{
  "best_global_step": 2000,
  "best_metric": 0.8704838022863507,
  "best_model_checkpoint": "./SALAMA_NEWMED9/checkpoint-2000",
  "epoch": 1.1983520599250936,
  "eval_steps": 2000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00599250936329588,
      "grad_norm": 3.4430904388427734,
      "learning_rate": 1.8e-07,
      "loss": 0.0305,
      "step": 10
    },
    {
      "epoch": 0.01198501872659176,
      "grad_norm": 1.3001453876495361,
      "learning_rate": 3.8e-07,
      "loss": 0.0217,
      "step": 20
    },
    {
      "epoch": 0.017977528089887642,
      "grad_norm": 2.323173999786377,
      "learning_rate": 5.800000000000001e-07,
      "loss": 0.0259,
      "step": 30
    },
    {
      "epoch": 0.02397003745318352,
      "grad_norm": 1.480966329574585,
      "learning_rate": 7.8e-07,
      "loss": 0.0274,
      "step": 40
    },
    {
      "epoch": 0.0299625468164794,
      "grad_norm": 0.8444674015045166,
      "learning_rate": 9.800000000000001e-07,
      "loss": 0.0204,
      "step": 50
    },
    {
      "epoch": 0.035955056179775284,
      "grad_norm": 4.492954254150391,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 0.0362,
      "step": 60
    },
    {
      "epoch": 0.041947565543071164,
      "grad_norm": 0.8774360418319702,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.0283,
      "step": 70
    },
    {
      "epoch": 0.04794007490636704,
      "grad_norm": 1.3773318529129028,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 0.0255,
      "step": 80
    },
    {
      "epoch": 0.05393258426966292,
      "grad_norm": 1.632542610168457,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.0186,
      "step": 90
    },
    {
      "epoch": 0.0599250936329588,
      "grad_norm": 2.4346158504486084,
      "learning_rate": 1.98e-06,
      "loss": 0.0311,
      "step": 100
    },
    {
      "epoch": 0.06591760299625468,
      "grad_norm": 1.4394844770431519,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.0279,
      "step": 110
    },
    {
      "epoch": 0.07191011235955057,
      "grad_norm": 2.2031307220458984,
      "learning_rate": 2.38e-06,
      "loss": 0.0223,
      "step": 120
    },
    {
      "epoch": 0.07790262172284644,
      "grad_norm": 1.1016514301300049,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.0159,
      "step": 130
    },
    {
      "epoch": 0.08389513108614233,
      "grad_norm": 2.271472454071045,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 0.0187,
      "step": 140
    },
    {
      "epoch": 0.0898876404494382,
      "grad_norm": 1.3717912435531616,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.0191,
      "step": 150
    },
    {
      "epoch": 0.09588014981273409,
      "grad_norm": 2.001311779022217,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.0245,
      "step": 160
    },
    {
      "epoch": 0.10187265917602996,
      "grad_norm": 2.1671388149261475,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 0.0207,
      "step": 170
    },
    {
      "epoch": 0.10786516853932585,
      "grad_norm": 2.7302298545837402,
      "learning_rate": 3.58e-06,
      "loss": 0.0251,
      "step": 180
    },
    {
      "epoch": 0.11385767790262172,
      "grad_norm": 2.360443115234375,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.0279,
      "step": 190
    },
    {
      "epoch": 0.1198501872659176,
      "grad_norm": 2.1473894119262695,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0166,
      "step": 200
    },
    {
      "epoch": 0.1258426966292135,
      "grad_norm": 2.1437275409698486,
      "learning_rate": 4.18e-06,
      "loss": 0.0296,
      "step": 210
    },
    {
      "epoch": 0.13183520599250936,
      "grad_norm": 1.155143141746521,
      "learning_rate": 4.38e-06,
      "loss": 0.0235,
      "step": 220
    },
    {
      "epoch": 0.13782771535580524,
      "grad_norm": 1.398034691810608,
      "learning_rate": 4.58e-06,
      "loss": 0.0256,
      "step": 230
    },
    {
      "epoch": 0.14382022471910114,
      "grad_norm": 1.704981803894043,
      "learning_rate": 4.78e-06,
      "loss": 0.0314,
      "step": 240
    },
    {
      "epoch": 0.149812734082397,
      "grad_norm": 2.0833070278167725,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.0292,
      "step": 250
    },
    {
      "epoch": 0.15580524344569288,
