patrickvonplaten's picture
up
ffb6ea4
raw
history blame
7.92 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 125.0,
"global_step": 4500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 2.78,
"learning_rate": 7.2e-06,
"loss": 22.7336,
"step": 100
},
{
"epoch": 5.56,
"learning_rate": 1.47e-05,
"loss": 8.3017,
"step": 200
},
{
"epoch": 8.33,
"learning_rate": 2.2199999999999998e-05,
"loss": 5.4371,
"step": 300
},
{
"epoch": 11.11,
"learning_rate": 2.97e-05,
"loss": 4.1583,
"step": 400
},
{
"epoch": 13.89,
"learning_rate": 3.7199999999999996e-05,
"loss": 3.6082,
"step": 500
},
{
"epoch": 13.89,
"eval_loss": 3.501119613647461,
"eval_runtime": 2.3778,
"eval_samples_per_second": 146.354,
"eval_steps_per_second": 4.626,
"eval_wer": 0.9825350621857635,
"step": 500
},
{
"epoch": 16.67,
"learning_rate": 4.4699999999999996e-05,
"loss": 3.4894,
"step": 600
},
{
"epoch": 19.44,
"learning_rate": 5.2199999999999995e-05,
"loss": 3.448,
"step": 700
},
{
"epoch": 22.22,
"learning_rate": 5.97e-05,
"loss": 3.4155,
"step": 800
},
{
"epoch": 25.0,
"learning_rate": 6.72e-05,
"loss": 3.3702,
"step": 900
},
{
"epoch": 27.78,
"learning_rate": 7.47e-05,
"loss": 3.2566,
"step": 1000
},
{
"epoch": 27.78,
"eval_loss": 2.9146299362182617,
"eval_runtime": 2.224,
"eval_samples_per_second": 156.477,
"eval_steps_per_second": 4.946,
"eval_wer": 0.9367557554908706,
"step": 1000
},
{
"epoch": 30.56,
"learning_rate": 7.336363636363636e-05,
"loss": 2.832,
"step": 1100
},
{
"epoch": 33.33,
"learning_rate": 7.16590909090909e-05,
"loss": 2.2848,
"step": 1200
},
{
"epoch": 36.11,
"learning_rate": 6.995454545454545e-05,
"loss": 1.9336,
"step": 1300
},
{
"epoch": 38.89,
"learning_rate": 6.824999999999999e-05,
"loss": 1.7366,
"step": 1400
},
{
"epoch": 41.67,
"learning_rate": 6.654545454545455e-05,
"loss": 1.5744,
"step": 1500
},
{
"epoch": 41.67,
"eval_loss": 0.6705893278121948,
"eval_runtime": 2.2526,
"eval_samples_per_second": 154.49,
"eval_steps_per_second": 4.883,
"eval_wer": 0.20706536120666844,
"step": 1500
},
{
"epoch": 44.44,
"learning_rate": 6.484090909090909e-05,
"loss": 1.4626,
"step": 1600
},
{
"epoch": 47.22,
"learning_rate": 6.313636363636363e-05,
"loss": 1.4203,
"step": 1700
},
{
"epoch": 50.0,
"learning_rate": 6.143181818181818e-05,
"loss": 1.3699,
"step": 1800
},
{
"epoch": 52.78,
"learning_rate": 5.9727272727272724e-05,
"loss": 1.3123,
"step": 1900
},
{
"epoch": 55.56,
"learning_rate": 5.802272727272727e-05,
"loss": 1.2744,
"step": 2000
},
{
"epoch": 55.56,
"eval_loss": 0.540650486946106,
"eval_runtime": 2.2596,
"eval_samples_per_second": 154.01,
"eval_steps_per_second": 4.868,
"eval_wer": 0.1749139984122784,
"step": 2000
},
{
"epoch": 58.33,
"learning_rate": 5.631818181818181e-05,
"loss": 1.2163,
"step": 2100
},
{
"epoch": 61.11,
"learning_rate": 5.463068181818181e-05,
"loss": 1.187,
"step": 2200
},
{
"epoch": 63.89,
"learning_rate": 5.292613636363636e-05,
"loss": 1.1699,
"step": 2300
},
{
"epoch": 66.67,
"learning_rate": 5.122159090909091e-05,
"loss": 1.1208,
