xls-r-300m-sv-robust / trainer_state.json
marinone94's picture
End of training
cf7ab1e
raw
history blame
29.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 49.99782451051487,
"global_step": 17200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.29,
"learning_rate": 3.7125e-06,
"loss": 12.2713,
"step": 100
},
{
"epoch": 0.58,
"learning_rate": 7.4625e-06,
"loss": 6.2026,
"step": 200
},
{
"epoch": 0.87,
"learning_rate": 1.1212499999999998e-05,
"loss": 4.1374,
"step": 300
},
{
"epoch": 1.16,
"learning_rate": 1.49625e-05,
"loss": 3.6755,
"step": 400
},
{
"epoch": 1.45,
"learning_rate": 1.8712499999999997e-05,
"loss": 3.3332,
"step": 500
},
{
"epoch": 1.45,
"eval_loss": 3.292029857635498,
"eval_runtime": 137.2974,
"eval_samples_per_second": 33.65,
"eval_steps_per_second": 4.21,
"eval_wer": 1.0,
"step": 500
},
{
"epoch": 1.74,
"learning_rate": 2.2462499999999997e-05,
"loss": 3.1418,
"step": 600
},
{
"epoch": 2.03,
"learning_rate": 2.6212499999999997e-05,
"loss": 3.0879,
"step": 700
},
{
"epoch": 2.32,
"learning_rate": 2.99625e-05,
"loss": 3.0216,
"step": 800
},
{
"epoch": 2.61,
"learning_rate": 3.37125e-05,
"loss": 2.9595,
"step": 900
},
{
"epoch": 2.91,
"learning_rate": 3.7462499999999996e-05,
"loss": 2.9269,
"step": 1000
},
{
"epoch": 2.91,
"eval_loss": 2.941540241241455,
"eval_runtime": 178.4388,
"eval_samples_per_second": 25.891,
"eval_steps_per_second": 3.239,
"eval_wer": 0.9966346448020559,
"step": 1000
},
{
"epoch": 3.2,
"learning_rate": 4.12125e-05,
"loss": 2.914,
"step": 1100
},
{
"epoch": 3.49,
"learning_rate": 4.4962499999999995e-05,
"loss": 2.8432,
"step": 1200
},
{
"epoch": 3.78,
"learning_rate": 4.871249999999999e-05,
"loss": 2.6828,
"step": 1300
},
{
"epoch": 4.07,
"learning_rate": 5.2462499999999994e-05,
"loss": 2.355,
"step": 1400
},
{
"epoch": 4.36,
"learning_rate": 5.62125e-05,
"loss": 2.0719,
"step": 1500
},
{
"epoch": 4.36,
"eval_loss": 1.164096713066101,
"eval_runtime": 134.2981,
"eval_samples_per_second": 34.401,
"eval_steps_per_second": 4.304,
"eval_wer": 0.8507923881784251,
"step": 1500
},
{
"epoch": 4.65,
"learning_rate": 5.9962499999999994e-05,
"loss": 1.944,
"step": 1600
},
{
"epoch": 4.94,
"learning_rate": 6.37125e-05,
"loss": 1.8772,
"step": 1700
},
{
"epoch": 5.23,
"learning_rate": 6.746249999999999e-05,
"loss": 1.807,
"step": 1800
},
{
"epoch": 5.52,
"learning_rate": 7.121249999999999e-05,
"loss": 1.7612,
"step": 1900
},
{
"epoch": 5.81,
"learning_rate": 7.492499999999999e-05,
"loss": 1.7404,
"step": 2000
},
{
"epoch": 5.81,
"eval_loss": 0.7280541658401489,
"eval_runtime": 174.7873,
"eval_samples_per_second": 26.432,
"eval_steps_per_second": 3.307,
"eval_wer": 0.6846356238144772,
"step": 2000
},
{
"epoch": 6.1,
"learning_rate": 7.451644736842105e-05,
"loss": 1.6883,
"step": 2100
},
{
"epoch": 6.39,
"learning_rate": 7.402302631578947e-05,
