{
  "best_metric": 3.4478375911712646,
  "best_model_checkpoint": "checkpoints/it5-small/checkpoint-17663",
  "epoch": 12.753068592057762,
  "eval_steps": 1039,
  "global_step": 17663,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.14945848375451262,
      "eval_g2l_cer": 293.7627,
      "eval_g2l_gen_len": 13.9075,
      "eval_g2l_rouge1": 17.4489,
      "eval_g2l_rouge2": 8.198,
      "eval_g2l_rougeL": 17.2001,
      "eval_g2l_rougeLsum": 17.1915,
      "eval_l2ex_cer": 136.8329,
      "eval_l2ex_gen_len": 54.1737,
      "eval_l2ex_rouge1": 17.812,
      "eval_l2ex_rouge2": 6.7548,
      "eval_l2ex_rougeL": 16.7136,
      "eval_l2ex_rougeLsum": 16.7013,
      "eval_l2g_cer": 154.2791,
      "eval_l2g_gen_len": 50.8855,
      "eval_l2g_rouge1": 13.7399,
      "eval_l2g_rouge2": 4.278,
      "eval_l2g_rougeL": 13.0474,
      "eval_l2g_rougeLsum": 13.0446,
      "eval_loss": 4.6176557540893555,
      "eval_runtime": 109.9862,
      "eval_samples_per_second": 90.202,
      "eval_steps_per_second": 0.709,
      "step": 207
    },
    {
      "epoch": 0.15018050541516245,
      "grad_norm": 99.67880249023438,
      "learning_rate": 0.0002003853564547206,
      "loss": 5.9292,
      "step": 208
    },
    {
      "epoch": 0.3003610108303249,
      "grad_norm": 195.41990661621094,
      "learning_rate": 0.0004007707129094412,
      "loss": 4.6698,
      "step": 416
    },
    {
      "epoch": 0.4505415162454874,
      "grad_norm": 274.79534912109375,
      "learning_rate": 0.0006011560693641619,
      "loss": 4.515,
      "step": 624
    },
    {
      "epoch": 0.6007220216606498,
      "grad_norm": 210.73873901367188,
      "learning_rate": 0.0008015414258188824,
      "loss": 4.4776,
      "step": 832
    },
    {
      "epoch": 0.7501805054151625,
      "eval_g2l_cer": 61.7538,
      "eval_g2l_gen_len": 3.2726,
      "eval_g2l_rouge1": 27.6616,
      "eval_g2l_rouge2": 19.0238,
      "eval_g2l_rougeL": 27.5754,
      "eval_g2l_rougeLsum": 27.5869,
      "eval_l2ex_cer": 88.1951,
      "eval_l2ex_gen_len": 16.4502,
      "eval_l2ex_rouge1": 28.4409,
      "eval_l2ex_rouge2": 14.714,
      "eval_l2ex_rougeL": 26.3524,
      "eval_l2ex_rougeLsum": 26.38,
      "eval_l2g_cer": 81.2622,
      "eval_l2g_gen_len": 12.7833,
      "eval_l2g_rouge1": 28.7842,
      "eval_l2g_rouge2": 15.662,
      "eval_l2g_rougeL": 27.3798,
      "eval_l2g_rougeLsum": 27.3872,
      "eval_loss": 4.086117744445801,
      "eval_runtime": 92.0779,
      "eval_samples_per_second": 107.746,
      "eval_steps_per_second": 0.847,
      "step": 1039
    },
    {
      "epoch": 0.7509025270758123,
      "grad_norm": 280.9309387207031,
      "learning_rate": 0.0009999999746484521,
      "loss": 4.4853,
      "step": 1040
    },
    {
      "epoch": 0.9010830324909748,
      "grad_norm": 278.86737060546875,
      "learning_rate": 0.000999720525233547,
      "loss": 4.4543,
      "step": 1248
    },
    {
      "epoch": 1.0512635379061372,
      "grad_norm": 347.6966857910156,
      "learning_rate": 0.0009988930279319957,
      "loss": 4.4011,
      "step": 1456
    },
    {
      "epoch": 1.2014440433212996,
      "grad_norm": 388.53009033203125,
      "learning_rate": 0.0009975183907215961,
      "loss": 4.3391,
      "step": 1664
    },
    {
      "epoch": 1.3516245487364622,
      "grad_norm": 374.0000915527344,
      "learning_rate": 0.0009955981219336306,
      "loss": 4.2954,
      "step": 1872
    },
    {
      "epoch": 1.500361010830325,
      "eval_g2l_cer": 60.8741,
      "eval_g2l_gen_len": 5.7528,
      "eval_g2l_rouge1": 34.1975,
      "eval_g2l_rouge2": 23.0519,
      "eval_g2l_rougeL": 33.7746,
      "eval_g2l_rougeLsum": 33.7503,
      "eval_l2ex_cer": 90.0364,
      "eval_l2ex_gen_len": 38.0339,
