loss": 0.2845,
+      "step": 870
+    },
+    {
+      "epoch": 0.9742729306487695,
+      "grad_norm": 0.4539712965488434,
+      "learning_rate": 4.2729378142345e-06,
+      "loss": 0.2829,
+      "step": 871
+    },
+    {
+      "epoch": 0.9753914988814317,
+      "grad_norm": 0.3986677825450897,
+      "learning_rate": 4.270641529307562e-06,
+      "loss": 0.2787,
+      "step": 872
+    },
+    {
+      "epoch": 0.9765100671140939,
+      "grad_norm": 0.363962322473526,
+      "learning_rate": 4.268342243037237e-06,
+      "loss": 0.263,
+      "step": 873
+    },
+    {
+      "epoch": 0.9776286353467561,
+      "grad_norm": 0.40558600425720215,
+      "learning_rate": 4.266039959320952e-06,
+      "loss": 0.2486,
+      "step": 874
+    },
+    {
+      "epoch": 0.9787472035794184,
+      "grad_norm": 0.359153687953949,
+      "learning_rate": 4.263734682061218e-06,
+      "loss": 0.2857,
+      "step": 875
+    },
+    {
+      "epoch": 0.9798657718120806,
+      "grad_norm": 0.3620988726615906,
+      "learning_rate": 4.261426415165617e-06,
+      "loss": 0.2758,
+      "step": 876
+    },
+    {
+      "epoch": 0.9809843400447428,
+      "grad_norm": 1.0297201871871948,
+      "learning_rate": 4.259115162546799e-06,
+      "loss": 0.2824,
+      "step": 877
+    },
+    {
+      "epoch": 0.9821029082774049,
+      "grad_norm": 0.38877785205841064,
+      "learning_rate": 4.2568009281224755e-06,
+      "loss": 0.275,
+      "step": 878
+    },
+    {
+      "epoch": 0.9832214765100671,
+      "grad_norm": 0.49828293919563293,
+      "learning_rate": 4.254483715815413e-06,
+      "loss": 0.4356,
+      "step": 879
+    },
+    {
+      "epoch": 0.9843400447427293,
+      "grad_norm": 0.428007036447525,
+      "learning_rate": 4.252163529553426e-06,
+      "loss": 0.2801,
+      "step": 880
+    },
+    {
+      "epoch": 0.9854586129753915,
+      "grad_norm": 0.40169861912727356,
+      "learning_rate": 4.249840373269367e-06,
+      "loss": 0.2396,
+      "step": 881
+    },
+    {
+      "epoch": 0.9865771812080537,
+      "grad_norm": 0.4257710874080658,
+      "learning_rate": 4.2475142509011255e-06,
+      "loss": 0.2555,
+      "step": 882
+    },
+    {
+      "epoch": 0.9876957494407159,
+      "grad_norm": 0.40037861466407776,
+      "learning_rate": 4.245185166391619e-06,
+      "loss": 0.2656,
+      "step": 883
+    },
+    {
+      "epoch": 0.9888143176733781,
+      "grad_norm": 0.46625956892967224,
+      "learning_rate": 4.242853123688785e-06,
+      "loss": 0.2558,
+      "step": 884
+    },
+    {
+      "epoch": 0.9899328859060402,
+      "grad_norm": 0.3978680372238159,
+      "learning_rate": 4.2405181267455755e-06,
+      "loss": 0.2646,
+      "step": 885
+    },
+    {
+      "epoch": 0.9910514541387024,
+      "grad_norm": 0.4757343828678131,
+      "learning_rate": 4.23818017951995e-06,
+      "loss": 0.4829,
+      "step": 886
+    },
+    {
+      "epoch": 0.9921700223713646,
+      "grad_norm": 0.37190476059913635,
+      "learning_rate": 4.235839285974869e-06,
+      "loss": 0.2468,
+      "step": 887
+    },
+    {
+      "epoch": 0.9932885906040269,
+      "grad_norm": 0.382260799407959,
+      "learning_rate": 4.233495450078286e-06,
+      "loss": 0.2951,
+      "step": 888
+    },
+    {
+      "epoch": 0.9944071588366891,
+      "grad_norm": 0.3903774619102478,
+      "learning_rate": 4.231148675803145e-06,
+      "loss": 0.2312,
+      "step": 889
+    },
+    {
+      "epoch": 0.9955257270693513,
+      "grad_norm": 0.4149327874183655,
+      "learning_rate": 4.228798967127368e-06,
+      "loss": 0.3111,
+      "step": 890
+    },
+    {
+      "epoch": 0.9966442953020134,
+      "grad_norm": 0.39577850699424744,
+      "learning_rate": 4.226446328033851e-06,
+      "loss": 0.2505,
+      "step": 891
+    },
+    {
+      "epoch": 0.9977628635346756,
+      "grad_norm": 0.4171232283115387,
+      "learning_rate": 4.22409076251046e-06,
+      "loss": 0.2816,
+      "step": 892
+    },
+    {
+      "epoch": 0.9988814317673378,
+      "grad_norm": 0.41486355662345886,
+      "learning_rate": 4.221732274550019e-06,
+      "loss": 0.2798,
+      "step": 893
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3876287043094635,
+      "learning_rate": 4.2193708681503056e-06,
+      "loss": 0.3919,
+      "step": 894
+    },
+    {
+      "epoch": 1.0011185682326622,
+      "grad_norm": 0.43928518891334534,
+      "learning_rate": 4.217006547314046e-06,
+      "loss": 0.2534,
+      "step": 895
+    },
+    {
+      "epoch": 1.0022371364653244,
+      "grad_norm": 0.4249671399593353,
+      "learning_rate": 4.214639316048903e-06,
+      "loss": 0.2827,
+      "step": 896
+    },
+    {
+      "epoch": 1.0033557046979866,
+      "grad_norm": 0.3880060911178589,
+      "learning_rate": 4.2122691783674786e-06,
+      "loss": 0.2781,
+      "step": 897
+    },
+    {
+      "epoch": 1.0044742729306488,
+      "grad_norm": 0.38975095748901367,
+      "learning_rate": 4.209896138287296e-06,
+      "loss": 0.2491,
+      "step": 898
+    },
+    {
+      "epoch": 1.005592841163311,
+      "grad_norm": 0.3789180815219879,
+      "learning_rate": 4.2075201998308005e-06,
+      "loss": 0.2677,
+      "step": 899
+    },
+    {
+      "epoch": 1.0067114093959733,
+      "grad_norm": 0.4408995509147644,
+      "learning_rate": 4.205141367025351e-06,
+      "loss": 0.3405,
+      "step": 900
+    },
+    {
+      "epoch": 1.0067114093959733,
+      "eval_loss": 0.28715062141418457,
+      "eval_runtime": 375.9525,
+      "eval_samples_per_second": 2.115,
+      "eval_steps_per_second": 2.115,
+      "step": 900
+    },
+    {
+      "epoch": 1.0078299776286352,
+      "grad_norm": 0.417765349149704,
+      "learning_rate": 4.20275964390321e-06,
+      "loss": 0.3074,
+      "step": 901
+    },
+    {
+      "epoch": 1.0089485458612975,
+      "grad_norm": 0.4237753450870514,
+      "learning_rate": 4.200375034501543e-06,
+      "loss": 0.2163,
+      "step": 902
+    },
+    {
+      "epoch": 1.0100671140939597,
+      "grad_norm": 0.3718244135379791,
+      "learning_rate": 4.197987542862403e-06,
+      "loss": 0.2521,
+      "step": 903
+    },
+    {
+      "epoch": 1.0111856823266219,
+      "grad_norm": 0.4825429916381836,
+      "learning_rate": 4.1955971730327335e-06,
+      "loss": 0.2658,
+      "step": 904
+    },
+    {
+      "epoch": 1.012304250559284,
+      "grad_norm": 0.38464468717575073,
+      "learning_rate": 4.1932039290643534e-06,
+      "loss": 0.2495,
+      "step": 905
+    },
+    {
+      "epoch": 1.0134228187919463,
+      "grad_norm": 0.45652639865875244,
+      "learning_rate": 4.190807815013954e-06,
+      "loss": 0.2642,
+      "step": 906
+    },
+    {
+      "epoch": 1.0145413870246085,
+      "grad_norm": 0.4322049915790558,
+      "learning_rate": 4.188408834943093e-06,
+      "loss": 0.2949,
+      "step": 907
+    },
+    {
+      "epoch": 1.0156599552572707,
+      "grad_norm": 0.46952658891677856,
+      "learning_rate": 4.186006992918185e-06,
+      "loss": 0.3128,
+      "step": 908
+    },
+    {
+      "epoch": 1.016778523489933,
+      "grad_norm": 0.35209888219833374,
+      "learning_rate": 4.1836022930104945e-06,
+      "loss": 0.2761,
+      "step": 909
+    },
+    {
+      "epoch": 1.0178970917225951,
+      "grad_norm": 0.35308632254600525,
+      "learning_rate": 4.181194739296135e-06,
+      "loss": 0.2381,
+      "step": 910
+    },
+    {
+      "epoch": 1.0190156599552573,
+      "grad_norm": 0.4179858863353729,
+      "learning_rate": 4.1787843358560515e-06,
+      "loss": 0.2602,
+      "step": 911
+    },
+    {
+      "epoch": 1.0201342281879195,
+      "grad_norm": 0.4116358160972595,
+      "learning_rate": 4.1763710867760235e-06,
+      "loss": 0.3425,
+      "step": 912
+    },
+    {
+      "epoch": 1.0212527964205818,
+      "grad_norm": 0.5124662518501282,
+      "learning_rate": 4.173954996146651e-06,
+      "loss": 0.3168,
+      "step": 913
+    },
+    {
+      "epoch": 1.0223713646532437,
+      "grad_norm": 0.42902371287345886,
+      "learning_rate": 4.1715360680633544e-06,
+      "loss": 0.2741,
+      "step": 914
+    },
+    {
+      "epoch": 1.023489932885906,
+      "grad_norm": 0.4505557119846344,
+      "learning_rate": 4.169114306626359e-06,
+      "loss": 0.3128,
+      "step": 915
+    },
+    {
+      "epoch": 1.0246085011185682,
+      "grad_norm": 0.4314258098602295,
+      "learning_rate": 4.166689715940699e-06,
+      "loss": 0.2391,
+      "step": 916
+    },
+    {
+      "epoch": 1.0257270693512304,
+      "grad_norm": 0.3705413043498993,
+      "learning_rate": 4.164262300116198e-06,
+      "loss": 0.2254,
+      "step": 917
+    },
+    {
+      "epoch": 1.0268456375838926,
+      "grad_norm": 0.412853479385376,
+      "learning_rate": 4.1618320632674715e-06,
+      "loss": 0.3066,
+      "step": 918
+    },
+    {
+      "epoch": 1.0279642058165548,
+      "grad_norm": 0.4166383743286133,
+      "learning_rate": 4.1593990095139165e-06,
+      "loss": 0.2943,
+      "step": 919
+    },
+    {
+      "epoch": 1.029082774049217,
+      "grad_norm": 0.4185318648815155,
+      "learning_rate": 4.156963142979706e-06,
+      "loss": 0.3954,
+      "step": 920
+    },
+    {
+      "epoch": 1.0302013422818792,
+      "grad_norm": 0.38008731603622437,
+      "learning_rate": 4.154524467793778e-06,
+      "loss": 0.3512,
+      "step": 921
+    },
+    {
+      "epoch": 1.0313199105145414,
+      "grad_norm": 0.48360851407051086,
+      "learning_rate": 4.152082988089834e-06,
+      "loss": 0.2704,
+      "step": 922
+    },
+    {
+      "epoch": 1.0324384787472036,
+      "grad_norm": 0.4031798243522644,
+      "learning_rate": 4.149638708006327e-06,
+      "loss": 0.2271,
+      "step": 923
+    },
+    {
+      "epoch": 1.0335570469798658,
+      "grad_norm": 0.4133947193622589,
+      "learning_rate": 4.147191631686459e-06,
+      "loss": 0.3012,
+      "step": 924
+    },
+    {
+      "epoch": 1.034675615212528,
+      "grad_norm": 0.36920469999313354,
+      "learning_rate": 4.144741763278172e-06,
+      "loss": 0.2366,
+      "step": 925
+    },
+    {
+      "epoch": 1.0357941834451903,
+      "grad_norm": 0.4654488265514374,
+      "learning_rate": 4.142289106934137e-06,
+      "loss": 0.3571,
+      "step": 926
+    },
+    {
+      "epoch": 1.0369127516778525,
+      "grad_norm": 0.40608105063438416,
+      "learning_rate": 4.139833666811755e-06,
+      "loss": 0.2592,
+      "step": 927
+    },
+    {
+      "epoch": 1.0380313199105144,
+      "grad_norm": 0.4193144142627716,
+      "learning_rate": 4.137375447073144e-06,
+      "loss": 0.2731,
+      "step": 928
+    },
+    {
+      "epoch": 1.0391498881431767,
+      "grad_norm": 0.38587087392807007,
+      "learning_rate": 4.134914451885133e-06,
+      "loss": 0.3534,
+      "step": 929
+    },
+    {
+      "epoch": 1.0402684563758389,
+      "grad_norm": 0.40791717171669006,
+      "learning_rate": 4.132450685419257e-06,
+      "loss": 0.3406,
+      "step": 930
+    },
+    {
+      "epoch": 1.041387024608501,
+      "grad_norm": 0.4273252785205841,
+      "learning_rate": 4.129984151851747e-06,
+      "loss": 0.2476,
+      "step": 931
+    },
+    {
+      "epoch": 1.0425055928411633,
+      "grad_norm": 0.41116195917129517,
+      "learning_rate": 4.127514855363524e-06,
+      "loss": 0.3042,
+      "step": 932
+    },
+    {
+      "epoch": 1.0436241610738255,
+      "grad_norm": 0.39274775981903076,
+      "learning_rate": 4.125042800140195e-06,
+      "loss": 0.2693,
+      "step": 933
+    },
+    {
+      "epoch": 1.0447427293064877,
+      "grad_norm": 0.4345043897628784,
+      "learning_rate": 4.122567990372041e-06,
+      "loss": 0.2833,
+      "step": 934
+    },
+    {
+      "epoch": 1.04586129753915,
+      "grad_norm": 0.37664178013801575,
+      "learning_rate": 4.120090430254013e-06,
+      "loss": 0.2499,
+      "step": 935
+    },
+    {
+      "epoch": 1.0469798657718121,
+      "grad_norm": 0.374865859746933,
+      "learning_rate": 4.117610123985724e-06,
+      "loss": 0.2158,
+      "step": 936
+    },
+    {
+      "epoch": 1.0480984340044743,
+      "grad_norm": 0.5587493181228638,
+      "learning_rate": 4.115127075771441e-06,
+      "loss": 0.2853,
+      "step": 937
+    },
+    {
+      "epoch": 1.0492170022371365,
+      "grad_norm": 0.3815159797668457,
+      "learning_rate": 4.112641289820078e-06,
+      "loss": 0.1987,
+      "step": 938
+    },
+    {
+      "epoch": 1.0503355704697988,
+      "grad_norm": 0.38109636306762695,
+      "learning_rate": 4.110152770345194e-06,
+      "loss": 0.2166,
+      "step": 939
+    },
+    {
+      "epoch": 1.0514541387024607,
+      "grad_norm": 0.468150794506073,
+      "learning_rate": 4.107661521564977e-06,
+      "loss": 0.301,
+      "step": 940
+    },
+    {
+      "epoch": 1.052572706935123,
+      "grad_norm": 0.44505491852760315,
+      "learning_rate": 4.105167547702242e-06,
+      "loss": 0.2883,
+      "step": 941
+    },
+    {
+      "epoch": 1.0536912751677852,
+      "grad_norm": 0.4302145540714264,
+      "learning_rate": 4.102670852984425e-06,
+      "loss": 0.2679,
+      "step": 942
+    },
+    {
+      "epoch": 1.0548098434004474,
+      "grad_norm": 0.4834899604320526,
+      "learning_rate": 4.1001714416435725e-06,
+      "loss": 0.3097,
+      "step": 943
+    },
+    {
+      "epoch": 1.0559284116331096,
+      "grad_norm": 0.4495421051979065,
+      "learning_rate": 4.0976693179163365e-06,
+      "loss": 0.3438,
+      "step": 944
+    },
+    {
+      "epoch": 1.0570469798657718,
+      "grad_norm": 0.5022640824317932,
+      "learning_rate": 4.095164486043967e-06,
+      "loss": 0.3253,
+      "step": 945
+    },
+    {
+      "epoch": 1.058165548098434,
+      "grad_norm": 0.4175395965576172,
+      "learning_rate": 4.092656950272304e-06,
+      "loss": 0.2956,
+      "step": 946
+    },
+    {
+      "epoch": 1.0592841163310962,
+      "grad_norm": 0.39942750334739685,
+      "learning_rate": 4.090146714851769e-06,
+      "loss": 0.2594,
+      "step": 947
+    },
+    {
+      "epoch": 1.0604026845637584,
+      "grad_norm": 0.3620707392692566,
+      "learning_rate": 4.087633784037364e-06,
+      "loss": 0.2385,
+      "step": 948
+    },
+    {
+      "epoch": 1.0615212527964206,
+      "grad_norm": 0.39491572976112366,
+      "learning_rate": 4.085118162088655e-06,
+      "loss": 0.2588,
+      "step": 949
+    },
+    {
+      "epoch": 1.0626398210290828,
+      "grad_norm": 0.410480797290802,
+      "learning_rate": 4.082599853269772e-06,
+      "loss": 0.2872,
+      "step": 950
+    },
+    {
+      "epoch": 1.0626398210290828,
+      "eval_loss": 0.28568652272224426,
+      "eval_runtime": 376.3199,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 2.113,
+      "step": 950
+    },
+    {
+      "epoch": 1.063758389261745,
+      "grad_norm": 0.5928165912628174,
+      "learning_rate": 4.080078861849401e-06,
+      "loss": 0.2618,
+      "step": 951
+    },
+    {
+      "epoch": 1.0648769574944073,
+      "grad_norm": 0.4022570848464966,
+      "learning_rate": 4.077555192100771e-06,
+      "loss": 0.3013,
+      "step": 952
+    },
+    {
+      "epoch": 1.0659955257270695,
+      "grad_norm": 0.3846629858016968,
+      "learning_rate": 4.075028848301655e-06,
+      "loss": 0.2504,
+      "step": 953
+    },
+    {
+      "epoch": 1.0671140939597314,
+      "grad_norm": 0.3668595254421234,
+      "learning_rate": 4.072499834734357e-06,
+      "loss": 0.2919,
+      "step": 954
+    },
+    {
+      "epoch": 1.0682326621923937,
+      "grad_norm": 0.31809625029563904,
+      "learning_rate": 4.069968155685705e-06,
+      "loss": 0.2179,
+      "step": 955
+    },
+    {