      "grad_norm": 1.6899584531784058,
      "learning_rate": 5.18e-06,
      "loss": 0.0226,
      "step": 260
    },
    {
      "epoch": 0.16179775280898875,
      "grad_norm": 1.5703374147415161,
      "learning_rate": 5.380000000000001e-06,
      "loss": 0.0251,
      "step": 270
    },
    {
      "epoch": 0.16779026217228465,
      "grad_norm": 0.9077961444854736,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.0193,
      "step": 280
    },
    {
      "epoch": 0.17378277153558053,
      "grad_norm": 2.0035812854766846,
      "learning_rate": 5.78e-06,
      "loss": 0.0231,
      "step": 290
    },
    {
      "epoch": 0.1797752808988764,
      "grad_norm": 2.8311734199523926,
      "learning_rate": 5.98e-06,
      "loss": 0.0326,
      "step": 300
    },
    {
      "epoch": 0.18576779026217227,
      "grad_norm": 1.6007928848266602,
      "learning_rate": 6.18e-06,
      "loss": 0.0231,
      "step": 310
    },
    {
      "epoch": 0.19176029962546817,
      "grad_norm": 1.3683286905288696,
      "learning_rate": 6.380000000000001e-06,
      "loss": 0.026,
      "step": 320
    },
    {
      "epoch": 0.19775280898876405,
      "grad_norm": 2.6587653160095215,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 0.0288,
      "step": 330
    },
    {
      "epoch": 0.20374531835205992,
      "grad_norm": 2.617583751678467,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.0297,
      "step": 340
    },
    {
      "epoch": 0.20973782771535582,
      "grad_norm": 1.4291918277740479,
      "learning_rate": 6.98e-06,
      "loss": 0.0172,
      "step": 350
    },
    {
      "epoch": 0.2157303370786517,
      "grad_norm": 1.9586079120635986,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.0413,
      "step": 360
    },
    {
      "epoch": 0.22172284644194756,
      "grad_norm": 2.081557273864746,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.0297,
      "step": 370
    },
    {
      "epoch": 0.22771535580524344,
      "grad_norm": 1.5032098293304443,
      "learning_rate": 7.58e-06,
      "loss": 0.027,
      "step": 380
    },
    {
      "epoch": 0.23370786516853934,
      "grad_norm": 2.316767454147339,
      "learning_rate": 7.78e-06,
      "loss": 0.0291,
      "step": 390
    },
    {
      "epoch": 0.2397003745318352,
      "grad_norm": 3.565624475479126,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0335,
      "step": 400
    },
    {
      "epoch": 0.24569288389513108,
      "grad_norm": 1.4883167743682861,
      "learning_rate": 8.18e-06,
      "loss": 0.0254,
      "step": 410
    },
    {
      "epoch": 0.251685393258427,
      "grad_norm": 2.2969090938568115,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.0401,
      "step": 420
    },
    {
      "epoch": 0.25767790262172285,
      "grad_norm": 1.9346531629562378,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.0254,
      "step": 430
    },
    {
      "epoch": 0.2636704119850187,
      "grad_norm": 0.6965954899787903,
      "learning_rate": 8.78e-06,
      "loss": 0.0248,
      "step": 440
    },
    {
      "epoch": 0.2696629213483146,
      "grad_norm": 3.862004041671753,
      "learning_rate": 8.98e-06,
      "loss": 0.0392,
      "step": 450
    },
    {
      "epoch": 0.27565543071161047,
      "grad_norm": 1.418031930923462,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.0364,
      "step": 460
    },
    {
      "epoch": 0.28164794007490634,
      "grad_norm": 1.2769050598144531,
      "learning_rate": 9.38e-06,
      "loss": 0.0239,
      "step": 470
    },
    {
      "epoch": 0.2876404494382023,
      "grad_norm": 2.03157901763916,
      "learning_rate": 9.58e-06,
      "loss": 0.0302,
      "step": 480
    },
    {
      "epoch": 0.29363295880149815,
      "grad_norm": 2.0723016262054443,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.0329,
      "step": 490
    },
    {
      "epoch": 0.299625468164794,
      "grad_norm": 1.585113763809204,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0268,
      "step": 500
    },
    {
      "epoch": 0.3056179775280899,
      "grad_norm": 2.0149030685424805,
      "learning_rate": 9.968287526427062e-06,
      "loss": 0.0321,
      "step": 510
    },
    {
      "epoch": 0.31161048689138576,
      "grad_norm": 1.5547727346420288,
      "learning_rate": 9.933051444679352e-06,