      "eval_l2ex_rouge1": 27.7928,
      "eval_l2ex_rouge2": 12.8688,
      "eval_l2ex_rougeL": 25.3603,
      "eval_l2ex_rougeLsum": 25.3866,
      "eval_l2g_cer": 95.9503,
      "eval_l2g_gen_len": 38.6107,
      "eval_l2g_rouge1": 31.3843,
      "eval_l2g_rouge2": 20.1215,
      "eval_l2g_rougeL": 30.0596,
      "eval_l2g_rougeLsum": 30.1295,
      "eval_loss": 4.214888095855713,
      "eval_runtime": 102.9376,
      "eval_samples_per_second": 96.379,
      "eval_steps_per_second": 0.758,
      "step": 2078
    },
    {
      "epoch": 1.5018050541516246,
      "grad_norm": 282.3067932128906,
      "learning_rate": 0.0009931343285978396,
      "loss": 4.2676,
      "step": 2080
    },
    {
      "epoch": 1.651985559566787,
      "grad_norm": 429.1705017089844,
      "learning_rate": 0.000990129714130466,
      "loss": 4.2425,
      "step": 2288
    },
    {
      "epoch": 1.8021660649819493,
      "grad_norm": 342.697509765625,
      "learning_rate": 0.0009865875753679114,
      "loss": 4.2198,
      "step": 2496
    },
    {
      "epoch": 1.952346570397112,
      "grad_norm": 287.99896240234375,
      "learning_rate": 0.0009825117989492565,
      "loss": 4.1951,
      "step": 2704
    },
    {
      "epoch": 2.1025270758122745,
      "grad_norm": 329.3447570800781,
      "learning_rate": 0.0009779068570516165,
      "loss": 4.137,
      "step": 2912
    },
    {
      "epoch": 2.2505415162454874,
      "eval_g2l_cer": 57.0175,
      "eval_g2l_gen_len": 3.4026,
      "eval_g2l_rouge1": 33.2248,
      "eval_g2l_rouge2": 24.7912,
      "eval_g2l_rougeL": 33.1481,
      "eval_g2l_rougeLsum": 33.1288,
      "eval_l2ex_cer": 83.9364,
      "eval_l2ex_gen_len": 17.4142,
      "eval_l2ex_rouge1": 30.4747,
      "eval_l2ex_rouge2": 15.2972,
      "eval_l2ex_rougeL": 27.443,
      "eval_l2ex_rougeLsum": 27.5452,
      "eval_l2g_cer": 80.6606,
      "eval_l2g_gen_len": 14.6152,
      "eval_l2g_rouge1": 34.2385,
      "eval_l2g_rouge2": 22.2422,
      "eval_l2g_rougeL": 32.5658,
      "eval_l2g_rougeLsum": 32.6354,
      "eval_loss": 3.8766722679138184,
      "eval_runtime": 93.2118,
      "eval_samples_per_second": 106.435,
      "eval_steps_per_second": 0.837,
      "step": 3117
    },
    {
      "epoch": 2.252707581227437,
      "grad_norm": 373.11468505859375,
      "learning_rate": 0.00097277780248301,
      "loss": 4.1204,
      "step": 3120
    },
    {
      "epoch": 2.4028880866425992,
      "grad_norm": 332.1923522949219,
      "learning_rate": 0.0009671302631381275,
      "loss": 4.1156,
      "step": 3328
    },
    {
      "epoch": 2.5530685920577616,
      "grad_norm": 439.4508361816406,
      "learning_rate": 0.0009609704358230791,
      "loss": 4.0976,
      "step": 3536
    },
    {
      "epoch": 2.7032490974729244,
      "grad_norm": 283.86175537109375,
      "learning_rate": 0.0009543050794559023,
      "loss": 4.0829,
      "step": 3744
    },
    {
      "epoch": 2.853429602888087,
      "grad_norm": 373.6076965332031,
      "learning_rate": 0.0009471415076502864,
      "loss": 4.0567,
      "step": 3952
    },
    {
      "epoch": 3.00072202166065,
      "eval_g2l_cer": 54.3375,
      "eval_g2l_gen_len": 5.1478,
      "eval_g2l_rouge1": 37.1423,
      "eval_g2l_rouge2": 25.8415,
      "eval_g2l_rougeL": 36.8263,
      "eval_g2l_rougeLsum": 36.799,
      "eval_l2ex_cer": 88.2942,
      "eval_l2ex_gen_len": 35.3597,
      "eval_l2ex_rouge1": 28.7362,
      "eval_l2ex_rouge2": 13.0176,
      "eval_l2ex_rougeL": 26.1133,
      "eval_l2ex_rougeLsum": 26.1476,
      "eval_l2g_cer": 85.0531,
      "eval_l2g_gen_len": 29.6997,
      "eval_l2g_rouge1": 35.6931,
      "eval_l2g_rouge2": 23.4243,
      "eval_l2g_rougeL": 34.012,
      "eval_l2g_rougeLsum": 34.0684,