+      "epoch": 1.0693512304250559,
+      "grad_norm": 0.4391195774078369,
+      "learning_rate": 4.067433815447047e-06,
+      "loss": 0.4113,
+      "step": 956
+    },
+    {
+      "epoch": 1.070469798657718,
+      "grad_norm": 0.4273832142353058,
+      "learning_rate": 4.064896818314242e-06,
+      "loss": 0.4426,
+      "step": 957
+    },
+    {
+      "epoch": 1.0715883668903803,
+      "grad_norm": 0.3945104777812958,
+      "learning_rate": 4.062357168587652e-06,
+      "loss": 0.282,
+      "step": 958
+    },
+    {
+      "epoch": 1.0727069351230425,
+      "grad_norm": 0.38819149136543274,
+      "learning_rate": 4.059814870572135e-06,
+      "loss": 0.2993,
+      "step": 959
+    },
+    {
+      "epoch": 1.0738255033557047,
+      "grad_norm": 0.41810351610183716,
+      "learning_rate": 4.057269928577038e-06,
+      "loss": 0.3307,
+      "step": 960
+    },
+    {
+      "epoch": 1.074944071588367,
+      "grad_norm": 0.3493334650993347,
+      "learning_rate": 4.054722346916191e-06,
+      "loss": 0.2358,
+      "step": 961
+    },
+    {
+      "epoch": 1.0760626398210291,
+      "grad_norm": 0.4229319393634796,
+      "learning_rate": 4.052172129907897e-06,
+      "loss": 0.3323,
+      "step": 962
+    },
+    {
+      "epoch": 1.0771812080536913,
+      "grad_norm": 0.4402555227279663,
+      "learning_rate": 4.049619281874926e-06,
+      "loss": 0.3527,
+      "step": 963
+    },
+    {
+      "epoch": 1.0782997762863535,
+      "grad_norm": 0.35863906145095825,
+      "learning_rate": 4.047063807144508e-06,
+      "loss": 0.2343,
+      "step": 964
+    },
+    {
+      "epoch": 1.0794183445190157,
+      "grad_norm": 0.40167996287345886,
+      "learning_rate": 4.044505710048327e-06,
+      "loss": 0.2631,
+      "step": 965
+    },
+    {
+      "epoch": 1.0805369127516777,
+      "grad_norm": 0.4514169692993164,
+      "learning_rate": 4.04194499492251e-06,
+      "loss": 0.2685,
+      "step": 966
+    },
+    {
+      "epoch": 1.08165548098434,
+      "grad_norm": 0.40112167596817017,
+      "learning_rate": 4.0393816661076215e-06,
+      "loss": 0.3761,
+      "step": 967
+    },
+    {
+      "epoch": 1.0827740492170022,
+      "grad_norm": 0.44660860300064087,
+      "learning_rate": 4.036815727948658e-06,
+      "loss": 0.3012,
+      "step": 968
+    },
+    {
+      "epoch": 1.0838926174496644,
+      "grad_norm": 0.4591975212097168,
+      "learning_rate": 4.034247184795039e-06,
+      "loss": 0.3133,
+      "step": 969
+    },
+    {
+      "epoch": 1.0850111856823266,
+      "grad_norm": 0.44080254435539246,
+      "learning_rate": 4.0316760410005985e-06,
+      "loss": 0.2504,
+      "step": 970
+    },
+    {
+      "epoch": 1.0861297539149888,
+      "grad_norm": 0.40203526616096497,
+      "learning_rate": 4.029102300923579e-06,
+      "loss": 0.2377,
+      "step": 971
+    },
+    {
+      "epoch": 1.087248322147651,
+      "grad_norm": 0.4447600841522217,
+      "learning_rate": 4.026525968926624e-06,
+      "loss": 0.2829,
+      "step": 972
+    },
+    {
+      "epoch": 1.0883668903803132,
+      "grad_norm": 0.4434783458709717,
+      "learning_rate": 4.02394704937677e-06,
+      "loss": 0.27,
+      "step": 973
+    },
+    {
+      "epoch": 1.0894854586129754,
+      "grad_norm": 0.45320531725883484,
+      "learning_rate": 4.02136554664544e-06,
+      "loss": 0.4428,
+      "step": 974
+    },
+    {
+      "epoch": 1.0906040268456376,
+      "grad_norm": 0.39350205659866333,
+      "learning_rate": 4.0187814651084375e-06,
+      "loss": 0.2443,
+      "step": 975
+    },
+    {
+      "epoch": 1.0917225950782998,
+      "grad_norm": 0.46188971400260925,
+      "learning_rate": 4.016194809145933e-06,
+      "loss": 0.2894,
+      "step": 976
+    },
+    {
+      "epoch": 1.092841163310962,
+      "grad_norm": 0.4228478968143463,
+      "learning_rate": 4.013605583142464e-06,
+      "loss": 0.2566,
+      "step": 977
+    },
+    {
+      "epoch": 1.0939597315436242,
+      "grad_norm": 0.4343189597129822,
+      "learning_rate": 4.0110137914869245e-06,
+      "loss": 0.3219,
+      "step": 978
+    },
+    {
+      "epoch": 1.0950782997762865,
+      "grad_norm": 0.5176400542259216,
+      "learning_rate": 4.008419438572555e-06,
+      "loss": 0.2774,
+      "step": 979
+    },
+    {
+      "epoch": 1.0961968680089484,
+      "grad_norm": 0.4388771653175354,
+      "learning_rate": 4.005822528796941e-06,
+      "loss": 0.265,
+      "step": 980
+    },
+    {
+      "epoch": 1.0973154362416107,
+      "grad_norm": 0.46949753165245056,
+      "learning_rate": 4.0032230665619976e-06,
+      "loss": 0.2451,
+      "step": 981
+    },
+    {
+      "epoch": 1.0984340044742729,
+      "grad_norm": 0.40662992000579834,
+      "learning_rate": 4.0006210562739715e-06,
+      "loss": 0.2446,
+      "step": 982
+    },
+    {
+      "epoch": 1.099552572706935,
+      "grad_norm": 0.42176052927970886,
+      "learning_rate": 3.998016502343422e-06,
+      "loss": 0.2524,
+      "step": 983
+    },
+    {
+      "epoch": 1.1006711409395973,
+      "grad_norm": 0.45420581102371216,
+      "learning_rate": 3.995409409185228e-06,
+      "loss": 0.2404,
+      "step": 984
+    },
+    {
+      "epoch": 1.1017897091722595,
+      "grad_norm": 0.38297194242477417,
+      "learning_rate": 3.992799781218567e-06,
+      "loss": 0.2882,
+      "step": 985
+    },
+    {
+      "epoch": 1.1029082774049217,
+      "grad_norm": 0.40031781792640686,
+      "learning_rate": 3.990187622866914e-06,
+      "loss": 0.2788,
+      "step": 986
+    },
+    {
+      "epoch": 1.104026845637584,
+      "grad_norm": 0.42089980840682983,
+      "learning_rate": 3.987572938558035e-06,
+      "loss": 0.2737,
+      "step": 987
+    },
+    {
+      "epoch": 1.1051454138702461,
+      "grad_norm": 0.4270496666431427,
+      "learning_rate": 3.984955732723975e-06,
+      "loss": 0.3267,
+      "step": 988
+    },
+    {
+      "epoch": 1.1062639821029083,
+      "grad_norm": 0.4366529583930969,
+      "learning_rate": 3.9823360098010565e-06,
+      "loss": 0.3141,
+      "step": 989
+    },
+    {
+      "epoch": 1.1073825503355705,
+      "grad_norm": 0.398361474275589,
+      "learning_rate": 3.979713774229864e-06,
+      "loss": 0.3189,
+      "step": 990
+    },
+    {
+      "epoch": 1.1085011185682327,
+      "grad_norm": 0.3778422474861145,
+      "learning_rate": 3.977089030455245e-06,
+      "loss": 0.2761,
+      "step": 991
+    },
+    {
+      "epoch": 1.109619686800895,
+      "grad_norm": 0.42596372961997986,
+      "learning_rate": 3.974461782926299e-06,
+      "loss": 0.2691,
+      "step": 992
+    },
+    {
+      "epoch": 1.110738255033557,
+      "grad_norm": 0.4279325306415558,
+      "learning_rate": 3.971832036096366e-06,
+      "loss": 0.2889,
+      "step": 993
+    },
+    {
+      "epoch": 1.1118568232662192,
+      "grad_norm": 0.5149924755096436,
+      "learning_rate": 3.969199794423024e-06,
+      "loss": 0.226,
+      "step": 994
+    },
+    {
+      "epoch": 1.1129753914988814,
+      "grad_norm": 0.47925782203674316,
+      "learning_rate": 3.9665650623680815e-06,
+      "loss": 0.3271,
+      "step": 995
+    },
+    {
+      "epoch": 1.1140939597315436,
+      "grad_norm": 0.3997594118118286,
+      "learning_rate": 3.963927844397566e-06,
+      "loss": 0.2284,
+      "step": 996
+    },
+    {
+      "epoch": 1.1152125279642058,
+      "grad_norm": 0.39924249053001404,
+      "learning_rate": 3.96128814498172e-06,
+      "loss": 0.3044,
+      "step": 997
+    },
+    {
+      "epoch": 1.116331096196868,
+      "grad_norm": 0.37642771005630493,
+      "learning_rate": 3.958645968594993e-06,
+      "loss": 0.267,
+      "step": 998
+    },
+    {
+      "epoch": 1.1174496644295302,
+      "grad_norm": 0.40999355912208557,
+      "learning_rate": 3.956001319716031e-06,
+      "loss": 0.3537,
+      "step": 999
+    },
+    {
+      "epoch": 1.1185682326621924,
+      "grad_norm": 0.5428104996681213,
+      "learning_rate": 3.953354202827673e-06,
+      "loss": 0.2921,
+      "step": 1000
+    },
+    {
+      "epoch": 1.1185682326621924,
+      "eval_loss": 0.2841748595237732,
+      "eval_runtime": 376.5128,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 1000
+    },
+    {
+      "epoch": 1.1196868008948546,
+      "grad_norm": 0.4338473975658417,
+      "learning_rate": 3.9507046224169394e-06,
+      "loss": 0.3259,
+      "step": 1001
+    },
+    {
+      "epoch": 1.1208053691275168,
+      "grad_norm": 0.39090994000434875,
+      "learning_rate": 3.9480525829750295e-06,
+      "loss": 0.2635,
+      "step": 1002
+    },
+    {
+      "epoch": 1.121923937360179,
+      "grad_norm": 0.4456063508987427,
+      "learning_rate": 3.945398088997306e-06,
+      "loss": 0.3348,
+      "step": 1003
+    },
+    {
+      "epoch": 1.1230425055928412,
+      "grad_norm": 0.4318437874317169,
+      "learning_rate": 3.942741144983298e-06,
+      "loss": 0.2593,
+      "step": 1004
+    },
+    {
+      "epoch": 1.1241610738255035,
+      "grad_norm": 0.4170302748680115,
+      "learning_rate": 3.940081755436685e-06,
+      "loss": 0.2704,
+      "step": 1005
+    },
+    {
+      "epoch": 1.1252796420581657,
+      "grad_norm": 0.45423558354377747,
+      "learning_rate": 3.937419924865289e-06,
+      "loss": 0.3639,
+      "step": 1006
+    },
+    {
+      "epoch": 1.1263982102908277,
+      "grad_norm": 0.4083240330219269,
+      "learning_rate": 3.934755657781072e-06,
+      "loss": 0.2648,
+      "step": 1007
+    },
+    {
+      "epoch": 1.1275167785234899,
+      "grad_norm": 0.39069411158561707,
+      "learning_rate": 3.932088958700129e-06,
+      "loss": 0.2433,
+      "step": 1008
+    },
+    {
+      "epoch": 1.128635346756152,
+      "grad_norm": 0.45195502042770386,
+      "learning_rate": 3.929419832142672e-06,
+      "loss": 0.3165,
+      "step": 1009
+    },
+    {
+      "epoch": 1.1297539149888143,
+      "grad_norm": 1.4624671936035156,
+      "learning_rate": 3.926748282633031e-06,
+      "loss": 0.2829,
+      "step": 1010
+    },
+    {
+      "epoch": 1.1308724832214765,
+      "grad_norm": 0.4014538526535034,
+      "learning_rate": 3.924074314699643e-06,
+      "loss": 0.3192,
+      "step": 1011
+    },
+    {
+      "epoch": 1.1319910514541387,
+      "grad_norm": 0.4161325991153717,
+      "learning_rate": 3.921397932875044e-06,
+      "loss": 0.2687,
+      "step": 1012
+    },
+    {
+      "epoch": 1.133109619686801,
+      "grad_norm": 0.44637060165405273,
+      "learning_rate": 3.91871914169586e-06,
+      "loss": 0.2731,
+      "step": 1013
+    },
+    {
+      "epoch": 1.1342281879194631,
+      "grad_norm": 0.4088190495967865,
+      "learning_rate": 3.916037945702803e-06,
+      "loss": 0.2479,
+      "step": 1014
+    },
+    {
+      "epoch": 1.1353467561521253,
+      "grad_norm": 0.4205600619316101,
+      "learning_rate": 3.913354349440661e-06,
+      "loss": 0.3115,
+      "step": 1015
+    },
+    {
+      "epoch": 1.1364653243847875,
+      "grad_norm": 0.4196604788303375,
+      "learning_rate": 3.9106683574582905e-06,
+      "loss": 0.2572,
+      "step": 1016
+    },
+    {
+      "epoch": 1.1375838926174497,
+      "grad_norm": 0.4075378179550171,
+      "learning_rate": 3.907979974308608e-06,
+      "loss": 0.2933,
+      "step": 1017
+    },
+    {
+      "epoch": 1.138702460850112,
+      "grad_norm": 0.4575662612915039,
+      "learning_rate": 3.905289204548585e-06,
+      "loss": 0.2622,
+      "step": 1018
+    },
+    {
+      "epoch": 1.139821029082774,
+      "grad_norm": 0.4366585910320282,
+      "learning_rate": 3.9025960527392355e-06,
+      "loss": 0.3377,
+      "step": 1019
+    },
+    {
+      "epoch": 1.1409395973154361,
+      "grad_norm": 0.4455721080303192,
+      "learning_rate": 3.899900523445615e-06,
+      "loss": 0.2732,
+      "step": 1020
+    },
+    {
+      "epoch": 1.1420581655480984,
+      "grad_norm": 0.47631779313087463,
+      "learning_rate": 3.897202621236806e-06,
+      "loss": 0.2309,
+      "step": 1021
+    },
+    {
+      "epoch": 1.1431767337807606,
+      "grad_norm": 0.3928377330303192,
+      "learning_rate": 3.894502350685913e-06,
+      "loss": 0.2815,
+      "step": 1022
+    },
+    {
+      "epoch": 1.1442953020134228,
+      "grad_norm": 0.40738141536712646,
+      "learning_rate": 3.8917997163700585e-06,
+      "loss": 0.286,
+      "step": 1023
+    },
+    {
+      "epoch": 1.145413870246085,
+      "grad_norm": 0.39984479546546936,
+      "learning_rate": 3.8890947228703686e-06,
+      "loss": 0.2575,
+      "step": 1024
+    },
+    {
+      "epoch": 1.1465324384787472,
+      "grad_norm": 0.3914332985877991,
+      "learning_rate": 3.88638737477197e-06,
+      "loss": 0.3588,
+      "step": 1025
+    },
+    {
+      "epoch": 1.1476510067114094,
+      "grad_norm": 0.41043785214424133,
+      "learning_rate": 3.883677676663977e-06,
+      "loss": 0.2658,
+      "step": 1026
+    },
+    {
+      "epoch": 1.1487695749440716,
+      "grad_norm": 0.46859249472618103,
+      "learning_rate": 3.880965633139493e-06,
+      "loss": 0.2093,
+      "step": 1027
+    },
+    {
+      "epoch": 1.1498881431767338,
+      "grad_norm": 0.40038394927978516,
+      "learning_rate": 3.878251248795593e-06,
+      "loss": 0.2604,
+      "step": 1028
+    },
+    {
+      "epoch": 1.151006711409396,
+      "grad_norm": 0.4512845277786255,
+      "learning_rate": 3.8755345282333215e-06,
+      "loss": 0.4159,
+      "step": 1029
+    },
+    {
+      "epoch": 1.1521252796420582,
+      "grad_norm": 0.4130917489528656,
+      "learning_rate": 3.872815476057682e-06,
+      "loss": 0.3089,
+      "step": 1030
+    },
+    {
+      "epoch": 1.1532438478747205,
+      "grad_norm": 0.47094887495040894,
+      "learning_rate": 3.8700940968776305e-06,
+      "loss": 0.3756,
+      "step": 1031
+    },
+    {
+      "epoch": 1.1543624161073827,
+      "grad_norm": 0.3684062957763672,
+      "learning_rate": 3.8673703953060685e-06,
+      "loss": 0.2441,
+      "step": 1032
+    },
+    {
+      "epoch": 1.1554809843400446,
+      "grad_norm": 0.5060502886772156,
+      "learning_rate": 3.8646443759598314e-06,
+      "loss": 0.2842,
+      "step": 1033
+    },
+    {
+      "epoch": 1.1565995525727069,
+      "grad_norm": 0.43973463773727417,
+      "learning_rate": 3.8619160434596866e-06,
+      "loss": 0.271,
+      "step": 1034
+    },
+    {
+      "epoch": 1.157718120805369,
+      "grad_norm": 0.46230003237724304,
+      "learning_rate": 3.859185402430322e-06,
+      "loss": 0.4077,
+      "step": 1035
+    },
+    {
+      "epoch": 1.1588366890380313,
+      "grad_norm": 0.43587398529052734,
+      "learning_rate": 3.856452457500336e-06,
+      "loss": 0.2675,
+      "step": 1036
+    },
+    {
+      "epoch": 1.1599552572706935,
+      "grad_norm": 0.41206982731819153,
+      "learning_rate": 3.853717213302234e-06,
+      "loss": 0.2413,
+      "step": 1037
+    },
+    {
+      "epoch": 1.1610738255033557,
+      "grad_norm": 0.4037029445171356,
+      "learning_rate": 3.850979674472418e-06,
+      "loss": 0.24,
+      "step": 1038
+    },
+    {
+      "epoch": 1.162192393736018,
+      "grad_norm": 0.3965253531932831,
+      "learning_rate": 3.848239845651182e-06,
+      "loss": 0.2597,
+      "step": 1039
+    },
+    {
+      "epoch": 1.1633109619686801,
+      "grad_norm": 0.40698757767677307,
+      "learning_rate": 3.8454977314827e-06,
+      "loss": 0.2638,
+      "step": 1040
+    },
+    {
+      "epoch": 1.1644295302013423,
+      "grad_norm": 0.4005014896392822,
+      "learning_rate": 3.842753336615017e-06,
+      "loss": 0.2411,
+      "step": 1041
+    },
+    {
+      "epoch": 1.1655480984340045,
+      "grad_norm": 0.4001332223415375,
