{
  "best_global_step": 2000,
  "best_metric": 0.6722360915468404,
  "best_model_checkpoint": "./SALAMA_NEW9/checkpoint-2000",
  "epoch": 1.0204133707578464,
  "eval_steps": 2000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0051033426894615975,
      "grad_norm": 0.5211380124092102,
      "learning_rate": 1.8e-07,
      "loss": 0.0083,
      "step": 10
    },
    {
      "epoch": 0.010206685378923195,
      "grad_norm": 1.0282593965530396,
      "learning_rate": 3.8e-07,
      "loss": 0.0078,
      "step": 20
    },
    {
      "epoch": 0.015310028068384792,
      "grad_norm": 0.4742085933685303,
      "learning_rate": 5.800000000000001e-07,
      "loss": 0.0082,
      "step": 30
    },
    {
      "epoch": 0.02041337075784639,
      "grad_norm": 0.7376335263252258,
      "learning_rate": 7.8e-07,
      "loss": 0.0056,
      "step": 40
    },
    {
      "epoch": 0.025516713447307986,
      "grad_norm": 1.5874927043914795,
      "learning_rate": 9.800000000000001e-07,
      "loss": 0.012,
      "step": 50
    },
    {
      "epoch": 0.030620056136769585,
      "grad_norm": 0.6289522051811218,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 0.0081,
      "step": 60
    },
    {
      "epoch": 0.035723398826231184,
      "grad_norm": 1.9904325008392334,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.018,
      "step": 70
    },
    {
      "epoch": 0.04082674151569278,
      "grad_norm": 2.7978858947753906,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 0.0083,
      "step": 80
    },
    {
      "epoch": 0.045930084205154376,
      "grad_norm": 1.7935121059417725,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.0107,
      "step": 90
    },
    {
      "epoch": 0.05103342689461597,
      "grad_norm": 0.3225713074207306,
      "learning_rate": 1.98e-06,
      "loss": 0.01,
      "step": 100
    },
    {
      "epoch": 0.056136769584077574,
      "grad_norm": 0.9831269979476929,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.0101,
      "step": 110
    },
    {
      "epoch": 0.06124011227353917,
      "grad_norm": 1.6235119104385376,
      "learning_rate": 2.38e-06,
      "loss": 0.0067,
      "step": 120
    },
    {
      "epoch": 0.06634345496300076,
      "grad_norm": 2.157205820083618,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.008,
      "step": 130
    },
    {
      "epoch": 0.07144679765246237,
      "grad_norm": 0.38939204812049866,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 0.0054,
      "step": 140
    },
    {
      "epoch": 0.07655014034192396,
      "grad_norm": 0.7885264754295349,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.0081,
      "step": 150
    },
    {
      "epoch": 0.08165348303138556,
      "grad_norm": 1.518344759941101,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.0088,
      "step": 160
    },
    {
      "epoch": 0.08675682572084716,
      "grad_norm": 0.5009278059005737,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 0.005,
      "step": 170
    },
    {
      "epoch": 0.09186016841030875,
      "grad_norm": 2.364478588104248,
      "learning_rate": 3.58e-06,
      "loss": 0.0085,
      "step": 180
    },
    {
      "epoch": 0.09696351109977035,
      "grad_norm": 0.31605347990989685,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.0073,
      "step": 190
    },
    {
      "epoch": 0.10206685378923194,
      "grad_norm": 1.516256332397461,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0081,
      "step": 200
    },
    {
      "epoch": 0.10717019647869354,
      "grad_norm": 1.2425719499588013,
      "learning_rate": 4.18e-06,
      "loss": 0.0072,
      "step": 210
    },
    {
      "epoch": 0.11227353916815515,
      "grad_norm": 2.350731611251831,
      "learning_rate": 4.38e-06,
      "loss": 0.0147,
      "step": 220
    },
    {
      "epoch": 0.11737688185761674,
      "grad_norm": 0.7598681449890137,
      "learning_rate": 4.58e-06,
      "loss": 0.0117,
      "step": 230
    },
    {
      "epoch": 0.12248022454707834,
      "grad_norm": 0.8952233791351318,
      "learning_rate": 4.78e-06,
      "loss": 0.0089,
      "step": 240
    },
    {
      "epoch": 0.12758356723653994,
      "grad_norm": 0.6309432983398438,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.0047,
      "step": 250
    },
    {
      "epoch": 0.13268690992600152,
