{
  "best_metric": 0.8880267143249512,
  "best_model_checkpoint": "miner_id_24/checkpoint-400",
  "epoch": 1.3722126929674099,
  "eval_steps": 100,
  "global_step": 400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003430531732418525,
      "grad_norm": 0.27824944257736206,
      "learning_rate": 1.4e-06,
      "loss": 1.2807,
      "step": 1
    },
    {
      "epoch": 0.003430531732418525,
      "eval_loss": 1.3408838510513306,
      "eval_runtime": 36.4521,
      "eval_samples_per_second": 13.47,
      "eval_steps_per_second": 3.374,
      "step": 1
    },
    {
      "epoch": 0.00686106346483705,
      "grad_norm": 0.2794307768344879,
      "learning_rate": 2.8e-06,
      "loss": 1.3341,
      "step": 2
    },
    {
      "epoch": 0.010291595197255575,
      "grad_norm": 0.29869186878204346,
      "learning_rate": 4.2e-06,
      "loss": 1.2984,
      "step": 3
    },
    {
      "epoch": 0.0137221269296741,
      "grad_norm": 0.30163782835006714,
      "learning_rate": 5.6e-06,
      "loss": 1.2416,
      "step": 4
    },
    {
      "epoch": 0.017152658662092625,
      "grad_norm": 0.31913354992866516,
      "learning_rate": 7e-06,
      "loss": 1.3119,
      "step": 5
    },
    {
      "epoch": 0.02058319039451115,
      "grad_norm": 0.31287121772766113,
      "learning_rate": 8.4e-06,
      "loss": 1.3023,
      "step": 6
    },
    {
      "epoch": 0.024013722126929673,
      "grad_norm": 0.3440784513950348,
      "learning_rate": 9.8e-06,
      "loss": 1.3269,
      "step": 7
    },
    {
      "epoch": 0.0274442538593482,
      "grad_norm": 0.30914536118507385,
      "learning_rate": 1.12e-05,
      "loss": 1.2841,
      "step": 8
    },
    {
      "epoch": 0.030874785591766724,
      "grad_norm": 0.32246437668800354,
      "learning_rate": 1.2599999999999998e-05,
      "loss": 1.2826,
      "step": 9
    },
    {
      "epoch": 0.03430531732418525,
      "grad_norm": 0.35504838824272156,
      "learning_rate": 1.4e-05,
      "loss": 1.2992,
      "step": 10
    },
    {
      "epoch": 0.03773584905660377,
      "grad_norm": 0.3929505944252014,
      "learning_rate": 1.5399999999999998e-05,
      "loss": 1.3377,
      "step": 11
    },
    {
      "epoch": 0.0411663807890223,
      "grad_norm": 0.3200545907020569,
      "learning_rate": 1.68e-05,
      "loss": 1.2655,
      "step": 12
    },
    {
      "epoch": 0.044596912521440824,
      "grad_norm": 0.3272934854030609,
      "learning_rate": 1.82e-05,
      "loss": 1.2907,
      "step": 13
    },
    {
      "epoch": 0.048027444253859346,
      "grad_norm": 0.3309055268764496,
      "learning_rate": 1.96e-05,
      "loss": 1.2089,
      "step": 14
    },
    {
      "epoch": 0.051457975986277875,
      "grad_norm": 0.3068203628063202,
      "learning_rate": 2.1e-05,
      "loss": 1.2526,
      "step": 15
    },
    {
      "epoch": 0.0548885077186964,
      "grad_norm": 0.30473190546035767,
      "learning_rate": 2.24e-05,
      "loss": 1.2913,
      "step": 16
    },
    {
      "epoch": 0.058319039451114926,
      "grad_norm": 0.31642964482307434,
      "learning_rate": 2.38e-05,
      "loss": 1.2156,
      "step": 17
    },
    {
      "epoch": 0.06174957118353345,
      "grad_norm": 0.27499595284461975,
      "learning_rate": 2.5199999999999996e-05,
      "loss": 1.2661,
      "step": 18
    },
    {
      "epoch": 0.06518010291595197,
      "grad_norm": 0.26945629715919495,
      "learning_rate": 2.66e-05,
      "loss": 1.2501,
      "step": 19
    },
    {
      "epoch": 0.0686106346483705,
      "grad_norm": 0.2821623682975769,
      "learning_rate": 2.8e-05,
      "loss": 1.2411,
      "step": 20
    },
    {
      "epoch": 0.07204116638078903,
      "grad_norm": 0.3003956079483032,
      "learning_rate": 2.9399999999999996e-05,
      "loss": 1.268,
      "step": 21
    },
    {
      "epoch": 0.07547169811320754,
      "grad_norm": 0.35937100648880005,
      "learning_rate": 3.0799999999999996e-05,
      "loss": 1.2288,
      "step": 22
    },
    {
      "epoch": 0.07890222984562607,
      "grad_norm": 0.3568710386753082,
      "learning_rate": 3.22e-05,
      "loss": 1.2481,
      "step": 23
    },
    {
      "epoch": 0.0823327615780446,
      "grad_norm": 0.37753623723983765,
      "learning_rate": 3.36e-05,
      "loss": 1.3478,
      "step": 24
    },
    {
      "epoch": 0.08576329331046312,
      "grad_norm": 0.3411467671394348,
      "learning_rate": 3.5e-05,
      "loss": 1.2498,
      "step": 25
    },
    {
      "epoch": 0.08919382504288165,
      "grad_norm": 0.3256891667842865,
      "learning_rate": 3.64e-05,
      "loss": 1.2086,
      "step": 26
    },
    {
      "epoch": 0.09262435677530018,
      "grad_norm": 0.31805384159088135,
      "learning_rate": 3.78e-05,
      "loss": 1.2715,
      "step": 27
    },
    {
      "epoch": 0.09605488850771869,
      "grad_norm": 0.31195560097694397,
      "learning_rate": 3.92e-05,
      "loss": 1.1789,
      "step": 28
    },
    {
      "epoch": 0.09948542024013722,
      "grad_norm": 0.3224494159221649,
      "learning_rate": 4.059999999999999e-05,
      "loss": 1.176,
      "step": 29
    },
    {
      "epoch": 0.10291595197255575,
      "grad_norm": 0.32157495617866516,
      "learning_rate": 4.2e-05,
      "loss": 1.1217,
      "step": 30
    },
    {
      "epoch": 0.10634648370497427,
      "grad_norm": 0.3384961783885956,
      "learning_rate": 4.34e-05,
      "loss": 1.1945,
      "step": 31
    },
    {
      "epoch": 0.1097770154373928,
      "grad_norm": 0.3938997983932495,
      "learning_rate": 4.48e-05,
      "loss": 1.1137,
      "step": 32
    },
    {
      "epoch": 0.11320754716981132,
      "grad_norm": 0.37928083539009094,
      "learning_rate": 4.62e-05,
      "loss": 1.1206,
      "step": 33
    },
    {
      "epoch": 0.11663807890222985,
      "grad_norm": 0.4108302891254425,
      "learning_rate": 4.76e-05,
      "loss": 1.177,
      "step": 34
    },
    {
      "epoch": 0.12006861063464837,
      "grad_norm": 0.40350550413131714,
      "learning_rate": 4.899999999999999e-05,
      "loss": 1.1136,
      "step": 35
    },
    {
      "epoch": 0.1234991423670669,
      "grad_norm": 0.44681158661842346,
      "learning_rate": 5.039999999999999e-05,
      "loss": 1.1296,
      "step": 36
    },
    {
      "epoch": 0.1269296740994854,
      "grad_norm": 0.4281773269176483,
      "learning_rate": 5.179999999999999e-05,
      "loss": 1.1422,
      "step": 37
    },
    {
      "epoch": 0.13036020583190394,
      "grad_norm": 0.4204246997833252,
      "learning_rate": 5.32e-05,
      "loss": 1.1203,
      "step": 38
    },
    {
      "epoch": 0.13379073756432247,
      "grad_norm": 0.495144248008728,
      "learning_rate": 5.46e-05,
      "loss": 1.1492,
      "step": 39
    },
    {
      "epoch": 0.137221269296741,
      "grad_norm": 0.38059696555137634,
      "learning_rate": 5.6e-05,
      "loss": 1.088,
      "step": 40
    },
    {
      "epoch": 0.14065180102915953,
      "grad_norm": 0.44585326313972473,
      "learning_rate": 5.739999999999999e-05,
      "loss": 1.104,
      "step": 41
    },
    {
      "epoch": 0.14408233276157806,
      "grad_norm": 0.45643752813339233,
      "learning_rate": 5.879999999999999e-05,
      "loss": 1.13,
      "step": 42
    },
    {
      "epoch": 0.14751286449399656,
      "grad_norm": 0.4543881118297577,
      "learning_rate": 6.019999999999999e-05,
      "loss": 1.0996,
      "step": 43
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 0.4942781627178192,
      "learning_rate": 6.159999999999999e-05,
      "loss": 1.1354,
      "step": 44
    },
    {
      "epoch": 0.15437392795883362,
      "grad_norm": 0.5626402497291565,
      "learning_rate": 6.3e-05,
      "loss": 1.1204,
      "step": 45
    },
    {
      "epoch": 0.15780445969125215,
      "grad_norm": 0.5675943493843079,
      "learning_rate": 6.44e-05,
      "loss": 1.1008,
      "step": 46
    },
    {
      "epoch": 0.16123499142367068,
      "grad_norm": 0.5674481987953186,
      "learning_rate": 6.579999999999999e-05,
      "loss": 1.0523,
      "step": 47
    },
    {
      "epoch": 0.1646655231560892,
      "grad_norm": 0.6175776720046997,
      "learning_rate": 6.72e-05,
      "loss": 1.1834,
      "step": 48
    },
    {
      "epoch": 0.1680960548885077,
      "grad_norm": 0.648908257484436,
      "learning_rate": 6.859999999999999e-05,
      "loss": 1.0875,
      "step": 49
    },
    {
      "epoch": 0.17152658662092624,
      "grad_norm": 0.8205939531326294,
      "learning_rate": 7e-05,
      "loss": 1.2005,
      "step": 50
    },
    {
      "epoch": 0.17495711835334476,
      "grad_norm": 0.617884635925293,
      "learning_rate": 6.999859006598048e-05,
      "loss": 1.2037,
      "step": 51
    },
    {
