{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5249619402593312,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002099847761037325,
      "grad_norm": 22925.884765625,
      "learning_rate": 1.9999999999999996e-07,
      "loss": 9.5004,
      "step": 10
    },
    {
      "epoch": 0.00419969552207465,
      "grad_norm": 1589.221435546875,
      "learning_rate": 1.2e-06,
      "loss": 8.7173,
      "step": 20
    },
    {
      "epoch": 0.006299543283111974,
      "grad_norm": 577.2943115234375,
      "learning_rate": 2.1999999999999997e-06,
      "loss": 7.6284,
      "step": 30
    },
    {
      "epoch": 0.0083993910441493,
      "grad_norm": 308.8800048828125,
      "learning_rate": 3.1999999999999994e-06,
      "loss": 7.2988,
      "step": 40
    },
    {
      "epoch": 0.010499238805186623,
      "grad_norm": 458.5738220214844,
      "learning_rate": 4.2e-06,
      "loss": 7.1397,
      "step": 50
    },
    {
      "epoch": 0.012599086566223949,
      "grad_norm": 2117.726806640625,
      "learning_rate": 5.199999999999999e-06,
      "loss": 7.0424,
      "step": 60
    },
    {
      "epoch": 0.014698934327261274,
      "grad_norm": 784.7866821289062,
      "learning_rate": 6.199999999999999e-06,
      "loss": 6.9896,
      "step": 70
    },
    {
      "epoch": 0.0167987820882986,
      "grad_norm": 719.2965087890625,
      "learning_rate": 7.2e-06,
      "loss": 6.9514,
      "step": 80
    },
    {
      "epoch": 0.018898629849335925,
      "grad_norm": 887.7239379882812,
      "learning_rate": 8.2e-06,
      "loss": 6.9188,
      "step": 90
    },
    {
      "epoch": 0.020998477610373247,
      "grad_norm": 657.1996459960938,
      "learning_rate": 9.199999999999998e-06,
      "loss": 6.9113,
      "step": 100
    },
    {
      "epoch": 0.023098325371410572,
      "grad_norm": 1241.19482421875,
      "learning_rate": 1.02e-05,
      "loss": 6.8953,
      "step": 110
    },
    {
      "epoch": 0.025198173132447897,
      "grad_norm": 411.5585021972656,
      "learning_rate": 1.12e-05,
      "loss": 6.8474,
      "step": 120
    },
    {
      "epoch": 0.027298020893485223,
      "grad_norm": 1062.7774658203125,
      "learning_rate": 1.2199999999999998e-05,
      "loss": 6.8514,
      "step": 130
    },
    {
      "epoch": 0.029397868654522548,
      "grad_norm": 148.0970001220703,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 6.8698,
      "step": 140
    },
    {
      "epoch": 0.03149771641555987,
      "grad_norm": 280.5868835449219,
      "learning_rate": 1.4199999999999998e-05,
      "loss": 6.8395,
      "step": 150
    },
    {
      "epoch": 0.0335975641765972,
      "grad_norm": 407.1022644042969,
      "learning_rate": 1.5199999999999998e-05,
      "loss": 6.8029,
      "step": 160
    },
    {
      "epoch": 0.035697411937634524,
      "grad_norm": 166.18153381347656,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 6.7946,
      "step": 170
    },
    {
      "epoch": 0.03779725969867185,
      "grad_norm": 234.69805908203125,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 6.7898,
      "step": 180
    },
    {
      "epoch": 0.03989710745970917,
      "grad_norm": 917.2261352539062,
      "learning_rate": 1.82e-05,
      "loss": 6.7996,
      "step": 190
    },
    {
      "epoch": 0.04199695522074649,
      "grad_norm": 237.35519409179688,
      "learning_rate": 1.92e-05,
      "loss": 6.7733,
      "step": 200
    },
    {
      "epoch": 0.04409680298178382,
      "grad_norm": 49.55766296386719,
      "learning_rate": 2.0199999999999996e-05,
      "loss": 6.7318,
      "step": 210
    },
    {
      "epoch": 0.046196650742821144,
      "grad_norm": 152.66554260253906,
      "learning_rate": 2.12e-05,
      "loss": 6.6867,
      "step": 220
    },
    {
      "epoch": 0.04829649850385847,
      "grad_norm": 284.306640625,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 6.6555,
      "step": 230
    },
    {
      "epoch": 0.050396346264895794,
      "grad_norm": 127.93234252929688,
      "learning_rate": 2.3199999999999998e-05,
      "loss": 6.6263,
      "step": 240
    },
    {
      "epoch": 0.05249619402593312,
      "grad_norm": 119.77033996582031,
      "learning_rate": 2.42e-05,
      "loss": 6.649,
      "step": 250
    },
    {
      "epoch": 0.054596041786970445,
      "grad_norm": 389.56536865234375,
      "learning_rate": 2.52e-05,
      "loss": 6.6383,
      "step": 260
    },
    {
      "epoch": 0.05669588954800777,
      "grad_norm": 427.01580810546875,
      "learning_rate": 2.6199999999999996e-05,
      "loss": 6.593,
      "step": 270
    },
    {
      "epoch": 0.058795737309045096,
      "grad_norm": 222.83114624023438,
      "learning_rate": 2.72e-05,
      "loss": 6.5701,
      "step": 280
    },
    {
      "epoch": 0.06089558507008242,
      "grad_norm": 92.09854125976562,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 6.5495,
      "step": 290
    },
    {
      "epoch": 0.06299543283111975,
      "grad_norm": 137.1016082763672,
      "learning_rate": 2.9199999999999995e-05,
      "loss": 6.5433,
      "step": 300
    },
    {
      "epoch": 0.06509528059215706,
      "grad_norm": 165.11099243164062,
      "learning_rate": 3.02e-05,
      "loss": 6.5528,
      "step": 310
    },
    {