      "grad_norm": 1.7359912395477295,
      "learning_rate": 5.18e-06,
      "loss": 0.0096,
      "step": 260
    },
    {
      "epoch": 0.13779025261546313,
      "grad_norm": 2.6553421020507812,
      "learning_rate": 5.380000000000001e-06,
      "loss": 0.0078,
      "step": 270
    },
    {
      "epoch": 0.14289359530492474,
      "grad_norm": 1.1739858388900757,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.0087,
      "step": 280
    },
    {
      "epoch": 0.14799693799438632,
      "grad_norm": 0.708070695400238,
      "learning_rate": 5.78e-06,
      "loss": 0.0075,
      "step": 290
    },
    {
      "epoch": 0.15310028068384793,
      "grad_norm": 0.8523297309875488,
      "learning_rate": 5.98e-06,
      "loss": 0.0085,
      "step": 300
    },
    {
      "epoch": 0.1582036233733095,
      "grad_norm": 2.7822892665863037,
      "learning_rate": 6.18e-06,
      "loss": 0.0092,
      "step": 310
    },
    {
      "epoch": 0.16330696606277112,
      "grad_norm": 3.9056875705718994,
      "learning_rate": 6.380000000000001e-06,
      "loss": 0.0172,
      "step": 320
    },
    {
      "epoch": 0.1684103087522327,
      "grad_norm": 1.9748897552490234,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 0.0086,
      "step": 330
    },
    {
      "epoch": 0.1735136514416943,
      "grad_norm": 1.313830852508545,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.0124,
      "step": 340
    },
    {
      "epoch": 0.17861699413115592,
      "grad_norm": 1.782209038734436,
      "learning_rate": 6.98e-06,
      "loss": 0.0166,
      "step": 350
    },
    {
      "epoch": 0.1837203368206175,
      "grad_norm": 2.647857427597046,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.0151,
      "step": 360
    },
    {
      "epoch": 0.1888236795100791,
      "grad_norm": 2.1850852966308594,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.0116,
      "step": 370
    },
    {
      "epoch": 0.1939270221995407,
      "grad_norm": 2.69811749458313,
      "learning_rate": 7.58e-06,
      "loss": 0.0118,
      "step": 380
    },
    {
      "epoch": 0.1990303648890023,
      "grad_norm": 3.1227176189422607,
      "learning_rate": 7.78e-06,
      "loss": 0.0137,
      "step": 390
    },
    {
      "epoch": 0.20413370757846389,
      "grad_norm": 3.615041494369507,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0113,
      "step": 400
    },
    {
      "epoch": 0.2092370502679255,
      "grad_norm": 2.067406177520752,
      "learning_rate": 8.18e-06,
      "loss": 0.0154,
      "step": 410
    },
    {
      "epoch": 0.21434039295738708,
      "grad_norm": 1.5332070589065552,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.0159,
      "step": 420
    },
    {
      "epoch": 0.2194437356468487,
      "grad_norm": 1.3139411211013794,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.0096,
      "step": 430
    },
    {
      "epoch": 0.2245470783363103,
      "grad_norm": 2.8063700199127197,
      "learning_rate": 8.78e-06,
      "loss": 0.0096,
      "step": 440
    },
    {
      "epoch": 0.22965042102577188,
      "grad_norm": 1.2194337844848633,
      "learning_rate": 8.98e-06,
      "loss": 0.0113,
      "step": 450
    },
    {
      "epoch": 0.2347537637152335,
      "grad_norm": 1.4826334714889526,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.01,
      "step": 460
    },
    {
      "epoch": 0.23985710640469507,
      "grad_norm": 1.4572652578353882,
      "learning_rate": 9.38e-06,
      "loss": 0.012,
      "step": 470
    },
    {
      "epoch": 0.24496044909415668,
      "grad_norm": 2.323155641555786,
      "learning_rate": 9.58e-06,
      "loss": 0.016,
      "step": 480
    },
    {
      "epoch": 0.25006379178361826,
      "grad_norm": 2.1754894256591797,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.0134,
      "step": 490
    },
    {
      "epoch": 0.25516713447307987,
      "grad_norm": 2.55068039894104,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0133,
      "step": 500
    },
    {
      "epoch": 0.2602704771625415,
      "grad_norm": 2.78916072845459,
      "learning_rate": 9.983271375464685e-06,
      "loss": 0.0102,
      "step": 510
    },
    {
      "epoch": 0.26537381985200303,
      "grad_norm": 3.6304802894592285,
      "learning_rate": 9.96468401486989e-06,