      "epoch": 0.1783876500857633,
      "grad_norm": 0.5661861896514893,
      "learning_rate": 6.999436037751703e-05,
      "loss": 1.2372,
      "step": 52
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.42782628536224365,
      "learning_rate": 6.998731127538574e-05,
      "loss": 1.1914,
      "step": 53
    },
    {
      "epoch": 0.18524871355060035,
      "grad_norm": 0.3282926678657532,
      "learning_rate": 6.997744332751624e-05,
      "loss": 1.1677,
      "step": 54
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 0.2774999737739563,
      "learning_rate": 6.996475732894601e-05,
      "loss": 1.166,
      "step": 55
    },
    {
      "epoch": 0.19210977701543738,
      "grad_norm": 0.2545096278190613,
      "learning_rate": 6.994925430175622e-05,
      "loss": 1.1396,
      "step": 56
    },
    {
      "epoch": 0.1955403087478559,
      "grad_norm": 0.2631377577781677,
      "learning_rate": 6.99309354949895e-05,
      "loss": 1.0787,
      "step": 57
    },
    {
      "epoch": 0.19897084048027444,
      "grad_norm": 0.2663518488407135,
      "learning_rate": 6.99098023845492e-05,
      "loss": 1.1618,
      "step": 58
    },
    {
      "epoch": 0.20240137221269297,
      "grad_norm": 0.2729957699775696,
      "learning_rate": 6.988585667308051e-05,
      "loss": 1.141,
      "step": 59
    },
    {
      "epoch": 0.2058319039451115,
      "grad_norm": 0.2546933889389038,
      "learning_rate": 6.985910028983336e-05,
      "loss": 1.1057,
      "step": 60
    },
    {
      "epoch": 0.20926243567753003,
      "grad_norm": 0.283761203289032,
      "learning_rate": 6.982953539050688e-05,
      "loss": 1.0742,
      "step": 61
    },
    {
      "epoch": 0.21269296740994853,
      "grad_norm": 0.27833765745162964,
      "learning_rate": 6.97971643570758e-05,
      "loss": 1.18,
      "step": 62
    },
    {
      "epoch": 0.21612349914236706,
      "grad_norm": 0.2722010314464569,
      "learning_rate": 6.976198979759845e-05,
      "loss": 1.115,
      "step": 63
    },
    {
      "epoch": 0.2195540308747856,
      "grad_norm": 0.279540091753006,
      "learning_rate": 6.972401454600672e-05,
      "loss": 1.0544,
      "step": 64
    },
    {
      "epoch": 0.22298456260720412,
      "grad_norm": 0.2840425372123718,
      "learning_rate": 6.968324166187771e-05,
      "loss": 1.0367,
      "step": 65
    },
    {
      "epoch": 0.22641509433962265,
      "grad_norm": 0.3138042390346527,
      "learning_rate": 6.963967443018721e-05,
      "loss": 0.973,
      "step": 66
    },
    {
      "epoch": 0.22984562607204118,
      "grad_norm": 0.29737910628318787,
      "learning_rate": 6.959331636104508e-05,
      "loss": 1.0418,
      "step": 67
    },
    {
      "epoch": 0.2332761578044597,
      "grad_norm": 0.3060307502746582,
      "learning_rate": 6.954417118941235e-05,
      "loss": 1.0985,
      "step": 68
    },
    {
      "epoch": 0.2367066895368782,
      "grad_norm": 0.3002816438674927,
      "learning_rate": 6.949224287480042e-05,
      "loss": 1.0626,
      "step": 69
    },
    {
      "epoch": 0.24013722126929674,
      "grad_norm": 0.32159680128097534,
      "learning_rate": 6.943753560095204e-05,
      "loss": 1.0734,
      "step": 70
    },
    {
      "epoch": 0.24356775300171526,
      "grad_norm": 0.3148500323295593,
      "learning_rate": 6.93800537755041e-05,
      "loss": 1.1352,
      "step": 71
    },
    {
      "epoch": 0.2469982847341338,
      "grad_norm": 0.3088507354259491,
      "learning_rate": 6.93198020296327e-05,
      "loss": 0.9925,
      "step": 72
    },
    {
      "epoch": 0.2504288164665523,
      "grad_norm": 0.3257371187210083,
      "learning_rate": 6.92567852176799e-05,
      "loss": 0.9782,
      "step": 73
    },
    {
      "epoch": 0.2538593481989708,
      "grad_norm": 0.31071141362190247,
      "learning_rate": 6.919100841676266e-05,
      "loss": 1.0096,
      "step": 74
    },
    {
      "epoch": 0.25728987993138935,
      "grad_norm": 0.34918615221977234,
      "learning_rate": 6.912247692636382e-05,
      "loss": 1.0163,
      "step": 75
    },
    {
      "epoch": 0.2607204116638079,
      "grad_norm": 0.3556749224662781,
      "learning_rate": 6.905119626790507e-05,
      "loss": 1.0961,
      "step": 76
    },
    {
      "epoch": 0.2641509433962264,
      "grad_norm": 0.3610279858112335,
      "learning_rate": 6.897717218430212e-05,
      "loss": 1.0246,
      "step": 77
    },
    {
      "epoch": 0.26758147512864494,
      "grad_norm": 0.3471314609050751,
      "learning_rate": 6.890041063950208e-05,
      "loss": 1.0174,
      "step": 78
    },
    {
      "epoch": 0.27101200686106347,
      "grad_norm": 0.3771928548812866,
      "learning_rate": 6.882091781800283e-05,
      "loss": 1.0467,
      "step": 79
    },
    {
      "epoch": 0.274442538593482,
      "grad_norm": 0.3704926669597626,
      "learning_rate": 6.873870012435486e-05,
      "loss": 1.0205,
      "step": 80
    },
    {
      "epoch": 0.27787307032590053,
      "grad_norm": 0.40470531582832336,
      "learning_rate": 6.86537641826452e-05,
      "loss": 1.0653,
      "step": 81
    },
    {
      "epoch": 0.28130360205831906,
      "grad_norm": 0.39418941736221313,
      "learning_rate": 6.856611683596379e-05,
      "loss": 0.994,
      "step": 82
    },
    {
      "epoch": 0.2847341337907376,
      "grad_norm": 0.4157142639160156,
      "learning_rate": 6.84757651458521e-05,
      "loss": 0.9998,
      "step": 83
    },
    {
      "epoch": 0.2881646655231561,
      "grad_norm": 0.42235854268074036,
      "learning_rate": 6.838271639173422e-05,
      "loss": 1.0191,
      "step": 84
    },
    {
      "epoch": 0.2915951972555746,
      "grad_norm": 0.40815040469169617,
      "learning_rate": 6.828697807033038e-05,
      "loss": 0.9527,
      "step": 85
    },
    {
      "epoch": 0.2950257289879931,
      "grad_norm": 0.440251886844635,
      "learning_rate": 6.818855789505291e-05,
      "loss": 1.0031,
      "step": 86
    },
    {
      "epoch": 0.29845626072041165,
      "grad_norm": 0.43153345584869385,
      "learning_rate": 6.80874637953849e-05,
      "loss": 0.9926,
      "step": 87
    },
    {
      "epoch": 0.3018867924528302,
      "grad_norm": 0.4722265899181366,
      "learning_rate": 6.798370391624122e-05,
      "loss": 1.0324,
      "step": 88
    },
    {
      "epoch": 0.3053173241852487,
      "grad_norm": 0.46640855073928833,
      "learning_rate": 6.787728661731232e-05,
      "loss": 1.052,
      "step": 89
    },
    {
      "epoch": 0.30874785591766724,
      "grad_norm": 0.4882723093032837,
      "learning_rate": 6.77682204723908e-05,
      "loss": 1.1014,
      "step": 90
    },
    {
      "epoch": 0.31217838765008576,
      "grad_norm": 0.46555763483047485,
      "learning_rate": 6.765651426868054e-05,
      "loss": 0.9989,
      "step": 91
    },
    {
      "epoch": 0.3156089193825043,
      "grad_norm": 0.5026487112045288,
      "learning_rate": 6.75421770060888e-05,
      "loss": 1.0579,
      "step": 92
    },
    {
      "epoch": 0.3190394511149228,
      "grad_norm": 0.47720423340797424,
      "learning_rate": 6.742521789650105e-05,
      "loss": 1.0396,
      "step": 93
    },
    {
      "epoch": 0.32246998284734135,
      "grad_norm": 0.53495192527771,
      "learning_rate": 6.73056463630389e-05,
      "loss": 0.9513,
      "step": 94
    },
    {
      "epoch": 0.3259005145797599,
      "grad_norm": 0.5590450167655945,
      "learning_rate": 6.718347203930078e-05,
      "loss": 1.051,
      "step": 95
    },
    {
      "epoch": 0.3293310463121784,
      "grad_norm": 0.5361358523368835,
      "learning_rate": 6.705870476858582e-05,
      "loss": 1.01,
      "step": 96
    },
    {
      "epoch": 0.33276157804459694,
      "grad_norm": 0.5490963459014893,
      "learning_rate": 6.69313546031009e-05,
      "loss": 0.9192,
      "step": 97
    },
    {
      "epoch": 0.3361921097770154,
      "grad_norm": 0.5866110920906067,
      "learning_rate": 6.68014318031506e-05,
      "loss": 0.982,
      "step": 98
    },
    {
      "epoch": 0.33962264150943394,
      "grad_norm": 0.7043346762657166,
      "learning_rate": 6.666894683631068e-05,
      "loss": 1.0057,
      "step": 99
    },
    {
      "epoch": 0.34305317324185247,
      "grad_norm": 0.8684700727462769,
      "learning_rate": 6.653391037658466e-05,
      "loss": 1.0552,
      "step": 100
    },
    {
      "epoch": 0.34305317324185247,
      "eval_loss": 1.0371321439743042,
      "eval_runtime": 36.9532,
      "eval_samples_per_second": 13.287,
      "eval_steps_per_second": 3.329,
      "step": 100
    },
    {
      "epoch": 0.346483704974271,
      "grad_norm": 0.4672603905200958,
      "learning_rate": 6.63963333035439e-05,
      "loss": 1.1285,
