{
  "best_global_step": 2000,
  "best_metric": 0.7323685598172008,
  "best_model_checkpoint": "./SALAMA_NEWMEDTT/checkpoint-2000",
  "epoch": 0.7664670658682635,
  "eval_steps": 2000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0038323353293413173,
      "grad_norm": 0.13999255001544952,
      "learning_rate": 1.8e-07,
      "loss": 0.0033,
      "step": 10
    },
    {
      "epoch": 0.007664670658682635,
      "grad_norm": 0.06949484348297119,
      "learning_rate": 3.8e-07,
      "loss": 0.0033,
      "step": 20
    },
    {
      "epoch": 0.011497005988023952,
      "grad_norm": 0.10336039215326309,
      "learning_rate": 5.800000000000001e-07,
      "loss": 0.0034,
      "step": 30
    },
    {
      "epoch": 0.01532934131736527,
      "grad_norm": 2.217076063156128,
      "learning_rate": 7.8e-07,
      "loss": 0.003,
      "step": 40
    },
    {
      "epoch": 0.019161676646706587,
      "grad_norm": 1.3674581050872803,
      "learning_rate": 9.800000000000001e-07,
      "loss": 0.0042,
      "step": 50
    },
    {
      "epoch": 0.022994011976047904,
      "grad_norm": 0.8775060176849365,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 0.0014,
      "step": 60
    },
    {
      "epoch": 0.02682634730538922,
      "grad_norm": 0.21147924661636353,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.0023,
      "step": 70
    },
    {
      "epoch": 0.03065868263473054,
      "grad_norm": 0.285540908575058,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 0.0038,
      "step": 80
    },
    {
      "epoch": 0.034491017964071856,
      "grad_norm": 0.5599580407142639,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.0036,
      "step": 90
    },
    {
      "epoch": 0.03832335329341317,
      "grad_norm": 0.7187970280647278,
      "learning_rate": 1.98e-06,
      "loss": 0.0054,
      "step": 100
    },
    {
      "epoch": 0.04215568862275449,
      "grad_norm": 1.0117833614349365,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.0035,
      "step": 110
    },
    {
      "epoch": 0.04598802395209581,
      "grad_norm": 0.15052905678749084,
      "learning_rate": 2.38e-06,
      "loss": 0.0032,
      "step": 120
    },
    {
      "epoch": 0.049820359281437125,
      "grad_norm": 0.3713392913341522,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.0025,
      "step": 130
    },
    {
      "epoch": 0.05365269461077844,
      "grad_norm": 0.11066653579473495,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 0.0017,
      "step": 140
    },
    {
      "epoch": 0.05748502994011976,
      "grad_norm": 0.7113040089607239,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.0035,
      "step": 150
    },
    {
      "epoch": 0.06131736526946108,
      "grad_norm": 0.19436658918857574,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.0029,
      "step": 160
    },
    {
      "epoch": 0.0651497005988024,
      "grad_norm": 0.10705593228340149,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 0.0015,
      "step": 170
    },
    {
      "epoch": 0.06898203592814371,
      "grad_norm": 0.28974607586860657,
      "learning_rate": 3.58e-06,
      "loss": 0.002,
      "step": 180
    },
    {
      "epoch": 0.07281437125748504,
      "grad_norm": 0.6417028903961182,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.002,
      "step": 190
    },
    {
      "epoch": 0.07664670658682635,
      "grad_norm": 0.4101906418800354,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0023,
      "step": 200
    },
    {
      "epoch": 0.08047904191616767,
      "grad_norm": 0.393304705619812,
      "learning_rate": 4.18e-06,
      "loss": 0.0029,
      "step": 210
    },
    {
      "epoch": 0.08431137724550898,
      "grad_norm": 2.8025639057159424,
      "learning_rate": 4.38e-06,
      "loss": 0.0063,
      "step": 220
    },
    {
      "epoch": 0.0881437125748503,
      "grad_norm": 0.30129295587539673,
      "learning_rate": 4.58e-06,
      "loss": 0.0043,
      "step": 230
    },
    {
      "epoch": 0.09197604790419162,
      "grad_norm": 0.8991917967796326,
      "learning_rate": 4.78e-06,
      "loss": 0.0041,
      "step": 240
    },
    {
      "epoch": 0.09580838323353294,
      "grad_norm": 0.5826700329780579,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.0052,
      "step": 250
    },
    {
      "epoch": 0.09964071856287425,