"loss": 1.6463,
"step": 2200
},
{
"epoch": 6.68,
"learning_rate": 7.352960526315788e-05,
"loss": 1.6216,
"step": 2300
},
{
"epoch": 6.97,
"learning_rate": 7.30361842105263e-05,
"loss": 1.586,
"step": 2400
},
{
"epoch": 7.27,
"learning_rate": 7.254276315789473e-05,
"loss": 1.5921,
"step": 2500
},
{
"epoch": 7.27,
"eval_loss": 0.5885757803916931,
"eval_runtime": 137.296,
"eval_samples_per_second": 33.65,
"eval_steps_per_second": 4.21,
"eval_wer": 0.5146545921801383,
"step": 2500
},
{
"epoch": 7.56,
"learning_rate": 7.204934210526316e-05,
"loss": 1.5694,
"step": 2600
},
{
"epoch": 7.85,
"learning_rate": 7.155592105263157e-05,
"loss": 1.5279,
"step": 2700
},
{
"epoch": 8.14,
"learning_rate": 7.10625e-05,
"loss": 1.5435,
"step": 2800
},
{
"epoch": 8.43,
"learning_rate": 7.056907894736841e-05,
"loss": 1.5163,
"step": 2900
},
{
"epoch": 8.72,
"learning_rate": 7.007565789473684e-05,
"loss": 1.4941,
"step": 3000
},
{
"epoch": 8.72,
"eval_loss": 0.5183178186416626,
"eval_runtime": 134.3566,
"eval_samples_per_second": 34.386,
"eval_steps_per_second": 4.302,
"eval_wer": 0.5063023924616044,
"step": 3000
},
{
"epoch": 9.01,
"learning_rate": 6.958223684210525e-05,
"loss": 1.5061,
"step": 3100
},
{
"epoch": 9.3,
"learning_rate": 6.908881578947368e-05,
"loss": 1.4551,
"step": 3200
},
{
"epoch": 9.59,
"learning_rate": 6.859539473684209e-05,
"loss": 1.4674,
"step": 3300
},
{
"epoch": 9.88,
"learning_rate": 6.810197368421052e-05,
"loss": 1.4691,
"step": 3400
},
{
"epoch": 10.17,
"learning_rate": 6.760855263157895e-05,
"loss": 1.4486,
"step": 3500
},
{
"epoch": 10.17,
"eval_loss": 0.47492966055870056,
"eval_runtime": 135.1792,
"eval_samples_per_second": 34.177,
"eval_steps_per_second": 4.276,
"eval_wer": 0.46760080768524753,
"step": 3500
},
{
"epoch": 10.46,
"learning_rate": 6.711513157894736e-05,
"loss": 1.4274,
"step": 3600
},
{
"epoch": 10.75,
"learning_rate": 6.66266447368421e-05,
"loss": 1.4277,
"step": 3700
},
{
"epoch": 11.05,
"learning_rate": 6.613322368421052e-05,
"loss": 1.4445,
"step": 3800
},
{
"epoch": 11.34,
"learning_rate": 6.563980263157894e-05,
"loss": 1.4101,
"step": 3900
},
{
"epoch": 11.63,
"learning_rate": 6.514638157894736e-05,
"loss": 1.3899,
"step": 4000
},
{
"epoch": 11.63,
"eval_loss": 0.4564875364303589,
"eval_runtime": 134.5257,
"eval_samples_per_second": 34.343,
"eval_steps_per_second": 4.297,
"eval_wer": 0.4432172795692345,
"step": 4000
},
{
"epoch": 11.92,
"learning_rate": 6.465296052631578e-05,
"loss": 1.404,
"step": 4100
},
{
"epoch": 12.21,
"learning_rate": 6.415953947368421e-05,
"loss": 1.3911,
"step": 4200
},
{
"epoch": 12.5,
"learning_rate": 6.366611842105262e-05,
"loss": 1.3873,
"step": 4300
},
{
"epoch": 12.79,
"learning_rate": 6.317269736842105e-05,
"loss": 1.3712,
"step": 4400
},
{
"epoch": 13.08,
"learning_rate": 6.267927631578946e-05,