+      "learning_rate": 3.840006665700049e-06,
+      "loss": 0.3343,
+      "step": 1042
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 0.41159456968307495,
+      "learning_rate": 3.837257723393567e-06,
+      "loss": 0.2826,
+      "step": 1043
+    },
+    {
+      "epoch": 1.167785234899329,
+      "grad_norm": 0.4403782784938812,
+      "learning_rate": 3.834506514355191e-06,
+      "loss": 0.2702,
+      "step": 1044
+    },
+    {
+      "epoch": 1.168903803131991,
+      "grad_norm": 0.4293852746486664,
+      "learning_rate": 3.831753043248389e-06,
+      "loss": 0.2544,
+      "step": 1045
+    },
+    {
+      "epoch": 1.1700223713646531,
+      "grad_norm": 0.4977855384349823,
+      "learning_rate": 3.828997314740456e-06,
+      "loss": 0.2181,
+      "step": 1046
+    },
+    {
+      "epoch": 1.1711409395973154,
+      "grad_norm": 0.43000197410583496,
+      "learning_rate": 3.826239333502518e-06,
+      "loss": 0.2655,
+      "step": 1047
+    },
+    {
+      "epoch": 1.1722595078299776,
+      "grad_norm": 0.49466222524642944,
+      "learning_rate": 3.823479104209517e-06,
+      "loss": 0.2804,
+      "step": 1048
+    },
+    {
+      "epoch": 1.1733780760626398,
+      "grad_norm": 0.4619726538658142,
+      "learning_rate": 3.820716631540209e-06,
+      "loss": 0.3081,
+      "step": 1049
+    },
+    {
+      "epoch": 1.174496644295302,
+      "grad_norm": 0.43511128425598145,
+      "learning_rate": 3.81795192017715e-06,
+      "loss": 0.2664,
+      "step": 1050
+    },
+    {
+      "epoch": 1.174496644295302,
+      "eval_loss": 0.2830018699169159,
+      "eval_runtime": 376.5304,
+      "eval_samples_per_second": 2.111,
+      "eval_steps_per_second": 2.111,
+      "step": 1050
+    },
+    {
+      "epoch": 1.1756152125279642,
+      "grad_norm": 0.4436204731464386,
+      "learning_rate": 3.81518497480669e-06,
+      "loss": 0.248,
+      "step": 1051
+    },
+    {
+      "epoch": 1.1767337807606264,
+      "grad_norm": 0.4732202887535095,
+      "learning_rate": 3.812415800118968e-06,
+      "loss": 0.2696,
+      "step": 1052
+    },
+    {
+      "epoch": 1.1778523489932886,
+      "grad_norm": 0.5298107266426086,
+      "learning_rate": 3.8096444008079005e-06,
+      "loss": 0.3134,
+      "step": 1053
+    },
+    {
+      "epoch": 1.1789709172259508,
+      "grad_norm": 0.46554410457611084,
+      "learning_rate": 3.806870781571175e-06,
+      "loss": 0.2705,
+      "step": 1054
+    },
+    {
+      "epoch": 1.180089485458613,
+      "grad_norm": 0.5077680945396423,
+      "learning_rate": 3.8040949471102427e-06,
+      "loss": 0.3107,
+      "step": 1055
+    },
+    {
+      "epoch": 1.1812080536912752,
+      "grad_norm": 0.4585459530353546,
+      "learning_rate": 3.8013169021303097e-06,
+      "loss": 0.2435,
+      "step": 1056
+    },
+    {
+      "epoch": 1.1823266219239374,
+      "grad_norm": 0.37901824712753296,
+      "learning_rate": 3.798536651340328e-06,
+      "loss": 0.2311,
+      "step": 1057
+    },
+    {
+      "epoch": 1.1834451901565997,
+      "grad_norm": 0.451414555311203,
+      "learning_rate": 3.795754199452988e-06,
+      "loss": 0.3908,
+      "step": 1058
+    },
+    {
+      "epoch": 1.1845637583892619,
+      "grad_norm": 0.4406803250312805,
+      "learning_rate": 3.792969551184713e-06,
+      "loss": 0.2512,
+      "step": 1059
+    },
+    {
+      "epoch": 1.1856823266219239,
+      "grad_norm": 0.5260480046272278,
+      "learning_rate": 3.7901827112556496e-06,
+      "loss": 0.3247,
+      "step": 1060
+    },
+    {
+      "epoch": 1.186800894854586,
+      "grad_norm": 0.38094276189804077,
+      "learning_rate": 3.787393684389657e-06,
+      "loss": 0.2269,
+      "step": 1061
+    },
+    {
+      "epoch": 1.1879194630872483,
+      "grad_norm": 0.4076838791370392,
+      "learning_rate": 3.784602475314303e-06,
+      "loss": 0.2758,
+      "step": 1062
+    },
+    {
+      "epoch": 1.1890380313199105,
+      "grad_norm": 0.6140595078468323,
+      "learning_rate": 3.781809088760855e-06,
+      "loss": 0.2499,
+      "step": 1063
+    },
+    {
+      "epoch": 1.1901565995525727,
+      "grad_norm": 0.4293782711029053,
+      "learning_rate": 3.779013529464268e-06,
+      "loss": 0.2705,
+      "step": 1064
+    },
+    {
+      "epoch": 1.191275167785235,
+      "grad_norm": 0.5008118152618408,
+      "learning_rate": 3.7762158021631863e-06,
+      "loss": 0.3425,
+      "step": 1065
+    },
+    {
+      "epoch": 1.192393736017897,
+      "grad_norm": 0.4591560661792755,
+      "learning_rate": 3.773415911599921e-06,
+      "loss": 0.302,
+      "step": 1066
+    },
+    {
+      "epoch": 1.1935123042505593,
+      "grad_norm": 0.4141213297843933,
+      "learning_rate": 3.770613862520457e-06,
+      "loss": 0.236,
+      "step": 1067
+    },
+    {
+      "epoch": 1.1946308724832215,
+      "grad_norm": 0.5093594789505005,
+      "learning_rate": 3.7678096596744328e-06,
+      "loss": 0.3498,
+      "step": 1068
+    },
+    {
+      "epoch": 1.1957494407158837,
+      "grad_norm": 0.5092828869819641,
+      "learning_rate": 3.7650033078151428e-06,
+      "loss": 0.2897,
+      "step": 1069
+    },
+    {
+      "epoch": 1.196868008948546,
+      "grad_norm": 0.4343785345554352,
+      "learning_rate": 3.76219481169952e-06,
+      "loss": 0.234,
+      "step": 1070
+    },
+    {
+      "epoch": 1.197986577181208,
+      "grad_norm": 0.42848721146583557,
+      "learning_rate": 3.7593841760881335e-06,
+      "loss": 0.2315,
+      "step": 1071
+    },
+    {
+      "epoch": 1.1991051454138701,
+      "grad_norm": 0.3831914961338043,
+      "learning_rate": 3.7565714057451786e-06,
+      "loss": 0.2312,
+      "step": 1072
+    },
+    {
+      "epoch": 1.2002237136465324,
+      "grad_norm": 0.38169366121292114,
+      "learning_rate": 3.75375650543847e-06,
+      "loss": 0.246,
+      "step": 1073
+    },
+    {
+      "epoch": 1.2013422818791946,
+      "grad_norm": 0.3888981342315674,
+      "learning_rate": 3.750939479939432e-06,
+      "loss": 0.2352,
+      "step": 1074
+    },
+    {
+      "epoch": 1.2024608501118568,
+      "grad_norm": 0.4558098316192627,
+      "learning_rate": 3.7481203340230914e-06,
+      "loss": 0.2914,
+      "step": 1075
+    },
+    {
+      "epoch": 1.203579418344519,
+      "grad_norm": 0.46701762080192566,
+      "learning_rate": 3.74529907246807e-06,
+      "loss": 0.2538,
+      "step": 1076
+    },
+    {
+      "epoch": 1.2046979865771812,
+      "grad_norm": 0.44844305515289307,
+      "learning_rate": 3.742475700056574e-06,
+      "loss": 0.2292,
+      "step": 1077
+    },
+    {
+      "epoch": 1.2058165548098434,
+      "grad_norm": 0.4744383990764618,
+      "learning_rate": 3.7396502215743893e-06,
+      "loss": 0.263,
+      "step": 1078
+    },
+    {
+      "epoch": 1.2069351230425056,
+      "grad_norm": 0.4899364411830902,
+      "learning_rate": 3.73682264181087e-06,
+      "loss": 0.2757,
+      "step": 1079
+    },
+    {
+      "epoch": 1.2080536912751678,
+      "grad_norm": 0.44050300121307373,
+      "learning_rate": 3.7339929655589345e-06,
+      "loss": 0.3652,
+      "step": 1080
+    },
+    {
+      "epoch": 1.20917225950783,
+      "grad_norm": 0.5100591778755188,
+      "learning_rate": 3.7311611976150523e-06,
+      "loss": 0.3371,
+      "step": 1081
+    },
+    {
+      "epoch": 1.2102908277404922,
+      "grad_norm": 0.4079134464263916,
+      "learning_rate": 3.7283273427792405e-06,
+      "loss": 0.24,
+      "step": 1082
+    },
+    {
+      "epoch": 1.2114093959731544,
+      "grad_norm": 0.42473068833351135,
+      "learning_rate": 3.7254914058550524e-06,
+      "loss": 0.339,
+      "step": 1083
+    },
+    {
+      "epoch": 1.2125279642058167,
+      "grad_norm": 0.4688520133495331,
+      "learning_rate": 3.7226533916495705e-06,
+      "loss": 0.3154,
+      "step": 1084
+    },
+    {
+      "epoch": 1.2136465324384789,
+      "grad_norm": 0.42565181851387024,
+      "learning_rate": 3.7198133049733988e-06,
+      "loss": 0.2511,
+      "step": 1085
+    },
+    {
+      "epoch": 1.2147651006711409,
+      "grad_norm": 0.39503219723701477,
+      "learning_rate": 3.7169711506406545e-06,
+      "loss": 0.2409,
+      "step": 1086
+    },
+    {
+      "epoch": 1.215883668903803,
+      "grad_norm": 0.4324930012226105,
+      "learning_rate": 3.7141269334689593e-06,
+      "loss": 0.2664,
+      "step": 1087
+    },
+    {
+      "epoch": 1.2170022371364653,
+      "grad_norm": 0.4191146790981293,
+      "learning_rate": 3.711280658279432e-06,
+      "loss": 0.2788,
+      "step": 1088
+    },
+    {
+      "epoch": 1.2181208053691275,
+      "grad_norm": 0.4261245131492615,
+      "learning_rate": 3.7084323298966786e-06,
+      "loss": 0.3008,
+      "step": 1089
+    },
+    {
+      "epoch": 1.2192393736017897,
+      "grad_norm": 0.3718254864215851,
+      "learning_rate": 3.705581953148787e-06,
+      "loss": 0.2669,
+      "step": 1090
+    },
+    {
+      "epoch": 1.220357941834452,
+      "grad_norm": 0.37301120162010193,
+      "learning_rate": 3.7027295328673153e-06,
+      "loss": 0.202,
+      "step": 1091
+    },
+    {
+      "epoch": 1.221476510067114,
+      "grad_norm": 0.3937848210334778,
+      "learning_rate": 3.699875073887288e-06,
+      "loss": 0.2632,
+      "step": 1092
+    },
+    {
+      "epoch": 1.2225950782997763,
+      "grad_norm": 0.39598923921585083,
+      "learning_rate": 3.697018581047182e-06,
+      "loss": 0.2649,
+      "step": 1093
+    },
+    {
+      "epoch": 1.2237136465324385,
+      "grad_norm": 0.3810122311115265,
+      "learning_rate": 3.694160059188926e-06,
+      "loss": 0.2435,
+      "step": 1094
+    },
+    {
+      "epoch": 1.2248322147651007,
+      "grad_norm": 0.5838693380355835,
+      "learning_rate": 3.6912995131578844e-06,
+      "loss": 0.2645,
+      "step": 1095
+    },
+    {
+      "epoch": 1.225950782997763,
+      "grad_norm": 0.4476895034313202,
+      "learning_rate": 3.6884369478028543e-06,
+      "loss": 0.2375,
+      "step": 1096
+    },
+    {
+      "epoch": 1.227069351230425,
+      "grad_norm": 0.45124199986457825,
+      "learning_rate": 3.685572367976055e-06,
+      "loss": 0.389,
+      "step": 1097
+    },
+    {
+      "epoch": 1.2281879194630871,
+      "grad_norm": 0.4733167588710785,
+      "learning_rate": 3.682705778533121e-06,
+      "loss": 0.284,
+      "step": 1098
+    },
+    {
+      "epoch": 1.2293064876957494,
+      "grad_norm": 0.39891883730888367,
+      "learning_rate": 3.679837184333093e-06,
+      "loss": 0.2539,
+      "step": 1099
+    },
+    {
+      "epoch": 1.2304250559284116,
+      "grad_norm": 0.45734989643096924,
+      "learning_rate": 3.67696659023841e-06,
+      "loss": 0.2782,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2304250559284116,
+      "eval_loss": 0.28197118639945984,
+      "eval_runtime": 376.4939,
+      "eval_samples_per_second": 2.112,
+      "eval_steps_per_second": 2.112,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2315436241610738,
+      "grad_norm": 0.4098396301269531,
+      "learning_rate": 3.6740940011149003e-06,
+      "loss": 0.2658,
+      "step": 1101
+    },
+    {
+      "epoch": 1.232662192393736,
+      "grad_norm": 0.5005487203598022,
+      "learning_rate": 3.6712194218317746e-06,
+      "loss": 0.2767,
+      "step": 1102
+    },
+    {
+      "epoch": 1.2337807606263982,
+      "grad_norm": 0.4187259376049042,
+      "learning_rate": 3.668342857261618e-06,
+      "loss": 0.2385,
+      "step": 1103
+    },
+    {
+      "epoch": 1.2348993288590604,
+      "grad_norm": 0.4496643841266632,
+      "learning_rate": 3.6654643122803785e-06,
+      "loss": 0.3185,
+      "step": 1104
+    },
+    {
+      "epoch": 1.2360178970917226,
+      "grad_norm": 0.46036508679389954,
+      "learning_rate": 3.662583791767362e-06,
+      "loss": 0.2874,
+      "step": 1105
+    },
+    {
+      "epoch": 1.2371364653243848,
+      "grad_norm": 0.4139268100261688,
+      "learning_rate": 3.659701300605224e-06,
+      "loss": 0.3835,
+      "step": 1106
+    },
+    {
+      "epoch": 1.238255033557047,
+      "grad_norm": 0.468631774187088,
+      "learning_rate": 3.6568168436799587e-06,
+      "loss": 0.2882,
+      "step": 1107
+    },
+    {
+      "epoch": 1.2393736017897092,
+      "grad_norm": 0.37668514251708984,
+      "learning_rate": 3.653930425880894e-06,
+      "loss": 0.2231,
+      "step": 1108
+    },
+    {
+      "epoch": 1.2404921700223714,
+      "grad_norm": 0.4401280879974365,
+      "learning_rate": 3.651042052100681e-06,
+      "loss": 0.2308,
+      "step": 1109
+    },
+    {
+      "epoch": 1.2416107382550337,
+      "grad_norm": 0.4293210506439209,
+      "learning_rate": 3.6481517272352846e-06,
+      "loss": 0.3656,
+      "step": 1110
+    },
+    {
+      "epoch": 1.2427293064876959,
+      "grad_norm": 0.4118444621562958,
+      "learning_rate": 3.6452594561839806e-06,
+      "loss": 0.189,
+      "step": 1111
+    },
+    {
+      "epoch": 1.2438478747203578,
+      "grad_norm": 0.4311460256576538,
+      "learning_rate": 3.64236524384934e-06,
+      "loss": 0.3029,
+      "step": 1112
+    },
+    {
+      "epoch": 1.24496644295302,
+      "grad_norm": 0.4397989809513092,
+      "learning_rate": 3.639469095137227e-06,
+      "loss": 0.2575,
+      "step": 1113
+    },
+    {
+      "epoch": 1.2460850111856823,
+      "grad_norm": 0.4245775043964386,
+      "learning_rate": 3.636571014956786e-06,
+      "loss": 0.3142,
+      "step": 1114
+    },
+    {
+      "epoch": 1.2472035794183445,
+      "grad_norm": 0.4065180718898773,
+      "learning_rate": 3.6336710082204385e-06,
+      "loss": 0.2477,
+      "step": 1115
+    },
+    {
+      "epoch": 1.2483221476510067,
+      "grad_norm": 0.5105834603309631,
+      "learning_rate": 3.6307690798438665e-06,
+      "loss": 0.2652,
+      "step": 1116
+    },
+    {
+      "epoch": 1.249440715883669,
+      "grad_norm": 0.396369993686676,
+      "learning_rate": 3.627865234746014e-06,
+      "loss": 0.237,
+      "step": 1117
+    },
+    {
+      "epoch": 1.250559284116331,
+      "grad_norm": 0.3931864798069,
+      "learning_rate": 3.624959477849072e-06,
+      "loss": 0.2563,
+      "step": 1118
+    },
+    {
+      "epoch": 1.2516778523489933,
+      "grad_norm": 0.40984857082366943,
+      "learning_rate": 3.622051814078473e-06,
+      "loss": 0.241,
+      "step": 1119
+    },
+    {
+      "epoch": 1.2527964205816555,
+      "grad_norm": 0.46675774455070496,
+      "learning_rate": 3.61914224836288e-06,
+      "loss": 0.2757,
+      "step": 1120
+    },
+    {
+      "epoch": 1.2539149888143177,
+      "grad_norm": 0.37578269839286804,
+      "learning_rate": 3.616230785634183e-06,
+      "loss": 0.2683,
+      "step": 1121
+    },
+    {
+      "epoch": 1.25503355704698,
+      "grad_norm": 0.44806206226348877,
+      "learning_rate": 3.6133174308274827e-06,
+      "loss": 0.23,
+      "step": 1122
+    },
+    {
+      "epoch": 1.256152125279642,
+      "grad_norm": 0.43839097023010254,
+      "learning_rate": 3.6104021888810927e-06,
+      "loss": 0.2564,
+      "step": 1123
+    },
+    {
+      "epoch": 1.2572706935123041,
+      "grad_norm": 0.46507710218429565,
+      "learning_rate": 3.60748506473652e-06,
+      "loss": 0.2702,
+      "step": 1124
+    },
+    {
+      "epoch": 1.2583892617449663,
+      "grad_norm": 0.426341712474823,
+      "learning_rate": 3.604566063338467e-06,
+      "loss": 0.2886,
+      "step": 1125
+    },
+    {
+      "epoch": 1.2595078299776286,
+      "grad_norm": 0.5875904560089111,
+      "learning_rate": 3.6016451896348146e-06,
+      "loss": 0.2735,
+      "step": 1126
+    },
+    {
+      "epoch": 1.2606263982102908,