      "epoch": 0.0671951283531944,
      "grad_norm": 193.72972106933594,
      "learning_rate": 3.119999999999999e-05,
      "loss": 6.4983,
      "step": 320
    },
    {
      "epoch": 0.06929497611423172,
      "grad_norm": 168.2730255126953,
      "learning_rate": 3.22e-05,
      "loss": 6.4875,
      "step": 330
    },
    {
      "epoch": 0.07139482387526905,
      "grad_norm": 204.45751953125,
      "learning_rate": 3.3199999999999994e-05,
      "loss": 6.4667,
      "step": 340
    },
    {
      "epoch": 0.07349467163630637,
      "grad_norm": 200.1185302734375,
      "learning_rate": 3.42e-05,
      "loss": 6.4163,
      "step": 350
    },
    {
      "epoch": 0.0755945193973437,
      "grad_norm": 212.8289337158203,
      "learning_rate": 3.5199999999999995e-05,
      "loss": 6.4009,
      "step": 360
    },
    {
      "epoch": 0.07769436715838102,
      "grad_norm": 67.6531982421875,
      "learning_rate": 3.62e-05,
      "loss": 6.3942,
      "step": 370
    },
    {
      "epoch": 0.07979421491941834,
      "grad_norm": 103.08153533935547,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 6.37,
      "step": 380
    },
    {
      "epoch": 0.08189406268045567,
      "grad_norm": 155.30517578125,
      "learning_rate": 3.8199999999999993e-05,
      "loss": 6.3541,
      "step": 390
    },
    {
      "epoch": 0.08399391044149299,
      "grad_norm": 78.574462890625,
      "learning_rate": 3.919999999999999e-05,
      "loss": 6.331,
      "step": 400
    },
    {
      "epoch": 0.08609375820253032,
      "grad_norm": 293.90155029296875,
      "learning_rate": 4.02e-05,
      "loss": 6.2978,
      "step": 410
    },
    {
      "epoch": 0.08819360596356764,
      "grad_norm": 133.43336486816406,
      "learning_rate": 4.12e-05,
      "loss": 6.2921,
      "step": 420
    },
    {
      "epoch": 0.09029345372460497,
      "grad_norm": 105.68527221679688,
      "learning_rate": 4.2199999999999996e-05,
      "loss": 6.2955,
      "step": 430
    },
    {
      "epoch": 0.09239330148564229,
      "grad_norm": 91.55309295654297,
      "learning_rate": 4.319999999999999e-05,
      "loss": 6.2592,
      "step": 440
    },
    {
      "epoch": 0.09449314924667962,
      "grad_norm": 72.15143585205078,
      "learning_rate": 4.42e-05,
      "loss": 6.2545,
      "step": 450
    },
    {
      "epoch": 0.09659299700771694,
      "grad_norm": 109.66728210449219,
      "learning_rate": 4.5199999999999994e-05,
      "loss": 6.2325,
      "step": 460
    },
    {
      "epoch": 0.09869284476875427,
      "grad_norm": 69.46297454833984,
      "learning_rate": 4.62e-05,
      "loss": 6.1691,
      "step": 470
    },
    {
      "epoch": 0.10079269252979159,
      "grad_norm": 137.57276916503906,
      "learning_rate": 4.7199999999999995e-05,
      "loss": 6.1748,
      "step": 480
    },
    {
      "epoch": 0.10289254029082892,
      "grad_norm": 142.85269165039062,
      "learning_rate": 4.82e-05,
      "loss": 6.1763,
      "step": 490
    },
    {
      "epoch": 0.10499238805186624,
      "grad_norm": 68.21001434326172,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 6.1683,
      "step": 500
    },
    {
      "epoch": 0.10709223581290356,
      "grad_norm": 190.91734313964844,
      "learning_rate": 5.0199999999999994e-05,
      "loss": 6.1276,
      "step": 510
    },
    {
      "epoch": 0.10919208357394089,
      "grad_norm": 96.1802978515625,
      "learning_rate": 5.119999999999999e-05,
      "loss": 6.1462,
      "step": 520
    },
    {
      "epoch": 0.11129193133497821,
      "grad_norm": 67.83528137207031,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 6.1067,
      "step": 530
    },
    {
      "epoch": 0.11339177909601554,
      "grad_norm": 42.651100158691406,
      "learning_rate": 5.32e-05,
      "loss": 6.0776,
      "step": 540
    },
    {
      "epoch": 0.11549162685705286,
      "grad_norm": 79.75711822509766,
      "learning_rate": 5.4199999999999996e-05,
      "loss": 6.0815,
      "step": 550
    },
    {
      "epoch": 0.11759147461809019,
      "grad_norm": 46.156211853027344,
      "learning_rate": 5.519999999999999e-05,
      "loss": 6.053,
      "step": 560
    },
    {
      "epoch": 0.11969132237912751,
      "grad_norm": 45.88727951049805,
      "learning_rate": 5.619999999999999e-05,
      "loss": 6.0348,
      "step": 570
    },
    {
      "epoch": 0.12179117014016484,
      "grad_norm": 46.44386672973633,
      "learning_rate": 5.72e-05,
      "loss": 6.0195,
      "step": 580
    },
    {
      "epoch": 0.12389101790120216,
      "grad_norm": 70.52837371826172,
      "learning_rate": 5.82e-05,
      "loss": 5.9979,
      "step": 590
    },
    {
      "epoch": 0.1259908656622395,
      "grad_norm": 59.506553649902344,
      "learning_rate": 5.9199999999999996e-05,
      "loss": 5.983,
      "step": 600
    },
    {
      "epoch": 0.12809071342327683,
      "grad_norm": 38.77995681762695,
      "learning_rate": 6.019999999999999e-05,
      "loss": 5.9576,
      "step": 610
    },
    {
      "epoch": 0.13019056118431413,
      "grad_norm": 62.45197677612305,
      "learning_rate": 6.12e-05,
      "loss": 5.951,
      "step": 620
    },
    {
      "epoch": 0.13229040894535146,
      "grad_norm": 46.94855499267578,
      "learning_rate": 6.22e-05,
      "loss": 5.9626,