"loss": 1.3881,
"step": 4500
},
{
"epoch": 13.08,
"eval_loss": 0.43156012892723083,
"eval_runtime": 134.756,
"eval_samples_per_second": 34.284,
"eval_steps_per_second": 4.289,
"eval_wer": 0.42278039527626504,
"step": 4500
},
{
"epoch": 13.37,
"learning_rate": 6.218585526315789e-05,
"loss": 1.3538,
"step": 4600
},
{
"epoch": 13.66,
"learning_rate": 6.16924342105263e-05,
"loss": 1.355,
"step": 4700
},
{
"epoch": 13.95,
"learning_rate": 6.119901315789473e-05,
"loss": 1.341,
"step": 4800
},
{
"epoch": 14.24,
"learning_rate": 6.070559210526316e-05,
"loss": 1.3474,
"step": 4900
},
{
"epoch": 14.53,
"learning_rate": 6.021217105263157e-05,
"loss": 1.3572,
"step": 5000
},
{
"epoch": 14.53,
"eval_loss": 0.4194825291633606,
"eval_runtime": 134.9338,
"eval_samples_per_second": 34.239,
"eval_steps_per_second": 4.284,
"eval_wer": 0.3834363335984825,
"step": 5000
},
{
"epoch": 14.82,
"learning_rate": 5.971875e-05,
"loss": 1.3408,
"step": 5100
},
{
"epoch": 15.12,
"learning_rate": 5.922532894736842e-05,
"loss": 1.3357,
"step": 5200
},
{
"epoch": 15.41,
"learning_rate": 5.873190789473683e-05,
"loss": 1.3288,
"step": 5300
},
{
"epoch": 15.7,
"learning_rate": 5.823848684210526e-05,
"loss": 1.3258,
"step": 5400
},
{
"epoch": 15.99,
"learning_rate": 5.7745065789473685e-05,
"loss": 1.3261,
"step": 5500
},
{
"epoch": 15.99,
"eval_loss": 0.3974045217037201,
"eval_runtime": 134.8844,
"eval_samples_per_second": 34.252,
"eval_steps_per_second": 4.285,
"eval_wer": 0.3606742948051153,
"step": 5500
},
{
"epoch": 16.28,
"learning_rate": 5.72516447368421e-05,
"loss": 1.3273,
"step": 5600
},
{
"epoch": 16.57,
"learning_rate": 5.675822368421052e-05,
"loss": 1.3131,
"step": 5700
},
{
"epoch": 16.86,
"learning_rate": 5.6264802631578946e-05,
"loss": 1.2739,
"step": 5800
},
{
"epoch": 17.15,
"learning_rate": 5.577138157894736e-05,
"loss": 1.3141,
"step": 5900
},
{
"epoch": 17.44,
"learning_rate": 5.5277960526315786e-05,
"loss": 1.2809,
"step": 6000
},
{
"epoch": 17.44,
"eval_loss": 0.3844749331474304,
"eval_runtime": 136.3592,
"eval_samples_per_second": 33.881,
"eval_steps_per_second": 4.239,
"eval_wer": 0.3466621795264027,
"step": 6000
},
{
"epoch": 17.73,
"learning_rate": 5.4784539473684206e-05,
"loss": 1.2847,
"step": 6100
},
{
"epoch": 18.02,
"learning_rate": 5.4291118421052626e-05,
"loss": 1.303,
"step": 6200
},
{
"epoch": 18.31,
"learning_rate": 5.3797697368421046e-05,
"loss": 1.2733,
"step": 6300
},
{
"epoch": 18.6,
"learning_rate": 5.330427631578947e-05,
"loss": 1.2707,
"step": 6400
},
{
"epoch": 18.89,
"learning_rate": 5.2810855263157886e-05,
"loss": 1.2713,
"step": 6500
},
{
"epoch": 18.89,
"eval_loss": 0.3831607401371002,
"eval_runtime": 138.2456,
"eval_samples_per_second": 33.419,
"eval_steps_per_second": 4.181,
"eval_wer": 0.3449795019274307,
"step": 6500
},
{
"epoch": 19.19,