      "grad_norm": 0.4652438759803772,
      "learning_rate": 5.18e-06,
      "loss": 0.0032,
      "step": 260
    },
    {
      "epoch": 0.10347305389221557,
      "grad_norm": 0.24232645332813263,
      "learning_rate": 5.380000000000001e-06,
      "loss": 0.003,
      "step": 270
    },
    {
      "epoch": 0.10730538922155688,
      "grad_norm": 0.2619079649448395,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.0035,
      "step": 280
    },
    {
      "epoch": 0.11113772455089821,
      "grad_norm": 0.36271339654922485,
      "learning_rate": 5.78e-06,
      "loss": 0.004,
      "step": 290
    },
    {
      "epoch": 0.11497005988023952,
      "grad_norm": 1.1893694400787354,
      "learning_rate": 5.98e-06,
      "loss": 0.0075,
      "step": 300
    },
    {
      "epoch": 0.11880239520958084,
      "grad_norm": 1.4707320928573608,
      "learning_rate": 6.18e-06,
      "loss": 0.0065,
      "step": 310
    },
    {
      "epoch": 0.12263473053892215,
      "grad_norm": 0.5402860045433044,
      "learning_rate": 6.380000000000001e-06,
      "loss": 0.0079,
      "step": 320
    },
    {
      "epoch": 0.12646706586826348,
      "grad_norm": 0.8445234894752502,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 0.0038,
      "step": 330
    },
    {
      "epoch": 0.1302994011976048,
      "grad_norm": 1.0490593910217285,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.0065,
      "step": 340
    },
    {
      "epoch": 0.1341317365269461,
      "grad_norm": 0.7944777011871338,
      "learning_rate": 6.98e-06,
      "loss": 0.0066,
      "step": 350
    },
    {
      "epoch": 0.13796407185628742,
      "grad_norm": 0.7534486055374146,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.0049,
      "step": 360
    },
    {
      "epoch": 0.14179640718562875,
      "grad_norm": 0.8252223134040833,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.0066,
      "step": 370
    },
    {
      "epoch": 0.14562874251497007,
      "grad_norm": 0.8292574286460876,
      "learning_rate": 7.58e-06,
      "loss": 0.0052,
      "step": 380
    },
    {
      "epoch": 0.14946107784431137,
      "grad_norm": 1.1227622032165527,
      "learning_rate": 7.78e-06,
      "loss": 0.0076,
      "step": 390
    },
    {
      "epoch": 0.1532934131736527,
      "grad_norm": 1.2978622913360596,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0065,
      "step": 400
    },
    {
      "epoch": 0.15712574850299402,
      "grad_norm": 2.281665325164795,
      "learning_rate": 8.18e-06,
      "loss": 0.0072,
      "step": 410
    },
    {
      "epoch": 0.16095808383233534,
      "grad_norm": 0.9634031653404236,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.0066,
      "step": 420
    },
    {
      "epoch": 0.16479041916167664,
      "grad_norm": 1.3554670810699463,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.007,
      "step": 430
    },
    {
      "epoch": 0.16862275449101796,
      "grad_norm": 1.5378248691558838,
      "learning_rate": 8.78e-06,
      "loss": 0.0082,
      "step": 440
    },
    {
      "epoch": 0.1724550898203593,
      "grad_norm": 2.1895182132720947,
      "learning_rate": 8.98e-06,
      "loss": 0.0103,
      "step": 450
    },
    {
      "epoch": 0.1762874251497006,
      "grad_norm": 0.634242057800293,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.0113,
      "step": 460
    },
    {
      "epoch": 0.1801197604790419,
      "grad_norm": 0.6717728972434998,
      "learning_rate": 9.38e-06,
      "loss": 0.0057,
      "step": 470
    },
    {
      "epoch": 0.18395209580838323,
      "grad_norm": 2.004511833190918,
      "learning_rate": 9.58e-06,
      "loss": 0.0147,
      "step": 480
    },
    {
      "epoch": 0.18778443113772456,
      "grad_norm": 8.510348320007324,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.012,
      "step": 490
    },
    {
      "epoch": 0.19161676646706588,
      "grad_norm": 1.1750833988189697,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0132,
      "step": 500
    },
    {
      "epoch": 0.19544910179640718,
      "grad_norm": 1.6581082344055176,
      "learning_rate": 9.980932203389831e-06,
      "loss": 0.0213,
      "step": 510
    },
    {
      "epoch": 0.1992814371257485,
      "grad_norm": 1.5398513078689575,
      "learning_rate": 9.959745762711866e-06,