      "loss": 0.0298,
      "step": 520
    },
    {
      "epoch": 0.31760299625468164,
      "grad_norm": 2.3159093856811523,
      "learning_rate": 9.897815362931644e-06,
      "loss": 0.0328,
      "step": 530
    },
    {
      "epoch": 0.3235955056179775,
      "grad_norm": 2.6754517555236816,
      "learning_rate": 9.862579281183932e-06,
      "loss": 0.0437,
      "step": 540
    },
    {
      "epoch": 0.3295880149812734,
      "grad_norm": 1.470944881439209,
      "learning_rate": 9.827343199436224e-06,
      "loss": 0.0324,
      "step": 550
    },
    {
      "epoch": 0.3355805243445693,
      "grad_norm": 1.461867094039917,
      "learning_rate": 9.792107117688514e-06,
      "loss": 0.0391,
      "step": 560
    },
    {
      "epoch": 0.3415730337078652,
      "grad_norm": 2.258100986480713,
      "learning_rate": 9.756871035940804e-06,
      "loss": 0.0382,
      "step": 570
    },
    {
      "epoch": 0.34756554307116105,
      "grad_norm": 2.2667694091796875,
      "learning_rate": 9.721634954193094e-06,
      "loss": 0.0393,
      "step": 580
    },
    {
      "epoch": 0.3535580524344569,
      "grad_norm": 2.1521716117858887,
      "learning_rate": 9.686398872445385e-06,
      "loss": 0.0242,
      "step": 590
    },
    {
      "epoch": 0.3595505617977528,
      "grad_norm": 2.3627607822418213,
      "learning_rate": 9.651162790697676e-06,
      "loss": 0.0303,
      "step": 600
    },
    {
      "epoch": 0.36554307116104867,
      "grad_norm": 2.5504517555236816,
      "learning_rate": 9.615926708949965e-06,
      "loss": 0.0384,
      "step": 610
    },
    {
      "epoch": 0.37153558052434454,
      "grad_norm": 2.1967527866363525,
      "learning_rate": 9.580690627202257e-06,
      "loss": 0.033,
      "step": 620
    },
    {
      "epoch": 0.3775280898876405,
      "grad_norm": 1.6752692461013794,
      "learning_rate": 9.545454545454547e-06,
      "loss": 0.0246,
      "step": 630
    },
    {
      "epoch": 0.38352059925093634,
      "grad_norm": 1.6692609786987305,
      "learning_rate": 9.510218463706837e-06,
      "loss": 0.0339,
      "step": 640
    },
    {
      "epoch": 0.3895131086142322,
      "grad_norm": 2.7334742546081543,
      "learning_rate": 9.474982381959127e-06,
      "loss": 0.0276,
      "step": 650
    },
    {
      "epoch": 0.3955056179775281,
      "grad_norm": 1.5142534971237183,
      "learning_rate": 9.439746300211417e-06,
      "loss": 0.0347,
      "step": 660
    },
    {
      "epoch": 0.40149812734082396,
      "grad_norm": 2.5328900814056396,
      "learning_rate": 9.404510218463707e-06,
      "loss": 0.0295,
      "step": 670
    },
    {
      "epoch": 0.40749063670411984,
      "grad_norm": 3.101393938064575,
      "learning_rate": 9.369274136715997e-06,
      "loss": 0.0409,
      "step": 680
    },
    {
      "epoch": 0.4134831460674157,
      "grad_norm": 2.7148630619049072,
      "learning_rate": 9.33403805496829e-06,
      "loss": 0.0311,
      "step": 690
    },
    {
      "epoch": 0.41947565543071164,
      "grad_norm": 1.6952073574066162,
      "learning_rate": 9.298801973220578e-06,
      "loss": 0.0341,
      "step": 700
    },
    {
      "epoch": 0.4254681647940075,
      "grad_norm": 2.545931577682495,
      "learning_rate": 9.26356589147287e-06,
      "loss": 0.036,
      "step": 710
    },
    {
      "epoch": 0.4314606741573034,
      "grad_norm": 2.246628999710083,
      "learning_rate": 9.22832980972516e-06,
      "loss": 0.0424,
      "step": 720
    },
    {
      "epoch": 0.43745318352059925,
      "grad_norm": 1.6698744297027588,
      "learning_rate": 9.19309372797745e-06,
      "loss": 0.0416,
      "step": 730
    },
    {
      "epoch": 0.4434456928838951,
      "grad_norm": 1.2881724834442139,
      "learning_rate": 9.15785764622974e-06,
      "loss": 0.0366,
      "step": 740
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 1.7771434783935547,
      "learning_rate": 9.12262156448203e-06,
      "loss": 0.0309,
      "step": 750
    },
    {
      "epoch": 0.45543071161048687,
      "grad_norm": 1.563678503036499,
      "learning_rate": 9.087385482734322e-06,
      "loss": 0.0376,
      "step": 760
    },
    {
      "epoch": 0.46142322097378274,
      "grad_norm": 2.24344801902771,
      "learning_rate": 9.05214940098661e-06,
      "loss": 0.0332,
      "step": 770