+      "grad_norm": 0.3921490013599396,
+      "learning_rate": 3.5987224485766194e-06,
+      "loss": 0.2128,
+      "step": 1127
+    },
+    {
+      "epoch": 1.261744966442953,
+      "grad_norm": 0.4550873935222626,
+      "learning_rate": 3.5957978451181014e-06,
+      "loss": 0.2175,
+      "step": 1128
+    },
+    {
+      "epoch": 1.2628635346756152,
+      "grad_norm": 0.41963446140289307,
+      "learning_rate": 3.592871384216639e-06,
+      "loss": 0.2164,
+      "step": 1129
+    },
+    {
+      "epoch": 1.2639821029082774,
+      "grad_norm": 0.44765520095825195,
+      "learning_rate": 3.5899430708327586e-06,
+      "loss": 0.2395,
+      "step": 1130
+    },
+    {
+      "epoch": 1.2651006711409396,
+      "grad_norm": 0.4832054674625397,
+      "learning_rate": 3.587012909930127e-06,
+      "loss": 0.2936,
+      "step": 1131
+    },
+    {
+      "epoch": 1.2662192393736018,
+      "grad_norm": 0.4616105556488037,
+      "learning_rate": 3.584080906475542e-06,
+      "loss": 0.2783,
+      "step": 1132
+    },
+    {
+      "epoch": 1.267337807606264,
+      "grad_norm": 0.5043412446975708,
+      "learning_rate": 3.5811470654389247e-06,
+      "loss": 0.2496,
+      "step": 1133
+    },
+    {
+      "epoch": 1.2684563758389262,
+      "grad_norm": 0.46124815940856934,
+      "learning_rate": 3.578211391793312e-06,
+      "loss": 0.3075,
+      "step": 1134
+    },
+    {
+      "epoch": 1.2695749440715884,
+      "grad_norm": 0.40088096261024475,
+      "learning_rate": 3.5752738905148455e-06,
+      "loss": 0.2313,
+      "step": 1135
+    },
+    {
+      "epoch": 1.2706935123042506,
+      "grad_norm": 0.39959943294525146,
+      "learning_rate": 3.572334566582766e-06,
+      "loss": 0.2521,
+      "step": 1136
+    },
+    {
+      "epoch": 1.2718120805369129,
+      "grad_norm": 0.4335862994194031,
+      "learning_rate": 3.569393424979405e-06,
+      "loss": 0.2272,
+      "step": 1137
+    },
+    {
+      "epoch": 1.272930648769575,
+      "grad_norm": 0.44511085748672485,
+      "learning_rate": 3.5664504706901715e-06,
+      "loss": 0.2541,
+      "step": 1138
+    },
+    {
+      "epoch": 1.274049217002237,
+      "grad_norm": 0.6216147541999817,
+      "learning_rate": 3.5635057087035513e-06,
+      "loss": 0.3711,
+      "step": 1139
+    },
+    {
+      "epoch": 1.2751677852348993,
+      "grad_norm": 0.4513746500015259,
+      "learning_rate": 3.5605591440110896e-06,
+      "loss": 0.2956,
+      "step": 1140
+    },
+    {
+      "epoch": 1.2762863534675615,
+      "grad_norm": 0.4705442786216736,
+      "learning_rate": 3.5576107816073926e-06,
+      "loss": 0.25,
+      "step": 1141
+    },
+    {
+      "epoch": 1.2774049217002237,
+      "grad_norm": 0.37419384717941284,
+      "learning_rate": 3.554660626490109e-06,
+      "loss": 0.2183,
+      "step": 1142
+    },
+    {
+      "epoch": 1.278523489932886,
+      "grad_norm": 0.48330289125442505,
+      "learning_rate": 3.551708683659929e-06,
+      "loss": 0.2639,
+      "step": 1143
+    },
+    {
+      "epoch": 1.279642058165548,
+      "grad_norm": 0.46695518493652344,
+      "learning_rate": 3.5487549581205727e-06,
+      "loss": 0.2962,
+      "step": 1144
+    },
+    {
+      "epoch": 1.2807606263982103,
+      "grad_norm": 0.42789673805236816,
+      "learning_rate": 3.5457994548787818e-06,
+      "loss": 0.2784,
+      "step": 1145
+    },
+    {
+      "epoch": 1.2818791946308725,
+      "grad_norm": 0.39100247621536255,
+      "learning_rate": 3.5428421789443114e-06,
+      "loss": 0.2326,
+      "step": 1146
+    },
+    {
+      "epoch": 1.2829977628635347,
+      "grad_norm": 0.4366264343261719,
+      "learning_rate": 3.5398831353299203e-06,
+      "loss": 0.2719,
+      "step": 1147
+    },
+    {
+      "epoch": 1.284116331096197,
+      "grad_norm": 0.5076384544372559,
+      "learning_rate": 3.536922329051365e-06,
+      "loss": 0.2774,
+      "step": 1148
+    },
+    {
+      "epoch": 1.285234899328859,
+      "grad_norm": 0.478981077671051,
+      "learning_rate": 3.5339597651273895e-06,
+      "loss": 0.3209,
+      "step": 1149
+    },
+    {
+      "epoch": 1.2863534675615211,
+      "grad_norm": 0.44728022813796997,
+      "learning_rate": 3.530995448579717e-06,
+      "loss": 0.2392,
+      "step": 1150
+    },
+    {
+      "epoch": 1.2863534675615211,
+      "eval_loss": 0.28084221482276917,
+      "eval_runtime": 376.4276,
+      "eval_samples_per_second": 2.112,
+      "eval_steps_per_second": 2.112,
+      "step": 1150
+    },
+    {
+      "epoch": 1.2874720357941833,
+      "grad_norm": 0.4280475080013275,
+      "learning_rate": 3.5280293844330416e-06,
+      "loss": 0.2766,
+      "step": 1151
+    },
+    {
+      "epoch": 1.2885906040268456,
+      "grad_norm": 0.46685996651649475,
+      "learning_rate": 3.52506157771502e-06,
+      "loss": 0.2739,
+      "step": 1152
+    },
+    {
+      "epoch": 1.2897091722595078,
+      "grad_norm": 0.41764935851097107,
+      "learning_rate": 3.522092033456261e-06,
+      "loss": 0.3506,
+      "step": 1153
+    },
+    {
+      "epoch": 1.29082774049217,
+      "grad_norm": 0.4733811318874359,
+      "learning_rate": 3.5191207566903206e-06,
+      "loss": 0.3456,
+      "step": 1154
+    },
+    {
+      "epoch": 1.2919463087248322,
+      "grad_norm": 0.4272392690181732,
+      "learning_rate": 3.5161477524536913e-06,
+      "loss": 0.2854,
+      "step": 1155
+    },
+    {
+      "epoch": 1.2930648769574944,
+      "grad_norm": 0.4370381236076355,
+      "learning_rate": 3.513173025785792e-06,
+      "loss": 0.2702,
+      "step": 1156
+    },
+    {
+      "epoch": 1.2941834451901566,
+      "grad_norm": 0.40805020928382874,
+      "learning_rate": 3.5101965817289634e-06,
+      "loss": 0.2875,
+      "step": 1157
+    },
+    {
+      "epoch": 1.2953020134228188,
+      "grad_norm": 0.4660682678222656,
+      "learning_rate": 3.5072184253284564e-06,
+      "loss": 0.2802,
+      "step": 1158
+    },
+    {
+      "epoch": 1.296420581655481,
+      "grad_norm": 0.4880177676677704,
+      "learning_rate": 3.5042385616324243e-06,
+      "loss": 0.2848,
+      "step": 1159
+    },
+    {
+      "epoch": 1.2975391498881432,
+      "grad_norm": 0.45903992652893066,
+      "learning_rate": 3.501256995691914e-06,
+      "loss": 0.2718,
+      "step": 1160
+    },
+    {
+      "epoch": 1.2986577181208054,
+      "grad_norm": 0.4915492534637451,
+      "learning_rate": 3.498273732560858e-06,
+      "loss": 0.3065,
+      "step": 1161
+    },
+    {
+      "epoch": 1.2997762863534676,
+      "grad_norm": 0.39194348454475403,
+      "learning_rate": 3.495288777296067e-06,
+      "loss": 0.3184,
+      "step": 1162
+    },
+    {
+      "epoch": 1.3008948545861299,
+      "grad_norm": 0.44880497455596924,
+      "learning_rate": 3.4923021349572183e-06,
+      "loss": 0.2652,
+      "step": 1163
+    },
+    {
+      "epoch": 1.302013422818792,
+      "grad_norm": 0.4580858647823334,
+      "learning_rate": 3.4893138106068497e-06,
+      "loss": 0.2477,
+      "step": 1164
+    },
+    {
+      "epoch": 1.3031319910514543,
+      "grad_norm": 0.5455112457275391,
+      "learning_rate": 3.486323809310349e-06,
+      "loss": 0.3024,
+      "step": 1165
+    },
+    {
+      "epoch": 1.3042505592841163,
+      "grad_norm": 0.42716994881629944,
+      "learning_rate": 3.4833321361359486e-06,
+      "loss": 0.303,
+      "step": 1166
+    },
+    {
+      "epoch": 1.3053691275167785,
+      "grad_norm": 0.418785035610199,
+      "learning_rate": 3.4803387961547123e-06,
+      "loss": 0.2475,
+      "step": 1167
+    },
+    {
+      "epoch": 1.3064876957494407,
+      "grad_norm": 0.5143936276435852,
+      "learning_rate": 3.477343794440532e-06,
+      "loss": 0.2711,
+      "step": 1168
+    },
+    {
+      "epoch": 1.307606263982103,
+      "grad_norm": 0.41924092173576355,
+      "learning_rate": 3.474347136070113e-06,
+      "loss": 0.2389,
+      "step": 1169
+    },
+    {
+      "epoch": 1.308724832214765,
+      "grad_norm": 0.42970436811447144,
+      "learning_rate": 3.4713488261229722e-06,
+      "loss": 0.2826,
+      "step": 1170
+    },
+    {
+      "epoch": 1.3098434004474273,
+      "grad_norm": 0.47588106989860535,
+      "learning_rate": 3.4683488696814245e-06,
+      "loss": 0.3085,
+      "step": 1171
+    },
+    {
+      "epoch": 1.3109619686800895,
+      "grad_norm": 0.5275523662567139,
+      "learning_rate": 3.465347271830575e-06,
+      "loss": 0.3025,
+      "step": 1172
+    },
+    {
+      "epoch": 1.3120805369127517,
+      "grad_norm": 0.5202552676200867,
+      "learning_rate": 3.4623440376583117e-06,
+      "loss": 0.3734,
+      "step": 1173
+    },
+    {
+      "epoch": 1.313199105145414,
+      "grad_norm": 0.4324059784412384,
+      "learning_rate": 3.4593391722552973e-06,
+      "loss": 0.27,
+      "step": 1174
+    },
+    {
+      "epoch": 1.3143176733780761,
+      "grad_norm": 0.4598994851112366,
+      "learning_rate": 3.4563326807149566e-06,
+      "loss": 0.2603,
+      "step": 1175
+    },
+    {
+      "epoch": 1.3154362416107381,
+      "grad_norm": 0.47085604071617126,
+      "learning_rate": 3.453324568133475e-06,
+      "loss": 0.3355,
+      "step": 1176
+    },
+    {
+      "epoch": 1.3165548098434003,
+      "grad_norm": 0.47546300292015076,
+      "learning_rate": 3.450314839609781e-06,
+      "loss": 0.2625,
+      "step": 1177
+    },
+    {
+      "epoch": 1.3176733780760626,
+      "grad_norm": 0.3994863033294678,
+      "learning_rate": 3.447303500245547e-06,
+      "loss": 0.2188,
+      "step": 1178
+    },
+    {
+      "epoch": 1.3187919463087248,
+      "grad_norm": 0.42841657996177673,
+      "learning_rate": 3.444290555145172e-06,
+      "loss": 0.2722,
+      "step": 1179
+    },
+    {
+      "epoch": 1.319910514541387,
+      "grad_norm": 0.44720879197120667,
+      "learning_rate": 3.441276009415778e-06,
+      "loss": 0.3199,
+      "step": 1180
+    },
+    {
+      "epoch": 1.3210290827740492,
+      "grad_norm": 0.47880569100379944,
+      "learning_rate": 3.4382598681672025e-06,
+      "loss": 0.2751,
+      "step": 1181
+    },
+    {
+      "epoch": 1.3221476510067114,
+      "grad_norm": 0.45834243297576904,
+      "learning_rate": 3.4352421365119837e-06,
+      "loss": 0.3132,
+      "step": 1182
+    },
+    {
+      "epoch": 1.3232662192393736,
+      "grad_norm": 0.7867944240570068,
+      "learning_rate": 3.4322228195653575e-06,
+      "loss": 0.2513,
+      "step": 1183
+    },
+    {
+      "epoch": 1.3243847874720358,
+      "grad_norm": 0.3896082043647766,
+      "learning_rate": 3.4292019224452472e-06,
+      "loss": 0.3146,
+      "step": 1184
+    },
+    {
+      "epoch": 1.325503355704698,
+      "grad_norm": 0.44636213779449463,
+      "learning_rate": 3.4261794502722547e-06,
+      "loss": 0.2572,
+      "step": 1185
+    },
+    {
+      "epoch": 1.3266219239373602,
+      "grad_norm": 0.4200538396835327,
+      "learning_rate": 3.4231554081696494e-06,
+      "loss": 0.3025,
+      "step": 1186
+    },
+    {
+      "epoch": 1.3277404921700224,
+      "grad_norm": 0.473981648683548,
+      "learning_rate": 3.4201298012633668e-06,
+      "loss": 0.3445,
+      "step": 1187
+    },
+    {
+      "epoch": 1.3288590604026846,
+      "grad_norm": 0.4472513794898987,
+      "learning_rate": 3.4171026346819893e-06,
+      "loss": 0.2689,
+      "step": 1188
+    },
+    {
+      "epoch": 1.3299776286353469,
+      "grad_norm": 0.4956374764442444,
+      "learning_rate": 3.4140739135567465e-06,
+      "loss": 0.2546,
+      "step": 1189
+    },
+    {
+      "epoch": 1.331096196868009,
+      "grad_norm": 0.5581144690513611,
+      "learning_rate": 3.4110436430215016e-06,
+      "loss": 0.3434,
+      "step": 1190
+    },
+    {
+      "epoch": 1.3322147651006713,
+      "grad_norm": 0.4358447194099426,
+      "learning_rate": 3.4080118282127445e-06,
+      "loss": 0.2508,
+      "step": 1191
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4419161379337311,
+      "learning_rate": 3.4049784742695835e-06,
+      "loss": 0.2792,
+      "step": 1192
+    },
+    {
+      "epoch": 1.3344519015659955,
+      "grad_norm": 0.5371615290641785,
+      "learning_rate": 3.4019435863337334e-06,
+      "loss": 0.3771,
+      "step": 1193
+    },
+    {
+      "epoch": 1.3355704697986577,
+      "grad_norm": 0.3891585171222687,
+      "learning_rate": 3.398907169549513e-06,
+      "loss": 0.2791,
+      "step": 1194
+    },
+    {
+      "epoch": 1.3366890380313199,
+      "grad_norm": 0.45763733983039856,
+      "learning_rate": 3.3958692290638295e-06,
+      "loss": 0.2898,
+      "step": 1195
+    },
+    {
+      "epoch": 1.337807606263982,
+      "grad_norm": 0.43887439370155334,
+      "learning_rate": 3.3928297700261735e-06,
+      "loss": 0.3189,
+      "step": 1196
+    },
+    {
+      "epoch": 1.3389261744966443,
+      "grad_norm": 0.45437106490135193,
+      "learning_rate": 3.3897887975886113e-06,
+      "loss": 0.2624,
+      "step": 1197
+    },
+    {
+      "epoch": 1.3400447427293065,
+      "grad_norm": 0.4571227729320526,
+      "learning_rate": 3.3867463169057726e-06,
+      "loss": 0.2707,
+      "step": 1198
+    },
+    {
+      "epoch": 1.3411633109619687,
+      "grad_norm": 1.0891879796981812,
+      "learning_rate": 3.3837023331348436e-06,
+      "loss": 0.2842,
+      "step": 1199
+    },
+    {
+      "epoch": 1.342281879194631,
+      "grad_norm": 0.44873741269111633,
+      "learning_rate": 3.38065685143556e-06,
+      "loss": 0.2835,
+      "step": 1200
+    },
+    {
+      "epoch": 1.342281879194631,
+      "eval_loss": 0.27979737520217896,
+      "eval_runtime": 376.2505,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 2.113,
+      "step": 1200
+    },
+    {
+      "epoch": 1.3434004474272931,
+      "grad_norm": 0.47640982270240784,
+      "learning_rate": 3.3776098769701944e-06,
+      "loss": 0.3103,
+      "step": 1201
+    },
+    {
+      "epoch": 1.3445190156599551,
+      "grad_norm": 0.4277811050415039,
+      "learning_rate": 3.3745614149035523e-06,
+      "loss": 0.2172,
+      "step": 1202
+    },
+    {
+      "epoch": 1.3456375838926173,
+      "grad_norm": 0.38408157229423523,
+      "learning_rate": 3.3715114704029583e-06,
+      "loss": 0.3037,
+      "step": 1203
+    },
+    {
+      "epoch": 1.3467561521252795,
+      "grad_norm": 0.4549292027950287,
+      "learning_rate": 3.368460048638251e-06,
+      "loss": 0.3602,
+      "step": 1204
+    },
+    {
+      "epoch": 1.3478747203579418,
+      "grad_norm": 0.43562158942222595,
+      "learning_rate": 3.3654071547817745e-06,
+      "loss": 0.2929,
+      "step": 1205
+    },
+    {
+      "epoch": 1.348993288590604,
+      "grad_norm": 0.4395628273487091,
+      "learning_rate": 3.3623527940083643e-06,
+      "loss": 0.2518,
+      "step": 1206
+    },
+    {
+      "epoch": 1.3501118568232662,
+      "grad_norm": 0.4159080386161804,
+      "learning_rate": 3.3592969714953464e-06,
+      "loss": 0.2261,
+      "step": 1207
+    },
+    {
+      "epoch": 1.3512304250559284,
+      "grad_norm": 0.5107335448265076,
+      "learning_rate": 3.3562396924225226e-06,
+      "loss": 0.2347,
+      "step": 1208
+    },
+    {
+      "epoch": 1.3523489932885906,
+      "grad_norm": 0.4724574387073517,
+      "learning_rate": 3.3531809619721632e-06,
+      "loss": 0.2917,
+      "step": 1209
+    },
+    {
+      "epoch": 1.3534675615212528,
+      "grad_norm": 0.427035391330719,
+      "learning_rate": 3.350120785329001e-06,
+      "loss": 0.3314,
+      "step": 1210
+    },
+    {
+      "epoch": 1.354586129753915,
+      "grad_norm": 0.5142832398414612,
+      "learning_rate": 3.347059167680216e-06,
+      "loss": 0.3263,
+      "step": 1211