      "loss": 0.0136,
      "step": 520
    },
    {
      "epoch": 0.20311377245508982,
      "grad_norm": 0.879449725151062,
      "learning_rate": 9.9385593220339e-06,
      "loss": 0.0119,
      "step": 530
    },
    {
      "epoch": 0.20694610778443115,
      "grad_norm": 2.3282341957092285,
      "learning_rate": 9.917372881355933e-06,
      "loss": 0.0143,
      "step": 540
    },
    {
      "epoch": 0.21077844311377245,
      "grad_norm": 2.3774726390838623,
      "learning_rate": 9.896186440677968e-06,
      "loss": 0.0159,
      "step": 550
    },
    {
      "epoch": 0.21461077844311377,
      "grad_norm": 1.5949875116348267,
      "learning_rate": 9.875000000000001e-06,
      "loss": 0.0101,
      "step": 560
    },
    {
      "epoch": 0.2184431137724551,
      "grad_norm": 1.4217886924743652,
      "learning_rate": 9.853813559322034e-06,
      "loss": 0.0137,
      "step": 570
    },
    {
      "epoch": 0.22227544910179642,
      "grad_norm": 1.0425392389297485,
      "learning_rate": 9.832627118644068e-06,
      "loss": 0.012,
      "step": 580
    },
    {
      "epoch": 0.22610778443113771,
      "grad_norm": 1.374623417854309,
      "learning_rate": 9.811440677966103e-06,
      "loss": 0.0128,
      "step": 590
    },
    {
      "epoch": 0.22994011976047904,
      "grad_norm": 2.8496975898742676,
      "learning_rate": 9.790254237288136e-06,
      "loss": 0.0122,
      "step": 600
    },
    {
      "epoch": 0.23377245508982036,
      "grad_norm": 1.7832646369934082,
      "learning_rate": 9.76906779661017e-06,
      "loss": 0.016,
      "step": 610
    },
    {
      "epoch": 0.2376047904191617,
      "grad_norm": 1.8113480806350708,
      "learning_rate": 9.747881355932204e-06,
      "loss": 0.016,
      "step": 620
    },
    {
      "epoch": 0.24143712574850298,
      "grad_norm": 1.0705426931381226,
      "learning_rate": 9.726694915254238e-06,
      "loss": 0.0149,
      "step": 630
    },
    {
      "epoch": 0.2452694610778443,
      "grad_norm": 1.5321799516677856,
      "learning_rate": 9.705508474576271e-06,
      "loss": 0.0158,
      "step": 640
    },
    {
      "epoch": 0.24910179640718563,
      "grad_norm": 2.2437734603881836,
      "learning_rate": 9.684322033898306e-06,
      "loss": 0.0176,
      "step": 650
    },
    {
      "epoch": 0.25293413173652696,
      "grad_norm": 0.9095500707626343,
      "learning_rate": 9.66313559322034e-06,
      "loss": 0.0153,
      "step": 660
    },
    {
      "epoch": 0.25676646706586825,
      "grad_norm": 1.800218105316162,
      "learning_rate": 9.641949152542374e-06,
      "loss": 0.0147,
      "step": 670
    },
    {
      "epoch": 0.2605988023952096,
      "grad_norm": 1.5202052593231201,
      "learning_rate": 9.620762711864408e-06,
      "loss": 0.0131,
      "step": 680
    },
    {
      "epoch": 0.2644311377245509,
      "grad_norm": 2.1265900135040283,
      "learning_rate": 9.59957627118644e-06,
      "loss": 0.0131,
      "step": 690
    },
    {
      "epoch": 0.2682634730538922,
      "grad_norm": 1.3656080961227417,
      "learning_rate": 9.578389830508476e-06,
      "loss": 0.0159,
      "step": 700
    },
    {
      "epoch": 0.27209580838323355,
      "grad_norm": 1.1013087034225464,
      "learning_rate": 9.557203389830509e-06,
      "loss": 0.0144,
      "step": 710
    },
    {
      "epoch": 0.27592814371257485,
      "grad_norm": 1.7320525646209717,
      "learning_rate": 9.536016949152544e-06,
      "loss": 0.013,
      "step": 720
    },
    {
      "epoch": 0.27976047904191614,
      "grad_norm": 0.7219749093055725,
      "learning_rate": 9.514830508474577e-06,
      "loss": 0.0134,
      "step": 730
    },
    {
      "epoch": 0.2835928143712575,
      "grad_norm": 1.914589285850525,
      "learning_rate": 9.49364406779661e-06,
      "loss": 0.0178,
      "step": 740
    },
    {
      "epoch": 0.2874251497005988,
      "grad_norm": 2.229616641998291,
      "learning_rate": 9.472457627118646e-06,
      "loss": 0.0163,
      "step": 750
    },
    {
      "epoch": 0.29125748502994014,
      "grad_norm": 0.7531014680862427,
      "learning_rate": 9.451271186440679e-06,
      "loss": 0.0097,
      "step": 760
    },
    {
      "epoch": 0.29508982035928144,
      "grad_norm": 1.1441811323165894,
      "learning_rate": 9.430084745762714e-06,
      "loss": 0.0121,
      "step": 770