      "step": 101
    },
    {
      "epoch": 0.34991423670668953,
      "grad_norm": 0.4868123531341553,
      "learning_rate": 6.625622670145098e-05,
      "loss": 1.1721,
      "step": 102
    },
    {
      "epoch": 0.35334476843910806,
      "grad_norm": 0.4107426702976227,
      "learning_rate": 6.611360185836676e-05,
      "loss": 1.1251,
      "step": 103
    },
    {
      "epoch": 0.3567753001715266,
      "grad_norm": 0.3443511426448822,
      "learning_rate": 6.596847026524084e-05,
      "loss": 1.1007,
      "step": 104
    },
    {
      "epoch": 0.3602058319039451,
      "grad_norm": 0.3096946179866791,
      "learning_rate": 6.582084361498583e-05,
      "loss": 1.143,
      "step": 105
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.2761515974998474,
      "learning_rate": 6.567073380153521e-05,
      "loss": 1.0956,
      "step": 106
    },
    {
      "epoch": 0.3670668953687822,
      "grad_norm": 0.2948932647705078,
      "learning_rate": 6.551815291888517e-05,
      "loss": 1.0838,
      "step": 107
    },
    {
      "epoch": 0.3704974271012007,
      "grad_norm": 0.2960832715034485,
      "learning_rate": 6.53631132601201e-05,
      "loss": 1.0319,
      "step": 108
    },
    {
      "epoch": 0.37392795883361923,
      "grad_norm": 0.3060498833656311,
      "learning_rate": 6.520562731642225e-05,
      "loss": 1.0636,
      "step": 109
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 0.2978265881538391,
      "learning_rate": 6.504570777606531e-05,
      "loss": 1.068,
      "step": 110
    },
    {
      "epoch": 0.38078902229845624,
      "grad_norm": 0.29292210936546326,
      "learning_rate": 6.488336752339214e-05,
      "loss": 1.0286,
      "step": 111
    },
    {
      "epoch": 0.38421955403087477,
      "grad_norm": 0.30305469036102295,
      "learning_rate": 6.471861963777677e-05,
      "loss": 0.9569,
      "step": 112
    },
    {
      "epoch": 0.3876500857632933,
      "grad_norm": 0.30470404028892517,
      "learning_rate": 6.455147739257053e-05,
      "loss": 0.9753,
      "step": 113
    },
    {
      "epoch": 0.3910806174957118,
      "grad_norm": 0.3086484968662262,
      "learning_rate": 6.438195425403269e-05,
      "loss": 0.9775,
      "step": 114
    },
    {
      "epoch": 0.39451114922813035,
      "grad_norm": 0.32541391253471375,
      "learning_rate": 6.421006388024559e-05,
      "loss": 1.0368,
      "step": 115
    },
    {
      "epoch": 0.3979416809605489,
      "grad_norm": 0.3314479887485504,
      "learning_rate": 6.403582012001409e-05,
      "loss": 1.0503,
      "step": 116
    },
    {
      "epoch": 0.4013722126929674,
      "grad_norm": 0.34121036529541016,
      "learning_rate": 6.385923701174992e-05,
      "loss": 0.9715,
      "step": 117
    },
    {
      "epoch": 0.40480274442538594,
      "grad_norm": 0.31866931915283203,
      "learning_rate": 6.368032878234061e-05,
      "loss": 1.0487,
      "step": 118
    },
    {
      "epoch": 0.40823327615780447,
      "grad_norm": 0.3176023066043854,
      "learning_rate": 6.349910984600323e-05,
      "loss": 0.9822,
      "step": 119
    },
    {
      "epoch": 0.411663807890223,
      "grad_norm": 0.33536022901535034,
      "learning_rate": 6.331559480312315e-05,
      "loss": 1.0269,
      "step": 120
    },
    {
      "epoch": 0.41509433962264153,
      "grad_norm": 0.3236567974090576,
      "learning_rate": 6.31297984390776e-05,
      "loss": 0.9503,
      "step": 121
    },
    {
      "epoch": 0.41852487135506006,
      "grad_norm": 0.34983325004577637,
      "learning_rate": 6.294173572304455e-05,
      "loss": 0.9403,
      "step": 122
    },
    {
      "epoch": 0.4219554030874786,
      "grad_norm": 0.363774836063385,
      "learning_rate": 6.275142180679663e-05,
      "loss": 1.0013,
      "step": 123
    },
    {
      "epoch": 0.42538593481989706,
      "grad_norm": 0.3461463451385498,
      "learning_rate": 6.255887202348041e-05,
      "loss": 0.999,
      "step": 124
    },
    {
      "epoch": 0.4288164665523156,
      "grad_norm": 0.36093324422836304,
      "learning_rate": 6.236410188638105e-05,
      "loss": 1.0242,
      "step": 125
    },
    {
      "epoch": 0.4322469982847341,
      "grad_norm": 0.3705779016017914,
      "learning_rate": 6.216712708767235e-05,
      "loss": 1.0526,
      "step": 126
    },
    {
      "epoch": 0.43567753001715265,
      "grad_norm": 0.3667316734790802,
      "learning_rate": 6.196796349715262e-05,
      "loss": 0.9446,
      "step": 127
    },
    {
      "epoch": 0.4391080617495712,
      "grad_norm": 0.3543902039527893,
      "learning_rate": 6.176662716096595e-05,
      "loss": 0.9858,
      "step": 128
    },
    {
      "epoch": 0.4425385934819897,
      "grad_norm": 0.39349791407585144,
      "learning_rate": 6.156313430030943e-05,
      "loss": 0.9398,
      "step": 129
    },
    {
      "epoch": 0.44596912521440824,
      "grad_norm": 0.36853110790252686,
      "learning_rate": 6.135750131012639e-05,
      "loss": 0.9709,
      "step": 130
    },
    {
      "epoch": 0.44939965694682676,
      "grad_norm": 0.36536240577697754,
      "learning_rate": 6.114974475778523e-05,
      "loss": 0.8983,
      "step": 131
    },
    {
      "epoch": 0.4528301886792453,
      "grad_norm": 0.37377798557281494,
      "learning_rate": 6.093988138174491e-05,
      "loss": 0.9184,
      "step": 132
    },
    {
      "epoch": 0.4562607204116638,
      "grad_norm": 0.39587923884391785,
      "learning_rate": 6.072792809020618e-05,
      "loss": 0.9228,
      "step": 133
    },
    {
      "epoch": 0.45969125214408235,
      "grad_norm": 0.40825745463371277,
      "learning_rate": 6.0513901959749396e-05,
      "loss": 0.95,
      "step": 134
    },
    {
      "epoch": 0.4631217838765009,
      "grad_norm": 0.41529640555381775,
      "learning_rate": 6.02978202339587e-05,
      "loss": 0.9542,
      "step": 135
    },
    {
      "epoch": 0.4665523156089194,
      "grad_norm": 0.4271698594093323,
      "learning_rate": 6.0079700322032724e-05,
      "loss": 0.9335,
      "step": 136
    },
    {
      "epoch": 0.4699828473413379,
      "grad_norm": 0.4341247081756592,
      "learning_rate": 5.985955979738199e-05,
      "loss": 1.0248,
      "step": 137
    },
    {
      "epoch": 0.4734133790737564,
      "grad_norm": 0.4385394752025604,
      "learning_rate": 5.963741639621307e-05,
      "loss": 1.0233,
      "step": 138
    },
    {
      "epoch": 0.47684391080617494,
      "grad_norm": 0.4260070323944092,
      "learning_rate": 5.941328801609958e-05,
      "loss": 0.9561,
      "step": 139
    },
    {
      "epoch": 0.48027444253859347,
      "grad_norm": 0.46535003185272217,
      "learning_rate": 5.918719271454026e-05,
      "loss": 0.9433,
      "step": 140
    },
    {
      "epoch": 0.483704974271012,
      "grad_norm": 0.45692041516304016,
      "learning_rate": 5.89591487075041e-05,
      "loss": 0.9409,
      "step": 141
    },
    {
      "epoch": 0.48713550600343053,
      "grad_norm": 0.46150150895118713,
      "learning_rate": 5.872917436796273e-05,
      "loss": 0.9387,
      "step": 142
    },
    {
      "epoch": 0.49056603773584906,
      "grad_norm": 0.4859018623828888,
      "learning_rate": 5.849728822441013e-05,
      "loss": 0.943,
      "step": 143
    },
    {
      "epoch": 0.4939965694682676,
      "grad_norm": 0.5400093197822571,
      "learning_rate": 5.8263508959369896e-05,
      "loss": 0.9913,
      "step": 144
    },
    {
      "epoch": 0.4974271012006861,
      "grad_norm": 0.5390109419822693,
      "learning_rate": 5.802785540788994e-05,
      "loss": 1.0125,
      "step": 145
    },
    {
      "epoch": 0.5008576329331046,
      "grad_norm": 0.5523272156715393,
      "learning_rate": 5.779034655602507e-05,
      "loss": 1.0892,
      "step": 146
    },
    {
      "epoch": 0.5042881646655232,
      "grad_norm": 0.578334391117096,
      "learning_rate": 5.75510015393073e-05,
      "loss": 0.9769,
      "step": 147
    },
    {
      "epoch": 0.5077186963979416,
      "grad_norm": 0.5656843185424805,
      "learning_rate": 5.7309839641204136e-05,
      "loss": 0.9271,
      "step": 148
    },
    {
      "epoch": 0.5111492281303602,
      "grad_norm": 0.6241077780723572,
      "learning_rate": 5.7066880291564976e-05,
      "loss": 0.9921,
      "step": 149
    },
    {
      "epoch": 0.5145797598627787,
      "grad_norm": 0.793802797794342,
      "learning_rate": 5.682214306505567e-05,
      "loss": 1.0064,
      "step": 150
    },
    {
      "epoch": 0.5180102915951973,
      "grad_norm": 0.3092074692249298,
      "learning_rate": 5.657564767958144e-05,
      "loss": 1.0916,
      "step": 151
    },
    {
      "epoch": 0.5214408233276158,
      "grad_norm": 0.3236907720565796,