      "loss": 0.0142,
      "step": 520
    },
    {
      "epoch": 0.27047716254146464,
      "grad_norm": 1.1248530149459839,
      "learning_rate": 9.946096654275093e-06,
      "loss": 0.0096,
      "step": 530
    },
    {
      "epoch": 0.27558050523092625,
      "grad_norm": 2.088334798812866,
      "learning_rate": 9.927509293680298e-06,
      "loss": 0.0189,
      "step": 540
    },
    {
      "epoch": 0.28068384792038786,
      "grad_norm": 2.035660982131958,
      "learning_rate": 9.908921933085503e-06,
      "loss": 0.0125,
      "step": 550
    },
    {
      "epoch": 0.2857871906098495,
      "grad_norm": 1.8379945755004883,
      "learning_rate": 9.890334572490708e-06,
      "loss": 0.01,
      "step": 560
    },
    {
      "epoch": 0.290890533299311,
      "grad_norm": 2.2616829872131348,
      "learning_rate": 9.871747211895911e-06,
      "loss": 0.0119,
      "step": 570
    },
    {
      "epoch": 0.29599387598877264,
      "grad_norm": 2.0087382793426514,
      "learning_rate": 9.853159851301116e-06,
      "loss": 0.0104,
      "step": 580
    },
    {
      "epoch": 0.30109721867823425,
      "grad_norm": 1.7350345849990845,
      "learning_rate": 9.83457249070632e-06,
      "loss": 0.0144,
      "step": 590
    },
    {
      "epoch": 0.30620056136769586,
      "grad_norm": 1.9522229433059692,
      "learning_rate": 9.815985130111524e-06,
      "loss": 0.0152,
      "step": 600
    },
    {
      "epoch": 0.31130390405715747,
      "grad_norm": 2.231642961502075,
      "learning_rate": 9.79739776951673e-06,
      "loss": 0.0138,
      "step": 610
    },
    {
      "epoch": 0.316407246746619,
      "grad_norm": 1.9675791263580322,
      "learning_rate": 9.778810408921934e-06,
      "loss": 0.0148,
      "step": 620
    },
    {
      "epoch": 0.32151058943608063,
      "grad_norm": 1.9099314212799072,
      "learning_rate": 9.76022304832714e-06,
      "loss": 0.0145,
      "step": 630
    },
    {
      "epoch": 0.32661393212554224,
      "grad_norm": 1.777403473854065,
      "learning_rate": 9.741635687732343e-06,
      "loss": 0.0164,
      "step": 640
    },
    {
      "epoch": 0.33171727481500385,
      "grad_norm": 3.160843849182129,
      "learning_rate": 9.723048327137548e-06,
      "loss": 0.0178,
      "step": 650
    },
    {
      "epoch": 0.3368206175044654,
      "grad_norm": 0.950518786907196,
      "learning_rate": 9.70446096654275e-06,
      "loss": 0.0154,
      "step": 660
    },
    {
      "epoch": 0.341923960193927,
      "grad_norm": 3.335329294204712,
      "learning_rate": 9.685873605947956e-06,
      "loss": 0.0149,
      "step": 670
    },
    {
      "epoch": 0.3470273028833886,
      "grad_norm": 2.4955413341522217,
      "learning_rate": 9.66728624535316e-06,
      "loss": 0.0095,
      "step": 680
    },
    {
      "epoch": 0.35213064557285023,
      "grad_norm": 1.4317821264266968,
      "learning_rate": 9.648698884758366e-06,
      "loss": 0.0143,
      "step": 690
    },
    {
      "epoch": 0.35723398826231184,
      "grad_norm": 1.6408915519714355,
      "learning_rate": 9.63011152416357e-06,
      "loss": 0.0262,
      "step": 700
    },
    {
      "epoch": 0.3623373309517734,
      "grad_norm": 5.888397693634033,
      "learning_rate": 9.611524163568774e-06,
      "loss": 0.0297,
      "step": 710
    },
    {
      "epoch": 0.367440673641235,
      "grad_norm": 2.526792526245117,
      "learning_rate": 9.592936802973979e-06,
      "loss": 0.0149,
      "step": 720
    },
    {
      "epoch": 0.3725440163306966,
      "grad_norm": 1.5245797634124756,
      "learning_rate": 9.574349442379182e-06,
      "loss": 0.0122,
      "step": 730
    },
    {
      "epoch": 0.3776473590201582,
      "grad_norm": 2.3768417835235596,
      "learning_rate": 9.555762081784387e-06,
      "loss": 0.0146,
      "step": 740
    },
    {
      "epoch": 0.3827507017096198,
      "grad_norm": 1.9545379877090454,
      "learning_rate": 9.537174721189592e-06,
      "loss": 0.0116,
      "step": 750
    },
    {
      "epoch": 0.3878540443990814,
      "grad_norm": 2.5888895988464355,
      "learning_rate": 9.518587360594797e-06,
      "loss": 0.0125,
      "step": 760
    },
    {
      "epoch": 0.392957387088543,
      "grad_norm": 2.554670810699463,
      "learning_rate": 9.5e-06,
      "loss": 0.0121,
      "step": 770