      "eval_loss": 3.9321906566619873,
      "eval_runtime": 100.3303,
      "eval_samples_per_second": 98.883,
      "eval_steps_per_second": 0.777,
      "step": 4156
    },
    {
      "epoch": 3.003610108303249,
      "grad_norm": 456.342529296875,
      "learning_rate": 0.0009394875806906537,
      "loss": 4.045,
      "step": 4160
    },
    {
      "epoch": 3.1537906137184115,
      "grad_norm": 336.9688720703125,
      "learning_rate": 0.0009313516969074018,
      "loss": 4.0002,
      "step": 4368
    },
    {
      "epoch": 3.303971119133574,
      "grad_norm": 354.8324890136719,
      "learning_rate": 0.0009227427834617708,
      "loss": 3.9884,
      "step": 4576
    },
    {
      "epoch": 3.4541516245487367,
      "grad_norm": 341.2556457519531,
      "learning_rate": 0.0009136702865504465,
      "loss": 3.9819,
      "step": 4784
    },
    {
      "epoch": 3.604332129963899,
      "grad_norm": 348.25225830078125,
      "learning_rate": 0.0009041441610406489,
      "loss": 3.9644,
      "step": 4992
    },
    {
      "epoch": 3.7509025270758123,
      "eval_g2l_cer": 54.7671,
      "eval_g2l_gen_len": 3.3446,
      "eval_g2l_rouge1": 34.9931,
      "eval_g2l_rouge2": 26.209,
      "eval_g2l_rougeL": 34.9234,
      "eval_g2l_rougeLsum": 34.913,
      "eval_l2ex_cer": 85.5203,
      "eval_l2ex_gen_len": 20.1264,
      "eval_l2ex_rouge1": 30.9385,
      "eval_l2ex_rouge2": 15.6373,
      "eval_l2ex_rougeL": 28.096,
      "eval_l2ex_rougeLsum": 28.1379,
      "eval_l2g_cer": 77.9784,
      "eval_l2g_gen_len": 14.0847,
      "eval_l2g_rouge1": 35.0014,
      "eval_l2g_rouge2": 23.2642,
      "eval_l2g_rougeL": 33.3426,
      "eval_l2g_rougeLsum": 33.385,
      "eval_loss": 3.767300844192505,
      "eval_runtime": 93.092,
      "eval_samples_per_second": 106.572,
      "eval_steps_per_second": 0.838,
      "step": 5195
    },
    {
      "epoch": 3.7545126353790614,
      "grad_norm": 282.4262390136719,
      "learning_rate": 0.0008941748595470763,
      "loss": 3.9561,
      "step": 5200
    },
    {
      "epoch": 3.904693140794224,
      "grad_norm": 309.5865783691406,
      "learning_rate": 0.0008837733209626942,
      "loss": 3.9365,
      "step": 5408
    },
    {
      "epoch": 4.054873646209386,
      "grad_norm": 314.7027893066406,
      "learning_rate": 0.0008729509584559499,
      "loss": 3.9144,
      "step": 5616
    },
    {
      "epoch": 4.205054151624549,
      "grad_norm": 354.8764953613281,
      "learning_rate": 0.0008617196469475859,
      "loss": 3.871,
      "step": 5824
    },
    {
      "epoch": 4.355234657039711,
      "grad_norm": 318.5345458984375,
      "learning_rate": 0.0008500917100807916,
      "loss": 3.8729,
      "step": 6032
    },
    {
      "epoch": 4.501083032490975,
      "eval_g2l_cer": 54.0839,
      "eval_g2l_gen_len": 5.259,
      "eval_g2l_rouge1": 37.7304,
      "eval_g2l_rouge2": 26.5008,
      "eval_g2l_rougeL": 37.4145,
      "eval_g2l_rougeLsum": 37.389,
      "eval_l2ex_cer": 87.7795,
      "eval_l2ex_gen_len": 34.444,
      "eval_l2ex_rouge1": 29.4565,
      "eval_l2ex_rouge2": 13.4704,
      "eval_l2ex_rougeL": 26.584,
      "eval_l2ex_rougeLsum": 26.6084,
      "eval_l2g_cer": 84.0775,
      "eval_l2g_gen_len": 27.3501,
      "eval_l2g_rouge1": 36.4264,
      "eval_l2g_rouge2": 23.9948,
      "eval_l2g_rougeL": 34.6234,
      "eval_l2g_rougeLsum": 34.6872,
      "eval_loss": 3.814581871032715,
      "eval_runtime": 99.5293,
      "eval_samples_per_second": 99.679,
      "eval_steps_per_second": 0.784,
      "step": 6234
    },
    {
      "epoch": 4.505415162454874,
      "grad_norm": 302.9512634277344,
      "learning_rate": 0.0008380799066989898,
      "loss": 3.8677,
      "step": 6240
    },
    {