"learning_rate": 5.231743421052631e-05,
"loss": 1.2562,
"step": 6600
},
{
"epoch": 19.48,
"learning_rate": 5.182894736842105e-05,
"loss": 1.2721,
"step": 6700
},
{
"epoch": 19.77,
"learning_rate": 5.133552631578947e-05,
"loss": 1.2536,
"step": 6800
},
{
"epoch": 20.06,
"learning_rate": 5.084210526315789e-05,
"loss": 1.2503,
"step": 6900
},
{
"epoch": 20.35,
"learning_rate": 5.0353618421052625e-05,
"loss": 1.257,
"step": 7000
},
{
"epoch": 20.35,
"eval_loss": 0.3778809607028961,
"eval_runtime": 138.2403,
"eval_samples_per_second": 33.42,
"eval_steps_per_second": 4.181,
"eval_wer": 0.33726977911032247,
"step": 7000
},
{
"epoch": 20.64,
"learning_rate": 4.9860197368421045e-05,
"loss": 1.2557,
"step": 7100
},
{
"epoch": 20.93,
"learning_rate": 4.936677631578947e-05,
"loss": 1.2397,
"step": 7200
},
{
"epoch": 21.22,
"learning_rate": 4.8873355263157886e-05,
"loss": 1.2527,
"step": 7300
},
{
"epoch": 21.51,
"learning_rate": 4.837993421052631e-05,
"loss": 1.2273,
"step": 7400
},
{
"epoch": 21.8,
"learning_rate": 4.788651315789473e-05,
"loss": 1.2298,
"step": 7500
},
{
"epoch": 21.8,
"eval_loss": 0.3743567168712616,
"eval_runtime": 136.1375,
"eval_samples_per_second": 33.936,
"eval_steps_per_second": 4.246,
"eval_wer": 0.33913602153827327,
"step": 7500
},
{
"epoch": 22.09,
"learning_rate": 4.739309210526315e-05,
"loss": 1.2444,
"step": 7600
},
{
"epoch": 22.38,
"learning_rate": 4.689967105263157e-05,
"loss": 1.2153,
"step": 7700
},
{
"epoch": 22.67,
"learning_rate": 4.640625e-05,
"loss": 1.2327,
"step": 7800
},
{
"epoch": 22.96,
"learning_rate": 4.591282894736841e-05,
"loss": 1.2404,
"step": 7900
},
{
"epoch": 23.26,
"learning_rate": 4.541940789473684e-05,
"loss": 1.2173,
"step": 8000
},
{
"epoch": 23.26,
"eval_loss": 0.3745496869087219,
"eval_runtime": 139.3565,
"eval_samples_per_second": 33.152,
"eval_steps_per_second": 4.148,
"eval_wer": 0.32619470109527016,
"step": 8000
},
{
"epoch": 23.55,
"learning_rate": 4.492598684210526e-05,
"loss": 1.222,
"step": 8100
},
{
"epoch": 23.84,
"learning_rate": 4.443256578947369e-05,
"loss": 1.1868,
"step": 8200
},
{
"epoch": 24.13,
"learning_rate": 4.39391447368421e-05,
"loss": 1.2272,
"step": 8300
},
{
"epoch": 24.42,
"learning_rate": 4.344572368421052e-05,
"loss": 1.2212,
"step": 8400
},
{
"epoch": 24.71,
"learning_rate": 4.295230263157895e-05,
"loss": 1.1966,
"step": 8500
},
{
"epoch": 24.71,
"eval_loss": 0.36800575256347656,
"eval_runtime": 137.1685,
"eval_samples_per_second": 33.681,
"eval_steps_per_second": 4.214,
"eval_wer": 0.32411429970017747,
"step": 8500
},
{
"epoch": 25.0,
"learning_rate": 4.245888157894736e-05,
"loss": 1.2128,
"step": 8600
},
{
"epoch": 25.29,
"learning_rate": 4.196546052631579e-05,
"loss": 1.2118,
"step": 8700
},
{
"epoch": 25.58,
"learning_rate": 4.147203947368421e-05,
"loss": 1.2009,
"step": 8800
},
{