+    },
+    {
+      "epoch": 1.3557046979865772,
+      "grad_norm": 0.45559144020080566,
+      "learning_rate": 3.343996114215436e-06,
+      "loss": 0.3229,
+      "step": 1212
+    },
+    {
+      "epoch": 1.3568232662192394,
+      "grad_norm": 0.4368569552898407,
+      "learning_rate": 3.340931630126718e-06,
+      "loss": 0.2739,
+      "step": 1213
+    },
+    {
+      "epoch": 1.3579418344519016,
+      "grad_norm": 0.5200837254524231,
+      "learning_rate": 3.3378657206085487e-06,
+      "loss": 0.298,
+      "step": 1214
+    },
+    {
+      "epoch": 1.3590604026845639,
+      "grad_norm": 0.41081973910331726,
+      "learning_rate": 3.334798390857825e-06,
+      "loss": 0.2585,
+      "step": 1215
+    },
+    {
+      "epoch": 1.360178970917226,
+      "grad_norm": 0.42855456471443176,
+      "learning_rate": 3.331729646073856e-06,
+      "loss": 0.2804,
+      "step": 1216
+    },
+    {
+      "epoch": 1.3612975391498883,
+      "grad_norm": 0.42450782656669617,
+      "learning_rate": 3.3286594914583483e-06,
+      "loss": 0.2393,
+      "step": 1217
+    },
+    {
+      "epoch": 1.3624161073825503,
+      "grad_norm": 0.46950003504753113,
+      "learning_rate": 3.3255879322153965e-06,
+      "loss": 0.2416,
+      "step": 1218
+    },
+    {
+      "epoch": 1.3635346756152125,
+      "grad_norm": 0.4398280680179596,
+      "learning_rate": 3.3225149735514785e-06,
+      "loss": 0.2718,
+      "step": 1219
+    },
+    {
+      "epoch": 1.3646532438478747,
+      "grad_norm": 0.4376338720321655,
+      "learning_rate": 3.319440620675442e-06,
+      "loss": 0.2376,
+      "step": 1220
+    },
+    {
+      "epoch": 1.3657718120805369,
+      "grad_norm": 0.4066784977912903,
+      "learning_rate": 3.3163648787985005e-06,
+      "loss": 0.2868,
+      "step": 1221
+    },
+    {
+      "epoch": 1.366890380313199,
+      "grad_norm": 0.4568162262439728,
+      "learning_rate": 3.31328775313422e-06,
+      "loss": 0.2824,
+      "step": 1222
+    },
+    {
+      "epoch": 1.3680089485458613,
+      "grad_norm": 0.47311756014823914,
+      "learning_rate": 3.310209248898513e-06,
+      "loss": 0.2975,
+      "step": 1223
+    },
+    {
+      "epoch": 1.3691275167785235,
+      "grad_norm": 0.48929595947265625,
+      "learning_rate": 3.3071293713096274e-06,
+      "loss": 0.3009,
+      "step": 1224
+    },
+    {
+      "epoch": 1.3702460850111857,
+      "grad_norm": 0.44714704155921936,
+      "learning_rate": 3.304048125588142e-06,
+      "loss": 0.2501,
+      "step": 1225
+    },
+    {
+      "epoch": 1.371364653243848,
+      "grad_norm": 0.422889769077301,
+      "learning_rate": 3.300965516956951e-06,
+      "loss": 0.2597,
+      "step": 1226
+    },
+    {
+      "epoch": 1.3724832214765101,
+      "grad_norm": 0.4930384159088135,
+      "learning_rate": 3.297881550641261e-06,
+      "loss": 0.2428,
+      "step": 1227
+    },
+    {
+      "epoch": 1.3736017897091721,
+      "grad_norm": 0.41760820150375366,
+      "learning_rate": 3.29479623186858e-06,
+      "loss": 0.2567,
+      "step": 1228
+    },
+    {
+      "epoch": 1.3747203579418343,
+      "grad_norm": 0.45922526717185974,
+      "learning_rate": 3.2917095658687077e-06,
+      "loss": 0.2815,
+      "step": 1229
+    },
+    {
+      "epoch": 1.3758389261744965,
+      "grad_norm": 0.4892861843109131,
+      "learning_rate": 3.2886215578737275e-06,
+      "loss": 0.2884,
+      "step": 1230
+    },
+    {
+      "epoch": 1.3769574944071588,
+      "grad_norm": 0.4022981524467468,
+      "learning_rate": 3.285532213117998e-06,
+      "loss": 0.3119,
+      "step": 1231
+    },
+    {
+      "epoch": 1.378076062639821,
+      "grad_norm": 0.461446613073349,
+      "learning_rate": 3.282441536838144e-06,
+      "loss": 0.2404,
+      "step": 1232
+    },
+    {
+      "epoch": 1.3791946308724832,
+      "grad_norm": 0.45386597514152527,
+      "learning_rate": 3.2793495342730466e-06,
+      "loss": 0.2885,
+      "step": 1233
+    },
+    {
+      "epoch": 1.3803131991051454,
+      "grad_norm": 0.44144392013549805,
+      "learning_rate": 3.276256210663834e-06,
+      "loss": 0.2836,
+      "step": 1234
+    },
+    {
+      "epoch": 1.3814317673378076,
+      "grad_norm": 0.4419529438018799,
+      "learning_rate": 3.273161571253877e-06,
+      "loss": 0.2816,
+      "step": 1235
+    },
+    {
+      "epoch": 1.3825503355704698,
+      "grad_norm": 0.5107296109199524,
+      "learning_rate": 3.2700656212887725e-06,
+      "loss": 0.322,
+      "step": 1236
+    },
+    {
+      "epoch": 1.383668903803132,
+      "grad_norm": 0.48052000999450684,
+      "learning_rate": 3.266968366016342e-06,
+      "loss": 0.2768,
+      "step": 1237
+    },
+    {
+      "epoch": 1.3847874720357942,
+      "grad_norm": 0.4279806613922119,
+      "learning_rate": 3.263869810686619e-06,
+      "loss": 0.3372,
+      "step": 1238
+    },
+    {
+      "epoch": 1.3859060402684564,
+      "grad_norm": 0.4076407253742218,
+      "learning_rate": 3.2607699605518408e-06,
+      "loss": 0.2428,
+      "step": 1239
+    },
+    {
+      "epoch": 1.3870246085011186,
+      "grad_norm": 0.4766339361667633,
+      "learning_rate": 3.2576688208664373e-06,
+      "loss": 0.2764,
+      "step": 1240
+    },
+    {
+      "epoch": 1.3881431767337808,
+      "grad_norm": 0.45896342396736145,
+      "learning_rate": 3.254566396887027e-06,
+      "loss": 0.3233,
+      "step": 1241
+    },
+    {
+      "epoch": 1.389261744966443,
+      "grad_norm": 0.524057924747467,
+      "learning_rate": 3.2514626938724035e-06,
+      "loss": 0.2821,
+      "step": 1242
+    },
+    {
+      "epoch": 1.3903803131991053,
+      "grad_norm": 0.4665883183479309,
+      "learning_rate": 3.24835771708353e-06,
+      "loss": 0.2624,
+      "step": 1243
+    },
+    {
+      "epoch": 1.3914988814317675,
+      "grad_norm": 0.4060594141483307,
+      "learning_rate": 3.2452514717835283e-06,
+      "loss": 0.3025,
+      "step": 1244
+    },
+    {
+      "epoch": 1.3926174496644295,
+      "grad_norm": 0.5079346895217896,
+      "learning_rate": 3.242143963237668e-06,
+      "loss": 0.2905,
+      "step": 1245
+    },
+    {
+      "epoch": 1.3937360178970917,
+      "grad_norm": 0.40741270780563354,
+      "learning_rate": 3.2390351967133644e-06,
+      "loss": 0.2175,
+      "step": 1246
+    },
+    {
+      "epoch": 1.3948545861297539,
+      "grad_norm": 0.42364364862442017,
+      "learning_rate": 3.235925177480162e-06,
+      "loss": 0.2302,
+      "step": 1247
+    },
+    {
+      "epoch": 1.395973154362416,
+      "grad_norm": 0.5337604880332947,
+      "learning_rate": 3.23281391080973e-06,
+      "loss": 0.2649,
+      "step": 1248
+    },
+    {
+      "epoch": 1.3970917225950783,
+      "grad_norm": 0.3975113332271576,
+      "learning_rate": 3.229701401975851e-06,
+      "loss": 0.2433,
+      "step": 1249
+    },
+    {
+      "epoch": 1.3982102908277405,
+      "grad_norm": 0.46082577109336853,
+      "learning_rate": 3.2265876562544142e-06,
+      "loss": 0.4772,
+      "step": 1250
+    },
+    {
+      "epoch": 1.3982102908277405,
+      "eval_loss": 0.279073566198349,
+      "eval_runtime": 376.2571,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 2.113,
+      "step": 1250
+    },
+    {
+      "epoch": 1.3993288590604027,
+      "grad_norm": 0.4870416522026062,
+      "learning_rate": 3.223472678923405e-06,
+      "loss": 0.2856,
+      "step": 1251
+    },
+    {
+      "epoch": 1.400447427293065,
+      "grad_norm": 0.4307974576950073,
+      "learning_rate": 3.2203564752628956e-06,
+      "loss": 0.319,
+      "step": 1252
+    },
+    {
+      "epoch": 1.4015659955257271,
+      "grad_norm": 0.44931602478027344,
+      "learning_rate": 3.2172390505550394e-06,
+      "loss": 0.2976,
+      "step": 1253
+    },
+    {
+      "epoch": 1.4026845637583891,
+      "grad_norm": 0.4705468714237213,
+      "learning_rate": 3.2141204100840566e-06,
+      "loss": 0.2592,
+      "step": 1254
+    },
+    {
+      "epoch": 1.4038031319910513,
+      "grad_norm": 0.4116649031639099,
+      "learning_rate": 3.2110005591362303e-06,
+      "loss": 0.2978,
+      "step": 1255
+    },
+    {
+      "epoch": 1.4049217002237135,
+      "grad_norm": 0.42026087641716003,
+      "learning_rate": 3.2078795029998934e-06,
+      "loss": 0.2711,
+      "step": 1256
+    },
+    {
+      "epoch": 1.4060402684563758,
+      "grad_norm": 0.39095547795295715,
+      "learning_rate": 3.2047572469654244e-06,
+      "loss": 0.274,
+      "step": 1257
+    },
+    {
+      "epoch": 1.407158836689038,
+      "grad_norm": 0.4151228964328766,
+      "learning_rate": 3.201633796325233e-06,
+      "loss": 0.2725,
+      "step": 1258
+    },
+    {
+      "epoch": 1.4082774049217002,
+      "grad_norm": 0.42332756519317627,
+      "learning_rate": 3.198509156373756e-06,
+      "loss": 0.2528,
+      "step": 1259
+    },
+    {
+      "epoch": 1.4093959731543624,
+      "grad_norm": 0.46915221214294434,
+      "learning_rate": 3.195383332407444e-06,
+      "loss": 0.2586,
+      "step": 1260
+    },
+    {
+      "epoch": 1.4105145413870246,
+      "grad_norm": 0.49486085772514343,
+      "learning_rate": 3.1922563297247568e-06,
+      "loss": 0.2529,
+      "step": 1261
+    },
+    {
+      "epoch": 1.4116331096196868,
+      "grad_norm": 0.49203628301620483,
+      "learning_rate": 3.1891281536261497e-06,
+      "loss": 0.3333,
+      "step": 1262
+    },
+    {
+      "epoch": 1.412751677852349,
+      "grad_norm": 0.5035507082939148,
+      "learning_rate": 3.1859988094140702e-06,
+      "loss": 0.3144,
+      "step": 1263
+    },
+    {
+      "epoch": 1.4138702460850112,
+      "grad_norm": 0.4640123248100281,
+      "learning_rate": 3.182868302392944e-06,
+      "loss": 0.2594,
+      "step": 1264
+    },
+    {
+      "epoch": 1.4149888143176734,
+      "grad_norm": 0.43496641516685486,
+      "learning_rate": 3.179736637869167e-06,
+      "loss": 0.2938,
+      "step": 1265
+    },
+    {
+      "epoch": 1.4161073825503356,
+      "grad_norm": 0.43234068155288696,
+      "learning_rate": 3.1766038211510975e-06,
+      "loss": 0.2925,
+      "step": 1266
+    },
+    {
+      "epoch": 1.4172259507829978,
+      "grad_norm": 0.41315048933029175,
+      "learning_rate": 3.1734698575490485e-06,
+      "loss": 0.263,
+      "step": 1267
+    },
+    {
+      "epoch": 1.41834451901566,
+      "grad_norm": 0.4369712471961975,
+      "learning_rate": 3.170334752375276e-06,
+      "loss": 0.256,
+      "step": 1268
+    },
+    {
+      "epoch": 1.4194630872483223,
+      "grad_norm": 0.4620123505592346,
+      "learning_rate": 3.1671985109439698e-06,
+      "loss": 0.2638,
+      "step": 1269
+    },
+    {
+      "epoch": 1.4205816554809845,
+      "grad_norm": 0.4419032335281372,
+      "learning_rate": 3.1640611385712472e-06,
+      "loss": 0.278,
+      "step": 1270
+    },
+    {
+      "epoch": 1.4217002237136465,
+      "grad_norm": 0.40368935465812683,
+      "learning_rate": 3.160922640575143e-06,
+      "loss": 0.3318,
+      "step": 1271
+    },
+    {
+      "epoch": 1.4228187919463087,
+      "grad_norm": 0.5157962441444397,
+      "learning_rate": 3.157783022275599e-06,
+      "loss": 0.3588,
+      "step": 1272
+    },
+    {
+      "epoch": 1.4239373601789709,
+      "grad_norm": 0.5941402316093445,
+      "learning_rate": 3.154642288994456e-06,
+      "loss": 0.345,
+      "step": 1273
+    },
+    {
+      "epoch": 1.425055928411633,
+      "grad_norm": 0.3772018849849701,
+      "learning_rate": 3.151500446055444e-06,
+      "loss": 0.2388,
+      "step": 1274
+    },
+    {
+      "epoch": 1.4261744966442953,
+      "grad_norm": 0.43480372428894043,
+      "learning_rate": 3.148357498784177e-06,
+      "loss": 0.2447,
+      "step": 1275
+    },
+    {
+      "epoch": 1.4272930648769575,
+      "grad_norm": 0.47598204016685486,
+      "learning_rate": 3.145213452508137e-06,
+      "loss": 0.2908,
+      "step": 1276
+    },
+    {
+      "epoch": 1.4284116331096197,
+      "grad_norm": 0.46050792932510376,
+      "learning_rate": 3.1420683125566703e-06,
+      "loss": 0.2681,
+      "step": 1277
+    },
+    {
+      "epoch": 1.429530201342282,
+      "grad_norm": 0.4539172947406769,
+      "learning_rate": 3.138922084260979e-06,
+      "loss": 0.2695,
+      "step": 1278
+    },
+    {
+      "epoch": 1.4306487695749441,
+      "grad_norm": 0.4421691596508026,
+      "learning_rate": 3.1357747729541065e-06,
+      "loss": 0.2418,
+      "step": 1279
+    },
+    {
+      "epoch": 1.4317673378076063,
+      "grad_norm": 0.39327266812324524,
+      "learning_rate": 3.132626383970936e-06,
+      "loss": 0.3114,
+      "step": 1280
+    },
+    {
+      "epoch": 1.4328859060402683,
+      "grad_norm": 0.3796561062335968,
+      "learning_rate": 3.129476922648174e-06,
+      "loss": 0.2312,
+      "step": 1281
+    },
+    {
+      "epoch": 1.4340044742729305,
+      "grad_norm": 0.5101112127304077,
+      "learning_rate": 3.126326394324346e-06,
+      "loss": 0.2655,
+      "step": 1282
+    },
+    {
+      "epoch": 1.4351230425055927,
+      "grad_norm": 0.4830431044101715,
+      "learning_rate": 3.1231748043397865e-06,
+      "loss": 0.2138,
+      "step": 1283
+    },
+    {
+      "epoch": 1.436241610738255,
+      "grad_norm": 0.4350200295448303,
+      "learning_rate": 3.120022158036629e-06,
+      "loss": 0.2199,
+      "step": 1284
+    },
+    {
+      "epoch": 1.4373601789709172,
+      "grad_norm": 0.4574565291404724,
+      "learning_rate": 3.116868460758798e-06,
+      "loss": 0.2236,
+      "step": 1285
+    },
+    {
+      "epoch": 1.4384787472035794,
+      "grad_norm": 0.41062912344932556,
+      "learning_rate": 3.1137137178519983e-06,
+      "loss": 0.2426,
+      "step": 1286
+    },
+    {
+      "epoch": 1.4395973154362416,
+      "grad_norm": 0.4689706265926361,
+      "learning_rate": 3.1105579346637083e-06,
+      "loss": 0.2215,
+      "step": 1287
+    },
+    {
+      "epoch": 1.4407158836689038,
+      "grad_norm": 0.4303361773490906,
+      "learning_rate": 3.1074011165431704e-06,
+      "loss": 0.2398,
+      "step": 1288
+    },
+    {
+      "epoch": 1.441834451901566,
+      "grad_norm": 0.4553805887699127,
+      "learning_rate": 3.10424326884138e-06,
+      "loss": 0.2957,
+      "step": 1289
+    },
+    {
+      "epoch": 1.4429530201342282,
+      "grad_norm": 0.4130866229534149,
+      "learning_rate": 3.101084396911077e-06,
+      "loss": 0.3306,
+      "step": 1290
+    },
+    {
+      "epoch": 1.4440715883668904,
+      "grad_norm": 0.44871020317077637,
+      "learning_rate": 3.0979245061067386e-06,
+      "loss": 0.2717,
+      "step": 1291
+    },
+    {
+      "epoch": 1.4451901565995526,
+      "grad_norm": 0.4335297644138336,
+      "learning_rate": 3.09476360178457e-06,
+      "loss": 0.2981,
+      "step": 1292
+    },
+    {
+      "epoch": 1.4463087248322148,
+      "grad_norm": 0.49229660630226135,
+      "learning_rate": 3.091601689302492e-06,
+      "loss": 0.3473,
+      "step": 1293
+    },
+    {
+      "epoch": 1.447427293064877,
+      "grad_norm": 0.4080130159854889,
+      "learning_rate": 3.088438774020136e-06,
+      "loss": 0.2694,
+      "step": 1294
+    },
+    {
+      "epoch": 1.4485458612975393,
+      "grad_norm": 0.4102078974246979,
+      "learning_rate": 3.0852748612988333e-06,
+      "loss": 0.2351,
+      "step": 1295
+    },
+    {
+      "epoch": 1.4496644295302015,
+      "grad_norm": 0.4815475046634674,
+      "learning_rate": 3.082109956501604e-06,
+      "loss": 0.2386,
+      "step": 1296
+    },
+    {
+      "epoch": 1.4507829977628635,
+      "grad_norm": 0.4592413902282715,
+      "learning_rate": 3.0789440649931525e-06,
+      "loss": 0.269,