    },
    {
      "epoch": 0.46741573033707867,
      "grad_norm": 3.014191150665283,
      "learning_rate": 9.016913319238902e-06,
      "loss": 0.0288,
      "step": 780
    },
    {
      "epoch": 0.47340823970037454,
      "grad_norm": 2.138899564743042,
      "learning_rate": 8.981677237491192e-06,
      "loss": 0.0332,
      "step": 790
    },
    {
      "epoch": 0.4794007490636704,
      "grad_norm": 2.4274239540100098,
      "learning_rate": 8.946441155743482e-06,
      "loss": 0.0426,
      "step": 800
    },
    {
      "epoch": 0.4853932584269663,
      "grad_norm": 1.720563292503357,
      "learning_rate": 8.911205073995772e-06,
      "loss": 0.0322,
      "step": 810
    },
    {
      "epoch": 0.49138576779026216,
      "grad_norm": 2.580158233642578,
      "learning_rate": 8.875968992248062e-06,
      "loss": 0.0337,
      "step": 820
    },
    {
      "epoch": 0.49737827715355803,
      "grad_norm": 1.918359398841858,
      "learning_rate": 8.840732910500353e-06,
      "loss": 0.0276,
      "step": 830
    },
    {
      "epoch": 0.503370786516854,
      "grad_norm": 3.0238075256347656,
      "learning_rate": 8.805496828752643e-06,
      "loss": 0.0363,
      "step": 840
    },
    {
      "epoch": 0.5093632958801498,
      "grad_norm": 1.9104974269866943,
      "learning_rate": 8.770260747004935e-06,
      "loss": 0.0372,
      "step": 850
    },
    {
      "epoch": 0.5153558052434457,
      "grad_norm": 4.188656806945801,
      "learning_rate": 8.735024665257225e-06,
      "loss": 0.0374,
      "step": 860
    },
    {
      "epoch": 0.5213483146067416,
      "grad_norm": 2.6487040519714355,
      "learning_rate": 8.699788583509515e-06,
      "loss": 0.0305,
      "step": 870
    },
    {
      "epoch": 0.5273408239700375,
      "grad_norm": 2.9171149730682373,
      "learning_rate": 8.664552501761805e-06,
      "loss": 0.0321,
      "step": 880
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 2.196499824523926,
      "learning_rate": 8.629316420014095e-06,
      "loss": 0.0497,
      "step": 890
    },
    {
      "epoch": 0.5393258426966292,
      "grad_norm": 2.324110507965088,
      "learning_rate": 8.594080338266385e-06,
      "loss": 0.0411,
      "step": 900
    },
    {
      "epoch": 0.5453183520599251,
      "grad_norm": 2.23406720161438,
      "learning_rate": 8.558844256518675e-06,
      "loss": 0.0302,
      "step": 910
    },
    {
      "epoch": 0.5513108614232209,
      "grad_norm": 2.158843994140625,
      "learning_rate": 8.523608174770967e-06,
      "loss": 0.0349,
      "step": 920
    },
    {
      "epoch": 0.5573033707865168,
      "grad_norm": 2.5448203086853027,
      "learning_rate": 8.488372093023256e-06,
      "loss": 0.0338,
      "step": 930
    },
    {
      "epoch": 0.5632958801498127,
      "grad_norm": 2.366697072982788,
      "learning_rate": 8.453136011275547e-06,
      "loss": 0.0307,
      "step": 940
    },
    {
      "epoch": 0.5692883895131086,
      "grad_norm": 1.7669707536697388,
      "learning_rate": 8.417899929527837e-06,
      "loss": 0.0374,
      "step": 950
    },
    {
      "epoch": 0.5752808988764045,
      "grad_norm": 1.8498706817626953,
      "learning_rate": 8.382663847780128e-06,
      "loss": 0.0368,
      "step": 960
    },
    {
      "epoch": 0.5812734082397004,
      "grad_norm": 1.1811097860336304,
      "learning_rate": 8.347427766032418e-06,
      "loss": 0.0319,
      "step": 970
    },
    {
      "epoch": 0.5872659176029963,
      "grad_norm": 1.306626796722412,
      "learning_rate": 8.312191684284708e-06,
      "loss": 0.0409,
      "step": 980
    },
    {
      "epoch": 0.5932584269662922,
      "grad_norm": 1.9113121032714844,
      "learning_rate": 8.276955602537e-06,
      "loss": 0.0329,
      "step": 990
    },
    {
      "epoch": 0.599250936329588,
      "grad_norm": 1.8065577745437622,
      "learning_rate": 8.241719520789288e-06,
      "loss": 0.0261,
      "step": 1000
    },
    {
      "epoch": 0.6052434456928839,
      "grad_norm": 1.644216775894165,
      "learning_rate": 8.20648343904158e-06,
      "loss": 0.0455,
      "step": 1010
    },
    {
      "epoch": 0.6112359550561798,
      "grad_norm": 2.524791955947876,
      "learning_rate": 8.17124735729387e-06,
      "loss": 0.0378,
      "step": 1020
    },
    {
      "epoch": 0.6172284644194757,
