wav2vec2-xls-r-300m-paper / trainer_state.json
hts98's picture
End of training
5bf9062
raw
history blame
16.7 kB
{
"best_metric": 0.7895313501358032,
"best_model_checkpoint": "./wav2vec2-xls-r-300m-paper/checkpoint-3015",
"epoch": 50.0,
"global_step": 16750,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"eval_loss": 3.7156972885131836,
"eval_runtime": 90.6265,
"eval_samples_per_second": 9.291,
"eval_steps_per_second": 1.17,
"eval_wer": 1.0,
"step": 335
},
{
"epoch": 1.49,
"learning_rate": 4.9770361298224126e-05,
"loss": 6.2976,
"step": 500
},
{
"epoch": 2.0,
"eval_loss": 3.364386558532715,
"eval_runtime": 105.7497,
"eval_samples_per_second": 7.962,
"eval_steps_per_second": 1.002,
"eval_wer": 1.0,
"step": 670
},
{
"epoch": 2.99,
"learning_rate": 4.8239436619718316e-05,
"loss": 3.2342,
"step": 1000
},
{
"epoch": 3.0,
"eval_loss": 2.459735155105591,
"eval_runtime": 100.2813,
"eval_samples_per_second": 8.396,
"eval_steps_per_second": 1.057,
"eval_wer": 0.9739311504731903,
"step": 1005
},
{
"epoch": 4.0,
"eval_loss": 1.4160194396972656,
"eval_runtime": 105.5456,
"eval_samples_per_second": 7.978,
"eval_steps_per_second": 1.004,
"eval_wer": 0.7444104809322317,
"step": 1340
},
{
"epoch": 4.48,
"learning_rate": 4.670851194121249e-05,
"loss": 1.2813,
"step": 1500
},
{
"epoch": 5.0,
"eval_loss": 1.1337928771972656,
"eval_runtime": 101.0094,
"eval_samples_per_second": 8.336,
"eval_steps_per_second": 1.049,
"eval_wer": 0.6543241186913518,
"step": 1675
},
{
"epoch": 5.97,
"learning_rate": 4.5177587262706676e-05,
"loss": 0.7279,
"step": 2000
},
{
"epoch": 6.0,
"eval_loss": 1.002017617225647,
"eval_runtime": 97.2569,
"eval_samples_per_second": 8.657,
"eval_steps_per_second": 1.09,
"eval_wer": 0.5855680286183377,
"step": 2010
},
{
"epoch": 7.0,
"eval_loss": 0.8434591293334961,
"eval_runtime": 101.7349,
"eval_samples_per_second": 8.276,
"eval_steps_per_second": 1.042,
"eval_wer": 0.48233378271954297,
"step": 2345
},
{
"epoch": 7.46,
"learning_rate": 4.364666258420086e-05,
"loss": 0.5226,
"step": 2500
},
{
"epoch": 8.0,
"eval_loss": 0.875725507736206,
"eval_runtime": 100.6138,
"eval_samples_per_second": 8.369,
"eval_steps_per_second": 1.054,
"eval_wer": 0.5078286637211847,
"step": 2680
},
{
"epoch": 8.96,
"learning_rate": 4.211573790569504e-05,
"loss": 0.4218,
"step": 3000
},
{
"epoch": 9.0,
"eval_loss": 0.7895313501358032,
"eval_runtime": 101.3795,
"eval_samples_per_second": 8.305,
"eval_steps_per_second": 1.046,
"eval_wer": 0.43980671943617605,
"step": 3015
},
{
"epoch": 10.0,
"eval_loss": 0.7991793155670166,
"eval_runtime": 87.9454,
"eval_samples_per_second": 9.574,
"eval_steps_per_second": 1.205,
"eval_wer": 0.42284127768063323,
"step": 3350
},
{
"epoch": 10.45,
"learning_rate": 4.058481322718922e-05,
"loss": 0.3421,
"step": 3500
},
{
"epoch": 11.0,
"eval_loss": 0.8118342757225037,
"eval_runtime": 105.2144,
"eval_samples_per_second": 8.003,
"eval_steps_per_second": 1.007,
"eval_wer": 0.43070331166490916,
"step": 3685
},
{
"epoch": 11.94,
"learning_rate": 3.905388854868341e-05,
"loss": 0.287,
"step": 4000
},
{
"epoch": 12.0,
"eval_loss": 0.8214530348777771,
"eval_runtime": 103.7557,
"eval_samples_per_second": 8.115,
"eval_steps_per_second": 1.022,
"eval_wer": 0.4248434934661025,
"step": 4020
},
{
"epoch": 13.0,