"step": 2400
},
{
"epoch": 69.44,
"learning_rate": 4.9517045454545456e-05,
"loss": 1.0734,
"step": 2500
},
{
"epoch": 69.44,
"eval_loss": 0.527560830116272,
"eval_runtime": 2.3157,
"eval_samples_per_second": 150.278,
"eval_steps_per_second": 4.75,
"eval_wer": 0.17610478962688542,
"step": 2500
},
{
"epoch": 72.22,
"learning_rate": 4.781249999999999e-05,
"loss": 1.0427,
"step": 2600
},
{
"epoch": 75.0,
"learning_rate": 4.610795454545454e-05,
"loss": 1.0227,
"step": 2700
},
{
"epoch": 77.78,
"learning_rate": 4.4420454545454544e-05,
"loss": 0.9966,
"step": 2800
},
{
"epoch": 80.56,
"learning_rate": 4.271590909090909e-05,
"loss": 0.9804,
"step": 2900
},
{
"epoch": 83.33,
"learning_rate": 4.101136363636364e-05,
"loss": 0.9451,
"step": 3000
},
{
"epoch": 83.33,
"eval_loss": 0.4845634400844574,
"eval_runtime": 2.2522,
"eval_samples_per_second": 154.518,
"eval_steps_per_second": 4.884,
"eval_wer": 0.17068007409367558,
"step": 3000
},
{
"epoch": 86.11,
"learning_rate": 3.9306818181818174e-05,
"loss": 0.9217,
"step": 3100
},
{
"epoch": 88.89,
"learning_rate": 3.760227272727272e-05,
"loss": 0.904,
"step": 3200
},
{
"epoch": 91.67,
"learning_rate": 3.591477272727273e-05,
"loss": 0.8857,
"step": 3300
},
{
"epoch": 94.44,
"learning_rate": 3.421022727272727e-05,
"loss": 0.88,
"step": 3400
},
{
"epoch": 97.22,
"learning_rate": 3.250568181818182e-05,
"loss": 0.8559,
"step": 3500
},
{
"epoch": 97.22,
"eval_loss": 0.494262158870697,
"eval_runtime": 2.0455,
"eval_samples_per_second": 170.13,
"eval_steps_per_second": 5.378,
"eval_wer": 0.1672400105848108,
"step": 3500
},
{
"epoch": 100.0,
"learning_rate": 3.0801136363636364e-05,
"loss": 0.819,
"step": 3600
},
{
"epoch": 102.78,
"learning_rate": 2.9096590909090905e-05,
"loss": 0.8242,
"step": 3700
},
{
"epoch": 105.56,
"learning_rate": 2.7392045454545453e-05,
"loss": 0.7983,
"step": 3800
},
{
"epoch": 108.33,
"learning_rate": 2.56875e-05,
"loss": 0.7911,
"step": 3900
},
{
"epoch": 111.11,
"learning_rate": 2.3982954545454542e-05,
"loss": 0.7725,
"step": 4000
},
{
"epoch": 111.11,
"eval_loss": 0.4919077455997467,
"eval_runtime": 2.1172,
"eval_samples_per_second": 164.365,
"eval_steps_per_second": 5.195,
"eval_wer": 0.1672400105848108,
"step": 4000
},
{
"epoch": 113.89,
"learning_rate": 2.227840909090909e-05,
"loss": 0.7691,
"step": 4100
},
{
"epoch": 116.67,
"learning_rate": 2.0573863636363634e-05,
"loss": 0.739,
"step": 4200
},
{
"epoch": 119.44,
"learning_rate": 1.886931818181818e-05,
"loss": 0.7452,
"step": 4300
},
{
"epoch": 122.22,
"learning_rate": 1.7164772727272726e-05,
"loss": 0.7218,
"step": 4400
},
{
"epoch": 125.0,
"learning_rate": 1.546022727272727e-05,
"loss": 0.7171,
"step": 4500
},
{
"epoch": 125.0,
"eval_loss": 0.5054941773414612,
"eval_runtime": 2.2494,
"eval_samples_per_second": 154.711,
"eval_steps_per_second": 4.89,
"eval_wer": 0.16115374437681926,
"step": 4500
}
],
"max_steps": 5400,
"num_train_epochs": 150,
"total_flos": 1.5292173814083879e+19,
"trial_name": null,
"trial_params": null
}