    },
    {
      "epoch": 0.3980607297780046,
      "grad_norm": 2.3274645805358887,
      "learning_rate": 9.481412639405206e-06,
      "loss": 0.0152,
      "step": 780
    },
    {
      "epoch": 0.4031640724674662,
      "grad_norm": 1.916551113128662,
      "learning_rate": 9.46282527881041e-06,
      "loss": 0.0181,
      "step": 790
    },
    {
      "epoch": 0.40826741515692777,
      "grad_norm": 2.7110981941223145,
      "learning_rate": 9.444237918215614e-06,
      "loss": 0.0201,
      "step": 800
    },
    {
      "epoch": 0.4133707578463894,
      "grad_norm": 2.66487193107605,
      "learning_rate": 9.425650557620819e-06,
      "loss": 0.0163,
      "step": 810
    },
    {
      "epoch": 0.418474100535851,
      "grad_norm": 4.3903303146362305,
      "learning_rate": 9.407063197026024e-06,
      "loss": 0.0203,
      "step": 820
    },
    {
      "epoch": 0.4235774432253126,
      "grad_norm": 0.8613393902778625,
      "learning_rate": 9.388475836431227e-06,
      "loss": 0.0101,
      "step": 830
    },
    {
      "epoch": 0.42868078591477415,
      "grad_norm": 14.285655975341797,
      "learning_rate": 9.369888475836432e-06,
      "loss": 0.0201,
      "step": 840
    },
    {
      "epoch": 0.43378412860423576,
      "grad_norm": 2.281245708465576,
      "learning_rate": 9.351301115241637e-06,
      "loss": 0.017,
      "step": 850
    },
    {
      "epoch": 0.4388874712936974,
      "grad_norm": 2.5612051486968994,
      "learning_rate": 9.33271375464684e-06,
      "loss": 0.0164,
      "step": 860
    },
    {
      "epoch": 0.443990813983159,
      "grad_norm": 3.728468894958496,
      "learning_rate": 9.314126394052045e-06,
      "loss": 0.0184,
      "step": 870
    },
    {
      "epoch": 0.4490941566726206,
      "grad_norm": 2.954237699508667,
      "learning_rate": 9.295539033457249e-06,
      "loss": 0.0216,
      "step": 880
    },
    {
      "epoch": 0.45419749936208215,
      "grad_norm": 2.5756335258483887,
      "learning_rate": 9.276951672862453e-06,
      "loss": 0.021,
      "step": 890
    },
    {
      "epoch": 0.45930084205154376,
      "grad_norm": 4.490197658538818,
      "learning_rate": 9.258364312267658e-06,
      "loss": 0.0138,
      "step": 900
    },
    {
      "epoch": 0.46440418474100537,
      "grad_norm": 1.9928340911865234,
      "learning_rate": 9.239776951672863e-06,
      "loss": 0.0158,
      "step": 910
    },
    {
      "epoch": 0.469507527430467,
      "grad_norm": 3.2016446590423584,
      "learning_rate": 9.221189591078068e-06,
      "loss": 0.0188,
      "step": 920
    },
    {
      "epoch": 0.47461087011992853,
      "grad_norm": 2.1624643802642822,
      "learning_rate": 9.202602230483272e-06,
      "loss": 0.0139,
      "step": 930
    },
    {
      "epoch": 0.47971421280939014,
      "grad_norm": 2.0108089447021484,
      "learning_rate": 9.184014869888477e-06,
      "loss": 0.0173,
      "step": 940
    },
    {
      "epoch": 0.48481755549885175,
      "grad_norm": 2.6266250610351562,
      "learning_rate": 9.16542750929368e-06,
      "loss": 0.0181,
      "step": 950
    },
    {
      "epoch": 0.48992089818831336,
      "grad_norm": 1.7041484117507935,
      "learning_rate": 9.146840148698885e-06,
      "loss": 0.0167,
      "step": 960
    },
    {
      "epoch": 0.49502424087777497,
      "grad_norm": 2.4042234420776367,
      "learning_rate": 9.12825278810409e-06,
      "loss": 0.017,
      "step": 970
    },
    {
      "epoch": 0.5001275835672365,
      "grad_norm": 1.770944595336914,
      "learning_rate": 9.109665427509295e-06,
      "loss": 0.01,
      "step": 980
    },
    {
      "epoch": 0.5052309262566982,
      "grad_norm": 2.101804256439209,
      "learning_rate": 9.0910780669145e-06,
      "loss": 0.0152,
      "step": 990
    },
    {
      "epoch": 0.5103342689461597,
      "grad_norm": 3.545254945755005,
      "learning_rate": 9.072490706319703e-06,
      "loss": 0.014,
      "step": 1000
    },
    {
      "epoch": 0.5154376116356213,
      "grad_norm": 2.445159912109375,
      "learning_rate": 9.053903345724908e-06,
      "loss": 0.0207,
      "step": 1010
    },
    {
      "epoch": 0.520540954325083,
      "grad_norm": 3.302297830581665,
      "learning_rate": 9.035315985130111e-06,
      "loss": 0.0212,
      "step": 1020
    },
    {