    },
    {
      "epoch": 0.29892215568862274,
      "grad_norm": 1.730209469795227,
      "learning_rate": 9.408898305084746e-06,
      "loss": 0.0176,
      "step": 780
    },
    {
      "epoch": 0.3027544910179641,
      "grad_norm": 2.434473752975464,
      "learning_rate": 9.38771186440678e-06,
      "loss": 0.0136,
      "step": 790
    },
    {
      "epoch": 0.3065868263473054,
      "grad_norm": 1.3024921417236328,
      "learning_rate": 9.366525423728814e-06,
      "loss": 0.013,
      "step": 800
    },
    {
      "epoch": 0.3104191616766467,
      "grad_norm": 1.1504980325698853,
      "learning_rate": 9.345338983050847e-06,
      "loss": 0.0115,
      "step": 810
    },
    {
      "epoch": 0.31425149700598803,
      "grad_norm": 1.0408570766448975,
      "learning_rate": 9.324152542372882e-06,
      "loss": 0.0146,
      "step": 820
    },
    {
      "epoch": 0.31808383233532933,
      "grad_norm": 1.8448822498321533,
      "learning_rate": 9.302966101694915e-06,
      "loss": 0.0137,
      "step": 830
    },
    {
      "epoch": 0.3219161676646707,
      "grad_norm": 1.2359811067581177,
      "learning_rate": 9.28177966101695e-06,
      "loss": 0.0115,
      "step": 840
    },
    {
      "epoch": 0.325748502994012,
      "grad_norm": 1.6629658937454224,
      "learning_rate": 9.260593220338984e-06,
      "loss": 0.0185,
      "step": 850
    },
    {
      "epoch": 0.3295808383233533,
      "grad_norm": 1.3565815687179565,
      "learning_rate": 9.239406779661017e-06,
      "loss": 0.0105,
      "step": 860
    },
    {
      "epoch": 0.3334131736526946,
      "grad_norm": 0.47145602107048035,
      "learning_rate": 9.218220338983052e-06,
      "loss": 0.0124,
      "step": 870
    },
    {
      "epoch": 0.3372455089820359,
      "grad_norm": 1.452635645866394,
      "learning_rate": 9.197033898305085e-06,
      "loss": 0.0143,
      "step": 880
    },
    {
      "epoch": 0.3410778443113772,
      "grad_norm": 1.3402444124221802,
      "learning_rate": 9.17584745762712e-06,
      "loss": 0.0158,
      "step": 890
    },
    {
      "epoch": 0.3449101796407186,
      "grad_norm": 0.9056028127670288,
      "learning_rate": 9.154661016949154e-06,
      "loss": 0.0132,
      "step": 900
    },
    {
      "epoch": 0.34874251497005987,
      "grad_norm": 2.6174232959747314,
      "learning_rate": 9.133474576271187e-06,
      "loss": 0.0104,
      "step": 910
    },
    {
      "epoch": 0.3525748502994012,
      "grad_norm": 1.692396879196167,
      "learning_rate": 9.112288135593222e-06,
      "loss": 0.0139,
      "step": 920
    },
    {
      "epoch": 0.3564071856287425,
      "grad_norm": 1.1301361322402954,
      "learning_rate": 9.091101694915255e-06,
      "loss": 0.0114,
      "step": 930
    },
    {
      "epoch": 0.3602395209580838,
      "grad_norm": 0.9672715663909912,
      "learning_rate": 9.069915254237288e-06,
      "loss": 0.0128,
      "step": 940
    },
    {
      "epoch": 0.36407185628742517,
      "grad_norm": 2.274716377258301,
      "learning_rate": 9.048728813559323e-06,
      "loss": 0.0141,
      "step": 950
    },
    {
      "epoch": 0.36790419161676646,
      "grad_norm": 1.0660594701766968,
      "learning_rate": 9.027542372881357e-06,
      "loss": 0.0113,
      "step": 960
    },
    {
      "epoch": 0.37173652694610776,
      "grad_norm": 1.8312493562698364,
      "learning_rate": 9.006355932203392e-06,
      "loss": 0.0131,
      "step": 970
    },
    {
      "epoch": 0.3755688622754491,
      "grad_norm": 1.8305320739746094,
      "learning_rate": 8.985169491525423e-06,
      "loss": 0.0146,
      "step": 980
    },
    {
      "epoch": 0.3794011976047904,
      "grad_norm": 1.281570553779602,
      "learning_rate": 8.963983050847458e-06,
      "loss": 0.0145,
      "step": 990
    },
    {
      "epoch": 0.38323353293413176,
      "grad_norm": 0.8759114146232605,
      "learning_rate": 8.942796610169492e-06,
      "loss": 0.0118,
      "step": 1000
    },
    {
      "epoch": 0.38706586826347306,
      "grad_norm": 1.1629912853240967,
      "learning_rate": 8.921610169491527e-06,
      "loss": 0.0152,
      "step": 1010
    },
    {
      "epoch": 0.39089820359281435,
      "grad_norm": 1.3534737825393677,
      "learning_rate": 8.90042372881356e-06,
      "loss": 0.0133,
      "step": 1020
    },
    {