      "learning_rate": 5.632741399469828e-05,
      "loss": 1.1536,
      "step": 152
    },
    {
      "epoch": 0.5248713550600344,
      "grad_norm": 0.3189297020435333,
      "learning_rate": 5.6077462010012874e-05,
      "loss": 1.0668,
      "step": 153
    },
    {
      "epoch": 0.5283018867924528,
      "grad_norm": 0.314827561378479,
      "learning_rate": 5.582581186357129e-05,
      "loss": 1.0863,
      "step": 154
    },
    {
      "epoch": 0.5317324185248714,
      "grad_norm": 0.3061763644218445,
      "learning_rate": 5.557248383023655e-05,
      "loss": 1.0706,
      "step": 155
    },
    {
      "epoch": 0.5351629502572899,
      "grad_norm": 0.2918294370174408,
      "learning_rate": 5.5317498320055066e-05,
      "loss": 1.0265,
      "step": 156
    },
    {
      "epoch": 0.5385934819897084,
      "grad_norm": 0.30654773116111755,
      "learning_rate": 5.506087587661228e-05,
      "loss": 1.0243,
      "step": 157
    },
    {
      "epoch": 0.5420240137221269,
      "grad_norm": 0.2902136743068695,
      "learning_rate": 5.4802637175377505e-05,
      "loss": 1.014,
      "step": 158
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.2913793921470642,
      "learning_rate": 5.4542803022038175e-05,
      "loss": 0.9565,
      "step": 159
    },
    {
      "epoch": 0.548885077186964,
      "grad_norm": 0.29561156034469604,
      "learning_rate": 5.4281394350823586e-05,
      "loss": 0.9876,
      "step": 160
    },
    {
      "epoch": 0.5523156089193825,
      "grad_norm": 0.30898797512054443,
      "learning_rate": 5.40184322228182e-05,
      "loss": 0.9693,
      "step": 161
    },
    {
      "epoch": 0.5557461406518011,
      "grad_norm": 0.3182258903980255,
      "learning_rate": 5.375393782426487e-05,
      "loss": 1.0828,
      "step": 162
    },
    {
      "epoch": 0.5591766723842195,
      "grad_norm": 0.2985084056854248,
      "learning_rate": 5.348793246485796e-05,
      "loss": 1.0298,
      "step": 163
    },
    {
      "epoch": 0.5626072041166381,
      "grad_norm": 0.3110847473144531,
      "learning_rate": 5.32204375760263e-05,
      "loss": 0.9457,
      "step": 164
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 0.30118316411972046,
      "learning_rate": 5.295147470920671e-05,
      "loss": 0.9951,
      "step": 165
    },
    {
      "epoch": 0.5694682675814752,
      "grad_norm": 0.3020259737968445,
      "learning_rate": 5.268106553410754e-05,
      "loss": 0.9737,
      "step": 166
    },
    {
      "epoch": 0.5728987993138936,
      "grad_norm": 0.30683499574661255,
      "learning_rate": 5.240923183696277e-05,
      "loss": 1.0024,
      "step": 167
    },
    {
      "epoch": 0.5763293310463122,
      "grad_norm": 0.32152485847473145,
      "learning_rate": 5.213599551877683e-05,
      "loss": 0.9837,
      "step": 168
    },
    {
      "epoch": 0.5797598627787307,
      "grad_norm": 0.31641682982444763,
      "learning_rate": 5.186137859356003e-05,
      "loss": 0.9473,
      "step": 169
    },
    {
      "epoch": 0.5831903945111492,
      "grad_norm": 0.326933890581131,
      "learning_rate": 5.1585403186554945e-05,
      "loss": 0.9939,
      "step": 170
    },
    {
      "epoch": 0.5866209262435678,
      "grad_norm": 0.3273809254169464,
      "learning_rate": 5.130809153245386e-05,
      "loss": 0.82,
      "step": 171
    },
    {
      "epoch": 0.5900514579759862,
      "grad_norm": 0.3486361801624298,
      "learning_rate": 5.1029465973607365e-05,
      "loss": 1.0137,
      "step": 172
    },
    {
      "epoch": 0.5934819897084048,
      "grad_norm": 0.35306909680366516,
      "learning_rate": 5.074954895822423e-05,
      "loss": 0.9206,
      "step": 173
    },
    {
      "epoch": 0.5969125214408233,
      "grad_norm": 0.3562031686306,
      "learning_rate": 5.0468363038562935e-05,
      "loss": 0.9296,
      "step": 174
    },
    {
      "epoch": 0.6003430531732419,
      "grad_norm": 0.363288551568985,
      "learning_rate": 5.0185930869114526e-05,
      "loss": 0.9342,
      "step": 175
    },
    {
      "epoch": 0.6037735849056604,
      "grad_norm": 0.3522874414920807,
      "learning_rate": 4.990227520477754e-05,
      "loss": 0.9254,
      "step": 176
    },
    {
      "epoch": 0.6072041166380789,
      "grad_norm": 0.3597092032432556,
      "learning_rate": 4.961741889902458e-05,
      "loss": 0.9445,
      "step": 177
    },
    {
      "epoch": 0.6106346483704974,
      "grad_norm": 0.37211406230926514,
      "learning_rate": 4.933138490206117e-05,
      "loss": 0.8982,
      "step": 178
    },
    {
      "epoch": 0.614065180102916,
      "grad_norm": 0.37037959694862366,
      "learning_rate": 4.904419625897659e-05,
      "loss": 0.9968,
      "step": 179
    },
    {
      "epoch": 0.6174957118353345,
      "grad_norm": 0.37710821628570557,
      "learning_rate": 4.8755876107887325e-05,
      "loss": 0.9233,
      "step": 180
    },
    {
      "epoch": 0.6209262435677531,
      "grad_norm": 0.379173219203949,
      "learning_rate": 4.846644767807276e-05,
      "loss": 0.9933,
      "step": 181
    },
    {
      "epoch": 0.6243567753001715,
      "grad_norm": 0.36022669076919556,
      "learning_rate": 4.8175934288103715e-05,
      "loss": 0.9296,
      "step": 182
    },
    {
      "epoch": 0.62778730703259,
      "grad_norm": 0.37822067737579346,
      "learning_rate": 4.788435934396373e-05,
      "loss": 0.8755,
      "step": 183
    },
    {
      "epoch": 0.6312178387650086,
      "grad_norm": 0.43253007531166077,
      "learning_rate": 4.7591746337163246e-05,
      "loss": 0.9862,
      "step": 184
    },
    {
      "epoch": 0.6346483704974271,
      "grad_norm": 0.435069739818573,
      "learning_rate": 4.7298118842846995e-05,
      "loss": 0.9024,
      "step": 185
    },
    {
      "epoch": 0.6380789022298456,
      "grad_norm": 0.3982680141925812,
      "learning_rate": 4.70035005178946e-05,
      "loss": 0.9238,
      "step": 186
    },
    {
      "epoch": 0.6415094339622641,
      "grad_norm": 0.4719775319099426,
      "learning_rate": 4.670791509901459e-05,
      "loss": 0.9862,
      "step": 187
    },
    {
      "epoch": 0.6449399656946827,
      "grad_norm": 0.456085205078125,
      "learning_rate": 4.641138640083196e-05,
      "loss": 0.9441,
      "step": 188
    },
    {
      "epoch": 0.6483704974271012,
      "grad_norm": 0.4457819163799286,
      "learning_rate": 4.611393831396955e-05,
      "loss": 0.9605,
      "step": 189
    },
    {
      "epoch": 0.6518010291595198,
      "grad_norm": 0.4855576753616333,
      "learning_rate": 4.581559480312316e-05,
      "loss": 0.9714,
      "step": 190
    },
    {
      "epoch": 0.6552315608919382,
      "grad_norm": 0.4760058522224426,
      "learning_rate": 4.5516379905130814e-05,
      "loss": 0.9397,
      "step": 191
    },
    {
      "epoch": 0.6586620926243568,
      "grad_norm": 0.4612843096256256,
      "learning_rate": 4.521631772703617e-05,
      "loss": 0.9389,
      "step": 192
    },
    {
      "epoch": 0.6620926243567753,
      "grad_norm": 0.48471036553382874,
      "learning_rate": 4.4915432444146227e-05,
      "loss": 0.9267,
      "step": 193
    },
    {
      "epoch": 0.6655231560891939,
      "grad_norm": 0.510553240776062,
      "learning_rate": 4.4613748298083655e-05,
      "loss": 0.9402,
      "step": 194
    },
    {
      "epoch": 0.6689536878216124,
      "grad_norm": 0.5086294412612915,
      "learning_rate": 4.4311289594833624e-05,
      "loss": 0.9198,
      "step": 195
    },
    {
      "epoch": 0.6723842195540308,
      "grad_norm": 0.5275819301605225,
      "learning_rate": 4.400808070278558e-05,
      "loss": 0.8963,
      "step": 196
    },
    {
      "epoch": 0.6758147512864494,
      "grad_norm": 0.5834717750549316,
      "learning_rate": 4.370414605076992e-05,
      "loss": 0.982,
      "step": 197
    },
    {
      "epoch": 0.6792452830188679,
      "grad_norm": 0.5746213793754578,
      "learning_rate": 4.33995101260898e-05,
      "loss": 0.9498,
      "step": 198
    },
    {
      "epoch": 0.6826758147512865,
      "grad_norm": 0.6132733821868896,
      "learning_rate": 4.309419747254832e-05,
      "loss": 0.9637,
      "step": 199
    },
    {
      "epoch": 0.6861063464837049,
      "grad_norm": 0.865511417388916,
      "learning_rate": 4.2788232688471e-05,
      "loss": 0.9599,
      "step": 200
    },
    {
      "epoch": 0.6861063464837049,
      "eval_loss": 0.9412220120429993,
      "eval_runtime": 36.697,
      "eval_samples_per_second": 13.38,
      "eval_steps_per_second": 3.352,
      "step": 200
    },
    {
      "epoch": 0.6895368782161235,
      "grad_norm": 0.28364285826683044,
      "learning_rate": 4.2481640424724e-05,
      "loss": 1.0507,
      "step": 201