      "epoch": 0.5256442970145445,
      "grad_norm": 4.689877510070801,
      "learning_rate": 9.016728624535316e-06,
      "loss": 0.025,
      "step": 1030
    },
    {
      "epoch": 0.5307476397040061,
      "grad_norm": 4.139590740203857,
      "learning_rate": 8.998141263940521e-06,
      "loss": 0.0158,
      "step": 1040
    },
    {
      "epoch": 0.5358509823934677,
      "grad_norm": 1.6236610412597656,
      "learning_rate": 8.979553903345726e-06,
      "loss": 0.0112,
      "step": 1050
    },
    {
      "epoch": 0.5409543250829293,
      "grad_norm": 2.6642770767211914,
      "learning_rate": 8.96096654275093e-06,
      "loss": 0.0226,
      "step": 1060
    },
    {
      "epoch": 0.546057667772391,
      "grad_norm": 2.012868642807007,
      "learning_rate": 8.942379182156135e-06,
      "loss": 0.0172,
      "step": 1070
    },
    {
      "epoch": 0.5511610104618525,
      "grad_norm": 1.9676612615585327,
      "learning_rate": 8.92379182156134e-06,
      "loss": 0.0131,
      "step": 1080
    },
    {
      "epoch": 0.5562643531513141,
      "grad_norm": 3.358045816421509,
      "learning_rate": 8.905204460966543e-06,
      "loss": 0.0168,
      "step": 1090
    },
    {
      "epoch": 0.5613676958407757,
      "grad_norm": 1.9890451431274414,
      "learning_rate": 8.886617100371748e-06,
      "loss": 0.0158,
      "step": 1100
    },
    {
      "epoch": 0.5664710385302373,
      "grad_norm": 2.1915857791900635,
      "learning_rate": 8.868029739776953e-06,
      "loss": 0.015,
      "step": 1110
    },
    {
      "epoch": 0.571574381219699,
      "grad_norm": 2.0204272270202637,
      "learning_rate": 8.849442379182158e-06,
      "loss": 0.0217,
      "step": 1120
    },
    {
      "epoch": 0.5766777239091605,
      "grad_norm": 1.8702834844589233,
      "learning_rate": 8.830855018587361e-06,
      "loss": 0.014,
      "step": 1130
    },
    {
      "epoch": 0.581781066598622,
      "grad_norm": 0.8649874925613403,
      "learning_rate": 8.812267657992566e-06,
      "loss": 0.0168,
      "step": 1140
    },
    {
      "epoch": 0.5868844092880837,
      "grad_norm": 2.020085334777832,
      "learning_rate": 8.79368029739777e-06,
      "loss": 0.0166,
      "step": 1150
    },
    {
      "epoch": 0.5919877519775453,
      "grad_norm": 0.6940491199493408,
      "learning_rate": 8.775092936802974e-06,
      "loss": 0.014,
      "step": 1160
    },
    {
      "epoch": 0.5970910946670069,
      "grad_norm": 1.6421513557434082,
      "learning_rate": 8.75650557620818e-06,
      "loss": 0.0149,
      "step": 1170
    },
    {
      "epoch": 0.6021944373564685,
      "grad_norm": 1.7957764863967896,
      "learning_rate": 8.737918215613384e-06,
      "loss": 0.0194,
      "step": 1180
    },
    {
      "epoch": 0.60729778004593,
      "grad_norm": 1.6488491296768188,
      "learning_rate": 8.719330855018588e-06,
      "loss": 0.0119,
      "step": 1190
    },
    {
      "epoch": 0.6124011227353917,
      "grad_norm": 1.9999263286590576,
      "learning_rate": 8.700743494423793e-06,
      "loss": 0.0165,
      "step": 1200
    },
    {
      "epoch": 0.6175044654248533,
      "grad_norm": 1.749192237854004,
      "learning_rate": 8.682156133828998e-06,
      "loss": 0.0193,
      "step": 1210
    },
    {
      "epoch": 0.6226078081143149,
      "grad_norm": 2.414264440536499,
      "learning_rate": 8.663568773234201e-06,
      "loss": 0.0199,
      "step": 1220
    },
    {
      "epoch": 0.6277111508037765,
      "grad_norm": 2.670834541320801,
      "learning_rate": 8.644981412639406e-06,
      "loss": 0.0178,
      "step": 1230
    },
    {
      "epoch": 0.632814493493238,
      "grad_norm": 3.2673842906951904,
      "learning_rate": 8.626394052044609e-06,
      "loss": 0.0161,
      "step": 1240
    },
    {
      "epoch": 0.6379178361826997,
      "grad_norm": 2.5664849281311035,
      "learning_rate": 8.607806691449814e-06,
      "loss": 0.0213,
      "step": 1250
    },
    {
      "epoch": 0.6430211788721613,
      "grad_norm": 2.350846290588379,
      "learning_rate": 8.589219330855019e-06,
      "loss": 0.0181,
      "step": 1260
    },
    {
      "epoch": 0.6481245215616228,
      "grad_norm": 2.494407892227173,
      "learning_rate": 8.570631970260224e-06,
      "loss": 0.0128,
      "step": 1270
    },
    {
      "epoch": 0.6532278642510845,