"epoch": 25.87,
"learning_rate": 4.097861842105263e-05,
"loss": 1.1684,
"step": 8900
},
{
"epoch": 26.16,
"learning_rate": 4.048519736842105e-05,
"loss": 1.1925,
"step": 9000
},
{
"epoch": 26.16,
"eval_loss": 0.3604615330696106,
"eval_runtime": 153.4534,
"eval_samples_per_second": 30.107,
"eval_steps_per_second": 3.767,
"eval_wer": 0.31713883619898425,
"step": 9000
},
{
"epoch": 26.45,
"learning_rate": 3.9991776315789475e-05,
"loss": 1.1759,
"step": 9100
},
{
"epoch": 26.74,
"learning_rate": 3.949835526315789e-05,
"loss": 1.1754,
"step": 9200
},
{
"epoch": 27.03,
"learning_rate": 3.9004934210526315e-05,
"loss": 1.198,
"step": 9300
},
{
"epoch": 27.32,
"learning_rate": 3.8511513157894735e-05,
"loss": 1.1602,
"step": 9400
},
{
"epoch": 27.61,
"learning_rate": 3.801809210526315e-05,
"loss": 1.1692,
"step": 9500
},
{
"epoch": 27.61,
"eval_loss": 0.3512294292449951,
"eval_runtime": 166.9619,
"eval_samples_per_second": 27.671,
"eval_steps_per_second": 3.462,
"eval_wer": 0.31472189928409716,
"step": 9500
},
{
"epoch": 27.91,
"learning_rate": 3.7524671052631575e-05,
"loss": 1.1495,
"step": 9600
},
{
"epoch": 28.2,
"learning_rate": 3.7031249999999995e-05,
"loss": 1.1722,
"step": 9700
},
{
"epoch": 28.49,
"learning_rate": 3.6537828947368416e-05,
"loss": 1.1443,
"step": 9800
},
{
"epoch": 28.78,
"learning_rate": 3.6044407894736836e-05,
"loss": 1.1535,
"step": 9900
},
{
"epoch": 29.07,
"learning_rate": 3.555098684210526e-05,
"loss": 1.1704,
"step": 10000
},
{
"epoch": 29.07,
"eval_loss": 0.3532196581363678,
"eval_runtime": 134.5452,
"eval_samples_per_second": 34.338,
"eval_steps_per_second": 4.296,
"eval_wer": 0.30976564890167047,
"step": 10000
},
{
"epoch": 29.36,
"learning_rate": 3.505756578947368e-05,
"loss": 1.1519,
"step": 10100
},
{
"epoch": 29.65,
"learning_rate": 3.45641447368421e-05,
"loss": 1.1507,
"step": 10200
},
{
"epoch": 29.94,
"learning_rate": 3.407072368421052e-05,
"loss": 1.1517,
"step": 10300
},
{
"epoch": 30.23,
"learning_rate": 3.357730263157894e-05,
"loss": 1.1447,
"step": 10400
},
{
"epoch": 30.52,
"learning_rate": 3.308388157894737e-05,
"loss": 1.1595,
"step": 10500
},
{
"epoch": 30.52,
"eval_loss": 0.34246256947517395,
"eval_runtime": 159.7393,
"eval_samples_per_second": 28.922,
"eval_steps_per_second": 3.618,
"eval_wer": 0.3038915743743499,
"step": 10500
},
{
"epoch": 30.81,
"learning_rate": 3.259046052631579e-05,
"loss": 1.1451,
"step": 10600
},
{
"epoch": 31.1,
"learning_rate": 3.2097039473684203e-05,
"loss": 1.1304,
"step": 10700
},
{
"epoch": 31.39,
"learning_rate": 3.160361842105263e-05,
"loss": 1.1316,
"step": 10800
},
{
"epoch": 31.68,
"learning_rate": 3.111019736842105e-05,
"loss": 1.1301,
"step": 10900
},
{
"epoch": 31.97,
"learning_rate": 3.061677631578947e-05,
"loss": 1.1433,
"step": 11000
},
{
"epoch": 31.97,
"eval_loss": 0.35683709383010864,