      "step": 630
    },
    {
      "epoch": 0.1343902567063888,
      "grad_norm": 54.95368194580078,
      "learning_rate": 6.319999999999999e-05,
      "loss": 5.9268,
      "step": 640
    },
    {
      "epoch": 0.1364901044674261,
      "grad_norm": 47.36136245727539,
      "learning_rate": 6.419999999999999e-05,
      "loss": 5.8941,
      "step": 650
    },
    {
      "epoch": 0.13858995222846343,
      "grad_norm": 40.393096923828125,
      "learning_rate": 6.519999999999999e-05,
      "loss": 5.8787,
      "step": 660
    },
    {
      "epoch": 0.14068979998950076,
      "grad_norm": 41.42562484741211,
      "learning_rate": 6.62e-05,
      "loss": 5.8632,
      "step": 670
    },
    {
      "epoch": 0.1427896477505381,
      "grad_norm": 41.52008056640625,
      "learning_rate": 6.72e-05,
      "loss": 5.8166,
      "step": 680
    },
    {
      "epoch": 0.1448894955115754,
      "grad_norm": 60.34001922607422,
      "learning_rate": 6.819999999999999e-05,
      "loss": 5.8412,
      "step": 690
    },
    {
      "epoch": 0.14698934327261273,
      "grad_norm": 45.22390365600586,
      "learning_rate": 6.919999999999999e-05,
      "loss": 5.7904,
      "step": 700
    },
    {
      "epoch": 0.14908919103365006,
      "grad_norm": 40.19221496582031,
      "learning_rate": 7.02e-05,
      "loss": 5.7796,
      "step": 710
    },
    {
      "epoch": 0.1511890387946874,
      "grad_norm": 38.9415283203125,
      "learning_rate": 7.12e-05,
      "loss": 5.7776,
      "step": 720
    },
    {
      "epoch": 0.1532888865557247,
      "grad_norm": 36.409732818603516,
      "learning_rate": 7.219999999999999e-05,
      "loss": 5.7442,
      "step": 730
    },
    {
      "epoch": 0.15538873431676203,
      "grad_norm": 33.082157135009766,
      "learning_rate": 7.319999999999999e-05,
      "loss": 5.7215,
      "step": 740
    },
    {
      "epoch": 0.15748858207779937,
      "grad_norm": 40.2638053894043,
      "learning_rate": 7.419999999999999e-05,
      "loss": 5.6798,
      "step": 750
    },
    {
      "epoch": 0.15958842983883667,
      "grad_norm": 36.72783660888672,
      "learning_rate": 7.519999999999998e-05,
      "loss": 5.6984,
      "step": 760
    },
    {
      "epoch": 0.161688277599874,
      "grad_norm": 40.32900619506836,
      "learning_rate": 7.62e-05,
      "loss": 5.6429,
      "step": 770
    },
    {
      "epoch": 0.16378812536091134,
      "grad_norm": 31.345502853393555,
      "learning_rate": 7.72e-05,
      "loss": 5.6839,
      "step": 780
    },
    {
      "epoch": 0.16588797312194867,
      "grad_norm": 42.13299560546875,
      "learning_rate": 7.819999999999999e-05,
      "loss": 5.641,
      "step": 790
    },
    {
      "epoch": 0.16798782088298597,
      "grad_norm": 42.9160270690918,
      "learning_rate": 7.92e-05,
      "loss": 5.6259,
      "step": 800
    },
    {
      "epoch": 0.1700876686440233,
      "grad_norm": 33.14772415161133,
      "learning_rate": 8.019999999999998e-05,
      "loss": 5.6016,
      "step": 810
    },
    {
      "epoch": 0.17218751640506064,
      "grad_norm": 39.43216323852539,
      "learning_rate": 8.12e-05,
      "loss": 5.5918,
      "step": 820
    },
    {
      "epoch": 0.17428736416609797,
      "grad_norm": 31.36737823486328,
      "learning_rate": 8.22e-05,
      "loss": 5.5341,
      "step": 830
    },
    {
      "epoch": 0.17638721192713527,
      "grad_norm": 44.43039321899414,
      "learning_rate": 8.319999999999999e-05,
      "loss": 5.5423,
      "step": 840
    },
    {
      "epoch": 0.1784870596881726,
      "grad_norm": 38.169437408447266,
      "learning_rate": 8.42e-05,
      "loss": 5.5218,
      "step": 850
    },
    {
      "epoch": 0.18058690744920994,
      "grad_norm": 29.236454010009766,
      "learning_rate": 8.519999999999998e-05,
      "loss": 5.4835,
      "step": 860
    },
    {
      "epoch": 0.18268675521024727,
      "grad_norm": 35.421852111816406,
      "learning_rate": 8.62e-05,
      "loss": 5.4431,
      "step": 870
    },
    {
      "epoch": 0.18478660297128457,
      "grad_norm": 34.888694763183594,
      "learning_rate": 8.719999999999999e-05,
      "loss": 5.4332,
      "step": 880
    },
    {
      "epoch": 0.1868864507323219,
      "grad_norm": 30.417287826538086,
      "learning_rate": 8.819999999999999e-05,
      "loss": 5.418,
      "step": 890
    },
    {
      "epoch": 0.18898629849335924,
      "grad_norm": 32.38473892211914,
      "learning_rate": 8.92e-05,
      "loss": 5.4061,
      "step": 900
    },
    {
      "epoch": 0.19108614625439654,
      "grad_norm": 34.51039123535156,
      "learning_rate": 9.02e-05,
      "loss": 5.3979,
      "step": 910
    },
    {
      "epoch": 0.19318599401543388,
      "grad_norm": 32.25777053833008,
      "learning_rate": 9.12e-05,
      "loss": 5.3694,
      "step": 920
    },
    {
      "epoch": 0.1952858417764712,
      "grad_norm": 37.39424514770508,
      "learning_rate": 9.219999999999999e-05,
      "loss": 5.3673,
      "step": 930
    },
    {
      "epoch": 0.19738568953750854,
      "grad_norm": 32.6727180480957,
      "learning_rate": 9.319999999999999e-05,
      "loss": 5.3439,
      "step": 940
    },
    {
      "epoch": 0.19948553729854585,
      "grad_norm": 35.477760314941406,