+      "step": 1297
+    },
+    {
+      "epoch": 1.4519015659955257,
+      "grad_norm": 0.45215940475463867,
+      "learning_rate": 3.0757771921398537e-06,
+      "loss": 0.4034,
+      "step": 1298
+    },
+    {
+      "epoch": 1.4530201342281879,
+      "grad_norm": 0.4564918577671051,
+      "learning_rate": 3.072609343309747e-06,
+      "loss": 0.2654,
+      "step": 1299
+    },
+    {
+      "epoch": 1.45413870246085,
+      "grad_norm": 0.4333513677120209,
+      "learning_rate": 3.069440523872526e-06,
+      "loss": 0.2462,
+      "step": 1300
+    },
+    {
+      "epoch": 1.45413870246085,
+      "eval_loss": 0.2780807316303253,
+      "eval_runtime": 376.2125,
+      "eval_samples_per_second": 2.113,
+      "eval_steps_per_second": 2.113,
+      "step": 1300
+    },
+    {
+      "epoch": 1.4552572706935123,
+      "grad_norm": 0.4474118947982788,
+      "learning_rate": 3.066270739199529e-06,
+      "loss": 0.2629,
+      "step": 1301
+    },
+    {
+      "epoch": 1.4563758389261745,
+      "grad_norm": 0.49198320508003235,
+      "learning_rate": 3.063099994663731e-06,
+      "loss": 0.3647,
+      "step": 1302
+    },
+    {
+      "epoch": 1.4574944071588367,
+      "grad_norm": 0.4739484190940857,
+      "learning_rate": 3.0599282956397335e-06,
+      "loss": 0.2351,
+      "step": 1303
+    },
+    {
+      "epoch": 1.458612975391499,
+      "grad_norm": 0.44590628147125244,
+      "learning_rate": 3.056755647503756e-06,
+      "loss": 0.3034,
+      "step": 1304
+    },
+    {
+      "epoch": 1.4597315436241611,
+      "grad_norm": 0.49475157260894775,
+      "learning_rate": 3.0535820556336278e-06,
+      "loss": 0.2731,
+      "step": 1305
+    },
+    {
+      "epoch": 1.4608501118568233,
+      "grad_norm": 0.49894434213638306,
+      "learning_rate": 3.050407525408776e-06,
+      "loss": 0.2749,
+      "step": 1306
+    },
+    {
+      "epoch": 1.4619686800894853,
+      "grad_norm": 0.6063686609268188,
+      "learning_rate": 3.0472320622102207e-06,
+      "loss": 0.3673,
+      "step": 1307
+    },
+    {
+      "epoch": 1.4630872483221475,
+      "grad_norm": 0.43170592188835144,
+      "learning_rate": 3.044055671420561e-06,
+      "loss": 0.2198,
+      "step": 1308
+    },
+    {
+      "epoch": 1.4642058165548097,
+      "grad_norm": 0.4351585805416107,
+      "learning_rate": 3.0408783584239705e-06,
+      "loss": 0.2999,
+      "step": 1309
+    },
+    {
+      "epoch": 1.465324384787472,
+      "grad_norm": 0.44208183884620667,
+      "learning_rate": 3.0377001286061836e-06,
+      "loss": 0.3575,
+      "step": 1310
+    },
+    {
+      "epoch": 1.4664429530201342,
+      "grad_norm": 0.4512241780757904,
+      "learning_rate": 3.0345209873544913e-06,
+      "loss": 0.2517,
+      "step": 1311
+    },
+    {
+      "epoch": 1.4675615212527964,
+      "grad_norm": 0.42034780979156494,
+      "learning_rate": 3.0313409400577265e-06,
+      "loss": 0.2332,
+      "step": 1312
+    },
+    {
+      "epoch": 1.4686800894854586,
+      "grad_norm": 0.43253034353256226,
+      "learning_rate": 3.0281599921062626e-06,
+      "loss": 0.2596,
+      "step": 1313
+    },
+    {
+      "epoch": 1.4697986577181208,
+      "grad_norm": 0.5201380252838135,
+      "learning_rate": 3.024978148891995e-06,
+      "loss": 0.2669,
+      "step": 1314
+    },
+    {
+      "epoch": 1.470917225950783,
+      "grad_norm": 0.47377583384513855,
+      "learning_rate": 3.0217954158083384e-06,
+      "loss": 0.2302,
+      "step": 1315
+    },
+    {
+      "epoch": 1.4720357941834452,
+      "grad_norm": 0.43780386447906494,
+      "learning_rate": 3.018611798250216e-06,
+      "loss": 0.2779,
+      "step": 1316
+    },
+    {
+      "epoch": 1.4731543624161074,
+      "grad_norm": 0.44429677724838257,
+      "learning_rate": 3.015427301614051e-06,
+      "loss": 0.2714,
+      "step": 1317
+    },
+    {
+      "epoch": 1.4742729306487696,
+      "grad_norm": 0.41376063227653503,
+      "learning_rate": 3.012241931297754e-06,
+      "loss": 0.2166,
+      "step": 1318
+    },
+    {
+      "epoch": 1.4753914988814318,
+      "grad_norm": 0.4291108548641205,
+      "learning_rate": 3.00905569270072e-06,
+      "loss": 0.306,
+      "step": 1319
+    },
+    {
+      "epoch": 1.476510067114094,
+      "grad_norm": 0.4623241722583771,
+      "learning_rate": 3.005868591223814e-06,
+      "loss": 0.2575,
+      "step": 1320
+    },
+    {
+      "epoch": 1.4776286353467563,
+      "grad_norm": 0.4402950406074524,
+      "learning_rate": 3.0026806322693625e-06,
+      "loss": 0.2863,
+      "step": 1321
+    },
+    {
+      "epoch": 1.4787472035794185,
+      "grad_norm": 0.41231775283813477,
+      "learning_rate": 2.9994918212411483e-06,
+      "loss": 0.2303,
+      "step": 1322
+    },
+    {
+      "epoch": 1.4798657718120805,
+      "grad_norm": 0.4511764645576477,
+      "learning_rate": 2.996302163544397e-06,
+      "loss": 0.232,
+      "step": 1323
+    },
+    {
+      "epoch": 1.4809843400447427,
+      "grad_norm": 0.478270947933197,
+      "learning_rate": 2.9931116645857683e-06,
+      "loss": 0.3482,
+      "step": 1324
+    },
+    {
+      "epoch": 1.4821029082774049,
+      "grad_norm": 0.4859148859977722,
+      "learning_rate": 2.989920329773349e-06,
+      "loss": 0.2553,
+      "step": 1325
+    },
+    {
+      "epoch": 1.483221476510067,
+      "grad_norm": 0.5420084595680237,
+      "learning_rate": 2.9867281645166433e-06,
+      "loss": 0.2735,
+      "step": 1326
+    },
+    {
+      "epoch": 1.4843400447427293,
+      "grad_norm": 0.4275536835193634,
+      "learning_rate": 2.983535174226562e-06,
+      "loss": 0.2274,
+      "step": 1327
+    },
+    {
+      "epoch": 1.4854586129753915,
+      "grad_norm": 0.46758130192756653,
+      "learning_rate": 2.9803413643154162e-06,
+      "loss": 0.2663,
+      "step": 1328
+    },
+    {
+      "epoch": 1.4865771812080537,
+      "grad_norm": 0.5383442044258118,
+      "learning_rate": 2.9771467401969017e-06,
+      "loss": 0.3537,
+      "step": 1329
+    },
+    {
+      "epoch": 1.487695749440716,
+      "grad_norm": 0.46814313530921936,
+      "learning_rate": 2.9739513072861005e-06,
+      "loss": 0.2943,
+      "step": 1330
+    },
+    {
+      "epoch": 1.4888143176733781,
+      "grad_norm": 0.4617949426174164,
+      "learning_rate": 2.970755070999461e-06,
+      "loss": 0.2875,
+      "step": 1331
+    },
+    {
+      "epoch": 1.4899328859060403,
+      "grad_norm": 0.42138421535491943,
+      "learning_rate": 2.967558036754796e-06,
+      "loss": 0.2172,
+      "step": 1332
+    },
+    {
+      "epoch": 1.4910514541387023,
+      "grad_norm": 0.4050348997116089,
+      "learning_rate": 2.9643602099712697e-06,
+      "loss": 0.2229,
+      "step": 1333
+    },
+    {
+      "epoch": 1.4921700223713645,
+      "grad_norm": 0.41779711842536926,
+      "learning_rate": 2.96116159606939e-06,
+      "loss": 0.2829,
+      "step": 1334
+    },
+    {
+      "epoch": 1.4932885906040267,
+      "grad_norm": 0.47670188546180725,
+      "learning_rate": 2.957962200470999e-06,
+      "loss": 0.3706,
+      "step": 1335
+    },
+    {
+      "epoch": 1.494407158836689,
+      "grad_norm": 0.4393787384033203,
+      "learning_rate": 2.954762028599263e-06,
+      "loss": 0.282,
+      "step": 1336
+    },
+    {
+      "epoch": 1.4955257270693512,
+      "grad_norm": 0.5207912921905518,
+      "learning_rate": 2.951561085878666e-06,
+      "loss": 0.3231,
+      "step": 1337
+    },
+    {
+      "epoch": 1.4966442953020134,
+      "grad_norm": 0.5009216070175171,
+      "learning_rate": 2.9483593777349967e-06,
+      "loss": 0.2538,
+      "step": 1338
+    },
+    {
+      "epoch": 1.4977628635346756,
+      "grad_norm": 0.45280393958091736,
+      "learning_rate": 2.9451569095953427e-06,
+      "loss": 0.2748,
+      "step": 1339
+    },
+    {
+      "epoch": 1.4988814317673378,
+      "grad_norm": 0.46338266134262085,
+      "learning_rate": 2.9419536868880783e-06,
+      "loss": 0.3031,
+      "step": 1340
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.38219305872917175,
+      "learning_rate": 2.9387497150428585e-06,
+      "loss": 0.2243,
+      "step": 1341
+    },
+    {
+      "epoch": 1.5011185682326622,
+      "grad_norm": 0.5096919536590576,
+      "learning_rate": 2.935544999490607e-06,
+      "loss": 0.3487,
+      "step": 1342
+    },
+    {
+      "epoch": 1.5022371364653244,
+      "grad_norm": 0.5015236735343933,
+      "learning_rate": 2.932339545663509e-06,
+      "loss": 0.2836,
+      "step": 1343
+    },
+    {
+      "epoch": 1.5033557046979866,
+      "grad_norm": 0.4565114378929138,
+      "learning_rate": 2.929133358995e-06,
+      "loss": 0.2527,
+      "step": 1344
+    },
+    {
+      "epoch": 1.5044742729306488,
+      "grad_norm": 0.47672224044799805,
+      "learning_rate": 2.9259264449197588e-06,
+      "loss": 0.3038,
+      "step": 1345
+    },
+    {
+      "epoch": 1.505592841163311,
+      "grad_norm": 0.4446094334125519,
+      "learning_rate": 2.922718808873696e-06,
+      "loss": 0.3938,
+      "step": 1346
+    },
+    {
+      "epoch": 1.5067114093959733,
+      "grad_norm": 0.47189855575561523,
+      "learning_rate": 2.9195104562939475e-06,
+      "loss": 0.2136,
+      "step": 1347
+    },
+    {
+      "epoch": 1.5078299776286355,
+      "grad_norm": 0.4714747369289398,
+      "learning_rate": 2.9163013926188633e-06,
+      "loss": 0.2284,
+      "step": 1348
+    },
+    {
+      "epoch": 1.5089485458612977,
+      "grad_norm": 0.4687177240848541,
+      "learning_rate": 2.9130916232879975e-06,
+      "loss": 0.2687,
+      "step": 1349
+    },
+    {
+      "epoch": 1.5100671140939599,
+      "grad_norm": 0.4531817138195038,
+      "learning_rate": 2.9098811537421017e-06,
+      "loss": 0.2154,
+      "step": 1350
+    },
+    {
+      "epoch": 1.5100671140939599,
+      "eval_loss": 0.2774081826210022,
+      "eval_runtime": 376.1445,
+      "eval_samples_per_second": 2.114,
+      "eval_steps_per_second": 2.114,
+      "step": 1350
+    },
+    {
+      "epoch": 1.5111856823266219,
+      "grad_norm": 0.5027119517326355,
+      "learning_rate": 2.9066699894231136e-06,
+      "loss": 0.2734,
+      "step": 1351
+    },
+    {
+      "epoch": 1.512304250559284,
+      "grad_norm": 0.46002259850502014,
+      "learning_rate": 2.9034581357741497e-06,
+      "loss": 0.2667,
+      "step": 1352
+    },
+    {
+      "epoch": 1.5134228187919463,
+      "grad_norm": 0.4395085573196411,
+      "learning_rate": 2.9002455982394946e-06,
+      "loss": 0.2492,
+      "step": 1353
+    },
+    {
+      "epoch": 1.5145413870246085,
+      "grad_norm": 0.4770921766757965,
+      "learning_rate": 2.8970323822645897e-06,
+      "loss": 0.2935,
+      "step": 1354
+    },
+    {
+      "epoch": 1.5156599552572707,
+      "grad_norm": 0.4344189763069153,
+      "learning_rate": 2.8938184932960293e-06,
+      "loss": 0.2888,
+      "step": 1355
+    },
+    {
+      "epoch": 1.516778523489933,
+      "grad_norm": 0.48144280910491943,
+      "learning_rate": 2.8906039367815486e-06,
+      "loss": 0.2541,
+      "step": 1356
+    },
+    {
+      "epoch": 1.5178970917225951,
+      "grad_norm": 0.49034619331359863,
+      "learning_rate": 2.8873887181700127e-06,
+      "loss": 0.2469,
+      "step": 1357
+    },
+    {
+      "epoch": 1.5190156599552571,
+      "grad_norm": 0.5554848909378052,
+      "learning_rate": 2.884172842911409e-06,
+      "loss": 0.2755,
+      "step": 1358
+    },
+    {
+      "epoch": 1.5201342281879193,
+      "grad_norm": 0.4101789593696594,
+      "learning_rate": 2.88095631645684e-06,
+      "loss": 0.3094,
+      "step": 1359
+    },
+    {
+      "epoch": 1.5212527964205815,
+      "grad_norm": 0.459759384393692,
+      "learning_rate": 2.877739144258509e-06,
+      "loss": 0.262,
+      "step": 1360
+    },
+    {
+      "epoch": 1.5223713646532437,
+      "grad_norm": 0.48272883892059326,
+      "learning_rate": 2.8745213317697164e-06,
+      "loss": 0.3004,
+      "step": 1361
+    },
+    {
+      "epoch": 1.523489932885906,
+      "grad_norm": 0.46975675225257874,
+      "learning_rate": 2.8713028844448467e-06,
+      "loss": 0.2796,
+      "step": 1362
+    },
+    {
+      "epoch": 1.5246085011185682,
+      "grad_norm": 0.4712791442871094,
+      "learning_rate": 2.8680838077393606e-06,
+      "loss": 0.2661,
+      "step": 1363
+    },
+    {
+      "epoch": 1.5257270693512304,
+      "grad_norm": 0.5124219655990601,
+      "learning_rate": 2.8648641071097867e-06,
+      "loss": 0.2975,
+      "step": 1364
+    },
+    {
+      "epoch": 1.5268456375838926,
+      "grad_norm": 0.48896124958992004,
+      "learning_rate": 2.8616437880137095e-06,
+      "loss": 0.4128,
+      "step": 1365
+    },
+    {
+      "epoch": 1.5279642058165548,
+      "grad_norm": 0.5106556415557861,
+      "learning_rate": 2.8584228559097625e-06,
+      "loss": 0.3077,
+      "step": 1366
+    },
+    {
+      "epoch": 1.529082774049217,
+      "grad_norm": 0.4693504869937897,
+      "learning_rate": 2.8552013162576186e-06,
+      "loss": 0.3336,
+      "step": 1367
+    },
+    {
+      "epoch": 1.5302013422818792,
+      "grad_norm": 0.48061493039131165,
+      "learning_rate": 2.851979174517981e-06,
+      "loss": 0.3096,
+      "step": 1368
+    },
+    {
+      "epoch": 1.5313199105145414,
+      "grad_norm": 0.5525304079055786,
+      "learning_rate": 2.8487564361525723e-06,
+      "loss": 0.3765,
+      "step": 1369
+    },
+    {
+      "epoch": 1.5324384787472036,
+      "grad_norm": 0.41975098848342896,
+      "learning_rate": 2.8455331066241266e-06,
+      "loss": 0.2095,
+      "step": 1370
+    },
+    {
+      "epoch": 1.5335570469798658,
+      "grad_norm": 0.4772942364215851,
+      "learning_rate": 2.8423091913963812e-06,
+      "loss": 0.3532,
+      "step": 1371
+    },
+    {
+      "epoch": 1.534675615212528,
+      "grad_norm": 0.58744877576828,
+      "learning_rate": 2.839084695934064e-06,
+      "loss": 0.2461,
+      "step": 1372
+    },
+    {
+      "epoch": 1.5357941834451903,
+      "grad_norm": 0.5044638514518738,
+      "learning_rate": 2.8358596257028898e-06,
+      "loss": 0.271,
+      "step": 1373
+    },
+    {
+      "epoch": 1.5369127516778525,
+      "grad_norm": 0.48165133595466614,
+      "learning_rate": 2.8326339861695447e-06,
+      "loss": 0.311,
+      "step": 1374
+    },
+    {
+      "epoch": 1.5380313199105147,
+      "grad_norm": 0.430262953042984,
+      "learning_rate": 2.8294077828016807e-06,
+      "loss": 0.341,
+      "step": 1375
+    },
+    {
+      "epoch": 1.5391498881431769,
+      "grad_norm": 0.48565196990966797,
+      "learning_rate": 2.826181021067906e-06,
+      "loss": 0.2613,
+      "step": 1376
+    },
+    {
+      "epoch": 1.540268456375839,
+      "grad_norm": 0.41178154945373535,
+      "learning_rate": 2.8229537064377736e-06,
+      "loss": 0.1784,
+      "step": 1377
+    },
+    {
+      "epoch": 1.541387024608501,
+      "grad_norm": 0.5487228035926819,
+      "learning_rate": 2.8197258443817777e-06,
+      "loss": 0.2458,
+      "step": 1378
+    },
+    {
+      "epoch": 1.5425055928411633,
+      "grad_norm": 0.5440687537193298,
+      "learning_rate": 2.816497440371335e-06,
+      "loss": 0.2926,
+      "step": 1379
+    },
+    {
+      "epoch": 1.5436241610738255,
+      "grad_norm": 0.5031837821006775,
+      "learning_rate": 2.8132684998787844e-06,
+      "loss": 0.2878,
+      "step": 1380
+    },
+    {
+      "epoch": 1.5447427293064877,
+      "grad_norm": 0.46020251512527466,
+      "learning_rate": 2.8100390283773743e-06,
+      "loss": 0.3035,
+      "step": 1381
+    },
+    {
+      "epoch": 1.54586129753915,
+      "grad_norm": 0.5148407816886902,