"eval_loss": 0.8603241443634033,
"eval_runtime": 98.5504,
"eval_samples_per_second": 8.544,
"eval_steps_per_second": 1.076,
"eval_wer": 0.40767783013201275,
"step": 4355
},
{
"epoch": 13.43,
"learning_rate": 3.752296387017759e-05,
"loss": 0.2415,
"step": 4500
},
{
"epoch": 14.0,
"eval_loss": 0.8329221606254578,
"eval_runtime": 106.1564,
"eval_samples_per_second": 7.932,
"eval_steps_per_second": 0.999,
"eval_wer": 0.3885766915386361,
"step": 4690
},
{
"epoch": 14.93,
"learning_rate": 3.5992039191671776e-05,
"loss": 0.2132,
"step": 5000
},
{
"epoch": 15.0,
"eval_loss": 0.8728479743003845,
"eval_runtime": 106.4415,
"eval_samples_per_second": 7.91,
"eval_steps_per_second": 0.996,
"eval_wer": 0.3955310543668326,
"step": 5025
},
{
"epoch": 16.0,
"eval_loss": 0.8741195797920227,
"eval_runtime": 105.2581,
"eval_samples_per_second": 7.999,
"eval_steps_per_second": 1.007,
"eval_wer": 0.39178023679538687,
"step": 5360
},
{
"epoch": 16.42,
"learning_rate": 3.446111451316595e-05,
"loss": 0.1857,
"step": 5500
},
{
"epoch": 17.0,
"eval_loss": 0.8632614612579346,
"eval_runtime": 107.8251,
"eval_samples_per_second": 7.809,
"eval_steps_per_second": 0.983,
"eval_wer": 0.3674733371597902,
"step": 5695
},
{
"epoch": 17.91,
"learning_rate": 3.2930189834660136e-05,
"loss": 0.1673,
"step": 6000
},
{
"epoch": 18.0,
"eval_loss": 0.8884367346763611,
"eval_runtime": 96.3592,
"eval_samples_per_second": 8.738,
"eval_steps_per_second": 1.1,
"eval_wer": 0.3803943030286851,
"step": 6030
},
{
"epoch": 19.0,
"eval_loss": 0.9140524864196777,
"eval_runtime": 99.6527,
"eval_samples_per_second": 8.449,
"eval_steps_per_second": 1.064,
"eval_wer": 0.3679271727378299,
"step": 6365
},
{
"epoch": 19.4,
"learning_rate": 3.139926515615432e-05,
"loss": 0.1479,
"step": 6500
},
{
"epoch": 20.0,
"eval_loss": 0.9567932486534119,
"eval_runtime": 99.4352,
"eval_samples_per_second": 8.468,
"eval_steps_per_second": 1.066,
"eval_wer": 0.36047893001588427,
"step": 6700
},
{
"epoch": 20.9,
"learning_rate": 2.98683404776485e-05,
"loss": 0.1386,
"step": 7000
},
{
"epoch": 21.0,
"eval_loss": 0.9340795874595642,
"eval_runtime": 103.1923,
"eval_samples_per_second": 8.16,
"eval_steps_per_second": 1.027,
"eval_wer": 0.36301507001081196,
"step": 7035
},
{
"epoch": 22.0,
"eval_loss": 0.9644697308540344,
"eval_runtime": 98.7834,
"eval_samples_per_second": 8.524,
"eval_steps_per_second": 1.073,
"eval_wer": 0.3536847444505252,
"step": 7370
},
{
"epoch": 22.39,
"learning_rate": 2.8337415799142685e-05,
"loss": 0.1233,
"step": 7500
},
{
"epoch": 23.0,
"eval_loss": 0.9728893041610718,
"eval_runtime": 102.9289,
"eval_samples_per_second": 8.18,
"eval_steps_per_second": 1.03,
"eval_wer": 0.35667472002349265,
"step": 7705
},
{
"epoch": 23.88,
"learning_rate": 2.680649112063687e-05,
"loss": 0.1177,
"step": 8000
},
{
"epoch": 24.0,
"eval_loss": 1.001348614692688,
"eval_runtime": 106.3343,
"eval_samples_per_second": 7.918,
"eval_steps_per_second": 0.997,
"eval_wer": 0.34538222299344606,
"step": 8040
},
{
"epoch": 25.0,
"eval_loss": 1.0323426723480225,
"eval_runtime": 102.2816,
"eval_samples_per_second": 8.232,
"eval_steps_per_second": 1.036,
"eval_wer": 0.35967804370169654,
"step": 8375
},
{
"epoch": 25.37,
"learning_rate": 2.527556644213105e-05,
"loss": 0.1061,
"step": 8500
},
{
"epoch": 26.0,