    },
    {
      "epoch": 0.692967409948542,
      "grad_norm": 0.30096933245658875,
      "learning_rate": 4.2174445382728067e-05,
      "loss": 1.0802,
      "step": 202
    },
    {
      "epoch": 0.6963979416809606,
      "grad_norm": 0.3202117681503296,
      "learning_rate": 4.186667231246833e-05,
      "loss": 1.0124,
      "step": 203
    },
    {
      "epoch": 0.6998284734133791,
      "grad_norm": 0.3037415146827698,
      "learning_rate": 4.1558346010500367e-05,
      "loss": 1.0186,
      "step": 204
    },
    {
      "epoch": 0.7032590051457976,
      "grad_norm": 0.3249450623989105,
      "learning_rate": 4.124949131795228e-05,
      "loss": 1.0409,
      "step": 205
    },
    {
      "epoch": 0.7066895368782161,
      "grad_norm": 0.3072739541530609,
      "learning_rate": 4.0940133118523404e-05,
      "loss": 0.9654,
      "step": 206
    },
    {
      "epoch": 0.7101200686106347,
      "grad_norm": 0.30447569489479065,
      "learning_rate": 4.063029633647944e-05,
      "loss": 1.0037,
      "step": 207
    },
    {
      "epoch": 0.7135506003430532,
      "grad_norm": 0.3089168965816498,
      "learning_rate": 4.032000593464436e-05,
      "loss": 1.0158,
      "step": 208
    },
    {
      "epoch": 0.7169811320754716,
      "grad_norm": 0.30727794766426086,
      "learning_rate": 4.000928691238918e-05,
      "loss": 0.8956,
      "step": 209
    },
    {
      "epoch": 0.7204116638078902,
      "grad_norm": 0.3099682927131653,
      "learning_rate": 3.969816430361794e-05,
      "loss": 0.9247,
      "step": 210
    },
    {
      "epoch": 0.7238421955403087,
      "grad_norm": 0.3134283125400543,
      "learning_rate": 3.938666317475065e-05,
      "loss": 0.953,
      "step": 211
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.29905807971954346,
      "learning_rate": 3.9074808622703795e-05,
      "loss": 0.8669,
      "step": 212
    },
    {
      "epoch": 0.7307032590051458,
      "grad_norm": 0.3141246438026428,
      "learning_rate": 3.876262577286837e-05,
      "loss": 1.0179,
      "step": 213
    },
    {
      "epoch": 0.7341337907375644,
      "grad_norm": 0.3220095932483673,
      "learning_rate": 3.845013977708552e-05,
      "loss": 0.9369,
      "step": 214
    },
    {
      "epoch": 0.7375643224699828,
      "grad_norm": 0.33446162939071655,
      "learning_rate": 3.813737581162017e-05,
      "loss": 0.8845,
      "step": 215
    },
    {
      "epoch": 0.7409948542024014,
      "grad_norm": 0.31734156608581543,
      "learning_rate": 3.78243590751326e-05,
      "loss": 0.896,
      "step": 216
    },
    {
      "epoch": 0.7444253859348199,
      "grad_norm": 0.34970560669898987,
      "learning_rate": 3.7511114786648266e-05,
      "loss": 0.9555,
      "step": 217
    },
    {
      "epoch": 0.7478559176672385,
      "grad_norm": 0.36380812525749207,
      "learning_rate": 3.719766818352597e-05,
      "loss": 0.9986,
      "step": 218
    },
    {
      "epoch": 0.7512864493996569,
      "grad_norm": 0.3587551712989807,
      "learning_rate": 3.6884044519424484e-05,
      "loss": 0.9198,
      "step": 219
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 0.36512261629104614,
      "learning_rate": 3.657026906226802e-05,
      "loss": 0.7804,
      "step": 220
    },
    {
      "epoch": 0.758147512864494,
      "grad_norm": 0.3511175811290741,
      "learning_rate": 3.625636709221038e-05,
      "loss": 0.9656,
      "step": 221
    },
    {
      "epoch": 0.7615780445969125,
      "grad_norm": 0.343990296125412,
      "learning_rate": 3.5942363899598195e-05,
      "loss": 0.8426,
      "step": 222
    },
    {
      "epoch": 0.7650085763293311,
      "grad_norm": 0.35831785202026367,
      "learning_rate": 3.56282847829334e-05,
      "loss": 0.8529,
      "step": 223
    },
    {
      "epoch": 0.7684391080617495,
      "grad_norm": 0.3546663522720337,
      "learning_rate": 3.5314155046834917e-05,
      "loss": 0.9023,
      "step": 224
    },
    {
      "epoch": 0.7718696397941681,
      "grad_norm": 0.3780810534954071,
      "learning_rate": 3.5e-05,
      "loss": 0.8917,
      "step": 225
    },
    {
      "epoch": 0.7753001715265866,
      "grad_norm": 0.3632754981517792,
      "learning_rate": 3.468584495316507e-05,
      "loss": 0.9039,
      "step": 226
    },
    {
      "epoch": 0.7787307032590052,
      "grad_norm": 0.3891032934188843,
      "learning_rate": 3.43717152170666e-05,
      "loss": 0.9462,
      "step": 227
    },
    {
      "epoch": 0.7821612349914236,
      "grad_norm": 0.4001498520374298,
      "learning_rate": 3.4057636100401806e-05,
      "loss": 0.9388,
      "step": 228
    },
    {
      "epoch": 0.7855917667238422,
      "grad_norm": 0.395388126373291,
      "learning_rate": 3.374363290778962e-05,
      "loss": 0.9392,
      "step": 229
    },
    {
      "epoch": 0.7890222984562607,
      "grad_norm": 0.40292492508888245,
      "learning_rate": 3.3429730937731987e-05,
      "loss": 0.9612,
      "step": 230
    },
    {
      "epoch": 0.7924528301886793,
      "grad_norm": 0.40302538871765137,
      "learning_rate": 3.311595548057551e-05,
      "loss": 0.8791,
      "step": 231
    },
    {
      "epoch": 0.7958833619210978,
      "grad_norm": 0.4243438243865967,
      "learning_rate": 3.280233181647403e-05,
      "loss": 0.9295,
      "step": 232
    },
    {
      "epoch": 0.7993138936535163,
      "grad_norm": 0.42024651169776917,
      "learning_rate": 3.248888521335172e-05,
      "loss": 0.9025,
      "step": 233
    },
    {
      "epoch": 0.8027444253859348,
      "grad_norm": 0.4553435742855072,
      "learning_rate": 3.2175640924867394e-05,
      "loss": 0.9679,
      "step": 234
    },
    {
      "epoch": 0.8061749571183533,
      "grad_norm": 0.44603100419044495,
      "learning_rate": 3.186262418837983e-05,
      "loss": 0.9312,
      "step": 235
    },
    {
      "epoch": 0.8096054888507719,
      "grad_norm": 0.46468275785446167,
      "learning_rate": 3.154986022291447e-05,
      "loss": 0.9202,
      "step": 236
    },
    {
      "epoch": 0.8130360205831904,
      "grad_norm": 0.43331602215766907,
      "learning_rate": 3.123737422713163e-05,
      "loss": 0.9126,
      "step": 237
    },
    {
      "epoch": 0.8164665523156089,
      "grad_norm": 0.4853699803352356,
      "learning_rate": 3.09251913772962e-05,
      "loss": 0.9592,
      "step": 238
    },
    {
      "epoch": 0.8198970840480274,
      "grad_norm": 0.47198018431663513,
      "learning_rate": 3.0613336825249346e-05,
      "loss": 0.8937,
      "step": 239
    },
    {
      "epoch": 0.823327615780446,
      "grad_norm": 0.4876589775085449,
      "learning_rate": 3.0301835696382067e-05,
      "loss": 0.935,
      "step": 240
    },
    {
      "epoch": 0.8267581475128645,
      "grad_norm": 0.5137282609939575,
      "learning_rate": 2.9990713087610816e-05,
      "loss": 0.8394,
      "step": 241
    },
    {
      "epoch": 0.8301886792452831,
      "grad_norm": 0.499109148979187,
      "learning_rate": 2.9679994065355647e-05,
      "loss": 0.9016,
      "step": 242
    },
    {
      "epoch": 0.8336192109777015,
      "grad_norm": 0.515414297580719,
      "learning_rate": 2.9369703663520558e-05,
      "loss": 0.8805,
      "step": 243
    },
    {
      "epoch": 0.8370497427101201,
      "grad_norm": 0.5474676489830017,
      "learning_rate": 2.9059866881476586e-05,
      "loss": 0.923,
      "step": 244
    },
    {
      "epoch": 0.8404802744425386,
      "grad_norm": 0.5066429972648621,
      "learning_rate": 2.8750508682047718e-05,
      "loss": 0.8451,
      "step": 245
    },
    {
      "epoch": 0.8439108061749572,
      "grad_norm": 0.6373265385627747,
      "learning_rate": 2.8441653989499627e-05,
      "loss": 0.9624,
      "step": 246
    },
    {
      "epoch": 0.8473413379073756,
      "grad_norm": 0.6143672466278076,
      "learning_rate": 2.8133327687531674e-05,
      "loss": 0.8682,
      "step": 247
    },
    {
      "epoch": 0.8507718696397941,
      "grad_norm": 0.7043458223342896,
      "learning_rate": 2.7825554617271934e-05,
      "loss": 0.9473,
      "step": 248
    },
    {
      "epoch": 0.8542024013722127,
      "grad_norm": 0.6915895342826843,
      "learning_rate": 2.7518359575275987e-05,
      "loss": 0.9472,
      "step": 249
    },
    {
      "epoch": 0.8576329331046312,
      "grad_norm": 0.8624438643455505,
      "learning_rate": 2.7211767311528996e-05,
      "loss": 0.9846,
      "step": 250
    },
    {
      "epoch": 0.8610634648370498,
      "grad_norm": 0.2864700257778168,
      "learning_rate": 2.6905802527451673e-05,
      "loss": 1.0976,
      "step": 251
    },
    {
      "epoch": 0.8644939965694682,
      "grad_norm": 0.2880098521709442,
      "learning_rate": 2.660048987391019e-05,