      "grad_norm": 4.8924880027771,
      "learning_rate": 8.13601127554616e-06,
      "loss": 0.0458,
      "step": 1030
    },
    {
      "epoch": 0.6232209737827715,
      "grad_norm": 2.6112005710601807,
      "learning_rate": 8.10077519379845e-06,
      "loss": 0.0388,
      "step": 1040
    },
    {
      "epoch": 0.6292134831460674,
      "grad_norm": 2.2493460178375244,
      "learning_rate": 8.06553911205074e-06,
      "loss": 0.0451,
      "step": 1050
    },
    {
      "epoch": 0.6352059925093633,
      "grad_norm": 1.854641079902649,
      "learning_rate": 8.03030303030303e-06,
      "loss": 0.0434,
      "step": 1060
    },
    {
      "epoch": 0.6411985018726591,
      "grad_norm": 2.749845027923584,
      "learning_rate": 7.99506694855532e-06,
      "loss": 0.0291,
      "step": 1070
    },
    {
      "epoch": 0.647191011235955,
      "grad_norm": 3.08630108833313,
      "learning_rate": 7.959830866807612e-06,
      "loss": 0.0256,
      "step": 1080
    },
    {
      "epoch": 0.6531835205992509,
      "grad_norm": 1.4188586473464966,
      "learning_rate": 7.924594785059903e-06,
      "loss": 0.0363,
      "step": 1090
    },
    {
      "epoch": 0.6591760299625468,
      "grad_norm": 1.34322988986969,
      "learning_rate": 7.889358703312193e-06,
      "loss": 0.0313,
      "step": 1100
    },
    {
      "epoch": 0.6651685393258427,
      "grad_norm": 1.7974929809570312,
      "learning_rate": 7.854122621564483e-06,
      "loss": 0.0245,
      "step": 1110
    },
    {
      "epoch": 0.6711610486891386,
      "grad_norm": 2.5316739082336426,
      "learning_rate": 7.818886539816773e-06,
      "loss": 0.0355,
      "step": 1120
    },
    {
      "epoch": 0.6771535580524345,
      "grad_norm": 3.0023632049560547,
      "learning_rate": 7.783650458069063e-06,
      "loss": 0.0344,
      "step": 1130
    },
    {
      "epoch": 0.6831460674157304,
      "grad_norm": 2.484605550765991,
      "learning_rate": 7.748414376321353e-06,
      "loss": 0.0442,
      "step": 1140
    },
    {
      "epoch": 0.6891385767790262,
      "grad_norm": 2.7692389488220215,
      "learning_rate": 7.713178294573645e-06,
      "loss": 0.0389,
      "step": 1150
    },
    {
      "epoch": 0.6951310861423221,
      "grad_norm": 1.1416544914245605,
      "learning_rate": 7.677942212825933e-06,
      "loss": 0.0357,
      "step": 1160
    },
    {
      "epoch": 0.701123595505618,
      "grad_norm": 2.1747801303863525,
      "learning_rate": 7.642706131078225e-06,
      "loss": 0.0329,
      "step": 1170
    },
    {
      "epoch": 0.7071161048689139,
      "grad_norm": 2.4001238346099854,
      "learning_rate": 7.607470049330515e-06,
      "loss": 0.0327,
      "step": 1180
    },
    {
      "epoch": 0.7131086142322097,
      "grad_norm": 2.489826202392578,
      "learning_rate": 7.572233967582805e-06,
      "loss": 0.038,
      "step": 1190
    },
    {
      "epoch": 0.7191011235955056,
      "grad_norm": 2.055882453918457,
      "learning_rate": 7.5369978858350956e-06,
      "loss": 0.0268,
      "step": 1200
    },
    {
      "epoch": 0.7250936329588015,
      "grad_norm": 1.705216646194458,
      "learning_rate": 7.5017618040873865e-06,
      "loss": 0.0258,
      "step": 1210
    },
    {
      "epoch": 0.7310861423220973,
      "grad_norm": 1.2729088068008423,
      "learning_rate": 7.466525722339677e-06,
      "loss": 0.0352,
      "step": 1220
    },
    {
      "epoch": 0.7370786516853932,
      "grad_norm": 4.454958438873291,
      "learning_rate": 7.431289640591967e-06,
      "loss": 0.0343,
      "step": 1230
    },
    {
      "epoch": 0.7430711610486891,
      "grad_norm": 1.8376761674880981,
      "learning_rate": 7.396053558844257e-06,
      "loss": 0.0337,
      "step": 1240
    },
    {
      "epoch": 0.7490636704119851,
      "grad_norm": 2.3845252990722656,
      "learning_rate": 7.360817477096548e-06,
      "loss": 0.0406,
      "step": 1250
    },
    {
      "epoch": 0.755056179775281,
      "grad_norm": 6.822844982147217,
      "learning_rate": 7.325581395348837e-06,
      "loss": 0.0417,
      "step": 1260
    },
    {
      "epoch": 0.7610486891385768,
      "grad_norm": 7.371566295623779,
      "learning_rate": 7.290345313601128e-06,
      "loss": 0.0271,
      "step": 1270
    },
    {
      "epoch": 0.7670411985018727,