      "epoch": 0.3947305389221557,
      "grad_norm": 1.538173794746399,
      "learning_rate": 8.879237288135593e-06,
      "loss": 0.0153,
      "step": 1030
    },
    {
      "epoch": 0.398562874251497,
      "grad_norm": 0.8408123254776001,
      "learning_rate": 8.858050847457628e-06,
      "loss": 0.0117,
      "step": 1040
    },
    {
      "epoch": 0.4023952095808383,
      "grad_norm": 1.6027411222457886,
      "learning_rate": 8.836864406779662e-06,
      "loss": 0.0107,
      "step": 1050
    },
    {
      "epoch": 0.40622754491017965,
      "grad_norm": 1.9977298974990845,
      "learning_rate": 8.815677966101695e-06,
      "loss": 0.0151,
      "step": 1060
    },
    {
      "epoch": 0.41005988023952095,
      "grad_norm": 1.1906282901763916,
      "learning_rate": 8.79449152542373e-06,
      "loss": 0.0133,
      "step": 1070
    },
    {
      "epoch": 0.4138922155688623,
      "grad_norm": 1.9095267057418823,
      "learning_rate": 8.773305084745763e-06,
      "loss": 0.0141,
      "step": 1080
    },
    {
      "epoch": 0.4177245508982036,
      "grad_norm": 1.7171216011047363,
      "learning_rate": 8.752118644067798e-06,
      "loss": 0.0122,
      "step": 1090
    },
    {
      "epoch": 0.4215568862275449,
      "grad_norm": 2.373567819595337,
      "learning_rate": 8.730932203389831e-06,
      "loss": 0.0165,
      "step": 1100
    },
    {
      "epoch": 0.42538922155688624,
      "grad_norm": 1.757907509803772,
      "learning_rate": 8.709745762711865e-06,
      "loss": 0.0167,
      "step": 1110
    },
    {
      "epoch": 0.42922155688622754,
      "grad_norm": 3.8954083919525146,
      "learning_rate": 8.6885593220339e-06,
      "loss": 0.0205,
      "step": 1120
    },
    {
      "epoch": 0.43305389221556884,
      "grad_norm": 1.8802250623703003,
      "learning_rate": 8.667372881355933e-06,
      "loss": 0.0106,
      "step": 1130
    },
    {
      "epoch": 0.4368862275449102,
      "grad_norm": 1.3269938230514526,
      "learning_rate": 8.646186440677968e-06,
      "loss": 0.0101,
      "step": 1140
    },
    {
      "epoch": 0.4407185628742515,
      "grad_norm": 0.8917752504348755,
      "learning_rate": 8.625000000000001e-06,
      "loss": 0.011,
      "step": 1150
    },
    {
      "epoch": 0.44455089820359284,
      "grad_norm": 1.179168939590454,
      "learning_rate": 8.603813559322035e-06,
      "loss": 0.0132,
      "step": 1160
    },
    {
      "epoch": 0.44838323353293413,
      "grad_norm": 1.326897382736206,
      "learning_rate": 8.582627118644068e-06,
      "loss": 0.0105,
      "step": 1170
    },
    {
      "epoch": 0.45221556886227543,
      "grad_norm": 1.5681304931640625,
      "learning_rate": 8.561440677966101e-06,
      "loss": 0.0107,
      "step": 1180
    },
    {
      "epoch": 0.4560479041916168,
      "grad_norm": 0.7448801398277283,
      "learning_rate": 8.540254237288136e-06,
      "loss": 0.0105,
      "step": 1190
    },
    {
      "epoch": 0.4598802395209581,
      "grad_norm": 1.8620606660842896,
      "learning_rate": 8.51906779661017e-06,
      "loss": 0.0149,
      "step": 1200
    },
    {
      "epoch": 0.4637125748502994,
      "grad_norm": 0.7713643312454224,
      "learning_rate": 8.497881355932204e-06,
      "loss": 0.0103,
      "step": 1210
    },
    {
      "epoch": 0.4675449101796407,
      "grad_norm": 1.0003689527511597,
      "learning_rate": 8.476694915254238e-06,
      "loss": 0.0131,
      "step": 1220
    },
    {
      "epoch": 0.471377245508982,
      "grad_norm": 1.3732171058654785,
      "learning_rate": 8.455508474576271e-06,
      "loss": 0.0147,
      "step": 1230
    },
    {
      "epoch": 0.4752095808383234,
      "grad_norm": 0.6933659911155701,
      "learning_rate": 8.434322033898306e-06,
      "loss": 0.0117,
      "step": 1240
    },
    {
      "epoch": 0.47904191616766467,
      "grad_norm": 1.7515251636505127,
      "learning_rate": 8.41313559322034e-06,
      "loss": 0.0146,
      "step": 1250
    },
    {
      "epoch": 0.48287425149700597,
      "grad_norm": 1.186540126800537,
      "learning_rate": 8.391949152542374e-06,
      "loss": 0.0094,
      "step": 1260
    },
    {
      "epoch": 0.4867065868263473,
      "grad_norm": 0.7318383455276489,
      "learning_rate": 8.370762711864408e-06,
      "loss": 0.0107,
      "step": 1270
    },
    {