      "learning_rate": 9.419999999999999e-05,
      "loss": 5.3101,
      "step": 950
    },
    {
      "epoch": 0.20158538505958318,
      "grad_norm": 26.56783103942871,
      "learning_rate": 9.52e-05,
      "loss": 5.3106,
      "step": 960
    },
    {
      "epoch": 0.2036852328206205,
      "grad_norm": 28.40528106689453,
      "learning_rate": 9.62e-05,
      "loss": 5.283,
      "step": 970
    },
    {
      "epoch": 0.20578508058165784,
      "grad_norm": 40.31626892089844,
      "learning_rate": 9.719999999999999e-05,
      "loss": 5.2491,
      "step": 980
    },
    {
      "epoch": 0.20788492834269515,
      "grad_norm": 30.382963180541992,
      "learning_rate": 9.819999999999999e-05,
      "loss": 5.2381,
      "step": 990
    },
    {
      "epoch": 0.20998477610373248,
      "grad_norm": 26.859519958496094,
      "learning_rate": 9.919999999999999e-05,
      "loss": 5.2041,
      "step": 1000
    },
    {
      "epoch": 0.2120846238647698,
      "grad_norm": 31.00687026977539,
      "learning_rate": 0.0001002,
      "loss": 5.1836,
      "step": 1010
    },
    {
      "epoch": 0.21418447162580712,
      "grad_norm": 29.516639709472656,
      "learning_rate": 0.0001012,
      "loss": 5.1701,
      "step": 1020
    },
    {
      "epoch": 0.21628431938684445,
      "grad_norm": 35.0793342590332,
      "learning_rate": 0.00010219999999999999,
      "loss": 5.1333,
      "step": 1030
    },
    {
      "epoch": 0.21838416714788178,
      "grad_norm": 29.049598693847656,
      "learning_rate": 0.00010319999999999999,
      "loss": 5.1305,
      "step": 1040
    },
    {
      "epoch": 0.2204840149089191,
      "grad_norm": 37.3223762512207,
      "learning_rate": 0.00010419999999999998,
      "loss": 5.1076,
      "step": 1050
    },
    {
      "epoch": 0.22258386266995642,
      "grad_norm": 34.8250846862793,
      "learning_rate": 0.0001052,
      "loss": 5.0842,
      "step": 1060
    },
    {
      "epoch": 0.22468371043099375,
      "grad_norm": 26.114519119262695,
      "learning_rate": 0.00010619999999999998,
      "loss": 5.0202,
      "step": 1070
    },
    {
      "epoch": 0.22678355819203108,
      "grad_norm": 27.96425437927246,
      "learning_rate": 0.00010719999999999999,
      "loss": 5.0492,
      "step": 1080
    },
    {
      "epoch": 0.22888340595306841,
      "grad_norm": 26.208778381347656,
      "learning_rate": 0.0001082,
      "loss": 5.0394,
      "step": 1090
    },
    {
      "epoch": 0.23098325371410572,
      "grad_norm": 25.2691650390625,
      "learning_rate": 0.00010919999999999998,
      "loss": 4.9898,
      "step": 1100
    },
    {
      "epoch": 0.23308310147514305,
      "grad_norm": 25.28044891357422,
      "learning_rate": 0.0001102,
      "loss": 4.9811,
      "step": 1110
    },
    {
      "epoch": 0.23518294923618038,
      "grad_norm": 25.568035125732422,
      "learning_rate": 0.00011119999999999998,
      "loss": 4.9602,
      "step": 1120
    },
    {
      "epoch": 0.23728279699721772,
      "grad_norm": 26.087724685668945,
      "learning_rate": 0.00011219999999999999,
      "loss": 4.9189,
      "step": 1130
    },
    {
      "epoch": 0.23938264475825502,
      "grad_norm": 28.299169540405273,
      "learning_rate": 0.0001132,
      "loss": 4.9193,
      "step": 1140
    },
    {
      "epoch": 0.24148249251929235,
      "grad_norm": 28.13056755065918,
      "learning_rate": 0.00011419999999999998,
      "loss": 4.8737,
      "step": 1150
    },
    {
      "epoch": 0.24358234028032968,
      "grad_norm": 31.378705978393555,
      "learning_rate": 0.0001152,
      "loss": 4.8834,
      "step": 1160
    },
    {
      "epoch": 0.245682188041367,
      "grad_norm": 25.29513931274414,
      "learning_rate": 0.00011619999999999998,
      "loss": 4.8287,
      "step": 1170
    },
    {
      "epoch": 0.24778203580240432,
      "grad_norm": 25.556110382080078,
      "learning_rate": 0.00011719999999999999,
      "loss": 4.7877,
      "step": 1180
    },
    {
      "epoch": 0.24988188356344165,
      "grad_norm": 24.92612075805664,
      "learning_rate": 0.0001182,
      "loss": 4.7733,
      "step": 1190
    },
    {
      "epoch": 0.251981731324479,
      "grad_norm": 25.60941505432129,
      "learning_rate": 0.00011919999999999998,
      "loss": 4.7554,
      "step": 1200
    },
    {
      "epoch": 0.2540815790855163,
      "grad_norm": 24.840194702148438,
      "learning_rate": 0.0001202,
      "loss": 4.7549,
      "step": 1210
    },
    {
      "epoch": 0.25618142684655365,
      "grad_norm": 25.710140228271484,
      "learning_rate": 0.00012119999999999999,
      "loss": 4.7128,
      "step": 1220
    },
    {
      "epoch": 0.2582812746075909,
      "grad_norm": 27.47252655029297,
      "learning_rate": 0.0001222,
      "loss": 4.6838,
      "step": 1230
    },
    {
      "epoch": 0.26038112236862826,
      "grad_norm": 27.68931007385254,
      "learning_rate": 0.00012319999999999999,
      "loss": 4.6814,
      "step": 1240
    },
    {
      "epoch": 0.2624809701296656,
      "grad_norm": 25.69074249267578,
      "learning_rate": 0.00012419999999999998,
      "loss": 4.636,
      "step": 1250
    },
    {
      "epoch": 0.2645808178907029,
      "grad_norm": 26.92357635498047,
      "learning_rate": 0.00012519999999999998,