"eval_loss": 1.0269230604171753,
"eval_runtime": 96.8319,
"eval_samples_per_second": 8.695,
"eval_steps_per_second": 1.095,
"eval_wer": 0.3456491850981753,
"step": 8710
},
{
"epoch": 26.87,
"learning_rate": 2.3744641763625232e-05,
"loss": 0.1028,
"step": 9000
},
{
"epoch": 27.0,
"eval_loss": 1.0042426586151123,
"eval_runtime": 94.8369,
"eval_samples_per_second": 8.878,
"eval_steps_per_second": 1.118,
"eval_wer": 0.3424189436309516,
"step": 9045
},
{
"epoch": 28.0,
"eval_loss": 1.0424461364746094,
"eval_runtime": 93.6728,
"eval_samples_per_second": 8.989,
"eval_steps_per_second": 1.132,
"eval_wer": 0.3394423161632206,
"step": 9380
},
{
"epoch": 28.36,
"learning_rate": 2.2213717085119412e-05,
"loss": 0.0961,
"step": 9500
},
{
"epoch": 29.0,
"eval_loss": 1.0599919557571411,
"eval_runtime": 104.5041,
"eval_samples_per_second": 8.057,
"eval_steps_per_second": 1.014,
"eval_wer": 0.3412309622649065,
"step": 9715
},
{
"epoch": 29.85,
"learning_rate": 2.0682792406613595e-05,
"loss": 0.0949,
"step": 10000
},
{
"epoch": 30.0,
"eval_loss": 1.051209568977356,
"eval_runtime": 104.5071,
"eval_samples_per_second": 8.057,
"eval_steps_per_second": 1.014,
"eval_wer": 0.33890839195376216,
"step": 10050
},
{
"epoch": 31.0,
"eval_loss": 1.0956796407699585,
"eval_runtime": 104.4359,
"eval_samples_per_second": 8.062,
"eval_steps_per_second": 1.015,
"eval_wer": 0.3389217400589986,
"step": 10385
},
{
"epoch": 31.34,
"learning_rate": 1.9154929577464788e-05,
"loss": 0.0878,
"step": 10500
},
{
"epoch": 32.0,
"eval_loss": 1.09244704246521,
"eval_runtime": 106.7914,
"eval_samples_per_second": 7.885,
"eval_steps_per_second": 0.993,
"eval_wer": 0.33107305417995914,
"step": 10720
},
{
"epoch": 32.84,
"learning_rate": 1.7627066748315983e-05,
"loss": 0.0852,
"step": 11000
},
{
"epoch": 33.0,
"eval_loss": 1.0858749151229858,
"eval_runtime": 104.2304,
"eval_samples_per_second": 8.078,
"eval_steps_per_second": 1.017,
"eval_wer": 0.3365991697478543,
"step": 11055
},
{
"epoch": 34.0,
"eval_loss": 1.1498078107833862,
"eval_runtime": 96.0047,
"eval_samples_per_second": 8.77,
"eval_steps_per_second": 1.104,
"eval_wer": 0.3450485203625345,
"step": 11390
},
{
"epoch": 34.33,
"learning_rate": 1.6096142069810167e-05,
"loss": 0.0837,
"step": 11500
},
{
"epoch": 35.0,
"eval_loss": 1.0844124555587769,
"eval_runtime": 95.1094,
"eval_samples_per_second": 8.853,
"eval_steps_per_second": 1.115,
"eval_wer": 0.332861700281645,
"step": 11725
},
{
"epoch": 35.82,
"learning_rate": 1.4565217391304348e-05,
"loss": 0.0814,
"step": 12000
},
{
"epoch": 36.0,
"eval_loss": 1.1050550937652588,
"eval_runtime": 105.0526,
"eval_samples_per_second": 8.015,
"eval_steps_per_second": 1.009,
"eval_wer": 0.3321008582831667,
"step": 12060
},
{
"epoch": 37.0,
"eval_loss": 1.0878149271011353,
"eval_runtime": 97.9103,
"eval_samples_per_second": 8.6,
"eval_steps_per_second": 1.083,
"eval_wer": 0.3310597060747227,
"step": 12395
},
{
"epoch": 37.31,
"learning_rate": 1.3034292712798532e-05,
"loss": 0.0793,
"step": 12500
},
{
"epoch": 38.0,
"eval_loss": 1.1377496719360352,
"eval_runtime": 101.9563,
"eval_samples_per_second": 8.258,
"eval_steps_per_second": 1.04,
"eval_wer": 0.32856361039550436,
"step": 12730
},
{
"epoch": 38.81,
"learning_rate": 1.1506429883649724e-05,
"loss": 0.0759,