+      "learning_rate": 2.806809031341251e-06,
+      "loss": 0.2549,
+      "step": 1382
+    },
+    {
+      "epoch": 1.5469798657718121,
+      "grad_norm": 0.454123854637146,
+      "learning_rate": 2.8035785142454535e-06,
+      "loss": 0.2432,
+      "step": 1383
+    },
+    {
+      "epoch": 1.548098434004474,
+      "grad_norm": 0.43279069662094116,
+      "learning_rate": 2.8003474825659016e-06,
+      "loss": 0.2499,
+      "step": 1384
+    },
+    {
+      "epoch": 1.5492170022371363,
+      "grad_norm": 0.5416368246078491,
+      "learning_rate": 2.7971159417793868e-06,
+      "loss": 0.2746,
+      "step": 1385
+    },
+    {
+      "epoch": 1.5503355704697985,
+      "grad_norm": 0.47241485118865967,
+      "learning_rate": 2.7938838973635656e-06,
+      "loss": 0.265,
+      "step": 1386
+    },
+    {
+      "epoch": 1.5514541387024607,
+      "grad_norm": 0.42764195799827576,
+      "learning_rate": 2.790651354796945e-06,
+      "loss": 0.2291,
+      "step": 1387
+    },
+    {
+      "epoch": 1.552572706935123,
+      "grad_norm": 0.5440428853034973,
+      "learning_rate": 2.7874183195588794e-06,
+      "loss": 0.2331,
+      "step": 1388
+    },
+    {
+      "epoch": 1.5536912751677852,
+      "grad_norm": 0.48521825671195984,
+      "learning_rate": 2.7841847971295564e-06,
+      "loss": 0.2497,
+      "step": 1389
+    },
+    {
+      "epoch": 1.5548098434004474,
+      "grad_norm": 0.46541547775268555,
+      "learning_rate": 2.7809507929899904e-06,
+      "loss": 0.2579,
+      "step": 1390
+    },
+    {
+      "epoch": 1.5559284116331096,
+      "grad_norm": 0.5091689825057983,
+      "learning_rate": 2.777716312622012e-06,
+      "loss": 0.3496,
+      "step": 1391
+    },
+    {
+      "epoch": 1.5570469798657718,
+      "grad_norm": 0.43868714570999146,
+      "learning_rate": 2.774481361508259e-06,
+      "loss": 0.229,
+      "step": 1392
+    },
+    {
+      "epoch": 1.558165548098434,
+      "grad_norm": 0.49231159687042236,
+      "learning_rate": 2.771245945132167e-06,
+      "loss": 0.2614,
+      "step": 1393
+    },
+    {
+      "epoch": 1.5592841163310962,
+      "grad_norm": 0.4468526840209961,
+      "learning_rate": 2.7680100689779606e-06,
+      "loss": 0.2557,
+      "step": 1394
+    },
+    {
+      "epoch": 1.5604026845637584,
+      "grad_norm": 0.4905088245868683,
+      "learning_rate": 2.7647737385306437e-06,
+      "loss": 0.2677,
+      "step": 1395
+    },
+    {
+      "epoch": 1.5615212527964206,
+      "grad_norm": 0.45887669920921326,
+      "learning_rate": 2.76153695927599e-06,
+      "loss": 0.2939,
+      "step": 1396
+    },
+    {
+      "epoch": 1.5626398210290828,
+      "grad_norm": 0.4739925265312195,
+      "learning_rate": 2.7582997367005337e-06,
+      "loss": 0.2549,
+      "step": 1397
+    },
+    {
+      "epoch": 1.563758389261745,
+      "grad_norm": 0.4784601628780365,
+      "learning_rate": 2.755062076291562e-06,
+      "loss": 0.3269,
+      "step": 1398
+    },
+    {
+      "epoch": 1.5648769574944073,
+      "grad_norm": 0.5115184187889099,
+      "learning_rate": 2.751823983537102e-06,
+      "loss": 0.3332,
+      "step": 1399
+    },
+    {
+      "epoch": 1.5659955257270695,
+      "grad_norm": 0.47996172308921814,
+      "learning_rate": 2.748585463925915e-06,
+      "loss": 0.3194,
+      "step": 1400
+    },
+    {
+      "epoch": 1.5659955257270695,
+      "eval_loss": 0.27672630548477173,
+      "eval_runtime": 374.797,
+      "eval_samples_per_second": 2.121,
+      "eval_steps_per_second": 2.121,
+      "step": 1400
+    },
+    {
+      "epoch": 1.5671140939597317,
+      "grad_norm": 0.46196073293685913,
+      "learning_rate": 2.745346522947486e-06,
+      "loss": 0.2494,
+      "step": 1401
+    },
+    {
+      "epoch": 1.5682326621923939,
+      "grad_norm": 0.5226995944976807,
+      "learning_rate": 2.7421071660920152e-06,
+      "loss": 0.3499,
+      "step": 1402
+    },
+    {
+      "epoch": 1.569351230425056,
+      "grad_norm": 0.49888554215431213,
+      "learning_rate": 2.7388673988504027e-06,
+      "loss": 0.2007,
+      "step": 1403
+    },
+    {
+      "epoch": 1.570469798657718,
+      "grad_norm": 0.4598582088947296,
+      "learning_rate": 2.7356272267142508e-06,
+      "loss": 0.3145,
+      "step": 1404
+    },
+    {
+      "epoch": 1.5715883668903803,
+      "grad_norm": 0.5614228844642639,
+      "learning_rate": 2.732386655175844e-06,
+      "loss": 0.3207,
+      "step": 1405
+    },
+    {
+      "epoch": 1.5727069351230425,
+      "grad_norm": 0.45056116580963135,
+      "learning_rate": 2.7291456897281454e-06,
+      "loss": 0.2458,
+      "step": 1406
+    },
+    {
+      "epoch": 1.5738255033557047,
+      "grad_norm": 0.47374850511550903,
+      "learning_rate": 2.7259043358647865e-06,
+      "loss": 0.263,
+      "step": 1407
+    },
+    {
+      "epoch": 1.574944071588367,
+      "grad_norm": 0.5121036767959595,
+      "learning_rate": 2.722662599080054e-06,
+      "loss": 0.2713,
+      "step": 1408
+    },
+    {
+      "epoch": 1.5760626398210291,
+      "grad_norm": 0.4344590902328491,
+      "learning_rate": 2.7194204848688876e-06,
+      "loss": 0.2212,
+      "step": 1409
+    },
+    {
+      "epoch": 1.5771812080536913,
+      "grad_norm": 0.4940078556537628,
+      "learning_rate": 2.7161779987268643e-06,
+      "loss": 0.3368,
+      "step": 1410
+    },
+    {
+      "epoch": 1.5782997762863533,
+      "grad_norm": 0.45780739188194275,
+      "learning_rate": 2.712935146150193e-06,
+      "loss": 0.2638,
+      "step": 1411
+    },
+    {
+      "epoch": 1.5794183445190155,
+      "grad_norm": 0.5181729197502136,
+      "learning_rate": 2.7096919326357012e-06,
+      "loss": 0.3233,
+      "step": 1412
+    },
+    {
+      "epoch": 1.5805369127516777,
+      "grad_norm": 0.4465189576148987,
+      "learning_rate": 2.7064483636808314e-06,
+      "loss": 0.2634,
+      "step": 1413
+    },
+    {
+      "epoch": 1.58165548098434,
+      "grad_norm": 0.5737823843955994,
+      "learning_rate": 2.7032044447836264e-06,
+      "loss": 0.2709,
+      "step": 1414
+    },
+    {
+      "epoch": 1.5827740492170022,
+      "grad_norm": 0.5013507008552551,
+      "learning_rate": 2.6999601814427232e-06,
+      "loss": 0.229,
+      "step": 1415
+    },
+    {
+      "epoch": 1.5838926174496644,
+      "grad_norm": 0.50819331407547,
+      "learning_rate": 2.696715579157343e-06,
+      "loss": 0.2926,
+      "step": 1416
+    },
+    {
+      "epoch": 1.5850111856823266,
+      "grad_norm": 0.4931088089942932,
+      "learning_rate": 2.6934706434272795e-06,
+      "loss": 0.3321,
+      "step": 1417
+    },
+    {
+      "epoch": 1.5861297539149888,
+      "grad_norm": 0.47347313165664673,
+      "learning_rate": 2.6902253797528945e-06,
+      "loss": 0.344,
+      "step": 1418
+    },
+    {
+      "epoch": 1.587248322147651,
+      "grad_norm": 0.41897106170654297,
+      "learning_rate": 2.686979793635103e-06,
+      "loss": 0.2577,
+      "step": 1419
+    },
+    {
+      "epoch": 1.5883668903803132,
+      "grad_norm": 0.43682315945625305,
+      "learning_rate": 2.6837338905753685e-06,
+      "loss": 0.3201,
+      "step": 1420
+    },
+    {
+      "epoch": 1.5894854586129754,
+      "grad_norm": 0.5034173727035522,
+      "learning_rate": 2.68048767607569e-06,
+      "loss": 0.2243,
+      "step": 1421
+    },
+    {
+      "epoch": 1.5906040268456376,
+      "grad_norm": 0.37273097038269043,
+      "learning_rate": 2.677241155638597e-06,
+      "loss": 0.218,
+      "step": 1422
+    },
+    {
+      "epoch": 1.5917225950782998,
+      "grad_norm": 0.4089590609073639,
+      "learning_rate": 2.673994334767135e-06,
+      "loss": 0.2194,
+      "step": 1423
+    },
+    {
+      "epoch": 1.592841163310962,
+      "grad_norm": 0.5197263360023499,
+      "learning_rate": 2.67074721896486e-06,
+      "loss": 0.2727,
+      "step": 1424
+    },
+    {
+      "epoch": 1.5939597315436242,
+      "grad_norm": 0.4723402261734009,
+      "learning_rate": 2.667499813735827e-06,
+      "loss": 0.2642,
+      "step": 1425
+    },
+    {
+      "epoch": 1.5950782997762865,
+      "grad_norm": 0.42288774251937866,
+      "learning_rate": 2.664252124584583e-06,
+      "loss": 0.2706,
+      "step": 1426
+    },
+    {
+      "epoch": 1.5961968680089487,
+      "grad_norm": 0.4269940257072449,
+      "learning_rate": 2.6610041570161565e-06,
+      "loss": 0.2573,
+      "step": 1427
+    },
+    {
+      "epoch": 1.5973154362416109,
+      "grad_norm": 0.5148060917854309,
+      "learning_rate": 2.657755916536045e-06,
+      "loss": 0.2852,
+      "step": 1428
+    },
+    {
+      "epoch": 1.598434004474273,
+      "grad_norm": 0.4956951141357422,
+      "learning_rate": 2.654507408650211e-06,
+      "loss": 0.2911,
+      "step": 1429
+    },
+    {
+      "epoch": 1.599552572706935,
+      "grad_norm": 0.4715212285518646,
+      "learning_rate": 2.65125863886507e-06,
+      "loss": 0.2682,
+      "step": 1430
+    },
+    {
+      "epoch": 1.6006711409395973,
+      "grad_norm": 0.46595996618270874,
+      "learning_rate": 2.6480096126874816e-06,
+      "loss": 0.274,
+      "step": 1431
+    },
+    {
+      "epoch": 1.6017897091722595,
+      "grad_norm": 0.5058538317680359,
+      "learning_rate": 2.6447603356247396e-06,
+      "loss": 0.2979,
+      "step": 1432
+    },
+    {
+      "epoch": 1.6029082774049217,
+      "grad_norm": 0.44332218170166016,
+      "learning_rate": 2.6415108131845636e-06,
+      "loss": 0.3167,
+      "step": 1433
+    },
+    {
+      "epoch": 1.604026845637584,
+      "grad_norm": 0.4438632130622864,
+      "learning_rate": 2.6382610508750873e-06,
+      "loss": 0.2543,
+      "step": 1434
+    },
+    {
+      "epoch": 1.6051454138702461,
+      "grad_norm": 0.49775663018226624,
+      "learning_rate": 2.6350110542048535e-06,
+      "loss": 0.2637,
+      "step": 1435
+    },
+    {
+      "epoch": 1.6062639821029083,
+      "grad_norm": 0.4145890176296234,
+      "learning_rate": 2.631760828682801e-06,
+      "loss": 0.2661,
+      "step": 1436
+    },
+    {
+      "epoch": 1.6073825503355703,
+      "grad_norm": 0.4427662193775177,
+      "learning_rate": 2.628510379818256e-06,
+      "loss": 0.2421,
+      "step": 1437
+    },
+    {
+      "epoch": 1.6085011185682325,
+      "grad_norm": 0.5245040059089661,
+      "learning_rate": 2.6252597131209245e-06,
+      "loss": 0.2858,
+      "step": 1438
+    },
+    {
+      "epoch": 1.6096196868008947,
+      "grad_norm": 0.4110528528690338,
+      "learning_rate": 2.622008834100881e-06,
+      "loss": 0.3516,
+      "step": 1439
+    },
+    {
+      "epoch": 1.610738255033557,
+      "grad_norm": 0.4899514615535736,
+      "learning_rate": 2.61875774826856e-06,
+      "loss": 0.3036,
+      "step": 1440
+    },
+    {
+      "epoch": 1.6118568232662192,
+      "grad_norm": 0.4851590096950531,
+      "learning_rate": 2.6155064611347462e-06,
+      "loss": 0.2318,
+      "step": 1441
+    },
+    {
+      "epoch": 1.6129753914988814,
+      "grad_norm": 0.4479272961616516,
+      "learning_rate": 2.612254978210567e-06,
+      "loss": 0.2696,
+      "step": 1442
+    },
+    {
+      "epoch": 1.6140939597315436,
+      "grad_norm": 0.41481825709342957,
+      "learning_rate": 2.6090033050074792e-06,
+      "loss": 0.2569,
+      "step": 1443
+    },
+    {
+      "epoch": 1.6152125279642058,
+      "grad_norm": 0.5684930086135864,
+      "learning_rate": 2.6057514470372642e-06,
+      "loss": 0.2618,
+      "step": 1444
+    },
+    {
+      "epoch": 1.616331096196868,
+      "grad_norm": 0.4883359372615814,
+      "learning_rate": 2.6024994098120165e-06,
+      "loss": 0.2731,
+      "step": 1445
+    },
+    {
+      "epoch": 1.6174496644295302,
+      "grad_norm": 0.474442720413208,
+      "learning_rate": 2.599247198844133e-06,
+      "loss": 0.2379,
+      "step": 1446
+    },
+    {
+      "epoch": 1.6185682326621924,
+      "grad_norm": 0.5488391518592834,
+      "learning_rate": 2.5959948196463058e-06,
+      "loss": 0.3668,
+      "step": 1447
+    },
+    {
+      "epoch": 1.6196868008948546,
+      "grad_norm": 0.439535528421402,
+      "learning_rate": 2.592742277731513e-06,
+      "loss": 0.3074,
+      "step": 1448
+    },
+    {
+      "epoch": 1.6208053691275168,
+      "grad_norm": 0.48842623829841614,
+      "learning_rate": 2.589489578613009e-06,
+      "loss": 0.2739,
+      "step": 1449
+    },
+    {
+      "epoch": 1.621923937360179,
+      "grad_norm": 0.5381295680999756,
+      "learning_rate": 2.5862367278043106e-06,
+      "loss": 0.275,
+      "step": 1450
+    },
+    {
+      "epoch": 1.621923937360179,
+      "eval_loss": 0.2759144604206085,
+      "eval_runtime": 375.7425,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 2.116,
+      "step": 1450
+    },
+    {
+      "epoch": 1.6230425055928412,
+      "grad_norm": 0.4498196244239807,
+      "learning_rate": 2.582983730819197e-06,
+      "loss": 0.2462,
+      "step": 1451
+    },
+    {
+      "epoch": 1.6241610738255035,
+      "grad_norm": 0.4534889757633209,
+      "learning_rate": 2.5797305931716927e-06,
+      "loss": 0.313,
+      "step": 1452
+    },
+    {
+      "epoch": 1.6252796420581657,
+      "grad_norm": 0.44035589694976807,
+      "learning_rate": 2.5764773203760595e-06,
+      "loss": 0.2358,
+      "step": 1453
+    },
+    {
+      "epoch": 1.6263982102908279,
+      "grad_norm": 0.4503270387649536,
+      "learning_rate": 2.5732239179467897e-06,
+      "loss": 0.2391,
+      "step": 1454
+    },
+    {
+      "epoch": 1.62751677852349,
+      "grad_norm": 0.4582473337650299,
+      "learning_rate": 2.569970391398596e-06,
+      "loss": 0.3332,
+      "step": 1455
+    },
+    {
+      "epoch": 1.6286353467561523,
+      "grad_norm": 0.4385569095611572,
+      "learning_rate": 2.5667167462464e-06,
+      "loss": 0.2944,
+      "step": 1456
+    },
+    {
+      "epoch": 1.6297539149888143,
+      "grad_norm": 0.4995662569999695,
+      "learning_rate": 2.563462988005326e-06,
+      "loss": 0.266,
+      "step": 1457
+    },
+    {
+      "epoch": 1.6308724832214765,
+      "grad_norm": 0.5043554306030273,
+      "learning_rate": 2.5602091221906878e-06,
+      "loss": 0.2693,
+      "step": 1458
+    },
+    {
+      "epoch": 1.6319910514541387,
+      "grad_norm": 0.5058841705322266,
+      "learning_rate": 2.5569551543179834e-06,
+      "loss": 0.2775,
+      "step": 1459
+    },
+    {
+      "epoch": 1.633109619686801,
+      "grad_norm": 0.47192928194999695,
+      "learning_rate": 2.5537010899028823e-06,
+      "loss": 0.3064,
+      "step": 1460
+    },
+    {
+      "epoch": 1.6342281879194631,
+      "grad_norm": 0.4847210645675659,
+      "learning_rate": 2.55044693446122e-06,
+      "loss": 0.39,
+      "step": 1461
+    },
+    {
+      "epoch": 1.6353467561521253,
+      "grad_norm": 0.4588189125061035,
+      "learning_rate": 2.5471926935089834e-06,
+      "loss": 0.2633,
+      "step": 1462
+    },
+    {
+      "epoch": 1.6364653243847873,
+      "grad_norm": 0.4846605360507965,
+      "learning_rate": 2.543938372562307e-06,
+      "loss": 0.3152,
+      "step": 1463
+    },
+    {
+      "epoch": 1.6375838926174495,
+      "grad_norm": 0.4680267572402954,
+      "learning_rate": 2.540683977137458e-06,
+      "loss": 0.3227,
+      "step": 1464
+    },
+    {
+      "epoch": 1.6387024608501117,
+      "grad_norm": 0.5017008185386658,
+      "learning_rate": 2.5374295127508333e-06,
+      "loss": 0.2963,
+      "step": 1465
+    },
+    {
+      "epoch": 1.639821029082774,
+      "grad_norm": 0.5527923703193665,
+      "learning_rate": 2.5341749849189428e-06,
+      "loss": 0.281,
+      "step": 1466
+    },
+    {
+      "epoch": 1.6409395973154361,