      "grad_norm": 2.3424453735351562,
      "learning_rate": 8.552044609665429e-06,
      "loss": 0.0127,
      "step": 1280
    },
    {
      "epoch": 0.658331206940546,
      "grad_norm": 2.1651947498321533,
      "learning_rate": 8.533457249070632e-06,
      "loss": 0.0229,
      "step": 1290
    },
    {
      "epoch": 0.6634345496300077,
      "grad_norm": 0.5863803029060364,
      "learning_rate": 8.514869888475837e-06,
      "loss": 0.0145,
      "step": 1300
    },
    {
      "epoch": 0.6685378923194693,
      "grad_norm": 1.3225018978118896,
      "learning_rate": 8.49628252788104e-06,
      "loss": 0.0149,
      "step": 1310
    },
    {
      "epoch": 0.6736412350089308,
      "grad_norm": 3.3000130653381348,
      "learning_rate": 8.477695167286246e-06,
      "loss": 0.0211,
      "step": 1320
    },
    {
      "epoch": 0.6787445776983925,
      "grad_norm": 2.677570104598999,
      "learning_rate": 8.45910780669145e-06,
      "loss": 0.0113,
      "step": 1330
    },
    {
      "epoch": 0.683847920387854,
      "grad_norm": 1.235533595085144,
      "learning_rate": 8.440520446096656e-06,
      "loss": 0.0132,
      "step": 1340
    },
    {
      "epoch": 0.6889512630773157,
      "grad_norm": 1.7336188554763794,
      "learning_rate": 8.42193308550186e-06,
      "loss": 0.0147,
      "step": 1350
    },
    {
      "epoch": 0.6940546057667772,
      "grad_norm": 3.8093788623809814,
      "learning_rate": 8.403345724907064e-06,
      "loss": 0.0168,
      "step": 1360
    },
    {
      "epoch": 0.6991579484562388,
      "grad_norm": 1.9721407890319824,
      "learning_rate": 8.384758364312269e-06,
      "loss": 0.0148,
      "step": 1370
    },
    {
      "epoch": 0.7042612911457005,
      "grad_norm": 4.275414943695068,
      "learning_rate": 8.366171003717472e-06,
      "loss": 0.0171,
      "step": 1380
    },
    {
      "epoch": 0.709364633835162,
      "grad_norm": 1.36530339717865,
      "learning_rate": 8.347583643122677e-06,
      "loss": 0.0157,
      "step": 1390
    },
    {
      "epoch": 0.7144679765246237,
      "grad_norm": 2.0768120288848877,
      "learning_rate": 8.328996282527882e-06,
      "loss": 0.0197,
      "step": 1400
    },
    {
      "epoch": 0.7195713192140852,
      "grad_norm": 3.6376969814300537,
      "learning_rate": 8.310408921933087e-06,
      "loss": 0.02,
      "step": 1410
    },
    {
      "epoch": 0.7246746619035468,
      "grad_norm": 4.029935836791992,
      "learning_rate": 8.29182156133829e-06,
      "loss": 0.0132,
      "step": 1420
    },
    {
      "epoch": 0.7297780045930085,
      "grad_norm": 3.0603153705596924,
      "learning_rate": 8.273234200743495e-06,
      "loss": 0.0124,
      "step": 1430
    },
    {
      "epoch": 0.73488134728247,
      "grad_norm": 0.8475554585456848,
      "learning_rate": 8.2546468401487e-06,
      "loss": 0.0124,
      "step": 1440
    },
    {
      "epoch": 0.7399846899719316,
      "grad_norm": 1.9978291988372803,
      "learning_rate": 8.236059479553904e-06,
      "loss": 0.0117,
      "step": 1450
    },
    {
      "epoch": 0.7450880326613932,
      "grad_norm": 1.5020562410354614,
      "learning_rate": 8.217472118959108e-06,
      "loss": 0.0167,
      "step": 1460
    },
    {
      "epoch": 0.7501913753508548,
      "grad_norm": 1.614305853843689,
      "learning_rate": 8.198884758364313e-06,
      "loss": 0.0149,
      "step": 1470
    },
    {
      "epoch": 0.7552947180403164,
      "grad_norm": 2.371570110321045,
      "learning_rate": 8.180297397769518e-06,
      "loss": 0.0142,
      "step": 1480
    },
    {
      "epoch": 0.760398060729778,
      "grad_norm": 1.5552469491958618,
      "learning_rate": 8.161710037174722e-06,
      "loss": 0.0134,
      "step": 1490
    },
    {
      "epoch": 0.7655014034192396,
      "grad_norm": 1.9674372673034668,
      "learning_rate": 8.143122676579927e-06,
      "loss": 0.0225,
      "step": 1500
    },
    {
      "epoch": 0.7706047461087012,
      "grad_norm": 1.94131600856781,
      "learning_rate": 8.12453531598513e-06,
      "loss": 0.0132,
      "step": 1510
    },
    {
      "epoch": 0.7757080887981628,
      "grad_norm": 2.533285140991211,
      "learning_rate": 8.105947955390335e-06,
      "loss": 0.0164,
      "step": 1520
    },
    {
      "epoch": 0.7808114314876244,