      "epoch": 4.6555956678700365,
      "grad_norm": 250.16696166992188,
      "learning_rate": 0.0008256974168460981,
      "loss": 3.8378,
      "step": 6448
    },
    {
      "epoch": 4.8057761732851985,
      "grad_norm": 348.5408935546875,
      "learning_rate": 0.0008129578273046243,
      "loss": 3.852,
      "step": 6656
    },
    {
      "epoch": 4.955956678700361,
      "grad_norm": 355.0418701171875,
      "learning_rate": 0.0007998751166874639,
      "loss": 3.845,
      "step": 6864
    },
    {
      "epoch": 5.106137184115523,
      "grad_norm": 349.153564453125,
      "learning_rate": 0.0007864636400997593,
      "loss": 3.799,
      "step": 7072
    },
    {
      "epoch": 5.251263537906137,
      "eval_g2l_cer": 52.3635,
      "eval_g2l_gen_len": 3.512,
      "eval_g2l_rouge1": 37.4372,
      "eval_g2l_rouge2": 27.515,
      "eval_g2l_rougeL": 37.3252,
      "eval_g2l_rougeLsum": 37.3643,
      "eval_l2ex_cer": 81.2205,
      "eval_l2ex_gen_len": 18.2107,
      "eval_l2ex_rouge1": 32.8709,
      "eval_l2ex_rouge2": 16.7456,
      "eval_l2ex_rougeL": 29.6388,
      "eval_l2ex_rougeLsum": 29.717,
      "eval_l2g_cer": 73.2203,
      "eval_l2g_gen_len": 13.7837,
      "eval_l2g_rouge1": 37.1711,
      "eval_l2g_rouge2": 24.5593,
      "eval_l2g_rougeL": 35.3326,
      "eval_l2g_rougeLsum": 35.3608,
      "eval_loss": 3.6153602600097656,
      "eval_runtime": 92.481,
      "eval_samples_per_second": 107.276,
      "eval_steps_per_second": 0.843,
      "step": 7273
    },
    {
      "epoch": 5.256317689530686,
      "grad_norm": 313.629638671875,
      "learning_rate": 0.0007727381133876502,
      "loss": 3.7717,
      "step": 7280
    },
    {
      "epoch": 5.406498194945849,
      "grad_norm": 366.9100036621094,
      "learning_rate": 0.0007587135969911982,
      "loss": 3.7698,
      "step": 7488
    },
    {
      "epoch": 5.556678700361011,
      "grad_norm": 405.91522216796875,
      "learning_rate": 0.000744405479419202,
      "loss": 3.7686,
      "step": 7696
    },
    {
      "epoch": 5.706859205776174,
      "grad_norm": 323.0384521484375,
      "learning_rate": 0.000729829460364038,
      "loss": 3.7567,
      "step": 7904
    },
    {
      "epoch": 5.8570397111913355,
      "grad_norm": 236.5808563232422,
      "learning_rate": 0.0007150015334750512,
      "loss": 3.7504,
      "step": 8112
    },
    {
      "epoch": 6.0014440433213,
      "eval_g2l_cer": 52.8799,
      "eval_g2l_gen_len": 4.9953,
      "eval_g2l_rouge1": 38.9469,
      "eval_g2l_rouge2": 27.4705,
      "eval_g2l_rougeL": 38.7013,
      "eval_g2l_rougeLsum": 38.6638,
      "eval_l2ex_cer": 83.8059,
      "eval_l2ex_gen_len": 29.591,
      "eval_l2ex_rouge1": 31.5224,
      "eval_l2ex_rouge2": 15.0063,
      "eval_l2ex_rougeL": 28.1947,
      "eval_l2ex_rougeLsum": 28.2288,
      "eval_l2g_cer": 80.9099,
      "eval_l2g_gen_len": 24.6797,
      "eval_l2g_rouge1": 37.5363,
      "eval_l2g_rouge2": 24.8417,
      "eval_l2g_rougeL": 35.6194,
      "eval_l2g_rougeLsum": 35.6546,
      "eval_loss": 3.7020649909973145,
      "eval_runtime": 98.5066,
      "eval_samples_per_second": 100.714,
      "eval_steps_per_second": 0.792,
      "step": 8312
    },
    {
      "epoch": 6.007220216606498,
      "grad_norm": 265.8843994140625,
      "learning_rate": 0.000699937968809401,
      "loss": 3.7525,
      "step": 8320
    },
    {
      "epoch": 6.15740072202166,
      "grad_norm": 291.2859802246094,
      "learning_rate": 0.0006846552949796146,
      "loss": 3.6905,
      "step": 8528
    },
    {
      "epoch": 6.307581227436823,
      "grad_norm": 268.9134826660156,
      "learning_rate": 0.0006691702810174398,
      "loss": 3.6939,
      "step": 8736
    },
    {
      "epoch": 6.457761732851986,