      "loss": 1.1026,
      "step": 252
    },
    {
      "epoch": 0.8679245283018868,
      "grad_norm": 0.3027153015136719,
      "learning_rate": 2.6295853949230086e-05,
      "loss": 0.9648,
      "step": 253
    },
    {
      "epoch": 0.8713550600343053,
      "grad_norm": 0.31133800745010376,
      "learning_rate": 2.599191929721442e-05,
      "loss": 1.011,
      "step": 254
    },
    {
      "epoch": 0.8747855917667239,
      "grad_norm": 0.32782459259033203,
      "learning_rate": 2.568871040516637e-05,
      "loss": 1.001,
      "step": 255
    },
    {
      "epoch": 0.8782161234991424,
      "grad_norm": 0.31751176714897156,
      "learning_rate": 2.5386251701916346e-05,
      "loss": 0.9752,
      "step": 256
    },
    {
      "epoch": 0.8816466552315609,
      "grad_norm": 0.3137318789958954,
      "learning_rate": 2.5084567555853778e-05,
      "loss": 0.9233,
      "step": 257
    },
    {
      "epoch": 0.8850771869639794,
      "grad_norm": 0.32128798961639404,
      "learning_rate": 2.4783682272963833e-05,
      "loss": 0.9362,
      "step": 258
    },
    {
      "epoch": 0.888507718696398,
      "grad_norm": 0.3337719142436981,
      "learning_rate": 2.448362009486918e-05,
      "loss": 1.0166,
      "step": 259
    },
    {
      "epoch": 0.8919382504288165,
      "grad_norm": 0.32666200399398804,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 1.0077,
      "step": 260
    },
    {
      "epoch": 0.8953687821612349,
      "grad_norm": 0.33111441135406494,
      "learning_rate": 2.3886061686030447e-05,
      "loss": 0.8519,
      "step": 261
    },
    {
      "epoch": 0.8987993138936535,
      "grad_norm": 0.35283076763153076,
      "learning_rate": 2.3588613599168032e-05,
      "loss": 0.9571,
      "step": 262
    },
    {
      "epoch": 0.902229845626072,
      "grad_norm": 0.32953575253486633,
      "learning_rate": 2.3292084900985412e-05,
      "loss": 0.9323,
      "step": 263
    },
    {
      "epoch": 0.9056603773584906,
      "grad_norm": 0.32089030742645264,
      "learning_rate": 2.29964994821054e-05,
      "loss": 0.8821,
      "step": 264
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.3504737317562103,
      "learning_rate": 2.2701881157153e-05,
      "loss": 0.8855,
      "step": 265
    },
    {
      "epoch": 0.9125214408233276,
      "grad_norm": 0.36239007115364075,
      "learning_rate": 2.2408253662836764e-05,
      "loss": 0.9553,
      "step": 266
    },
    {
      "epoch": 0.9159519725557461,
      "grad_norm": 0.3552619218826294,
      "learning_rate": 2.2115640656036272e-05,
      "loss": 0.9501,
      "step": 267
    },
    {
      "epoch": 0.9193825042881647,
      "grad_norm": 0.39280280470848083,
      "learning_rate": 2.182406571189628e-05,
      "loss": 0.8483,
      "step": 268
    },
    {
      "epoch": 0.9228130360205832,
      "grad_norm": 0.3870380222797394,
      "learning_rate": 2.1533552321927244e-05,
      "loss": 0.8857,
      "step": 269
    },
    {
      "epoch": 0.9262435677530018,
      "grad_norm": 0.3588116765022278,
      "learning_rate": 2.1244123892112673e-05,
      "loss": 0.8503,
      "step": 270
    },
    {
      "epoch": 0.9296740994854202,
      "grad_norm": 0.40483033657073975,
      "learning_rate": 2.09558037410234e-05,
      "loss": 0.8276,
      "step": 271
    },
    {
      "epoch": 0.9331046312178388,
      "grad_norm": 0.41076287627220154,
      "learning_rate": 2.0668615097938837e-05,
      "loss": 0.8986,
      "step": 272
    },
    {
      "epoch": 0.9365351629502573,
      "grad_norm": 0.43502911925315857,
      "learning_rate": 2.0382581100975413e-05,
      "loss": 0.8615,
      "step": 273
    },
    {
      "epoch": 0.9399656946826758,
      "grad_norm": 0.4127068519592285,
      "learning_rate": 2.0097724795222454e-05,
      "loss": 0.8874,
      "step": 274
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 0.42638474702835083,
      "learning_rate": 1.9814069130885465e-05,
      "loss": 0.9345,
      "step": 275
    },
    {
      "epoch": 0.9468267581475128,
      "grad_norm": 0.4107501208782196,
      "learning_rate": 1.9531636961437072e-05,
      "loss": 0.8836,
      "step": 276
    },
    {
      "epoch": 0.9502572898799314,
      "grad_norm": 0.4086393415927887,
      "learning_rate": 1.9250451041775757e-05,
      "loss": 0.8445,
      "step": 277
    },
    {
      "epoch": 0.9536878216123499,
      "grad_norm": 0.46297699213027954,
      "learning_rate": 1.897053402639264e-05,
      "loss": 0.9266,
      "step": 278
    },
    {
      "epoch": 0.9571183533447685,
      "grad_norm": 0.4402381181716919,
      "learning_rate": 1.869190846754614e-05,
      "loss": 0.9002,
      "step": 279
    },
    {
      "epoch": 0.9605488850771869,
      "grad_norm": 0.4369219243526459,
      "learning_rate": 1.8414596813445046e-05,
      "loss": 0.892,
      "step": 280
    },
    {
      "epoch": 0.9639794168096055,
      "grad_norm": 0.46288469433784485,
      "learning_rate": 1.8138621406439958e-05,
      "loss": 0.8939,
      "step": 281
    },
    {
      "epoch": 0.967409948542024,
      "grad_norm": 0.52180415391922,
      "learning_rate": 1.7864004481223176e-05,
      "loss": 0.9288,
      "step": 282
    },
    {
      "epoch": 0.9708404802744426,
      "grad_norm": 0.4818165600299835,
      "learning_rate": 1.7590768163037234e-05,
      "loss": 0.8919,
      "step": 283
    },
    {
      "epoch": 0.9742710120068611,
      "grad_norm": 0.4901525378227234,
      "learning_rate": 1.7318934465892462e-05,
      "loss": 0.8918,
      "step": 284
    },
    {
      "epoch": 0.9777015437392796,
      "grad_norm": 0.5124236941337585,
      "learning_rate": 1.7048525290793288e-05,
      "loss": 0.9144,
      "step": 285
    },
    {
      "epoch": 0.9811320754716981,
      "grad_norm": 0.5014034509658813,
      "learning_rate": 1.6779562423973706e-05,
      "loss": 0.8403,
      "step": 286
    },
    {
      "epoch": 0.9845626072041166,
      "grad_norm": 0.5504285097122192,
      "learning_rate": 1.6512067535142047e-05,
      "loss": 0.8671,
      "step": 287
    },
    {
      "epoch": 0.9879931389365352,
      "grad_norm": 0.564166247844696,
      "learning_rate": 1.6246062175735108e-05,
      "loss": 0.967,
      "step": 288
    },
    {
      "epoch": 0.9914236706689536,
      "grad_norm": 0.637752115726471,
      "learning_rate": 1.598156777718181e-05,
      "loss": 1.0031,
      "step": 289
    },
    {
      "epoch": 0.9948542024013722,
      "grad_norm": 0.6041699647903442,
      "learning_rate": 1.5718605649176414e-05,
      "loss": 0.8875,
      "step": 290
    },
    {
      "epoch": 0.9982847341337907,
      "grad_norm": 0.6616570949554443,
      "learning_rate": 1.5457196977961805e-05,
      "loss": 0.8621,
      "step": 291
    },
    {
      "epoch": 1.0017152658662092,
      "grad_norm": 1.1894855499267578,
      "learning_rate": 1.5197362824622492e-05,
      "loss": 1.2629,
      "step": 292
    },
    {
      "epoch": 1.0051457975986278,
      "grad_norm": 0.27210918068885803,
      "learning_rate": 1.4939124123387724e-05,
      "loss": 1.077,
      "step": 293
    },
    {
      "epoch": 1.0085763293310464,
      "grad_norm": 0.27636784315109253,
      "learning_rate": 1.4682501679944924e-05,
      "loss": 1.0727,
      "step": 294
    },
    {
      "epoch": 1.012006861063465,
      "grad_norm": 0.27114996314048767,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 0.9625,
      "step": 295
    },
    {
      "epoch": 1.0154373927958833,
      "grad_norm": 0.2962283790111542,
      "learning_rate": 1.4174188136428706e-05,
      "loss": 0.9278,
      "step": 296
    },
    {
      "epoch": 1.0188679245283019,
      "grad_norm": 0.3061296343803406,
      "learning_rate": 1.3922537989987123e-05,
      "loss": 0.9629,
      "step": 297
    },
    {
      "epoch": 1.0222984562607205,
      "grad_norm": 0.31669551134109497,
      "learning_rate": 1.3672586005301718e-05,
      "loss": 0.936,
      "step": 298
    },
    {
      "epoch": 1.0257289879931388,
      "grad_norm": 0.3030271530151367,
      "learning_rate": 1.3424352320418558e-05,
      "loss": 0.972,
      "step": 299
    },
    {
      "epoch": 1.0291595197255574,
      "grad_norm": 0.30286943912506104,
      "learning_rate": 1.3177856934944328e-05,
      "loss": 0.9446,
      "step": 300
    },
    {
      "epoch": 1.0291595197255574,
      "eval_loss": 0.902333676815033,
      "eval_runtime": 36.4331,
      "eval_samples_per_second": 13.477,
      "eval_steps_per_second": 3.376,
      "step": 300
    },
    {
      "epoch": 1.032590051457976,
      "grad_norm": 0.30614227056503296,
      "learning_rate": 1.2933119708435013e-05,
      "loss": 0.9224,
      "step": 301
    },
    {