+      "grad_norm": 0.5300388932228088,
+      "learning_rate": 2.5309203991584076e-06,
+      "loss": 0.2664,
+      "step": 1467
+    },
+    {
+      "epoch": 1.6420581655480984,
+      "grad_norm": 0.5231226086616516,
+      "learning_rate": 2.527665760985945e-06,
+      "loss": 0.2508,
+      "step": 1468
+    },
+    {
+      "epoch": 1.6431767337807606,
+      "grad_norm": 0.5145414471626282,
+      "learning_rate": 2.524411075918361e-06,
+      "loss": 0.2706,
+      "step": 1469
+    },
+    {
+      "epoch": 1.6442953020134228,
+      "grad_norm": 0.46745577454566956,
+      "learning_rate": 2.5211563494725424e-06,
+      "loss": 0.2379,
+      "step": 1470
+    },
+    {
+      "epoch": 1.645413870246085,
+      "grad_norm": 0.421368271112442,
+      "learning_rate": 2.517901587165445e-06,
+      "loss": 0.279,
+      "step": 1471
+    },
+    {
+      "epoch": 1.6465324384787472,
+      "grad_norm": 0.4670732021331787,
+      "learning_rate": 2.514646794514085e-06,
+      "loss": 0.2141,
+      "step": 1472
+    },
+    {
+      "epoch": 1.6476510067114094,
+      "grad_norm": 0.5940575003623962,
+      "learning_rate": 2.5113919770355316e-06,
+      "loss": 0.3001,
+      "step": 1473
+    },
+    {
+      "epoch": 1.6487695749440716,
+      "grad_norm": 0.5621731281280518,
+      "learning_rate": 2.5081371402468955e-06,
+      "loss": 0.2596,
+      "step": 1474
+    },
+    {
+      "epoch": 1.6498881431767338,
+      "grad_norm": 0.4713788628578186,
+      "learning_rate": 2.50488228966532e-06,
+      "loss": 0.3097,
+      "step": 1475
+    },
+    {
+      "epoch": 1.651006711409396,
+      "grad_norm": 0.4422723948955536,
+      "learning_rate": 2.501627430807971e-06,
+      "loss": 0.3222,
+      "step": 1476
+    },
+    {
+      "epoch": 1.6521252796420582,
+      "grad_norm": 0.44848915934562683,
+      "learning_rate": 2.4983725691920296e-06,
+      "loss": 0.217,
+      "step": 1477
+    },
+    {
+      "epoch": 1.6532438478747205,
+      "grad_norm": 0.4025619328022003,
+      "learning_rate": 2.49511771033468e-06,
+      "loss": 0.2525,
+      "step": 1478
+    },
+    {
+      "epoch": 1.6543624161073827,
+      "grad_norm": 0.477193683385849,
+      "learning_rate": 2.4918628597531045e-06,
+      "loss": 0.2437,
+      "step": 1479
+    },
+    {
+      "epoch": 1.6554809843400449,
+      "grad_norm": 0.42993927001953125,
+      "learning_rate": 2.4886080229644684e-06,
+      "loss": 0.2551,
+      "step": 1480
+    },
+    {
+      "epoch": 1.656599552572707,
+      "grad_norm": 0.4778929650783539,
+      "learning_rate": 2.485353205485915e-06,
+      "loss": 0.2783,
+      "step": 1481
+    },
+    {
+      "epoch": 1.6577181208053693,
+      "grad_norm": 0.43891024589538574,
+      "learning_rate": 2.482098412834556e-06,
+      "loss": 0.2552,
+      "step": 1482
+    },
+    {
+      "epoch": 1.6588366890380313,
+      "grad_norm": 0.47965750098228455,
+      "learning_rate": 2.478843650527458e-06,
+      "loss": 0.2637,
+      "step": 1483
+    },
+    {
+      "epoch": 1.6599552572706935,
+      "grad_norm": 0.5575161576271057,
+      "learning_rate": 2.4755889240816397e-06,
+      "loss": 0.2817,
+      "step": 1484
+    },
+    {
+      "epoch": 1.6610738255033557,
+      "grad_norm": 0.42932045459747314,
+      "learning_rate": 2.472334239014056e-06,
+      "loss": 0.2434,
+      "step": 1485
+    },
+    {
+      "epoch": 1.662192393736018,
+      "grad_norm": 0.5370996594429016,
+      "learning_rate": 2.4690796008415928e-06,
+      "loss": 0.2664,
+      "step": 1486
+    },
+    {
+      "epoch": 1.6633109619686801,
+      "grad_norm": 0.4575931429862976,
+      "learning_rate": 2.4658250150810576e-06,
+      "loss": 0.2542,
+      "step": 1487
+    },
+    {
+      "epoch": 1.6644295302013423,
+      "grad_norm": 0.4369608461856842,
+      "learning_rate": 2.4625704872491676e-06,
+      "loss": 0.2964,
+      "step": 1488
+    },
+    {
+      "epoch": 1.6655480984340043,
+      "grad_norm": 0.4544220566749573,
+      "learning_rate": 2.4593160228625423e-06,
+      "loss": 0.2316,
+      "step": 1489
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.46925288438796997,
+      "learning_rate": 2.456061627437694e-06,
+      "loss": 0.2676,
+      "step": 1490
+    },
+    {
+      "epoch": 1.6677852348993287,
+      "grad_norm": 0.5094079971313477,
+      "learning_rate": 2.4528073064910175e-06,
+      "loss": 0.3158,
+      "step": 1491
+    },
+    {
+      "epoch": 1.668903803131991,
+      "grad_norm": 0.47183501720428467,
+      "learning_rate": 2.4495530655387813e-06,
+      "loss": 0.2619,
+      "step": 1492
+    },
+    {
+      "epoch": 1.6700223713646531,
+      "grad_norm": 0.5022105574607849,
+      "learning_rate": 2.446298910097119e-06,
+      "loss": 0.3357,
+      "step": 1493
+    },
+    {
+      "epoch": 1.6711409395973154,
+      "grad_norm": 0.5041405558586121,
+      "learning_rate": 2.443044845682017e-06,
+      "loss": 0.2393,
+      "step": 1494
+    },
+    {
+      "epoch": 1.6722595078299776,
+      "grad_norm": 0.495934396982193,
+      "learning_rate": 2.4397908778093126e-06,
+      "loss": 0.2777,
+      "step": 1495
+    },
+    {
+      "epoch": 1.6733780760626398,
+      "grad_norm": 0.4248402416706085,
+      "learning_rate": 2.4365370119946746e-06,
+      "loss": 0.2144,
+      "step": 1496
+    },
+    {
+      "epoch": 1.674496644295302,
+      "grad_norm": 0.46098002791404724,
+      "learning_rate": 2.4332832537536003e-06,
+      "loss": 0.2413,
+      "step": 1497
+    },
+    {
+      "epoch": 1.6756152125279642,
+      "grad_norm": 0.4943296015262604,
+      "learning_rate": 2.4300296086014043e-06,
+      "loss": 0.2593,
+      "step": 1498
+    },
+    {
+      "epoch": 1.6767337807606264,
+      "grad_norm": 0.5300238728523254,
+      "learning_rate": 2.4267760820532107e-06,
+      "loss": 0.2777,
+      "step": 1499
+    },
+    {
+      "epoch": 1.6778523489932886,
+      "grad_norm": 0.4185192584991455,
+      "learning_rate": 2.4235226796239414e-06,
+      "loss": 0.2676,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6778523489932886,
+      "eval_loss": 0.275237500667572,
+      "eval_runtime": 375.6796,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 2.116,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6789709172259508,
+      "grad_norm": 0.4575055241584778,
+      "learning_rate": 2.420269406828308e-06,
+      "loss": 0.3427,
+      "step": 1501
+    },
+    {
+      "epoch": 1.680089485458613,
+      "grad_norm": 0.48837757110595703,
+      "learning_rate": 2.4170162691808034e-06,
+      "loss": 0.2545,
+      "step": 1502
+    },
+    {
+      "epoch": 1.6812080536912752,
+      "grad_norm": 0.44575124979019165,
+      "learning_rate": 2.41376327219569e-06,
+      "loss": 0.2165,
+      "step": 1503
+    },
+    {
+      "epoch": 1.6823266219239374,
+      "grad_norm": 0.47158029675483704,
+      "learning_rate": 2.4105104213869925e-06,
+      "loss": 0.2831,
+      "step": 1504
+    },
+    {
+      "epoch": 1.6834451901565997,
+      "grad_norm": 0.3921125829219818,
+      "learning_rate": 2.4072577222684874e-06,
+      "loss": 0.2413,
+      "step": 1505
+    },
+    {
+      "epoch": 1.6845637583892619,
+      "grad_norm": 0.5641574859619141,
+      "learning_rate": 2.404005180353695e-06,
+      "loss": 0.3595,
+      "step": 1506
+    },
+    {
+      "epoch": 1.685682326621924,
+      "grad_norm": 0.4749375879764557,
+      "learning_rate": 2.400752801155868e-06,
+      "loss": 0.2849,
+      "step": 1507
+    },
+    {
+      "epoch": 1.6868008948545863,
+      "grad_norm": 0.5459949970245361,
+      "learning_rate": 2.397500590187985e-06,
+      "loss": 0.2635,
+      "step": 1508
+    },
+    {
+      "epoch": 1.6879194630872483,
+      "grad_norm": 0.42180827260017395,
+      "learning_rate": 2.394248552962737e-06,
+      "loss": 0.3156,
+      "step": 1509
+    },
+    {
+      "epoch": 1.6890380313199105,
+      "grad_norm": 0.48319974541664124,
+      "learning_rate": 2.3909966949925224e-06,
+      "loss": 0.258,
+      "step": 1510
+    },
+    {
+      "epoch": 1.6901565995525727,
+      "grad_norm": 0.5040591955184937,
+      "learning_rate": 2.3877450217894336e-06,
+      "loss": 0.3478,
+      "step": 1511
+    },
+    {
+      "epoch": 1.691275167785235,
+      "grad_norm": 0.5428957343101501,
+      "learning_rate": 2.3844935388652546e-06,
+      "loss": 0.2652,
+      "step": 1512
+    },
+    {
+      "epoch": 1.692393736017897,
+      "grad_norm": 0.4393710196018219,
+      "learning_rate": 2.381242251731441e-06,
+      "loss": 0.2434,
+      "step": 1513
+    },
+    {
+      "epoch": 1.6935123042505593,
+      "grad_norm": 0.48532846570014954,
+      "learning_rate": 2.37799116589912e-06,
+      "loss": 0.3033,
+      "step": 1514
+    },
+    {
+      "epoch": 1.6946308724832215,
+      "grad_norm": 0.49175703525543213,
+      "learning_rate": 2.3747402868790763e-06,
+      "loss": 0.2941,
+      "step": 1515
+    },
+    {
+      "epoch": 1.6957494407158835,
+      "grad_norm": 0.539387583732605,
+      "learning_rate": 2.3714896201817446e-06,
+      "loss": 0.281,
+      "step": 1516
+    },
+    {
+      "epoch": 1.6968680089485457,
+      "grad_norm": 0.46241310238838196,
+      "learning_rate": 2.3682391713172e-06,
+      "loss": 0.4632,
+      "step": 1517
+    },
+    {
+      "epoch": 1.697986577181208,
+      "grad_norm": 0.4993603527545929,
+      "learning_rate": 2.3649889457951473e-06,
+      "loss": 0.3288,
+      "step": 1518
+    },
+    {
+      "epoch": 1.6991051454138701,
+      "grad_norm": 0.47136905789375305,
+      "learning_rate": 2.361738949124913e-06,
+      "loss": 0.3277,
+      "step": 1519
+    },
+    {
+      "epoch": 1.7002237136465324,
+      "grad_norm": 0.5480743646621704,
+      "learning_rate": 2.3584891868154377e-06,
+      "loss": 0.2968,
+      "step": 1520
+    },
+    {
+      "epoch": 1.7013422818791946,
+      "grad_norm": 0.5108246803283691,
+      "learning_rate": 2.355239664375261e-06,
+      "loss": 0.2294,
+      "step": 1521
+    },
+    {
+      "epoch": 1.7024608501118568,
+      "grad_norm": 0.4914730191230774,
+      "learning_rate": 2.3519903873125192e-06,
+      "loss": 0.2832,
+      "step": 1522
+    },
+    {
+      "epoch": 1.703579418344519,
+      "grad_norm": 0.5753250122070312,
+      "learning_rate": 2.3487413611349307e-06,
+      "loss": 0.3134,
+      "step": 1523
+    },
+    {
+      "epoch": 1.7046979865771812,
+      "grad_norm": 0.4844762682914734,
+      "learning_rate": 2.3454925913497905e-06,
+      "loss": 0.3104,
+      "step": 1524
+    },
+    {
+      "epoch": 1.7058165548098434,
+      "grad_norm": 0.5251889228820801,
+      "learning_rate": 2.3422440834639565e-06,
+      "loss": 0.27,
+      "step": 1525
+    },
+    {
+      "epoch": 1.7069351230425056,
+      "grad_norm": 0.46541234850883484,
+      "learning_rate": 2.3389958429838448e-06,
+      "loss": 0.2145,
+      "step": 1526
+    },
+    {
+      "epoch": 1.7080536912751678,
+      "grad_norm": 0.480028361082077,
+      "learning_rate": 2.335747875415417e-06,
+      "loss": 0.3166,
+      "step": 1527
+    },
+    {
+      "epoch": 1.70917225950783,
+      "grad_norm": 0.4876837730407715,
+      "learning_rate": 2.3325001862641732e-06,
+      "loss": 0.2157,
+      "step": 1528
+    },
+    {
+      "epoch": 1.7102908277404922,
+      "grad_norm": 0.4283449053764343,
+      "learning_rate": 2.3292527810351404e-06,
+      "loss": 0.2397,
+      "step": 1529
+    },
+    {
+      "epoch": 1.7114093959731544,
+      "grad_norm": 0.5044905543327332,
+      "learning_rate": 2.3260056652328654e-06,
+      "loss": 0.2726,
+      "step": 1530
+    },
+    {
+      "epoch": 1.7125279642058167,
+      "grad_norm": 0.5471357107162476,
+      "learning_rate": 2.3227588443614034e-06,
+      "loss": 0.278,
+      "step": 1531
+    },
+    {
+      "epoch": 1.7136465324384789,
+      "grad_norm": 0.43691202998161316,
+      "learning_rate": 2.31951232392431e-06,
+      "loss": 0.2479,
+      "step": 1532
+    },
+    {
+      "epoch": 1.714765100671141,
+      "grad_norm": 0.4888100028038025,
+      "learning_rate": 2.3162661094246324e-06,
+      "loss": 0.2334,
+      "step": 1533
+    },
+    {
+      "epoch": 1.7158836689038033,
+      "grad_norm": 0.4282808303833008,
+      "learning_rate": 2.3130202063648975e-06,
+      "loss": 0.2327,
+      "step": 1534
+    },
+    {
+      "epoch": 1.7170022371364653,
+      "grad_norm": 0.4824657142162323,
+      "learning_rate": 2.3097746202471063e-06,
+      "loss": 0.2534,
+      "step": 1535
+    },
+    {
+      "epoch": 1.7181208053691275,
+      "grad_norm": 0.4524305462837219,
+      "learning_rate": 2.3065293565727214e-06,
+      "loss": 0.2319,
+      "step": 1536
+    },
+    {
+      "epoch": 1.7192393736017897,
+      "grad_norm": 0.479215532541275,
+      "learning_rate": 2.303284420842658e-06,
+      "loss": 0.2724,
+      "step": 1537
+    },
+    {
+      "epoch": 1.720357941834452,
+      "grad_norm": 0.42658016085624695,
+      "learning_rate": 2.300039818557277e-06,
+      "loss": 0.3037,
+      "step": 1538
+    },
+    {
+      "epoch": 1.721476510067114,
+      "grad_norm": 0.4995690584182739,
+      "learning_rate": 2.2967955552163744e-06,
+      "loss": 0.2744,
+      "step": 1539
+    },
+    {
+      "epoch": 1.7225950782997763,
+      "grad_norm": 0.46471908688545227,
+      "learning_rate": 2.2935516363191695e-06,
+      "loss": 0.2996,
+      "step": 1540
+    },
+    {
+      "epoch": 1.7237136465324385,
+      "grad_norm": 0.4902131259441376,
+      "learning_rate": 2.2903080673643e-06,
+      "loss": 0.2777,
+      "step": 1541
+    },
+    {
+      "epoch": 1.7248322147651005,
+      "grad_norm": 0.5179248452186584,
+      "learning_rate": 2.2870648538498086e-06,
+      "loss": 0.288,
+      "step": 1542
+    },
+    {
+      "epoch": 1.7259507829977627,
+      "grad_norm": 0.49520888924598694,
+      "learning_rate": 2.2838220012731365e-06,
+      "loss": 0.2456,
+      "step": 1543
+    },
+    {
+      "epoch": 1.727069351230425,
+      "grad_norm": 0.5106678009033203,
+      "learning_rate": 2.2805795151311124e-06,
+      "loss": 0.2928,
+      "step": 1544
+    },
+    {
+      "epoch": 1.7281879194630871,
+      "grad_norm": 0.42379066348075867,
+      "learning_rate": 2.277337400919946e-06,
+      "loss": 0.2237,
+      "step": 1545
+    },
+    {
+      "epoch": 1.7293064876957494,
+      "grad_norm": 0.4539119005203247,
+      "learning_rate": 2.274095664135214e-06,
+      "loss": 0.2524,
+      "step": 1546
+    },
+    {
+      "epoch": 1.7304250559284116,
+      "grad_norm": 0.4954555928707123,
+      "learning_rate": 2.2708543102718546e-06,
+      "loss": 0.3193,
+      "step": 1547
+    },
+    {
+      "epoch": 1.7315436241610738,
+      "grad_norm": 0.4650874733924866,
+      "learning_rate": 2.2676133448241565e-06,
+      "loss": 0.2528,
+      "step": 1548
+    },
+    {
+      "epoch": 1.732662192393736,
+      "grad_norm": 0.4710632860660553,
+      "learning_rate": 2.2643727732857496e-06,
+      "loss": 0.2198,
+      "step": 1549
+    },
+    {
+      "epoch": 1.7337807606263982,
+      "grad_norm": 0.5317122340202332,
+      "learning_rate": 2.2611326011495977e-06,
+      "loss": 0.2641,
+      "step": 1550
+    },
+    {
+      "epoch": 1.7337807606263982,
+      "eval_loss": 0.2745862603187561,
+      "eval_runtime": 376.1405,
+      "eval_samples_per_second": 2.114,
+      "eval_steps_per_second": 2.114,
+      "step": 1550
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2682,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.8563938867981824e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}