      "grad_norm": 292.39154052734375,
      "learning_rate": 0.0006534999179738952,
      "loss": 3.7019,
      "step": 8944
    },
    {
      "epoch": 6.607942238267148,
      "grad_norm": 298.4355773925781,
      "learning_rate": 0.0006376614002757093,
      "loss": 3.6892,
      "step": 9152
    },
    {
      "epoch": 6.751624548736462,
      "eval_g2l_cer": 51.8358,
      "eval_g2l_gen_len": 3.5613,
      "eval_g2l_rouge1": 37.7745,
      "eval_g2l_rouge2": 27.9843,
      "eval_g2l_rougeL": 37.6788,
      "eval_g2l_rougeLsum": 37.6728,
      "eval_l2ex_cer": 84.3919,
      "eval_l2ex_gen_len": 20.5961,
      "eval_l2ex_rouge1": 32.2651,
      "eval_l2ex_rouge2": 16.2957,
      "eval_l2ex_rougeL": 28.9932,
      "eval_l2ex_rougeLsum": 29.0868,
      "eval_l2g_cer": 75.9064,
      "eval_l2g_gen_len": 14.9821,
      "eval_l2g_rouge1": 37.3147,
      "eval_l2g_rouge2": 24.7819,
      "eval_l2g_rougeL": 35.4462,
      "eval_l2g_rougeLsum": 35.4671,
      "eval_loss": 3.5767345428466797,
      "eval_runtime": 93.796,
      "eval_samples_per_second": 105.772,
      "eval_steps_per_second": 0.832,
      "step": 9351
    },
    {
      "epoch": 6.758122743682311,
      "grad_norm": 257.7994689941406,
      "learning_rate": 0.0006216721068586019,
      "loss": 3.6827,
      "step": 9360
    },
    {
      "epoch": 6.908303249097473,
      "grad_norm": 264.89263916015625,
      "learning_rate": 0.0006055495820981136,
      "loss": 3.6662,
      "step": 9568
    },
    {
      "epoch": 7.058483754512635,
      "grad_norm": 289.5307312011719,
      "learning_rate": 0.0005893115165589047,
      "loss": 3.6482,
      "step": 9776
    },
    {
      "epoch": 7.208664259927798,
      "grad_norm": 254.08291625976562,
      "learning_rate": 0.0005729757275836455,
      "loss": 3.6198,
      "step": 9984
    },
    {
      "epoch": 7.35884476534296,
      "grad_norm": 181.30072021484375,
      "learning_rate": 0.0005565601397428012,
      "loss": 3.6104,
      "step": 10192
    },
    {
      "epoch": 7.501805054151625,
      "eval_g2l_cer": 52.672,
      "eval_g2l_gen_len": 4.8856,
      "eval_g2l_rouge1": 39.792,
      "eval_g2l_rouge2": 28.4504,
      "eval_g2l_rougeL": 39.5411,
      "eval_g2l_rougeLsum": 39.532,
      "eval_l2ex_cer": 87.7735,
      "eval_l2ex_gen_len": 32.8952,
      "eval_l2ex_rouge1": 30.3306,
      "eval_l2ex_rouge2": 14.2701,
      "eval_l2ex_rougeL": 27.0806,
      "eval_l2ex_rougeLsum": 27.0923,
      "eval_l2g_cer": 83.3818,
      "eval_l2g_gen_len": 25.6912,
      "eval_l2g_rouge1": 37.5685,
      "eval_l2g_rouge2": 25.026,
      "eval_l2g_rougeL": 35.6593,
      "eval_l2g_rougeLsum": 35.7335,
      "eval_loss": 3.6247808933258057,
      "eval_runtime": 97.6949,
      "eval_samples_per_second": 101.551,
      "eval_steps_per_second": 0.798,
      "step": 10390
    },
    {
      "epoch": 7.509025270758123,
      "grad_norm": 213.4962158203125,
      "learning_rate": 0.0005400827651667568,
      "loss": 3.6116,
      "step": 10400
    },
    {
      "epoch": 7.659205776173286,
      "grad_norm": 200.51080322265625,
      "learning_rate": 0.0005235616837818695,
      "loss": 3.613,
      "step": 10608
    },
    {
      "epoch": 7.809386281588448,
      "grad_norm": 210.73484802246094,
      "learning_rate": 0.0005070150234721305,
      "loss": 3.5998,
      "step": 10816
    },
    {
      "epoch": 7.95956678700361,
      "grad_norm": 245.5376434326172,
      "learning_rate": 0.0004904609401882021,
      "loss": 3.6003,
      "step": 11024
    },
    {
      "epoch": 8.109747292418772,
      "grad_norm": 234.57310485839844,
      "learning_rate": 0.00047391759802566346,
      "loss": 3.5769,
      "step": 11232
    },
    {
      "epoch": 8.251985559566787,