      "epoch": 0.4905389221556886,
      "grad_norm": 1.4842137098312378,
      "learning_rate": 8.349576271186441e-06,
      "loss": 0.016,
      "step": 1280
    },
    {
      "epoch": 0.4943712574850299,
      "grad_norm": 0.6988366842269897,
      "learning_rate": 8.328389830508476e-06,
      "loss": 0.0114,
      "step": 1290
    },
    {
      "epoch": 0.49820359281437127,
      "grad_norm": 2.0333776473999023,
      "learning_rate": 8.30720338983051e-06,
      "loss": 0.0139,
      "step": 1300
    },
    {
      "epoch": 0.5020359281437126,
      "grad_norm": 1.5975853204727173,
      "learning_rate": 8.286016949152543e-06,
      "loss": 0.0132,
      "step": 1310
    },
    {
      "epoch": 0.5058682634730539,
      "grad_norm": 1.2636547088623047,
      "learning_rate": 8.264830508474577e-06,
      "loss": 0.0131,
      "step": 1320
    },
    {
      "epoch": 0.5097005988023952,
      "grad_norm": 1.1417256593704224,
      "learning_rate": 8.24364406779661e-06,
      "loss": 0.0129,
      "step": 1330
    },
    {
      "epoch": 0.5135329341317365,
      "grad_norm": 1.3969192504882812,
      "learning_rate": 8.222457627118646e-06,
      "loss": 0.0077,
      "step": 1340
    },
    {
      "epoch": 0.5173652694610779,
      "grad_norm": 0.5558528900146484,
      "learning_rate": 8.201271186440679e-06,
      "loss": 0.0122,
      "step": 1350
    },
    {
      "epoch": 0.5211976047904192,
      "grad_norm": 1.0696184635162354,
      "learning_rate": 8.180084745762712e-06,
      "loss": 0.0099,
      "step": 1360
    },
    {
      "epoch": 0.5250299401197605,
      "grad_norm": 1.4563461542129517,
      "learning_rate": 8.158898305084746e-06,
      "loss": 0.0106,
      "step": 1370
    },
    {
      "epoch": 0.5288622754491018,
      "grad_norm": 0.995069146156311,
      "learning_rate": 8.13771186440678e-06,
      "loss": 0.0117,
      "step": 1380
    },
    {
      "epoch": 0.5326946107784432,
      "grad_norm": 1.1932368278503418,
      "learning_rate": 8.116525423728814e-06,
      "loss": 0.0119,
      "step": 1390
    },
    {
      "epoch": 0.5365269461077844,
      "grad_norm": 0.5526638031005859,
      "learning_rate": 8.095338983050847e-06,
      "loss": 0.0164,
      "step": 1400
    },
    {
      "epoch": 0.5403592814371257,
      "grad_norm": 1.2106008529663086,
      "learning_rate": 8.074152542372882e-06,
      "loss": 0.0117,
      "step": 1410
    },
    {
      "epoch": 0.5441916167664671,
      "grad_norm": 1.3328733444213867,
      "learning_rate": 8.052966101694916e-06,
      "loss": 0.0103,
      "step": 1420
    },
    {
      "epoch": 0.5480239520958083,
      "grad_norm": 1.2352383136749268,
      "learning_rate": 8.031779661016949e-06,
      "loss": 0.0132,
      "step": 1430
    },
    {
      "epoch": 0.5518562874251497,
      "grad_norm": 0.8932350277900696,
      "learning_rate": 8.010593220338984e-06,
      "loss": 0.0094,
      "step": 1440
    },
    {
      "epoch": 0.555688622754491,
      "grad_norm": 1.1897926330566406,
      "learning_rate": 7.989406779661017e-06,
      "loss": 0.0083,
      "step": 1450
    },
    {
      "epoch": 0.5595209580838323,
      "grad_norm": 3.584073543548584,
      "learning_rate": 7.968220338983052e-06,
      "loss": 0.0107,
      "step": 1460
    },
    {
      "epoch": 0.5633532934131736,
      "grad_norm": 2.2174160480499268,
      "learning_rate": 7.947033898305085e-06,
      "loss": 0.0143,
      "step": 1470
    },
    {
      "epoch": 0.567185628742515,
      "grad_norm": 1.3436836004257202,
      "learning_rate": 7.925847457627119e-06,
      "loss": 0.016,
      "step": 1480
    },
    {
      "epoch": 0.5710179640718562,
      "grad_norm": 2.2885513305664062,
      "learning_rate": 7.904661016949154e-06,
      "loss": 0.0097,
      "step": 1490
    },
    {
      "epoch": 0.5748502994011976,
      "grad_norm": 1.111118197441101,
      "learning_rate": 7.883474576271187e-06,
      "loss": 0.012,
      "step": 1500
    },
    {
      "epoch": 0.5786826347305389,
      "grad_norm": 1.1496660709381104,
      "learning_rate": 7.862288135593222e-06,
      "loss": 0.0112,
      "step": 1510
    },
    {
      "epoch": 0.5825149700598803,
      "grad_norm": 1.069238305091858,
      "learning_rate": 7.841101694915255e-06,
      "loss": 0.0143,
      "step": 1520
    },
    {
      "epoch": 0.5863473053892215,