"eval_runtime": 158.7503,
"eval_samples_per_second": 29.102,
"eval_steps_per_second": 3.641,
"eval_wer": 0.30257602643333537,
"step": 11000
},
{
"epoch": 32.27,
"learning_rate": 3.0123355263157894e-05,
"loss": 1.1337,
"step": 11100
},
{
"epoch": 32.56,
"learning_rate": 2.9629934210526314e-05,
"loss": 1.1227,
"step": 11200
},
{
"epoch": 32.85,
"learning_rate": 2.913651315789473e-05,
"loss": 1.1281,
"step": 11300
},
{
"epoch": 33.14,
"learning_rate": 2.8643092105263154e-05,
"loss": 1.1203,
"step": 11400
},
{
"epoch": 33.43,
"learning_rate": 2.8149671052631574e-05,
"loss": 1.1295,
"step": 11500
},
{
"epoch": 33.43,
"eval_loss": 0.34605443477630615,
"eval_runtime": 160.1749,
"eval_samples_per_second": 28.843,
"eval_steps_per_second": 3.609,
"eval_wer": 0.2992106712353913,
"step": 11500
},
{
"epoch": 33.72,
"learning_rate": 2.7656249999999998e-05,
"loss": 1.0915,
"step": 11600
},
{
"epoch": 34.01,
"learning_rate": 2.7162828947368418e-05,
"loss": 1.1312,
"step": 11700
},
{
"epoch": 34.3,
"learning_rate": 2.6669407894736838e-05,
"loss": 1.1146,
"step": 11800
},
{
"epoch": 34.59,
"learning_rate": 2.6175986842105262e-05,
"loss": 1.1184,
"step": 11900
},
{
"epoch": 34.88,
"learning_rate": 2.56875e-05,
"loss": 1.1131,
"step": 12000
},
{
"epoch": 34.88,
"eval_loss": 0.3348712623119354,
"eval_runtime": 157.6811,
"eval_samples_per_second": 29.3,
"eval_steps_per_second": 3.666,
"eval_wer": 0.29422382671480146,
"step": 12000
},
{
"epoch": 35.17,
"learning_rate": 2.5194078947368418e-05,
"loss": 1.1025,
"step": 12100
},
{
"epoch": 35.46,
"learning_rate": 2.4700657894736838e-05,
"loss": 1.1069,
"step": 12200
},
{
"epoch": 35.75,
"learning_rate": 2.420723684210526e-05,
"loss": 1.076,
"step": 12300
},
{
"epoch": 36.05,
"learning_rate": 2.371381578947368e-05,
"loss": 1.111,
"step": 12400
},
{
"epoch": 36.34,
"learning_rate": 2.32203947368421e-05,
"loss": 1.1015,
"step": 12500
},
{
"epoch": 36.34,
"eval_loss": 0.3378337025642395,
"eval_runtime": 160.9202,
"eval_samples_per_second": 28.71,
"eval_steps_per_second": 3.592,
"eval_wer": 0.29612066328091535,
"step": 12500
},
{
"epoch": 36.63,
"learning_rate": 2.2726973684210525e-05,
"loss": 1.1047,
"step": 12600
},
{
"epoch": 36.92,
"learning_rate": 2.2233552631578945e-05,
"loss": 1.102,
"step": 12700
},
{
"epoch": 37.21,
"learning_rate": 2.1740131578947365e-05,
"loss": 1.095,
"step": 12800
},
{
"epoch": 37.5,
"learning_rate": 2.124671052631579e-05,
"loss": 1.0709,
"step": 12900
},
{
"epoch": 37.79,
"learning_rate": 2.075328947368421e-05,
"loss": 1.0835,
"step": 13000
},
{
"epoch": 37.79,
"eval_loss": 0.3281959593296051,
"eval_runtime": 159.3279,
"eval_samples_per_second": 28.997,
"eval_steps_per_second": 3.628,
"eval_wer": 0.2865141038976932,
"step": 13000
},
{
"epoch": 38.08,
"learning_rate": 2.0259868421052632e-05,
"loss": 1.0846,