      "eval_g2l_cer": 51.2806,
      "eval_g2l_gen_len": 4.0062,
      "eval_g2l_rouge1": 39.5838,
      "eval_g2l_rouge2": 28.8731,
      "eval_g2l_rougeL": 39.3964,
      "eval_g2l_rougeLsum": 39.3929,
      "eval_l2ex_cer": 81.177,
      "eval_l2ex_gen_len": 21.2641,
      "eval_l2ex_rouge1": 33.1446,
      "eval_l2ex_rouge2": 16.6249,
      "eval_l2ex_rougeL": 29.723,
      "eval_l2ex_rougeLsum": 29.7703,
      "eval_l2g_cer": 73.5184,
      "eval_l2g_gen_len": 15.4612,
      "eval_l2g_rouge1": 38.7817,
      "eval_l2g_rouge2": 25.5669,
      "eval_l2g_rougeL": 36.728,
      "eval_l2g_rougeLsum": 36.7794,
      "eval_loss": 3.502086639404297,
      "eval_runtime": 94.3872,
      "eval_samples_per_second": 105.11,
      "eval_steps_per_second": 0.826,
      "step": 11429
    },
    {
      "epoch": 8.259927797833935,
      "grad_norm": 212.4471435546875,
      "learning_rate": 0.00045740314929431514,
      "loss": 3.5588,
      "step": 11440
    },
    {
      "epoch": 8.410108303249098,
      "grad_norm": 227.54701232910156,
      "learning_rate": 0.000440935714600417,
      "loss": 3.5575,
      "step": 11648
    },
    {
      "epoch": 8.56028880866426,
      "grad_norm": 175.19906616210938,
      "learning_rate": 0.00042453336296371426,
      "loss": 3.5425,
      "step": 11856
    },
    {
      "epoch": 8.710469314079422,
      "grad_norm": 197.85183715820312,
      "learning_rate": 0.0004082140919910653,
      "loss": 3.5516,
      "step": 12064
    },
    {
      "epoch": 8.860649819494585,
      "grad_norm": 174.2976531982422,
      "learning_rate": 0.00039199580812842894,
      "loss": 3.5262,
      "step": 12272
    },
    {
      "epoch": 9.00216606498195,
      "eval_g2l_cer": 52.3453,
      "eval_g2l_gen_len": 4.6585,
      "eval_g2l_rouge1": 39.8411,
      "eval_g2l_rouge2": 28.7875,
      "eval_g2l_rougeL": 39.7115,
      "eval_g2l_rougeLsum": 39.6779,
      "eval_l2ex_cer": 84.2856,
      "eval_l2ex_gen_len": 28.0884,
      "eval_l2ex_rouge1": 31.9562,
      "eval_l2ex_rouge2": 15.2366,
      "eval_l2ex_rougeL": 28.4418,
      "eval_l2ex_rougeLsum": 28.4937,
      "eval_l2g_cer": 82.0714,
      "eval_l2g_gen_len": 22.9518,
      "eval_l2g_rouge1": 38.3811,
      "eval_l2g_rouge2": 25.3667,
      "eval_l2g_rougeL": 36.2921,
      "eval_l2g_rougeLsum": 36.3599,
      "eval_loss": 3.542821168899536,
      "eval_runtime": 97.8166,
      "eval_samples_per_second": 101.424,
      "eval_steps_per_second": 0.797,
      "step": 12468
    },
    {
      "epoch": 9.010830324909747,
      "grad_norm": 148.35972595214844,
      "learning_rate": 0.00037589630701287847,
      "loss": 3.5412,
      "step": 12480
    },
    {
      "epoch": 9.16101083032491,
      "grad_norm": 181.6919708251953,
      "learning_rate": 0.00035993325394620217,
      "loss": 3.4914,
      "step": 12688
    },
    {
      "epoch": 9.311191335740073,
      "grad_norm": 218.32742309570312,
      "learning_rate": 0.00034412416451151385,
      "loss": 3.5073,
      "step": 12896
    },
    {
      "epoch": 9.461371841155234,
      "grad_norm": 144.4810791015625,
      "learning_rate": 0.0003284863853541441,
      "loss": 3.5023,
      "step": 13104
    },
    {
      "epoch": 9.611552346570397,
      "grad_norm": 204.1243896484375,
      "learning_rate": 0.00031303707514790077,
      "loss": 3.4959,
      "step": 13312
    },
    {
      "epoch": 9.752346570397112,
      "eval_g2l_cer": 50.963,
      "eval_g2l_gen_len": 3.9938,
      "eval_g2l_rouge1": 39.9737,
      "eval_g2l_rouge2": 29.4419,
      "eval_g2l_rougeL": 39.8196,
      "eval_g2l_rougeLsum": 39.8028,
      "eval_l2ex_cer": 81.2748,
      "eval_l2ex_gen_len": 22.4378,
      "eval_l2ex_rouge1": 33.2444,
      "eval_l2ex_rouge2": 16.7607,