      "grad_norm": 0.8200716972351074,
      "learning_rate": 7.819915254237289e-06,
      "loss": 0.0122,
      "step": 1530
    },
    {
      "epoch": 0.5901796407185629,
      "grad_norm": 0.911482036113739,
      "learning_rate": 7.798728813559324e-06,
      "loss": 0.0114,
      "step": 1540
    },
    {
      "epoch": 0.5940119760479042,
      "grad_norm": 0.8424916863441467,
      "learning_rate": 7.777542372881357e-06,
      "loss": 0.0111,
      "step": 1550
    },
    {
      "epoch": 0.5978443113772455,
      "grad_norm": 1.4030300378799438,
      "learning_rate": 7.75635593220339e-06,
      "loss": 0.0126,
      "step": 1560
    },
    {
      "epoch": 0.6016766467065868,
      "grad_norm": 1.4150911569595337,
      "learning_rate": 7.735169491525423e-06,
      "loss": 0.0083,
      "step": 1570
    },
    {
      "epoch": 0.6055089820359282,
      "grad_norm": 1.5504539012908936,
      "learning_rate": 7.713983050847458e-06,
      "loss": 0.0104,
      "step": 1580
    },
    {
      "epoch": 0.6093413173652694,
      "grad_norm": 2.837200403213501,
      "learning_rate": 7.692796610169492e-06,
      "loss": 0.0148,
      "step": 1590
    },
    {
      "epoch": 0.6131736526946108,
      "grad_norm": 1.2483636140823364,
      "learning_rate": 7.671610169491525e-06,
      "loss": 0.0124,
      "step": 1600
    },
    {
      "epoch": 0.6170059880239521,
      "grad_norm": 1.9936306476593018,
      "learning_rate": 7.65042372881356e-06,
      "loss": 0.0111,
      "step": 1610
    },
    {
      "epoch": 0.6208383233532934,
      "grad_norm": 1.3132163286209106,
      "learning_rate": 7.629237288135593e-06,
      "loss": 0.0122,
      "step": 1620
    },
    {
      "epoch": 0.6246706586826347,
      "grad_norm": 1.5041028261184692,
      "learning_rate": 7.6080508474576275e-06,
      "loss": 0.0114,
      "step": 1630
    },
    {
      "epoch": 0.6285029940119761,
      "grad_norm": 1.1258797645568848,
      "learning_rate": 7.586864406779662e-06,
      "loss": 0.0177,
      "step": 1640
    },
    {
      "epoch": 0.6323353293413174,
      "grad_norm": 1.5175997018814087,
      "learning_rate": 7.565677966101696e-06,
      "loss": 0.0112,
      "step": 1650
    },
    {
      "epoch": 0.6361676646706587,
      "grad_norm": 0.8304109573364258,
      "learning_rate": 7.544491525423729e-06,
      "loss": 0.0132,
      "step": 1660
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6195735931396484,
      "learning_rate": 7.523305084745763e-06,
      "loss": 0.008,
      "step": 1670
    },
    {
      "epoch": 0.6438323353293414,
      "grad_norm": 0.660678505897522,
      "learning_rate": 7.502118644067797e-06,
      "loss": 0.0095,
      "step": 1680
    },
    {
      "epoch": 0.6476646706586826,
      "grad_norm": 1.7854886054992676,
      "learning_rate": 7.4809322033898315e-06,
      "loss": 0.0154,
      "step": 1690
    },
    {
      "epoch": 0.651497005988024,
      "grad_norm": 1.3273577690124512,
      "learning_rate": 7.459745762711866e-06,
      "loss": 0.0094,
      "step": 1700
    },
    {
      "epoch": 0.6553293413173653,
      "grad_norm": 0.7021352648735046,
      "learning_rate": 7.438559322033899e-06,
      "loss": 0.0097,
      "step": 1710
    },
    {
      "epoch": 0.6591616766467066,
      "grad_norm": 1.966849446296692,
      "learning_rate": 7.417372881355933e-06,
      "loss": 0.0104,
      "step": 1720
    },
    {
      "epoch": 0.6629940119760479,
      "grad_norm": 1.0740894079208374,
      "learning_rate": 7.396186440677967e-06,
      "loss": 0.0098,
      "step": 1730
    },
    {
      "epoch": 0.6668263473053893,
      "grad_norm": 1.1839215755462646,
      "learning_rate": 7.375000000000001e-06,
      "loss": 0.0173,
      "step": 1740
    },
    {
      "epoch": 0.6706586826347305,
      "grad_norm": 1.8997067213058472,
      "learning_rate": 7.353813559322035e-06,
      "loss": 0.0104,
      "step": 1750
    },
    {
      "epoch": 0.6744910179640718,
      "grad_norm": 1.172722339630127,
      "learning_rate": 7.332627118644068e-06,
      "loss": 0.0118,
      "step": 1760
    },
    {
      "epoch": 0.6783233532934132,
      "grad_norm": 1.0804802179336548,
      "learning_rate": 7.311440677966102e-06,
      "loss": 0.0125,
      "step": 1770
    },
    {
      "epoch": 0.6821556886227544,
      "grad_norm": 1.2899483442306519,