"step": 13100
},
{
"epoch": 38.37,
"learning_rate": 1.9766447368421053e-05,
"loss": 1.0933,
"step": 13200
},
{
"epoch": 38.66,
"learning_rate": 1.927302631578947e-05,
"loss": 1.0661,
"step": 13300
},
{
"epoch": 38.95,
"learning_rate": 1.8779605263157893e-05,
"loss": 1.0902,
"step": 13400
},
{
"epoch": 39.24,
"learning_rate": 1.8286184210526313e-05,
"loss": 1.083,
"step": 13500
},
{
"epoch": 39.24,
"eval_loss": 0.3181643486022949,
"eval_runtime": 158.3527,
"eval_samples_per_second": 29.175,
"eval_steps_per_second": 3.65,
"eval_wer": 0.2825674600746497,
"step": 13500
},
{
"epoch": 39.53,
"learning_rate": 1.7792763157894736e-05,
"loss": 1.0548,
"step": 13600
},
{
"epoch": 39.82,
"learning_rate": 1.7299342105263156e-05,
"loss": 1.0735,
"step": 13700
},
{
"epoch": 40.12,
"learning_rate": 1.6805921052631577e-05,
"loss": 1.085,
"step": 13800
},
{
"epoch": 40.41,
"learning_rate": 1.63125e-05,
"loss": 1.0531,
"step": 13900
},
{
"epoch": 40.7,
"learning_rate": 1.581907894736842e-05,
"loss": 1.0819,
"step": 14000
},
{
"epoch": 40.7,
"eval_loss": 0.32643991708755493,
"eval_runtime": 159.1099,
"eval_samples_per_second": 29.037,
"eval_steps_per_second": 3.633,
"eval_wer": 0.28504558526586304,
"step": 14000
},
{
"epoch": 40.99,
"learning_rate": 1.532565789473684e-05,
"loss": 1.0679,
"step": 14100
},
{
"epoch": 41.28,
"learning_rate": 1.4832236842105262e-05,
"loss": 1.0773,
"step": 14200
},
{
"epoch": 41.57,
"learning_rate": 1.4338815789473682e-05,
"loss": 1.0587,
"step": 14300
},
{
"epoch": 41.86,
"learning_rate": 1.3845394736842104e-05,
"loss": 1.0287,
"step": 14400
},
{
"epoch": 42.15,
"learning_rate": 1.3351973684210524e-05,
"loss": 1.072,
"step": 14500
},
{
"epoch": 42.15,
"eval_loss": 0.32792535424232483,
"eval_runtime": 155.2521,
"eval_samples_per_second": 29.758,
"eval_steps_per_second": 3.723,
"eval_wer": 0.2817108242060821,
"step": 14500
},
{
"epoch": 42.44,
"learning_rate": 1.2863486842105262e-05,
"loss": 1.0508,
"step": 14600
},
{
"epoch": 42.73,
"learning_rate": 1.2370065789473684e-05,
"loss": 1.0408,
"step": 14700
},
{
"epoch": 43.02,
"learning_rate": 1.1876644736842105e-05,
"loss": 1.0725,
"step": 14800
},
{
"epoch": 43.31,
"learning_rate": 1.1383223684210525e-05,
"loss": 1.0381,
"step": 14900
},
{
"epoch": 43.6,
"learning_rate": 1.0889802631578946e-05,
"loss": 1.0456,
"step": 15000
},
{
"epoch": 43.6,
"eval_loss": 0.323445200920105,
"eval_runtime": 137.3413,
"eval_samples_per_second": 33.639,
"eval_steps_per_second": 4.208,
"eval_wer": 0.2792632931530319,
"step": 15000
},
{
"epoch": 43.89,
"learning_rate": 1.0396381578947367e-05,
"loss": 1.0655,
"step": 15100
},
{
"epoch": 44.19,
"learning_rate": 9.90296052631579e-06,
"loss": 1.0509,
"step": 15200
},
{
"epoch": 44.48,
"learning_rate": 9.40953947368421e-06,
"loss": 1.0357,
"step": 15300
},
{
"epoch": 44.77,