      "loss": 4.6422,
      "step": 1260
    },
    {
      "epoch": 0.26668066565174026,
      "grad_norm": 24.645177841186523,
      "learning_rate": 0.0001262,
      "loss": 4.6017,
      "step": 1270
    },
    {
      "epoch": 0.2687805134127776,
      "grad_norm": 23.585927963256836,
      "learning_rate": 0.00012719999999999997,
      "loss": 4.5689,
      "step": 1280
    },
    {
      "epoch": 0.2708803611738149,
      "grad_norm": 26.848289489746094,
      "learning_rate": 0.0001282,
      "loss": 4.5541,
      "step": 1290
    },
    {
      "epoch": 0.2729802089348522,
      "grad_norm": 27.28962516784668,
      "learning_rate": 0.00012919999999999997,
      "loss": 4.553,
      "step": 1300
    },
    {
      "epoch": 0.27508005669588953,
      "grad_norm": 23.665647506713867,
      "learning_rate": 0.0001302,
      "loss": 4.4997,
      "step": 1310
    },
    {
      "epoch": 0.27717990445692686,
      "grad_norm": 24.5865535736084,
      "learning_rate": 0.0001312,
      "loss": 4.5219,
      "step": 1320
    },
    {
      "epoch": 0.2792797522179642,
      "grad_norm": 24.61684226989746,
      "learning_rate": 0.0001322,
      "loss": 4.4479,
      "step": 1330
    },
    {
      "epoch": 0.2813795999790015,
      "grad_norm": 22.41622543334961,
      "learning_rate": 0.00013319999999999999,
      "loss": 4.4379,
      "step": 1340
    },
    {
      "epoch": 0.28347944774003886,
      "grad_norm": 23.269062042236328,
      "learning_rate": 0.0001342,
      "loss": 4.4129,
      "step": 1350
    },
    {
      "epoch": 0.2855792955010762,
      "grad_norm": 24.822797775268555,
      "learning_rate": 0.00013519999999999998,
      "loss": 4.4118,
      "step": 1360
    },
    {
      "epoch": 0.2876791432621135,
      "grad_norm": 27.6806640625,
      "learning_rate": 0.0001362,
      "loss": 4.3816,
      "step": 1370
    },
    {
      "epoch": 0.2897789910231508,
      "grad_norm": 25.281164169311523,
      "learning_rate": 0.00013719999999999997,
      "loss": 4.3618,
      "step": 1380
    },
    {
      "epoch": 0.29187883878418813,
      "grad_norm": 23.890789031982422,
      "learning_rate": 0.0001382,
      "loss": 4.3054,
      "step": 1390
    },
    {
      "epoch": 0.29397868654522546,
      "grad_norm": 21.517894744873047,
      "learning_rate": 0.0001392,
      "loss": 4.3151,
      "step": 1400
    },
    {
      "epoch": 0.2960785343062628,
      "grad_norm": 24.384906768798828,
      "learning_rate": 0.0001402,
      "loss": 4.2983,
      "step": 1410
    },
    {
      "epoch": 0.29817838206730013,
      "grad_norm": 23.395402908325195,
      "learning_rate": 0.0001412,
      "loss": 4.2543,
      "step": 1420
    },
    {
      "epoch": 0.30027822982833746,
      "grad_norm": 24.19387435913086,
      "learning_rate": 0.0001422,
      "loss": 4.2149,
      "step": 1430
    },
    {
      "epoch": 0.3023780775893748,
      "grad_norm": 21.88194465637207,
      "learning_rate": 0.00014319999999999998,
      "loss": 4.1945,
      "step": 1440
    },
    {
      "epoch": 0.30447792535041207,
      "grad_norm": 24.345176696777344,
      "learning_rate": 0.00014419999999999998,
      "loss": 4.2174,
      "step": 1450
    },
    {
      "epoch": 0.3065777731114494,
      "grad_norm": 56.81325149536133,
      "learning_rate": 0.00014519999999999998,
      "loss": 4.1764,
      "step": 1460
    },
    {
      "epoch": 0.30867762087248674,
      "grad_norm": 25.25181770324707,
      "learning_rate": 0.0001462,
      "loss": 4.1153,
      "step": 1470
    },
    {
      "epoch": 0.31077746863352407,
      "grad_norm": 22.449996948242188,
      "learning_rate": 0.00014719999999999997,
      "loss": 4.1213,
      "step": 1480
    },
    {
      "epoch": 0.3128773163945614,
      "grad_norm": 23.437353134155273,
      "learning_rate": 0.0001482,
      "loss": 4.1414,
      "step": 1490
    },
    {
      "epoch": 0.31497716415559873,
      "grad_norm": 23.401479721069336,
      "learning_rate": 0.0001492,
      "loss": 4.0593,
      "step": 1500
    },
    {
      "epoch": 0.31707701191663606,
      "grad_norm": 21.885251998901367,
      "learning_rate": 0.0001502,
      "loss": 4.0705,
      "step": 1510
    },
    {
      "epoch": 0.31917685967767334,
      "grad_norm": 22.176816940307617,
      "learning_rate": 0.0001512,
      "loss": 4.0434,
      "step": 1520
    },
    {
      "epoch": 0.3212767074387107,
      "grad_norm": 21.512222290039062,
      "learning_rate": 0.00015219999999999999,
      "loss": 3.9969,
      "step": 1530
    },
    {
      "epoch": 0.323376555199748,
      "grad_norm": 21.887107849121094,
      "learning_rate": 0.0001532,
      "loss": 3.9678,
      "step": 1540
    },
    {
      "epoch": 0.32547640296078534,
      "grad_norm": 21.871145248413086,
      "learning_rate": 0.00015419999999999998,
      "loss": 3.9544,
      "step": 1550
    },
    {
      "epoch": 0.32757625072182267,
      "grad_norm": 21.275279998779297,
      "learning_rate": 0.00015519999999999998,
      "loss": 3.935,
      "step": 1560
    },
    {
      "epoch": 0.32967609848286,
      "grad_norm": 22.149080276489258,
      "learning_rate": 0.0001562,
      "loss": 3.9628,
      "step": 1570
    },
    {
      "epoch": 0.33177594624389734,
      "grad_norm": 21.588594436645508,