      "eval_l2ex_rougeL": 29.9288,
      "eval_l2ex_rougeLsum": 29.9861,
      "eval_l2g_cer": 74.1059,
      "eval_l2g_gen_len": 15.8848,
      "eval_l2g_rouge1": 38.9415,
      "eval_l2g_rouge2": 25.7151,
      "eval_l2g_rougeL": 36.9163,
      "eval_l2g_rougeLsum": 36.9791,
      "eval_loss": 3.475815534591675,
      "eval_runtime": 94.3857,
      "eval_samples_per_second": 105.111,
      "eval_steps_per_second": 0.826,
      "step": 13507
    },
    {
      "epoch": 9.76173285198556,
      "grad_norm": 156.18637084960938,
      "learning_rate": 0.0002977931857675812,
      "loss": 3.4928,
      "step": 13520
    },
    {
      "epoch": 9.911913357400723,
      "grad_norm": 116.77749633789062,
      "learning_rate": 0.00028277144368839654,
      "loss": 3.4805,
      "step": 13728
    },
    {
      "epoch": 10.062093862815885,
      "grad_norm": 147.93405151367188,
      "learning_rate": 0.00026798833163271856,
      "loss": 3.4859,
      "step": 13936
    },
    {
      "epoch": 10.212274368231046,
      "grad_norm": 129.8922882080078,
      "learning_rate": 0.0002534600704842848,
      "loss": 3.4512,
      "step": 14144
    },
    {
      "epoch": 10.36245487364621,
      "grad_norm": 129.17861938476562,
      "learning_rate": 0.00023920260148970935,
      "loss": 3.4679,
      "step": 14352
    },
    {
      "epoch": 10.502527075812274,
      "eval_g2l_cer": 51.589,
      "eval_g2l_gen_len": 4.615,
      "eval_g2l_rouge1": 40.3572,
      "eval_g2l_rouge2": 29.0165,
      "eval_g2l_rougeL": 40.1696,
      "eval_g2l_rougeLsum": 40.127,
      "eval_l2ex_cer": 83.3939,
      "eval_l2ex_gen_len": 27.2939,
      "eval_l2ex_rouge1": 32.0,
      "eval_l2ex_rouge2": 15.2047,
      "eval_l2ex_rougeL": 28.4221,
      "eval_l2ex_rougeLsum": 28.4944,
      "eval_l2g_cer": 78.1523,
      "eval_l2g_gen_len": 21.5805,
      "eval_l2g_rouge1": 38.8011,
      "eval_l2g_rouge2": 25.5645,
      "eval_l2g_rougeL": 36.5851,
      "eval_l2g_rougeLsum": 36.6686,
      "eval_loss": 3.490727663040161,
      "eval_runtime": 97.0968,
      "eval_samples_per_second": 102.176,
      "eval_steps_per_second": 0.803,
      "step": 14546
    },
    {
      "epoch": 10.512635379061372,
      "grad_norm": 120.92171478271484,
      "learning_rate": 0.00022523156876682566,
      "loss": 3.4559,
      "step": 14560
    },
    {
      "epoch": 10.662815884476535,
      "grad_norm": 72.50321960449219,
      "learning_rate": 0.00021156230213905934,
      "loss": 3.4527,
      "step": 14768
    },
    {
      "epoch": 10.812996389891698,
      "grad_norm": 29.285614013671875,
      "learning_rate": 0.0001982098003146605,
      "loss": 3.445,
      "step": 14976
    },
    {
      "epoch": 10.963176895306859,
      "grad_norm": 139.87692260742188,
      "learning_rate": 0.00018518871442925475,
      "loss": 3.4474,
      "step": 15184
    },
    {
      "epoch": 11.113357400722021,
      "grad_norm": 115.22322845458984,
      "learning_rate": 0.0001725133319697725,
      "loss": 3.444,
      "step": 15392
    },
    {
      "epoch": 11.252707581227437,
      "eval_g2l_cer": 50.9127,
      "eval_g2l_gen_len": 4.3921,
      "eval_g2l_rouge1": 41.0139,
      "eval_g2l_rouge2": 29.6136,
      "eval_g2l_rougeL": 40.8151,
      "eval_g2l_rougeLsum": 40.8006,
      "eval_l2ex_cer": 85.5529,
      "eval_l2ex_gen_len": 28.0987,
      "eval_l2ex_rouge1": 31.8781,
      "eval_l2ex_rouge2": 15.2454,
      "eval_l2ex_rougeL": 28.3874,
      "eval_l2ex_rougeLsum": 28.4594,
      "eval_l2g_cer": 79.5578,
      "eval_l2g_gen_len": 20.9953,
      "eval_l2g_rouge1": 38.8463,
      "eval_l2g_rouge2": 25.6991,
      "eval_l2g_rougeL": 36.7062,
      "eval_l2g_rougeLsum": 36.784,
      "eval_loss": 3.4728221893310547,