      "grad_norm": 1.3738433122634888,
      "learning_rate": 7.255109231853419e-06,
      "loss": 0.0296,
      "step": 1280
    },
    {
      "epoch": 0.7730337078651686,
      "grad_norm": 1.8998222351074219,
      "learning_rate": 7.219873150105708e-06,
      "loss": 0.0317,
      "step": 1290
    },
    {
      "epoch": 0.7790262172284644,
      "grad_norm": 2.569655418395996,
      "learning_rate": 7.184637068357999e-06,
      "loss": 0.0305,
      "step": 1300
    },
    {
      "epoch": 0.7850187265917603,
      "grad_norm": 1.7254374027252197,
      "learning_rate": 7.1494009866102894e-06,
      "loss": 0.0327,
      "step": 1310
    },
    {
      "epoch": 0.7910112359550562,
      "grad_norm": 2.161518096923828,
      "learning_rate": 7.1141649048625796e-06,
      "loss": 0.0329,
      "step": 1320
    },
    {
      "epoch": 0.797003745318352,
      "grad_norm": 1.4740070104599,
      "learning_rate": 7.07892882311487e-06,
      "loss": 0.0346,
      "step": 1330
    },
    {
      "epoch": 0.8029962546816479,
      "grad_norm": 2.4593379497528076,
      "learning_rate": 7.043692741367161e-06,
      "loss": 0.0267,
      "step": 1340
    },
    {
      "epoch": 0.8089887640449438,
      "grad_norm": 1.8585057258605957,
      "learning_rate": 7.008456659619452e-06,
      "loss": 0.0262,
      "step": 1350
    },
    {
      "epoch": 0.8149812734082397,
      "grad_norm": 2.7434608936309814,
      "learning_rate": 6.973220577871741e-06,
      "loss": 0.039,
      "step": 1360
    },
    {
      "epoch": 0.8209737827715355,
      "grad_norm": 2.6166908740997314,
      "learning_rate": 6.937984496124032e-06,
      "loss": 0.0346,
      "step": 1370
    },
    {
      "epoch": 0.8269662921348314,
      "grad_norm": 1.4991202354431152,
      "learning_rate": 6.902748414376322e-06,
      "loss": 0.0355,
      "step": 1380
    },
    {
      "epoch": 0.8329588014981273,
      "grad_norm": 1.4986398220062256,
      "learning_rate": 6.867512332628612e-06,
      "loss": 0.0332,
      "step": 1390
    },
    {
      "epoch": 0.8389513108614233,
      "grad_norm": 2.284785509109497,
      "learning_rate": 6.832276250880902e-06,
      "loss": 0.0301,
      "step": 1400
    },
    {
      "epoch": 0.8449438202247191,
      "grad_norm": 1.3910906314849854,
      "learning_rate": 6.797040169133193e-06,
      "loss": 0.024,
      "step": 1410
    },
    {
      "epoch": 0.850936329588015,
      "grad_norm": 2.0242066383361816,
      "learning_rate": 6.7618040873854825e-06,
      "loss": 0.0359,
      "step": 1420
    },
    {
      "epoch": 0.8569288389513109,
      "grad_norm": 2.430185317993164,
      "learning_rate": 6.7265680056377734e-06,
      "loss": 0.0226,
      "step": 1430
    },
    {
      "epoch": 0.8629213483146068,
      "grad_norm": 0.8888330459594727,
      "learning_rate": 6.691331923890064e-06,
      "loss": 0.0289,
      "step": 1440
    },
    {
      "epoch": 0.8689138576779026,
      "grad_norm": 1.8304884433746338,
      "learning_rate": 6.656095842142354e-06,
      "loss": 0.0238,
      "step": 1450
    },
    {
      "epoch": 0.8749063670411985,
      "grad_norm": 2.193533182144165,
      "learning_rate": 6.620859760394645e-06,
      "loss": 0.0385,
      "step": 1460
    },
    {
      "epoch": 0.8808988764044944,
      "grad_norm": 2.304433822631836,
      "learning_rate": 6.585623678646935e-06,
      "loss": 0.0341,
      "step": 1470
    },
    {
      "epoch": 0.8868913857677903,
      "grad_norm": 1.412208080291748,
      "learning_rate": 6.550387596899226e-06,
      "loss": 0.0262,
      "step": 1480
    },
    {
      "epoch": 0.8928838951310861,
      "grad_norm": 0.917147696018219,
      "learning_rate": 6.515151515151516e-06,
      "loss": 0.0219,
      "step": 1490
    },
    {
      "epoch": 0.898876404494382,
      "grad_norm": 1.979040265083313,
      "learning_rate": 6.479915433403806e-06,
      "loss": 0.0396,
      "step": 1500
    },
    {
      "epoch": 0.9048689138576779,
      "grad_norm": 1.5810662508010864,
      "learning_rate": 6.444679351656097e-06,
      "loss": 0.0274,
      "step": 1510
    },
    {
      "epoch": 0.9108614232209737,
      "grad_norm": 2.405066728591919,
      "learning_rate": 6.409443269908386e-06,
      "loss": 0.0388,
      "step": 1520
    },
    {
      "epoch": 0.9168539325842696,