      "learning_rate": 0.0001572,
      "loss": 3.8729,
      "step": 1580
    },
    {
      "epoch": 0.33387579400493467,
      "grad_norm": 22.29717254638672,
      "learning_rate": 0.00015819999999999997,
      "loss": 3.8577,
      "step": 1590
    },
    {
      "epoch": 0.33597564176597194,
      "grad_norm": 21.18522834777832,
      "learning_rate": 0.00015919999999999997,
      "loss": 3.8638,
      "step": 1600
    },
    {
      "epoch": 0.3380754895270093,
      "grad_norm": 21.79401969909668,
      "learning_rate": 0.0001602,
      "loss": 3.7835,
      "step": 1610
    },
    {
      "epoch": 0.3401753372880466,
      "grad_norm": 21.65855598449707,
      "learning_rate": 0.0001612,
      "loss": 3.8225,
      "step": 1620
    },
    {
      "epoch": 0.34227518504908394,
      "grad_norm": 20.30379295349121,
      "learning_rate": 0.00016219999999999999,
      "loss": 3.7654,
      "step": 1630
    },
    {
      "epoch": 0.3443750328101213,
      "grad_norm": 21.457090377807617,
      "learning_rate": 0.0001632,
      "loss": 3.7415,
      "step": 1640
    },
    {
      "epoch": 0.3464748805711586,
      "grad_norm": 20.24045753479004,
      "learning_rate": 0.00016419999999999998,
      "loss": 3.7371,
      "step": 1650
    },
    {
      "epoch": 0.34857472833219594,
      "grad_norm": 20.66719627380371,
      "learning_rate": 0.00016519999999999998,
      "loss": 3.6917,
      "step": 1660
    },
    {
      "epoch": 0.3506745760932332,
      "grad_norm": 21.13929557800293,
      "learning_rate": 0.0001662,
      "loss": 3.7044,
      "step": 1670
    },
    {
      "epoch": 0.35277442385427055,
      "grad_norm": 19.64665412902832,
      "learning_rate": 0.0001672,
      "loss": 3.6972,
      "step": 1680
    },
    {
      "epoch": 0.3548742716153079,
      "grad_norm": 21.007585525512695,
      "learning_rate": 0.00016819999999999997,
      "loss": 3.6806,
      "step": 1690
    },
    {
      "epoch": 0.3569741193763452,
      "grad_norm": 19.849245071411133,
      "learning_rate": 0.00016919999999999997,
      "loss": 3.6044,
      "step": 1700
    },
    {
      "epoch": 0.35907396713738254,
      "grad_norm": 20.355571746826172,
      "learning_rate": 0.0001702,
      "loss": 3.6336,
      "step": 1710
    },
    {
      "epoch": 0.3611738148984199,
      "grad_norm": 19.361953735351562,
      "learning_rate": 0.0001712,
      "loss": 3.5691,
      "step": 1720
    },
    {
      "epoch": 0.3632736626594572,
      "grad_norm": 19.103567123413086,
      "learning_rate": 0.00017219999999999998,
      "loss": 3.5452,
      "step": 1730
    },
    {
      "epoch": 0.36537351042049454,
      "grad_norm": 19.732566833496094,
      "learning_rate": 0.00017319999999999998,
      "loss": 3.516,
      "step": 1740
    },
    {
      "epoch": 0.3674733581815318,
      "grad_norm": 20.030908584594727,
      "learning_rate": 0.00017419999999999998,
      "loss": 3.5212,
      "step": 1750
    },
    {
      "epoch": 0.36957320594256915,
      "grad_norm": 22.157352447509766,
      "learning_rate": 0.00017519999999999998,
      "loss": 3.4927,
      "step": 1760
    },
    {
      "epoch": 0.3716730537036065,
      "grad_norm": 19.714481353759766,
      "learning_rate": 0.0001762,
      "loss": 3.4625,
      "step": 1770
    },
    {
      "epoch": 0.3737729014646438,
      "grad_norm": 20.90492820739746,
      "learning_rate": 0.0001772,
      "loss": 3.4752,
      "step": 1780
    },
    {
      "epoch": 0.37587274922568115,
      "grad_norm": 18.985624313354492,
      "learning_rate": 0.00017819999999999997,
      "loss": 3.4626,
      "step": 1790
    },
    {
      "epoch": 0.3779725969867185,
      "grad_norm": 20.272470474243164,
      "learning_rate": 0.0001792,
      "loss": 3.3595,
      "step": 1800
    },
    {
      "epoch": 0.3800724447477558,
      "grad_norm": 20.29452896118164,
      "learning_rate": 0.0001802,
      "loss": 3.3535,
      "step": 1810
    },
    {
      "epoch": 0.3821722925087931,
      "grad_norm": 19.74087142944336,
      "learning_rate": 0.00018119999999999999,
      "loss": 3.3843,
      "step": 1820
    },
    {
      "epoch": 0.3842721402698304,
      "grad_norm": 20.133838653564453,
      "learning_rate": 0.00018219999999999996,
      "loss": 3.3159,
      "step": 1830
    },
    {
      "epoch": 0.38637198803086775,
      "grad_norm": 21.46741485595703,
      "learning_rate": 0.00018319999999999998,
      "loss": 3.3751,
      "step": 1840
    },
    {
      "epoch": 0.3884718357919051,
      "grad_norm": 20.637907028198242,
      "learning_rate": 0.00018419999999999998,
      "loss": 3.313,
      "step": 1850
    },
    {
      "epoch": 0.3905716835529424,
      "grad_norm": 18.241500854492188,
      "learning_rate": 0.00018519999999999998,
      "loss": 3.2968,
      "step": 1860
    },
    {
      "epoch": 0.39267153131397975,
      "grad_norm": 18.708087921142578,
      "learning_rate": 0.0001862,
      "loss": 3.2791,
      "step": 1870
    },
    {
      "epoch": 0.3947713790750171,
      "grad_norm": 20.34215545654297,
      "learning_rate": 0.0001872,
      "loss": 3.2397,
      "step": 1880
    },
    {
      "epoch": 0.3968712268360544,
      "grad_norm": 19.339441299438477,
      "learning_rate": 0.00018819999999999997,