      "epoch": 1.0360205831903946,
      "grad_norm": 0.33084699511528015,
      "learning_rate": 1.2690160358795858e-05,
      "loss": 0.9067,
      "step": 302
    },
    {
      "epoch": 1.039451114922813,
      "grad_norm": 0.31589779257774353,
      "learning_rate": 1.2448998460692702e-05,
      "loss": 0.9176,
      "step": 303
    },
    {
      "epoch": 1.0428816466552315,
      "grad_norm": 0.3340657949447632,
      "learning_rate": 1.2209653443974924e-05,
      "loss": 0.9584,
      "step": 304
    },
    {
      "epoch": 1.0463121783876501,
      "grad_norm": 0.34587377309799194,
      "learning_rate": 1.1972144592110058e-05,
      "loss": 0.9104,
      "step": 305
    },
    {
      "epoch": 1.0497427101200687,
      "grad_norm": 0.30994516611099243,
      "learning_rate": 1.173649104063011e-05,
      "loss": 0.8132,
      "step": 306
    },
    {
      "epoch": 1.053173241852487,
      "grad_norm": 0.3400658965110779,
      "learning_rate": 1.150271177558986e-05,
      "loss": 0.8421,
      "step": 307
    },
    {
      "epoch": 1.0566037735849056,
      "grad_norm": 0.35720840096473694,
      "learning_rate": 1.1270825632037277e-05,
      "loss": 0.8611,
      "step": 308
    },
    {
      "epoch": 1.0600343053173242,
      "grad_norm": 0.3368900716304779,
      "learning_rate": 1.1040851292495893e-05,
      "loss": 0.7947,
      "step": 309
    },
    {
      "epoch": 1.0634648370497426,
      "grad_norm": 0.34533587098121643,
      "learning_rate": 1.0812807285459736e-05,
      "loss": 0.8882,
      "step": 310
    },
    {
      "epoch": 1.0668953687821612,
      "grad_norm": 0.3583405911922455,
      "learning_rate": 1.0586711983900424e-05,
      "loss": 0.8356,
      "step": 311
    },
    {
      "epoch": 1.0703259005145798,
      "grad_norm": 0.37022221088409424,
      "learning_rate": 1.0362583603786926e-05,
      "loss": 0.7715,
      "step": 312
    },
    {
      "epoch": 1.0737564322469983,
      "grad_norm": 0.3587871491909027,
      "learning_rate": 1.0140440202618006e-05,
      "loss": 0.8419,
      "step": 313
    },
    {
      "epoch": 1.0771869639794167,
      "grad_norm": 0.37034180760383606,
      "learning_rate": 9.92029967796727e-06,
      "loss": 0.8245,
      "step": 314
    },
    {
      "epoch": 1.0806174957118353,
      "grad_norm": 0.3774360120296478,
      "learning_rate": 9.7021797660413e-06,
      "loss": 0.898,
      "step": 315
    },
    {
      "epoch": 1.0840480274442539,
      "grad_norm": 0.35760581493377686,
      "learning_rate": 9.486098040250603e-06,
      "loss": 0.8397,
      "step": 316
    },
    {
      "epoch": 1.0874785591766725,
      "grad_norm": 0.37607425451278687,
      "learning_rate": 9.272071909793812e-06,
      "loss": 0.83,
      "step": 317
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.38920724391937256,
      "learning_rate": 9.060118618255087e-06,
      "loss": 0.8629,
      "step": 318
    },
    {
      "epoch": 1.0943396226415094,
      "grad_norm": 0.38250231742858887,
      "learning_rate": 8.85025524221476e-06,
      "loss": 0.8374,
      "step": 319
    },
    {
      "epoch": 1.097770154373928,
      "grad_norm": 0.37902170419692993,
      "learning_rate": 8.64249868987362e-06,
      "loss": 0.8281,
      "step": 320
    },
    {
      "epoch": 1.1012006861063466,
      "grad_norm": 0.4074465036392212,
      "learning_rate": 8.43686569969056e-06,
      "loss": 0.7924,
      "step": 321
    },
    {
      "epoch": 1.104631217838765,
      "grad_norm": 0.40183237195014954,
      "learning_rate": 8.233372839034056e-06,
      "loss": 0.8825,
      "step": 322
    },
    {
      "epoch": 1.1080617495711835,
      "grad_norm": 0.40959110856056213,
      "learning_rate": 8.03203650284738e-06,
      "loss": 0.8316,
      "step": 323
    },
    {
      "epoch": 1.1114922813036021,
      "grad_norm": 0.4111100733280182,
      "learning_rate": 7.83287291232765e-06,
      "loss": 0.8548,
      "step": 324
    },
    {
      "epoch": 1.1149228130360207,
      "grad_norm": 0.43431782722473145,
      "learning_rate": 7.635898113618957e-06,
      "loss": 0.8036,
      "step": 325
    },
    {
      "epoch": 1.118353344768439,
      "grad_norm": 0.4242084324359894,
      "learning_rate": 7.441127976519579e-06,
      "loss": 0.7887,
      "step": 326
    },
    {
      "epoch": 1.1217838765008576,
      "grad_norm": 0.4480046331882477,
      "learning_rate": 7.2485781932033645e-06,
      "loss": 0.8647,
      "step": 327
    },
    {
      "epoch": 1.1252144082332762,
      "grad_norm": 0.42903417348861694,
      "learning_rate": 7.058264276955453e-06,
      "loss": 0.7581,
      "step": 328
    },
    {
      "epoch": 1.1286449399656946,
      "grad_norm": 0.47604501247406006,
      "learning_rate": 6.870201560922393e-06,
      "loss": 0.8808,
      "step": 329
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 0.47266608476638794,
      "learning_rate": 6.684405196876842e-06,
      "loss": 0.8516,
      "step": 330
    },
    {
      "epoch": 1.1355060034305318,
      "grad_norm": 0.49949389696121216,
      "learning_rate": 6.500890153996754e-06,
      "loss": 0.8614,
      "step": 331
    },
    {
      "epoch": 1.1389365351629503,
      "grad_norm": 0.4845390021800995,
      "learning_rate": 6.319671217659384e-06,
      "loss": 0.8862,
      "step": 332
    },
    {
      "epoch": 1.1423670668953687,
      "grad_norm": 0.5104158520698547,
      "learning_rate": 6.140762988250079e-06,
      "loss": 0.8593,
      "step": 333
    },
    {
      "epoch": 1.1457975986277873,
      "grad_norm": 0.5185325741767883,
      "learning_rate": 5.964179879985916e-06,
      "loss": 0.8589,
      "step": 334
    },
    {
      "epoch": 1.1492281303602059,
      "grad_norm": 0.508443295955658,
      "learning_rate": 5.789936119754407e-06,
      "loss": 0.7782,
      "step": 335
    },
    {
      "epoch": 1.1526586620926245,
      "grad_norm": 0.5443438291549683,
      "learning_rate": 5.618045745967295e-06,
      "loss": 0.8328,
      "step": 336
    },
    {
      "epoch": 1.1560891938250428,
      "grad_norm": 0.5635469555854797,
      "learning_rate": 5.4485226074294795e-06,
      "loss": 0.8997,
      "step": 337
    },
    {
      "epoch": 1.1595197255574614,
      "grad_norm": 0.5843603610992432,
      "learning_rate": 5.2813803622232276e-06,
      "loss": 0.7866,
      "step": 338
    },
    {
      "epoch": 1.16295025728988,
      "grad_norm": 0.5807653665542603,
      "learning_rate": 5.116632476607844e-06,
      "loss": 0.8116,
      "step": 339
    },
    {
      "epoch": 1.1663807890222984,
      "grad_norm": 0.6621823310852051,
      "learning_rate": 4.954292223934686e-06,
      "loss": 0.7836,
      "step": 340
    },
    {
      "epoch": 1.169811320754717,
      "grad_norm": 0.8782541751861572,
      "learning_rate": 4.794372683577743e-06,
      "loss": 0.851,
      "step": 341
    },
    {
      "epoch": 1.1732418524871355,
      "grad_norm": 0.686014711856842,
      "learning_rate": 4.636886739879885e-06,
      "loss": 0.8096,
      "step": 342
    },
    {
      "epoch": 1.1766723842195541,
      "grad_norm": 0.3087368309497833,
      "learning_rate": 4.48184708111481e-06,
      "loss": 1.1842,
      "step": 343
    },
    {
      "epoch": 1.1801029159519725,
      "grad_norm": 0.2577741742134094,
      "learning_rate": 4.329266198464782e-06,
      "loss": 0.9624,
      "step": 344
    },
    {
      "epoch": 1.183533447684391,
      "grad_norm": 0.3151746690273285,
      "learning_rate": 4.179156385014169e-06,
      "loss": 1.0816,
      "step": 345
    },
    {
      "epoch": 1.1869639794168096,
      "grad_norm": 0.2938729226589203,
      "learning_rate": 4.0315297347591515e-06,
      "loss": 0.9021,
      "step": 346
    },
    {
      "epoch": 1.1903945111492282,
      "grad_norm": 0.3388144373893738,
      "learning_rate": 3.886398141633238e-06,
      "loss": 1.0032,
      "step": 347
    },
    {
      "epoch": 1.1938250428816466,
      "grad_norm": 0.3290029466152191,
      "learning_rate": 3.743773298549014e-06,
      "loss": 0.8824,
      "step": 348
    },
    {
      "epoch": 1.1972555746140652,
      "grad_norm": 0.32908475399017334,
      "learning_rate": 3.6036666964560957e-06,
      "loss": 0.9527,
      "step": 349
    },
    {
      "epoch": 1.2006861063464838,
      "grad_norm": 0.37448111176490784,
      "learning_rate": 3.4660896234153336e-06,
      "loss": 1.0084,
      "step": 350
    },
    {
      "epoch": 1.2041166380789021,
      "grad_norm": 0.3365894854068756,
      "learning_rate": 3.3310531636893224e-06,
      "loss": 0.9195,
      "step": 351
    },
    {
      "epoch": 1.2075471698113207,
      "grad_norm": 0.34142544865608215,
      "learning_rate": 3.1985681968493928e-06,