      "grad_norm": 2.646991014480591,
      "learning_rate": 6.374207188160677e-06,
      "loss": 0.0392,
      "step": 1530
    },
    {
      "epoch": 0.9228464419475655,
      "grad_norm": 2.1389219760894775,
      "learning_rate": 6.338971106412968e-06,
      "loss": 0.0345,
      "step": 1540
    },
    {
      "epoch": 0.9288389513108615,
      "grad_norm": 1.683204174041748,
      "learning_rate": 6.3037350246652574e-06,
      "loss": 0.0354,
      "step": 1550
    },
    {
      "epoch": 0.9348314606741573,
      "grad_norm": 2.031403064727783,
      "learning_rate": 6.268498942917548e-06,
      "loss": 0.0263,
      "step": 1560
    },
    {
      "epoch": 0.9408239700374532,
      "grad_norm": 2.195925712585449,
      "learning_rate": 6.2332628611698385e-06,
      "loss": 0.028,
      "step": 1570
    },
    {
      "epoch": 0.9468164794007491,
      "grad_norm": 1.9090993404388428,
      "learning_rate": 6.198026779422129e-06,
      "loss": 0.0339,
      "step": 1580
    },
    {
      "epoch": 0.952808988764045,
      "grad_norm": 2.7698168754577637,
      "learning_rate": 6.162790697674419e-06,
      "loss": 0.0412,
      "step": 1590
    },
    {
      "epoch": 0.9588014981273408,
      "grad_norm": 2.477466106414795,
      "learning_rate": 6.12755461592671e-06,
      "loss": 0.0307,
      "step": 1600
    },
    {
      "epoch": 0.9647940074906367,
      "grad_norm": 1.7105556726455688,
      "learning_rate": 6.092318534179001e-06,
      "loss": 0.0378,
      "step": 1610
    },
    {
      "epoch": 0.9707865168539326,
      "grad_norm": 1.9348987340927124,
      "learning_rate": 6.05708245243129e-06,
      "loss": 0.0312,
      "step": 1620
    },
    {
      "epoch": 0.9767790262172285,
      "grad_norm": 2.8063669204711914,
      "learning_rate": 6.021846370683581e-06,
      "loss": 0.0321,
      "step": 1630
    },
    {
      "epoch": 0.9827715355805243,
      "grad_norm": 1.3787035942077637,
      "learning_rate": 5.986610288935871e-06,
      "loss": 0.0204,
      "step": 1640
    },
    {
      "epoch": 0.9887640449438202,
      "grad_norm": 1.0348024368286133,
      "learning_rate": 5.951374207188161e-06,
      "loss": 0.0269,
      "step": 1650
    },
    {
      "epoch": 0.9947565543071161,
      "grad_norm": 2.3629918098449707,
      "learning_rate": 5.916138125440451e-06,
      "loss": 0.0289,
      "step": 1660
    },
    {
      "epoch": 1.0005992509363295,
      "grad_norm": 1.635030746459961,
      "learning_rate": 5.880902043692742e-06,
      "loss": 0.0271,
      "step": 1670
    },
    {
      "epoch": 1.0065917602996255,
      "grad_norm": 1.2064359188079834,
      "learning_rate": 5.8456659619450316e-06,
      "loss": 0.0068,
      "step": 1680
    },
    {
      "epoch": 1.0125842696629213,
      "grad_norm": 1.9273971319198608,
      "learning_rate": 5.8104298801973225e-06,
      "loss": 0.0074,
      "step": 1690
    },
    {
      "epoch": 1.0185767790262172,
      "grad_norm": 1.0188418626785278,
      "learning_rate": 5.7751937984496135e-06,
      "loss": 0.0124,
      "step": 1700
    },
    {
      "epoch": 1.024569288389513,
      "grad_norm": 0.795347273349762,
      "learning_rate": 5.739957716701903e-06,
      "loss": 0.0102,
      "step": 1710
    },
    {
      "epoch": 1.030561797752809,
      "grad_norm": 0.6345170736312866,
      "learning_rate": 5.704721634954194e-06,
      "loss": 0.009,
      "step": 1720
    },
    {
      "epoch": 1.036554307116105,
      "grad_norm": 0.6679102778434753,
      "learning_rate": 5.669485553206484e-06,
      "loss": 0.0086,
      "step": 1730
    },
    {
      "epoch": 1.0425468164794007,
      "grad_norm": 2.5230817794799805,
      "learning_rate": 5.634249471458775e-06,
      "loss": 0.0076,
      "step": 1740
    },
    {
      "epoch": 1.0485393258426967,
      "grad_norm": 0.9390186667442322,
      "learning_rate": 5.599013389711064e-06,
      "loss": 0.0056,
      "step": 1750
    },
    {
      "epoch": 1.0545318352059925,
      "grad_norm": 1.2290983200073242,
      "learning_rate": 5.563777307963355e-06,
      "loss": 0.0067,
      "step": 1760
    },
    {
      "epoch": 1.0605243445692885,
      "grad_norm": 1.8171833753585815,
      "learning_rate": 5.528541226215646e-06,
      "loss": 0.009,
      "step": 1770
    },
    {
      "epoch": 1.0665168539325842,