      "loss": 3.2633,
      "step": 1890
    },
    {
      "epoch": 0.3989710745970917,
      "grad_norm": 18.52556800842285,
      "learning_rate": 0.0001892,
      "loss": 3.2064,
      "step": 1900
    },
    {
      "epoch": 0.401070922358129,
      "grad_norm": 19.54344940185547,
      "learning_rate": 0.0001902,
      "loss": 3.1975,
      "step": 1910
    },
    {
      "epoch": 0.40317077011916636,
      "grad_norm": 18.180559158325195,
      "learning_rate": 0.00019119999999999999,
      "loss": 3.2184,
      "step": 1920
    },
    {
      "epoch": 0.4052706178802037,
      "grad_norm": 18.006343841552734,
      "learning_rate": 0.0001922,
      "loss": 3.2137,
      "step": 1930
    },
    {
      "epoch": 0.407370465641241,
      "grad_norm": 18.950502395629883,
      "learning_rate": 0.00019319999999999998,
      "loss": 3.1627,
      "step": 1940
    },
    {
      "epoch": 0.40947031340227835,
      "grad_norm": 19.233171463012695,
      "learning_rate": 0.00019419999999999998,
      "loss": 3.1599,
      "step": 1950
    },
    {
      "epoch": 0.4115701611633157,
      "grad_norm": 20.070098876953125,
      "learning_rate": 0.00019519999999999997,
      "loss": 3.1319,
      "step": 1960
    },
    {
      "epoch": 0.41367000892435296,
      "grad_norm": 17.84896469116211,
      "learning_rate": 0.0001962,
      "loss": 3.0938,
      "step": 1970
    },
    {
      "epoch": 0.4157698566853903,
      "grad_norm": 21.04241943359375,
      "learning_rate": 0.00019719999999999997,
      "loss": 3.1118,
      "step": 1980
    },
    {
      "epoch": 0.4178697044464276,
      "grad_norm": 17.594589233398438,
      "learning_rate": 0.00019819999999999997,
      "loss": 3.0804,
      "step": 1990
    },
    {
      "epoch": 0.41996955220746496,
      "grad_norm": 18.5731258392334,
      "learning_rate": 0.0001992,
      "loss": 3.0664,
      "step": 2000
    },
    {
      "epoch": 0.4220693999685023,
      "grad_norm": 18.234962463378906,
      "learning_rate": 0.0002002,
      "loss": 3.0466,
      "step": 2010
    },
    {
      "epoch": 0.4241692477295396,
      "grad_norm": 18.205047607421875,
      "learning_rate": 0.00020119999999999998,
      "loss": 3.0454,
      "step": 2020
    },
    {
      "epoch": 0.42626909549057695,
      "grad_norm": 18.562522888183594,
      "learning_rate": 0.0002022,
      "loss": 3.0035,
      "step": 2030
    },
    {
      "epoch": 0.42836894325161423,
      "grad_norm": 19.143112182617188,
      "learning_rate": 0.00020319999999999998,
      "loss": 3.0265,
      "step": 2040
    },
    {
      "epoch": 0.43046879101265156,
      "grad_norm": 18.834646224975586,
      "learning_rate": 0.00020419999999999998,
      "loss": 2.9659,
      "step": 2050
    },
    {
      "epoch": 0.4325686387736889,
      "grad_norm": 18.57697868347168,
      "learning_rate": 0.0002052,
      "loss": 2.9717,
      "step": 2060
    },
    {
      "epoch": 0.43466848653472623,
      "grad_norm": 18.24930763244629,
      "learning_rate": 0.0002062,
      "loss": 2.9772,
      "step": 2070
    },
    {
      "epoch": 0.43676833429576356,
      "grad_norm": 19.47739601135254,
      "learning_rate": 0.00020719999999999997,
      "loss": 2.9644,
      "step": 2080
    },
    {
      "epoch": 0.4388681820568009,
      "grad_norm": 20.972169876098633,
      "learning_rate": 0.00020819999999999996,
      "loss": 2.9397,
      "step": 2090
    },
    {
      "epoch": 0.4409680298178382,
      "grad_norm": 18.49868392944336,
      "learning_rate": 0.0002092,
      "loss": 2.9347,
      "step": 2100
    },
    {
      "epoch": 0.44306787757887556,
      "grad_norm": 17.23448371887207,
      "learning_rate": 0.0002102,
      "loss": 2.902,
      "step": 2110
    },
    {
      "epoch": 0.44516772533991283,
      "grad_norm": 18.497655868530273,
      "learning_rate": 0.00021119999999999996,
      "loss": 2.8292,
      "step": 2120
    },
    {
      "epoch": 0.44726757310095017,
      "grad_norm": 19.482572555541992,
      "learning_rate": 0.00021219999999999998,
      "loss": 2.8823,
      "step": 2130
    },
    {
      "epoch": 0.4493674208619875,
      "grad_norm": 19.21995735168457,
      "learning_rate": 0.00021319999999999998,
      "loss": 2.8764,
      "step": 2140
    },
    {
      "epoch": 0.45146726862302483,
      "grad_norm": 54.40972137451172,
      "learning_rate": 0.00021419999999999998,
      "loss": 2.8654,
      "step": 2150
    },
    {
      "epoch": 0.45356711638406216,
      "grad_norm": 17.572521209716797,
      "learning_rate": 0.0002152,
      "loss": 2.8347,
      "step": 2160
    },
    {
      "epoch": 0.4556669641450995,
      "grad_norm": 17.291915893554688,
      "learning_rate": 0.0002162,
      "loss": 2.8724,
      "step": 2170
    },
    {
      "epoch": 0.45776681190613683,
      "grad_norm": 17.69890785217285,
      "learning_rate": 0.00021719999999999997,
      "loss": 2.8069,
      "step": 2180
    },
    {
      "epoch": 0.4598666596671741,
      "grad_norm": 17.662452697753906,
      "learning_rate": 0.0002182,
      "loss": 2.8119,
      "step": 2190
    },
    {
      "epoch": 0.46196650742821144,
      "grad_norm": 16.374813079833984,
      "learning_rate": 0.0002192,
      "loss": 2.7883,
      "step": 2200
    },
    {