"learning_rate": 8.916118421052631e-06,
"loss": 1.0467,
"step": 15400
},
{
"epoch": 45.06,
"learning_rate": 8.422697368421051e-06,
"loss": 1.0581,
"step": 15500
},
{
"epoch": 45.06,
"eval_loss": 0.32200726866722107,
"eval_runtime": 138.9222,
"eval_samples_per_second": 33.256,
"eval_steps_per_second": 4.161,
"eval_wer": 0.27791715107385423,
"step": 15500
},
{
"epoch": 45.35,
"learning_rate": 7.929276315789473e-06,
"loss": 1.0512,
"step": 15600
},
{
"epoch": 45.64,
"learning_rate": 7.435855263157894e-06,
"loss": 1.0397,
"step": 15700
},
{
"epoch": 45.93,
"learning_rate": 6.9473684210526315e-06,
"loss": 1.0455,
"step": 15800
},
{
"epoch": 46.22,
"learning_rate": 6.453947368421052e-06,
"loss": 1.0347,
"step": 15900
},
{
"epoch": 46.51,
"learning_rate": 5.9605263157894735e-06,
"loss": 1.0406,
"step": 16000
},
{
"epoch": 46.51,
"eval_loss": 0.32076749205589294,
"eval_runtime": 134.5334,
"eval_samples_per_second": 34.341,
"eval_steps_per_second": 4.296,
"eval_wer": 0.2762344734748822,
"step": 16000
},
{
"epoch": 46.8,
"learning_rate": 5.467105263157894e-06,
"loss": 1.0528,
"step": 16100
},
{
"epoch": 47.09,
"learning_rate": 4.973684210526316e-06,
"loss": 1.045,
"step": 16200
},
{
"epoch": 47.38,
"learning_rate": 4.480263157894736e-06,
"loss": 1.0129,
"step": 16300
},
{
"epoch": 47.67,
"learning_rate": 3.986842105263157e-06,
"loss": 1.0102,
"step": 16400
},
{
"epoch": 47.96,
"learning_rate": 3.4934210526315787e-06,
"loss": 1.0422,
"step": 16500
},
{
"epoch": 47.96,
"eval_loss": 0.3183736503124237,
"eval_runtime": 137.2442,
"eval_samples_per_second": 33.663,
"eval_steps_per_second": 4.211,
"eval_wer": 0.2751942727773359,
"step": 16500
},
{
"epoch": 48.26,
"learning_rate": 2.9999999999999997e-06,
"loss": 1.0323,
"step": 16600
},
{
"epoch": 48.55,
"learning_rate": 2.506578947368421e-06,
"loss": 1.0292,
"step": 16700
},
{
"epoch": 48.84,
"learning_rate": 2.013157894736842e-06,
"loss": 1.0378,
"step": 16800
},
{
"epoch": 49.13,
"learning_rate": 1.519736842105263e-06,
"loss": 1.035,
"step": 16900
},
{
"epoch": 49.42,
"learning_rate": 1.0263157894736841e-06,
"loss": 1.0099,
"step": 17000
},
{
"epoch": 49.42,
"eval_loss": 0.31810811161994934,
"eval_runtime": 134.533,
"eval_samples_per_second": 34.341,
"eval_steps_per_second": 4.296,
"eval_wer": 0.2734810010402007,
"step": 17000
},
{
"epoch": 49.71,
"learning_rate": 5.328947368421052e-07,
"loss": 1.0338,
"step": 17100
},
{
"epoch": 50.0,
"learning_rate": 3.947368421052631e-08,
"loss": 1.03,
"step": 17200
},
{
"epoch": 50.0,
"step": 17200,
"total_flos": 5.074861809449127e+19,
"train_loss": 1.4555730460410894,
"train_runtime": 29339.4334,
"train_samples_per_second": 18.797,
"train_steps_per_second": 0.586
}
],
"max_steps": 17200,
"num_train_epochs": 50,
"total_flos": 5.074861809449127e+19,
"trial_name": null,
"trial_params": null
}