      "grad_norm": 1.3471766710281372,
      "learning_rate": 5.493305144467935e-06,
      "loss": 0.0061,
      "step": 1780
    },
    {
      "epoch": 1.0725093632958802,
      "grad_norm": 0.8515785336494446,
      "learning_rate": 5.458069062720226e-06,
      "loss": 0.0059,
      "step": 1790
    },
    {
      "epoch": 1.078501872659176,
      "grad_norm": 1.4226453304290771,
      "learning_rate": 5.422832980972516e-06,
      "loss": 0.0098,
      "step": 1800
    },
    {
      "epoch": 1.084494382022472,
      "grad_norm": 2.040360927581787,
      "learning_rate": 5.3875968992248065e-06,
      "loss": 0.009,
      "step": 1810
    },
    {
      "epoch": 1.0904868913857677,
      "grad_norm": 1.0943787097930908,
      "learning_rate": 5.352360817477097e-06,
      "loss": 0.0107,
      "step": 1820
    },
    {
      "epoch": 1.0964794007490637,
      "grad_norm": 2.117577314376831,
      "learning_rate": 5.317124735729388e-06,
      "loss": 0.0082,
      "step": 1830
    },
    {
      "epoch": 1.1024719101123595,
      "grad_norm": 1.5584527254104614,
      "learning_rate": 5.281888653981677e-06,
      "loss": 0.0058,
      "step": 1840
    },
    {
      "epoch": 1.1084644194756554,
      "grad_norm": 1.2435089349746704,
      "learning_rate": 5.246652572233968e-06,
      "loss": 0.0073,
      "step": 1850
    },
    {
      "epoch": 1.1144569288389512,
      "grad_norm": 1.123214602470398,
      "learning_rate": 5.211416490486259e-06,
      "loss": 0.0081,
      "step": 1860
    },
    {
      "epoch": 1.1204494382022472,
      "grad_norm": 0.5520896315574646,
      "learning_rate": 5.176180408738549e-06,
      "loss": 0.009,
      "step": 1870
    },
    {
      "epoch": 1.1264419475655432,
      "grad_norm": 1.1572812795639038,
      "learning_rate": 5.140944326990839e-06,
      "loss": 0.0062,
      "step": 1880
    },
    {
      "epoch": 1.132434456928839,
      "grad_norm": 1.0478726625442505,
      "learning_rate": 5.105708245243129e-06,
      "loss": 0.0057,
      "step": 1890
    },
    {
      "epoch": 1.138426966292135,
      "grad_norm": 0.7916937470436096,
      "learning_rate": 5.07047216349542e-06,
      "loss": 0.0068,
      "step": 1900
    },
    {
      "epoch": 1.1444194756554307,
      "grad_norm": 0.804720401763916,
      "learning_rate": 5.0352360817477094e-06,
      "loss": 0.0052,
      "step": 1910
    },
    {
      "epoch": 1.1504119850187267,
      "grad_norm": 0.8383046984672546,
      "learning_rate": 5e-06,
      "loss": 0.0103,
      "step": 1920
    },
    {
      "epoch": 1.1564044943820224,
      "grad_norm": 0.44970327615737915,
      "learning_rate": 4.9647639182522905e-06,
      "loss": 0.0086,
      "step": 1930
    },
    {
      "epoch": 1.1623970037453184,
      "grad_norm": 1.1988434791564941,
      "learning_rate": 4.9295278365045815e-06,
      "loss": 0.0091,
      "step": 1940
    },
    {
      "epoch": 1.1683895131086142,
      "grad_norm": 0.7889288067817688,
      "learning_rate": 4.894291754756872e-06,
      "loss": 0.0057,
      "step": 1950
    },
    {
      "epoch": 1.1743820224719101,
      "grad_norm": 0.5470920205116272,
      "learning_rate": 4.859055673009162e-06,
      "loss": 0.0072,
      "step": 1960
    },
    {
      "epoch": 1.180374531835206,
      "grad_norm": 1.7936517000198364,
      "learning_rate": 4.823819591261452e-06,
      "loss": 0.0048,
      "step": 1970
    },
    {
      "epoch": 1.186367041198502,
      "grad_norm": 1.2022855281829834,
      "learning_rate": 4.788583509513742e-06,
      "loss": 0.0066,
      "step": 1980
    },
    {
      "epoch": 1.1923595505617977,
      "grad_norm": 0.5040098428726196,
      "learning_rate": 4.753347427766033e-06,
      "loss": 0.0067,
      "step": 1990
    },
    {
      "epoch": 1.1983520599250936,
      "grad_norm": 0.6506677269935608,
      "learning_rate": 4.718111346018323e-06,
      "loss": 0.0045,
      "step": 2000
    },
    {
      "epoch": 1.1983520599250936,
      "eval_loss": 0.011116987094283104,
      "eval_runtime": 9693.0647,
      "eval_samples_per_second": 1.377,
      "eval_steps_per_second": 0.172,
      "eval_wer": 0.8704838022863507,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 3338,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.531054924201984e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}