"step": 13000
},
{
"epoch": 39.0,
"eval_loss": 1.1136152744293213,
"eval_runtime": 98.218,
"eval_samples_per_second": 8.573,
"eval_steps_per_second": 1.079,
"eval_wer": 0.3246125712455117,
"step": 13065
},
{
"epoch": 40.0,
"eval_loss": 1.1215593814849854,
"eval_runtime": 107.5832,
"eval_samples_per_second": 7.826,
"eval_steps_per_second": 0.985,
"eval_wer": 0.32682835671476435,
"step": 13400
},
{
"epoch": 40.3,
"learning_rate": 9.975505205143907e-06,
"loss": 0.0726,
"step": 13500
},
{
"epoch": 41.0,
"eval_loss": 1.1300030946731567,
"eval_runtime": 106.9995,
"eval_samples_per_second": 7.869,
"eval_steps_per_second": 0.991,
"eval_wer": 0.3252666284020983,
"step": 13735
},
{
"epoch": 41.79,
"learning_rate": 8.444580526638089e-06,
"loss": 0.0715,
"step": 14000
},
{
"epoch": 42.0,
"eval_loss": 1.1506843566894531,
"eval_runtime": 93.8834,
"eval_samples_per_second": 8.969,
"eval_steps_per_second": 1.129,
"eval_wer": 0.32622769197912355,
"step": 14070
},
{
"epoch": 43.0,
"eval_loss": 1.1561784744262695,
"eval_runtime": 104.7418,
"eval_samples_per_second": 8.039,
"eval_steps_per_second": 1.012,
"eval_wer": 0.3275491543975333,
"step": 14405
},
{
"epoch": 43.28,
"learning_rate": 6.913655848132272e-06,
"loss": 0.0711,
"step": 14500
},
{
"epoch": 44.0,
"eval_loss": 1.1485936641693115,
"eval_runtime": 104.0283,
"eval_samples_per_second": 8.094,
"eval_steps_per_second": 1.019,
"eval_wer": 0.3218895577772735,
"step": 14740
},
{
"epoch": 44.78,
"learning_rate": 5.382731169626455e-06,
"loss": 0.0699,
"step": 15000
},
{
"epoch": 45.0,
"eval_loss": 1.1580160856246948,
"eval_runtime": 98.4301,
"eval_samples_per_second": 8.554,
"eval_steps_per_second": 1.077,
"eval_wer": 0.31940681020329165,
"step": 15075
},
{
"epoch": 46.0,
"eval_loss": 1.1580368280410767,
"eval_runtime": 96.4383,
"eval_samples_per_second": 8.731,
"eval_steps_per_second": 1.099,
"eval_wer": 0.3195269431504198,
"step": 15410
},
{
"epoch": 46.27,
"learning_rate": 3.851806491120637e-06,
"loss": 0.0667,
"step": 15500
},
{
"epoch": 47.0,
"eval_loss": 1.1504408121109009,
"eval_runtime": 99.6568,
"eval_samples_per_second": 8.449,
"eval_steps_per_second": 1.064,
"eval_wer": 0.321208804410214,
"step": 15745
},
{
"epoch": 47.76,
"learning_rate": 2.3208818126148195e-06,
"loss": 0.0667,
"step": 16000
},
{
"epoch": 48.0,
"eval_loss": 1.1580157279968262,
"eval_runtime": 110.9917,
"eval_samples_per_second": 7.586,
"eval_steps_per_second": 0.955,
"eval_wer": 0.3202610889384252,
"step": 16080
},
{
"epoch": 49.0,
"eval_loss": 1.1697617769241333,
"eval_runtime": 107.1691,
"eval_samples_per_second": 7.857,
"eval_steps_per_second": 0.989,
"eval_wer": 0.3192065886247447,
"step": 16415
},
{
"epoch": 49.25,
"learning_rate": 7.899571341090019e-07,
"loss": 0.0664,
"step": 16500
},
{
"epoch": 50.0,
"eval_loss": 1.17439603805542,
"eval_runtime": 107.0438,
"eval_samples_per_second": 7.866,
"eval_steps_per_second": 0.99,
"eval_wer": 0.3192332848352176,
"step": 16750
},
{
"epoch": 50.0,
"step": 16750,
"total_flos": 1.4087701795050537e+20,
"train_loss": 0.4734274124600994,
"train_runtime": 45905.6833,
"train_samples_per_second": 3.649,
"train_steps_per_second": 0.365
}
],
"max_steps": 16750,
"num_train_epochs": 50,
"total_flos": 1.4087701795050537e+20,
"trial_name": null,
"trial_params": null
}