      "grad_norm": 1.7931355237960815,
      "learning_rate": 8.08736059479554e-06,
      "loss": 0.0145,
      "step": 1530
    },
    {
      "epoch": 0.785914774177086,
      "grad_norm": 1.5637154579162598,
      "learning_rate": 8.068773234200745e-06,
      "loss": 0.0131,
      "step": 1540
    },
    {
      "epoch": 0.7910181168665475,
      "grad_norm": 1.0649983882904053,
      "learning_rate": 8.050185873605948e-06,
      "loss": 0.0317,
      "step": 1550
    },
    {
      "epoch": 0.7961214595560092,
      "grad_norm": 1.9837394952774048,
      "learning_rate": 8.031598513011153e-06,
      "loss": 0.0168,
      "step": 1560
    },
    {
      "epoch": 0.8012248022454708,
      "grad_norm": 3.6585099697113037,
      "learning_rate": 8.013011152416358e-06,
      "loss": 0.0131,
      "step": 1570
    },
    {
      "epoch": 0.8063281449349324,
      "grad_norm": 2.7953765392303467,
      "learning_rate": 7.994423791821561e-06,
      "loss": 0.0162,
      "step": 1580
    },
    {
      "epoch": 0.811431487624394,
      "grad_norm": 2.3890202045440674,
      "learning_rate": 7.975836431226766e-06,
      "loss": 0.0158,
      "step": 1590
    },
    {
      "epoch": 0.8165348303138555,
      "grad_norm": 2.073019504547119,
      "learning_rate": 7.95724907063197e-06,
      "loss": 0.0159,
      "step": 1600
    },
    {
      "epoch": 0.8216381730033172,
      "grad_norm": 2.4629039764404297,
      "learning_rate": 7.938661710037175e-06,
      "loss": 0.0117,
      "step": 1610
    },
    {
      "epoch": 0.8267415156927788,
      "grad_norm": 1.4736220836639404,
      "learning_rate": 7.92007434944238e-06,
      "loss": 0.0145,
      "step": 1620
    },
    {
      "epoch": 0.8318448583822403,
      "grad_norm": 3.2814719676971436,
      "learning_rate": 7.901486988847585e-06,
      "loss": 0.0143,
      "step": 1630
    },
    {
      "epoch": 0.836948201071702,
      "grad_norm": 2.1625795364379883,
      "learning_rate": 7.88289962825279e-06,
      "loss": 0.0118,
      "step": 1640
    },
    {
      "epoch": 0.8420515437611635,
      "grad_norm": 1.660874605178833,
      "learning_rate": 7.864312267657993e-06,
      "loss": 0.0127,
      "step": 1650
    },
    {
      "epoch": 0.8471548864506252,
      "grad_norm": 1.7518630027770996,
      "learning_rate": 7.845724907063198e-06,
      "loss": 0.0085,
      "step": 1660
    },
    {
      "epoch": 0.8522582291400868,
      "grad_norm": 1.452298879623413,
      "learning_rate": 7.827137546468401e-06,
      "loss": 0.0151,
      "step": 1670
    },
    {
      "epoch": 0.8573615718295483,
      "grad_norm": 1.8911986351013184,
      "learning_rate": 7.808550185873606e-06,
      "loss": 0.0166,
      "step": 1680
    },
    {
      "epoch": 0.86246491451901,
      "grad_norm": 3.8515708446502686,
      "learning_rate": 7.789962825278811e-06,
      "loss": 0.0221,
      "step": 1690
    },
    {
      "epoch": 0.8675682572084715,
      "grad_norm": 2.210042953491211,
      "learning_rate": 7.771375464684016e-06,
      "loss": 0.0208,
      "step": 1700
    },
    {
      "epoch": 0.8726715998979332,
      "grad_norm": 2.0735044479370117,
      "learning_rate": 7.75278810408922e-06,
      "loss": 0.0142,
      "step": 1710
    },
    {
      "epoch": 0.8777749425873947,
      "grad_norm": 2.415004253387451,
      "learning_rate": 7.734200743494424e-06,
      "loss": 0.0179,
      "step": 1720
    },
    {
      "epoch": 0.8828782852768563,
      "grad_norm": 2.272406816482544,
      "learning_rate": 7.71561338289963e-06,
      "loss": 0.0142,
      "step": 1730
    },
    {
      "epoch": 0.887981627966318,
      "grad_norm": 1.2048219442367554,
      "learning_rate": 7.697026022304833e-06,
      "loss": 0.0102,
      "step": 1740
    },
    {
      "epoch": 0.8930849706557795,
      "grad_norm": 1.414962887763977,
      "learning_rate": 7.678438661710038e-06,
      "loss": 0.0112,
      "step": 1750
    },
    {
      "epoch": 0.8981883133452412,
      "grad_norm": 0.9970257878303528,
      "learning_rate": 7.659851301115243e-06,
      "loss": 0.0116,
      "step": 1760
    },
    {
      "epoch": 0.9032916560347027,
      "grad_norm": 1.7614041566848755,
      "learning_rate": 7.641263940520448e-06,
      "loss": 0.0198,
      "step": 1770
    },
    {
      "epoch": 0.9083949987241643,