      "eval_runtime": 95.7364,
      "eval_samples_per_second": 103.628,
      "eval_steps_per_second": 0.815,
      "step": 15585
    },
    {
      "epoch": 11.263537906137184,
      "grad_norm": 90.60981750488281,
      "learning_rate": 0.000160197561097393,
      "loss": 3.4291,
      "step": 15600
    },
    {
      "epoch": 11.413718411552347,
      "grad_norm": 84.94570922851562,
      "learning_rate": 0.00014825491538670787,
      "loss": 3.4392,
      "step": 15808
    },
    {
      "epoch": 11.56389891696751,
      "grad_norm": 74.72296142578125,
      "learning_rate": 0.00013669849899784683,
      "loss": 3.4338,
      "step": 16016
    },
    {
      "epoch": 11.714079422382671,
      "grad_norm": 73.01377868652344,
      "learning_rate": 0.0001255409922978368,
      "loss": 3.4147,
      "step": 16224
    },
    {
      "epoch": 11.864259927797834,
      "grad_norm": 29.206298828125,
      "learning_rate": 0.00011479463794697153,
      "loss": 3.4327,
      "step": 16432
    },
    {
      "epoch": 12.0028880866426,
      "eval_g2l_cer": 51.0704,
      "eval_g2l_gen_len": 4.428,
      "eval_g2l_rouge1": 40.8383,
      "eval_g2l_rouge2": 29.5544,
      "eval_g2l_rougeL": 40.6824,
      "eval_g2l_rougeLsum": 40.648,
      "eval_l2ex_cer": 85.2521,
      "eval_l2ex_gen_len": 27.6208,
      "eval_l2ex_rouge1": 31.8072,
      "eval_l2ex_rouge2": 15.1802,
      "eval_l2ex_rougeL": 28.2533,
      "eval_l2ex_rougeLsum": 28.3249,
      "eval_l2g_cer": 78.2445,
      "eval_l2g_gen_len": 20.4718,
      "eval_l2g_rouge1": 38.9972,
      "eval_l2g_rouge2": 25.8126,
      "eval_l2g_rougeL": 36.8133,
      "eval_l2g_rougeLsum": 36.8913,
      "eval_loss": 3.460233211517334,
      "eval_runtime": 96.6773,
      "eval_samples_per_second": 102.62,
      "eval_steps_per_second": 0.807,
      "step": 16624
    },
    {
      "epoch": 12.014440433212997,
      "grad_norm": 75.82231140136719,
      "learning_rate": 0.00010447122746545834,
      "loss": 3.4103,
      "step": 16640
    },
    {
      "epoch": 12.16462093862816,
      "grad_norm": 68.91098022460938,
      "learning_rate": 9.458208829508076e-05,
      "loss": 3.4193,
      "step": 16848
    },
    {
      "epoch": 12.31480144404332,
      "grad_norm": 53.81922149658203,
      "learning_rate": 8.513807137007677e-05,
      "loss": 3.3993,
      "step": 17056
    },
    {
      "epoch": 12.464981949458483,
      "grad_norm": 9.850875854492188,
      "learning_rate": 7.61495392108672e-05,
      "loss": 3.401,
      "step": 17264
    },
    {
      "epoch": 12.615162454873646,
      "grad_norm": 36.068912506103516,
      "learning_rate": 6.762635455369967e-05,
      "loss": 3.4024,
      "step": 17472
    },
    {
      "epoch": 12.753068592057762,
      "eval_g2l_cer": 50.4581,
      "eval_g2l_gen_len": 4.2626,
      "eval_g2l_rouge1": 40.828,
      "eval_g2l_rouge2": 29.681,
      "eval_g2l_rougeL": 40.7031,
      "eval_g2l_rougeLsum": 40.6328,
      "eval_l2ex_cer": 83.6513,
      "eval_l2ex_gen_len": 25.1151,
      "eval_l2ex_rouge1": 32.7247,
      "eval_l2ex_rouge2": 15.913,
      "eval_l2ex_rougeL": 29.2133,
      "eval_l2ex_rougeLsum": 29.2625,
      "eval_l2g_cer": 76.3987,
      "eval_l2g_gen_len": 18.6672,
      "eval_l2g_rouge1": 39.0308,
      "eval_l2g_rouge2": 25.8759,
      "eval_l2g_rougeL": 36.9076,
      "eval_l2g_rougeLsum": 36.965,
      "eval_loss": 3.4478375911712646,
      "eval_runtime": 94.7036,
      "eval_samples_per_second": 104.758,
      "eval_steps_per_second": 0.824,
      "step": 17663
    }
  ],
  "logging_steps": 208,
  "max_steps": 20775,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 1039,
  "total_flos": 1.0504606235964211e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}