      "loss": 0.8443,
      "step": 352
    },
    {
      "epoch": 1.2109777015437393,
      "grad_norm": 0.33907952904701233,
      "learning_rate": 3.0686453968990917e-06,
      "loss": 0.9509,
      "step": 353
    },
    {
      "epoch": 1.2144082332761579,
      "grad_norm": 0.3465336859226227,
      "learning_rate": 2.941295231414174e-06,
      "loss": 0.8515,
      "step": 354
    },
    {
      "epoch": 1.2178387650085762,
      "grad_norm": 0.3430810272693634,
      "learning_rate": 2.8165279606992298e-06,
      "loss": 0.8187,
      "step": 355
    },
    {
      "epoch": 1.2212692967409948,
      "grad_norm": 0.3591018319129944,
      "learning_rate": 2.6943536369610948e-06,
      "loss": 0.8756,
      "step": 356
    },
    {
      "epoch": 1.2246998284734134,
      "grad_norm": 0.3418238162994385,
      "learning_rate": 2.574782103498939e-06,
      "loss": 0.8075,
      "step": 357
    },
    {
      "epoch": 1.228130360205832,
      "grad_norm": 0.37161850929260254,
      "learning_rate": 2.4578229939112028e-06,
      "loss": 0.8725,
      "step": 358
    },
    {
      "epoch": 1.2315608919382504,
      "grad_norm": 0.36024555563926697,
      "learning_rate": 2.3434857313194522e-06,
      "loss": 0.8085,
      "step": 359
    },
    {
      "epoch": 1.234991423670669,
      "grad_norm": 0.3657310903072357,
      "learning_rate": 2.2317795276091974e-06,
      "loss": 0.8603,
      "step": 360
    },
    {
      "epoch": 1.2384219554030875,
      "grad_norm": 0.3807387351989746,
      "learning_rate": 2.1227133826876807e-06,
      "loss": 0.9165,
      "step": 361
    },
    {
      "epoch": 1.241852487135506,
      "grad_norm": 0.36893582344055176,
      "learning_rate": 2.0162960837587843e-06,
      "loss": 0.8543,
      "step": 362
    },
    {
      "epoch": 1.2452830188679245,
      "grad_norm": 0.3827870190143585,
      "learning_rate": 1.91253620461509e-06,
      "loss": 0.888,
      "step": 363
    },
    {
      "epoch": 1.248713550600343,
      "grad_norm": 0.3683498203754425,
      "learning_rate": 1.8114421049470751e-06,
      "loss": 0.8347,
      "step": 364
    },
    {
      "epoch": 1.2521440823327616,
      "grad_norm": 0.38052716851234436,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 0.866,
      "step": 365
    },
    {
      "epoch": 1.2555746140651802,
      "grad_norm": 0.3877822756767273,
      "learning_rate": 1.617283608265781e-06,
      "loss": 0.8293,
      "step": 366
    },
    {
      "epoch": 1.2590051457975986,
      "grad_norm": 0.39266732335090637,
      "learning_rate": 1.5242348541478972e-06,
      "loss": 0.8322,
      "step": 367
    },
    {
      "epoch": 1.2624356775300172,
      "grad_norm": 0.42105159163475037,
      "learning_rate": 1.4338831640362052e-06,
      "loss": 0.8821,
      "step": 368
    },
    {
      "epoch": 1.2658662092624358,
      "grad_norm": 0.42221251130104065,
      "learning_rate": 1.3462358173547898e-06,
      "loss": 0.9076,
      "step": 369
    },
    {
      "epoch": 1.2692967409948541,
      "grad_norm": 0.39558228850364685,
      "learning_rate": 1.2612998756451365e-06,
      "loss": 0.8062,
      "step": 370
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.40436068177223206,
      "learning_rate": 1.1790821819971635e-06,
      "loss": 0.7965,
      "step": 371
    },
    {
      "epoch": 1.2761578044596913,
      "grad_norm": 0.42944255471229553,
      "learning_rate": 1.0995893604979122e-06,
      "loss": 0.9084,
      "step": 372
    },
    {
      "epoch": 1.2795883361921097,
      "grad_norm": 0.40103161334991455,
      "learning_rate": 1.0228278156978653e-06,
      "loss": 0.8577,
      "step": 373
    },
    {
      "epoch": 1.2830188679245282,
      "grad_norm": 0.4177500605583191,
      "learning_rate": 9.488037320949271e-07,
      "loss": 0.8069,
      "step": 374
    },
    {
      "epoch": 1.2864493996569468,
      "grad_norm": 0.4498164653778076,
      "learning_rate": 8.775230736361732e-07,
      "loss": 0.8458,
      "step": 375
    },
    {
      "epoch": 1.2898799313893654,
      "grad_norm": 0.4338374137878418,
      "learning_rate": 8.08991583237335e-07,
      "loss": 0.8328,
      "step": 376
    },
    {
      "epoch": 1.293310463121784,
      "grad_norm": 0.4457828104496002,
      "learning_rate": 7.432147823201001e-07,
      "loss": 0.8506,
      "step": 377
    },
    {
      "epoch": 1.2967409948542024,
      "grad_norm": 0.449349582195282,
      "learning_rate": 6.801979703672994e-07,
      "loss": 0.7733,
      "step": 378
    },
    {
      "epoch": 1.300171526586621,
      "grad_norm": 0.46898260712623596,
      "learning_rate": 6.199462244958986e-07,
      "loss": 0.8709,
      "step": 379
    },
    {
      "epoch": 1.3036020583190395,
      "grad_norm": 0.46654513478279114,
      "learning_rate": 5.624643990479616e-07,
      "loss": 0.8177,
      "step": 380
    },
    {
      "epoch": 1.3070325900514579,
      "grad_norm": 0.46529287099838257,
      "learning_rate": 5.07757125199566e-07,
      "loss": 0.8263,
      "step": 381
    },
    {
      "epoch": 1.3104631217838765,
      "grad_norm": 0.4870767295360565,
      "learning_rate": 4.5582881058765176e-07,
      "loss": 0.8359,
      "step": 382
    },
    {
      "epoch": 1.313893653516295,
      "grad_norm": 0.5216002464294434,
      "learning_rate": 4.0668363895492663e-07,
      "loss": 0.8749,
      "step": 383
    },
    {
      "epoch": 1.3173241852487134,
      "grad_norm": 0.4754326343536377,
      "learning_rate": 3.6032556981277885e-07,
      "loss": 0.7423,
      "step": 384
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 0.5112282633781433,
      "learning_rate": 3.1675833812228545e-07,
      "loss": 0.7929,
      "step": 385
    },
    {
      "epoch": 1.3241852487135506,
      "grad_norm": 0.550466001033783,
      "learning_rate": 2.759854539932782e-07,
      "loss": 0.8225,
      "step": 386
    },
    {
      "epoch": 1.3276157804459692,
      "grad_norm": 0.5394650101661682,
      "learning_rate": 2.380102024015479e-07,
      "loss": 0.7906,
      "step": 387
    },
    {
      "epoch": 1.3310463121783878,
      "grad_norm": 0.5764582753181458,
      "learning_rate": 2.028356429241995e-07,
      "loss": 0.8316,
      "step": 388
    },
    {
      "epoch": 1.3344768439108061,
      "grad_norm": 0.6128343343734741,
      "learning_rate": 1.7046460949310724e-07,
      "loss": 0.8267,
      "step": 389
    },
    {
      "epoch": 1.3379073756432247,
      "grad_norm": 0.6350001692771912,
      "learning_rate": 1.408997101666326e-07,
      "loss": 0.8195,
      "step": 390
    },
    {
      "epoch": 1.3413379073756433,
      "grad_norm": 0.7249876856803894,
      "learning_rate": 1.1414332691948913e-07,
      "loss": 0.7925,
      "step": 391
    },
    {
      "epoch": 1.3447684391080617,
      "grad_norm": 0.553812563419342,
      "learning_rate": 9.019761545080828e-08,
      "loss": 0.8482,
      "step": 392
    },
    {
      "epoch": 1.3481989708404802,
      "grad_norm": 0.2474144697189331,
      "learning_rate": 6.906450501049543e-08,
      "loss": 1.049,
      "step": 393
    },
    {
      "epoch": 1.3516295025728988,
      "grad_norm": 0.26705402135849,
      "learning_rate": 5.074569824376762e-08,
      "loss": 1.0023,
      "step": 394
    },
    {
      "epoch": 1.3550600343053174,
      "grad_norm": 0.27517229318618774,
      "learning_rate": 3.524267105398937e-08,
      "loss": 0.9524,
      "step": 395
    },
    {
      "epoch": 1.3584905660377358,
      "grad_norm": 0.29240480065345764,
      "learning_rate": 2.2556672483752258e-08,
      "loss": 0.9617,
      "step": 396
    },
    {
      "epoch": 1.3619210977701544,
      "grad_norm": 0.2889561057090759,
      "learning_rate": 1.2688724614260404e-08,
      "loss": 0.9433,
      "step": 397
    },
    {
      "epoch": 1.365351629502573,
      "grad_norm": 0.29089388251304626,
      "learning_rate": 5.639622482963568e-09,
      "loss": 0.8261,
      "step": 398
    },
    {
      "epoch": 1.3687821612349915,
      "grad_norm": 0.29852592945098877,
      "learning_rate": 1.409934019511727e-09,
      "loss": 0.8461,
      "step": 399
    },
    {
      "epoch": 1.3722126929674099,
      "grad_norm": 0.31576597690582275,
      "learning_rate": 0.0,
      "loss": 0.8767,
      "step": 400
    },
    {
      "epoch": 1.3722126929674099,
      "eval_loss": 0.8880267143249512,
      "eval_runtime": 36.7109,
      "eval_samples_per_second": 13.375,
      "eval_steps_per_second": 3.351,
      "step": 400
    }
  ],
  "logging_steps": 1,
  "max_steps": 400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 4,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.529926614646784e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}