      "grad_norm": 2.285222291946411,
      "learning_rate": 7.622676579925651e-06,
      "loss": 0.0146,
      "step": 1780
    },
    {
      "epoch": 0.913498341413626,
      "grad_norm": 2.238495111465454,
      "learning_rate": 7.604089219330856e-06,
      "loss": 0.0102,
      "step": 1790
    },
    {
      "epoch": 0.9186016841030875,
      "grad_norm": 0.7516927123069763,
      "learning_rate": 7.58550185873606e-06,
      "loss": 0.0134,
      "step": 1800
    },
    {
      "epoch": 0.9237050267925491,
      "grad_norm": 1.6228662729263306,
      "learning_rate": 7.566914498141265e-06,
      "loss": 0.0182,
      "step": 1810
    },
    {
      "epoch": 0.9288083694820107,
      "grad_norm": 1.0676440000534058,
      "learning_rate": 7.548327137546469e-06,
      "loss": 0.0117,
      "step": 1820
    },
    {
      "epoch": 0.9339117121714723,
      "grad_norm": 2.345280170440674,
      "learning_rate": 7.529739776951673e-06,
      "loss": 0.0116,
      "step": 1830
    },
    {
      "epoch": 0.939015054860934,
      "grad_norm": 2.056405782699585,
      "learning_rate": 7.511152416356878e-06,
      "loss": 0.0181,
      "step": 1840
    },
    {
      "epoch": 0.9441183975503955,
      "grad_norm": 1.5895274877548218,
      "learning_rate": 7.492565055762082e-06,
      "loss": 0.0143,
      "step": 1850
    },
    {
      "epoch": 0.9492217402398571,
      "grad_norm": 3.693983554840088,
      "learning_rate": 7.473977695167287e-06,
      "loss": 0.0139,
      "step": 1860
    },
    {
      "epoch": 0.9543250829293187,
      "grad_norm": 1.7493189573287964,
      "learning_rate": 7.455390334572491e-06,
      "loss": 0.012,
      "step": 1870
    },
    {
      "epoch": 0.9594284256187803,
      "grad_norm": 6.353549957275391,
      "learning_rate": 7.436802973977696e-06,
      "loss": 0.0182,
      "step": 1880
    },
    {
      "epoch": 0.9645317683082419,
      "grad_norm": 3.067734956741333,
      "learning_rate": 7.4182156133829e-06,
      "loss": 0.0116,
      "step": 1890
    },
    {
      "epoch": 0.9696351109977035,
      "grad_norm": 2.4685025215148926,
      "learning_rate": 7.399628252788105e-06,
      "loss": 0.0153,
      "step": 1900
    },
    {
      "epoch": 0.974738453687165,
      "grad_norm": 2.9748520851135254,
      "learning_rate": 7.38104089219331e-06,
      "loss": 0.0196,
      "step": 1910
    },
    {
      "epoch": 0.9798417963766267,
      "grad_norm": 1.787302017211914,
      "learning_rate": 7.362453531598514e-06,
      "loss": 0.0139,
      "step": 1920
    },
    {
      "epoch": 0.9849451390660883,
      "grad_norm": 2.998495101928711,
      "learning_rate": 7.343866171003719e-06,
      "loss": 0.0117,
      "step": 1930
    },
    {
      "epoch": 0.9900484817555499,
      "grad_norm": 2.461190938949585,
      "learning_rate": 7.325278810408922e-06,
      "loss": 0.0188,
      "step": 1940
    },
    {
      "epoch": 0.9951518244450115,
      "grad_norm": 2.0859811305999756,
      "learning_rate": 7.306691449814127e-06,
      "loss": 0.0144,
      "step": 1950
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.421244621276855,
      "learning_rate": 7.288104089219331e-06,
      "loss": 0.0214,
      "step": 1960
    },
    {
      "epoch": 1.0051033426894616,
      "grad_norm": 1.6903966665267944,
      "learning_rate": 7.269516728624536e-06,
      "loss": 0.0052,
      "step": 1970
    },
    {
      "epoch": 1.010206685378923,
      "grad_norm": 1.4454400539398193,
      "learning_rate": 7.25092936802974e-06,
      "loss": 0.0054,
      "step": 1980
    },
    {
      "epoch": 1.0153100280683849,
      "grad_norm": 1.46286141872406,
      "learning_rate": 7.2323420074349444e-06,
      "loss": 0.0036,
      "step": 1990
    },
    {
      "epoch": 1.0204133707578464,
      "grad_norm": 1.4223207235336304,
      "learning_rate": 7.213754646840149e-06,
      "loss": 0.0041,
      "step": 2000
    },
    {
      "epoch": 1.0204133707578464,
      "eval_loss": 0.007851608097553253,
      "eval_runtime": 5932.2159,
      "eval_samples_per_second": 2.642,
      "eval_steps_per_second": 0.33,
      "eval_wer": 0.6722360915468404,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 5880,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 1.846311674167296e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}