      "epoch": 0.46406635518924877,
      "grad_norm": 17.873462677001953,
      "learning_rate": 0.00022019999999999999,
      "loss": 2.7752,
      "step": 2210
    },
    {
      "epoch": 0.4661662029502861,
      "grad_norm": 18.1060733795166,
      "learning_rate": 0.00022119999999999996,
      "loss": 2.7642,
      "step": 2220
    },
    {
      "epoch": 0.46826605071132343,
      "grad_norm": 17.768138885498047,
      "learning_rate": 0.00022219999999999998,
      "loss": 2.7926,
      "step": 2230
    },
    {
      "epoch": 0.47036589847236077,
      "grad_norm": 18.01146125793457,
      "learning_rate": 0.00022319999999999998,
      "loss": 2.8,
      "step": 2240
    },
    {
      "epoch": 0.4724657462333981,
      "grad_norm": 19.251340866088867,
      "learning_rate": 0.00022419999999999997,
      "loss": 2.7525,
      "step": 2250
    },
    {
      "epoch": 0.47456559399443543,
      "grad_norm": 18.650836944580078,
      "learning_rate": 0.0002252,
      "loss": 2.7113,
      "step": 2260
    },
    {
      "epoch": 0.4766654417554727,
      "grad_norm": 17.6818904876709,
      "learning_rate": 0.00022619999999999997,
      "loss": 2.7266,
      "step": 2270
    },
    {
      "epoch": 0.47876528951651004,
      "grad_norm": 17.18866539001465,
      "learning_rate": 0.00022719999999999997,
      "loss": 2.7273,
      "step": 2280
    },
    {
      "epoch": 0.48086513727754737,
      "grad_norm": 16.871488571166992,
      "learning_rate": 0.0002282,
      "loss": 2.7053,
      "step": 2290
    },
    {
      "epoch": 0.4829649850385847,
      "grad_norm": 17.364320755004883,
      "learning_rate": 0.0002292,
      "loss": 2.6935,
      "step": 2300
    },
    {
      "epoch": 0.48506483279962204,
      "grad_norm": 16.227325439453125,
      "learning_rate": 0.00023019999999999998,
      "loss": 2.6832,
      "step": 2310
    },
    {
      "epoch": 0.48716468056065937,
      "grad_norm": 18.579835891723633,
      "learning_rate": 0.0002312,
      "loss": 2.6319,
      "step": 2320
    },
    {
      "epoch": 0.4892645283216967,
      "grad_norm": 17.750078201293945,
      "learning_rate": 0.00023219999999999998,
      "loss": 2.6687,
      "step": 2330
    },
    {
      "epoch": 0.491364376082734,
      "grad_norm": 16.852567672729492,
      "learning_rate": 0.00023319999999999998,
      "loss": 2.6348,
      "step": 2340
    },
    {
      "epoch": 0.4934642238437713,
      "grad_norm": 16.808338165283203,
      "learning_rate": 0.00023419999999999997,
      "loss": 2.6376,
      "step": 2350
    },
    {
      "epoch": 0.49556407160480864,
      "grad_norm": 16.15403175354004,
      "learning_rate": 0.0002352,
      "loss": 2.6347,
      "step": 2360
    },
    {
      "epoch": 0.497663919365846,
      "grad_norm": 16.643596649169922,
      "learning_rate": 0.00023619999999999997,
      "loss": 2.6669,
      "step": 2370
    },
    {
      "epoch": 0.4997637671268833,
      "grad_norm": 17.631317138671875,
      "learning_rate": 0.00023719999999999997,
      "loss": 2.6264,
      "step": 2380
    },
    {
      "epoch": 0.5018636148879206,
      "grad_norm": 17.213918685913086,
      "learning_rate": 0.0002382,
      "loss": 2.5763,
      "step": 2390
    },
    {
      "epoch": 0.503963462648958,
      "grad_norm": 17.565031051635742,
      "learning_rate": 0.0002392,
      "loss": 2.5686,
      "step": 2400
    },
    {
      "epoch": 0.5060633104099953,
      "grad_norm": 17.955644607543945,
      "learning_rate": 0.00024019999999999996,
      "loss": 2.5984,
      "step": 2410
    },
    {
      "epoch": 0.5081631581710326,
      "grad_norm": 17.599609375,
      "learning_rate": 0.00024119999999999998,
      "loss": 2.5992,
      "step": 2420
    },
    {
      "epoch": 0.51026300593207,
      "grad_norm": 18.456865310668945,
      "learning_rate": 0.00024219999999999998,
      "loss": 2.6249,
      "step": 2430
    },
    {
      "epoch": 0.5123628536931073,
      "grad_norm": 18.269716262817383,
      "learning_rate": 0.00024319999999999998,
      "loss": 2.5443,
      "step": 2440
    },
    {
      "epoch": 0.5144627014541445,
      "grad_norm": 15.301117897033691,
      "learning_rate": 0.00024419999999999997,
      "loss": 2.6064,
      "step": 2450
    },
    {
      "epoch": 0.5165625492151819,
      "grad_norm": 16.484882354736328,
      "learning_rate": 0.0002452,
      "loss": 2.5898,
      "step": 2460
    },
    {
      "epoch": 0.5186623969762192,
      "grad_norm": 18.93348503112793,
      "learning_rate": 0.00024619999999999997,
      "loss": 2.5501,
      "step": 2470
    },
    {
      "epoch": 0.5207622447372565,
      "grad_norm": 17.024005889892578,
      "learning_rate": 0.0002472,
      "loss": 2.5847,
      "step": 2480
    },
    {
      "epoch": 0.5228620924982939,
      "grad_norm": 15.91640567779541,
      "learning_rate": 0.00024819999999999996,
      "loss": 2.5018,
      "step": 2490
    },
    {
      "epoch": 0.5249619402593312,
      "grad_norm": 16.563140869140625,
      "learning_rate": 0.0002492,
      "loss": 2.5097,
      "step": 2500
    }
  ],
  "logging_steps": 10,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 21,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.628136559017984e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}