      "learning_rate": 7.290254237288135e-06,
      "loss": 0.0113,
      "step": 1780
    },
    {
      "epoch": 0.6859880239520958,
      "grad_norm": 1.2753748893737793,
      "learning_rate": 7.2690677966101696e-06,
      "loss": 0.0076,
      "step": 1790
    },
    {
      "epoch": 0.6898203592814371,
      "grad_norm": 1.0572164058685303,
      "learning_rate": 7.247881355932204e-06,
      "loss": 0.0146,
      "step": 1800
    },
    {
      "epoch": 0.6936526946107785,
      "grad_norm": 0.9690531492233276,
      "learning_rate": 7.226694915254238e-06,
      "loss": 0.0106,
      "step": 1810
    },
    {
      "epoch": 0.6974850299401197,
      "grad_norm": 0.37759602069854736,
      "learning_rate": 7.205508474576271e-06,
      "loss": 0.0087,
      "step": 1820
    },
    {
      "epoch": 0.7013173652694611,
      "grad_norm": 1.595554232597351,
      "learning_rate": 7.184322033898305e-06,
      "loss": 0.0121,
      "step": 1830
    },
    {
      "epoch": 0.7051497005988024,
      "grad_norm": 5.621078014373779,
      "learning_rate": 7.1631355932203394e-06,
      "loss": 0.0097,
      "step": 1840
    },
    {
      "epoch": 0.7089820359281437,
      "grad_norm": 0.313541978597641,
      "learning_rate": 7.141949152542374e-06,
      "loss": 0.0105,
      "step": 1850
    },
    {
      "epoch": 0.712814371257485,
      "grad_norm": 0.7415321469306946,
      "learning_rate": 7.120762711864408e-06,
      "loss": 0.013,
      "step": 1860
    },
    {
      "epoch": 0.7166467065868264,
      "grad_norm": 0.7116707563400269,
      "learning_rate": 7.099576271186441e-06,
      "loss": 0.0089,
      "step": 1870
    },
    {
      "epoch": 0.7204790419161676,
      "grad_norm": 2.017526388168335,
      "learning_rate": 7.078389830508475e-06,
      "loss": 0.0123,
      "step": 1880
    },
    {
      "epoch": 0.724311377245509,
      "grad_norm": 1.4061137437820435,
      "learning_rate": 7.057203389830509e-06,
      "loss": 0.0077,
      "step": 1890
    },
    {
      "epoch": 0.7281437125748503,
      "grad_norm": 1.1427956819534302,
      "learning_rate": 7.0360169491525435e-06,
      "loss": 0.0074,
      "step": 1900
    },
    {
      "epoch": 0.7319760479041916,
      "grad_norm": 4.021897792816162,
      "learning_rate": 7.014830508474577e-06,
      "loss": 0.0109,
      "step": 1910
    },
    {
      "epoch": 0.7358083832335329,
      "grad_norm": 0.36056721210479736,
      "learning_rate": 6.993644067796611e-06,
      "loss": 0.0102,
      "step": 1920
    },
    {
      "epoch": 0.7396407185628743,
      "grad_norm": 0.8428685665130615,
      "learning_rate": 6.972457627118645e-06,
      "loss": 0.0084,
      "step": 1930
    },
    {
      "epoch": 0.7434730538922155,
      "grad_norm": 0.6571751236915588,
      "learning_rate": 6.951271186440679e-06,
      "loss": 0.0075,
      "step": 1940
    },
    {
      "epoch": 0.7473053892215569,
      "grad_norm": 1.2280906438827515,
      "learning_rate": 6.930084745762713e-06,
      "loss": 0.0141,
      "step": 1950
    },
    {
      "epoch": 0.7511377245508982,
      "grad_norm": 0.9602510333061218,
      "learning_rate": 6.908898305084746e-06,
      "loss": 0.0092,
      "step": 1960
    },
    {
      "epoch": 0.7549700598802396,
      "grad_norm": 0.7274242043495178,
      "learning_rate": 6.88771186440678e-06,
      "loss": 0.0061,
      "step": 1970
    },
    {
      "epoch": 0.7588023952095808,
      "grad_norm": 1.538460373878479,
      "learning_rate": 6.866525423728814e-06,
      "loss": 0.0106,
      "step": 1980
    },
    {
      "epoch": 0.7626347305389222,
      "grad_norm": 0.8836348056793213,
      "learning_rate": 6.845338983050847e-06,
      "loss": 0.0119,
      "step": 1990
    },
    {
      "epoch": 0.7664670658682635,
      "grad_norm": 1.2621474266052246,
      "learning_rate": 6.8241525423728815e-06,
      "loss": 0.0118,
      "step": 2000
    },
    {
      "epoch": 0.7664670658682635,
      "eval_loss": 0.006921224296092987,
      "eval_runtime": 15262.8627,
      "eval_samples_per_second": 1.368,
      "eval_steps_per_second": 0.171,
      "eval_wer": 0.7323685598172008,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 5220,
  "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": 6.531871408128e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}