infinitejoy's picture
End of training
c790bfc
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 120.0,
"global_step": 10080,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.19,
"learning_rate": 3.3949999999999997e-06,
"loss": 16.6111,
"step": 100
},
{
"epoch": 2.38,
"learning_rate": 6.895e-06,
"loss": 8.8994,
"step": 200
},
{
"epoch": 3.57,
"learning_rate": 1.0394999999999998e-05,
"loss": 5.2091,
"step": 300
},
{
"epoch": 4.76,
"learning_rate": 1.3895e-05,
"loss": 4.3004,
"step": 400
},
{
"epoch": 5.95,
"learning_rate": 1.7395e-05,
"loss": 3.6606,
"step": 500
},
{
"epoch": 7.14,
"learning_rate": 2.0894999999999996e-05,
"loss": 3.2383,
"step": 600
},
{
"epoch": 8.33,
"learning_rate": 2.4394999999999996e-05,
"loss": 3.0354,
"step": 700
},
{
"epoch": 9.52,
"learning_rate": 2.7895e-05,
"loss": 2.9742,
"step": 800
},
{
"epoch": 10.71,
"learning_rate": 3.1395e-05,
"loss": 2.9316,
"step": 900
},
{
"epoch": 11.9,
"learning_rate": 3.4895e-05,
"loss": 2.8914,
"step": 1000
},
{
"epoch": 13.1,
"learning_rate": 3.8394999999999994e-05,
"loss": 2.8328,
"step": 1100
},
{
"epoch": 14.29,
"learning_rate": 4.1895e-05,
"loss": 2.8058,
"step": 1200
},
{
"epoch": 15.48,
"learning_rate": 4.5394999999999995e-05,
"loss": 2.7434,
"step": 1300
},
{
"epoch": 16.67,
"learning_rate": 4.8895e-05,
"loss": 2.2153,
"step": 1400
},
{
"epoch": 17.86,
"learning_rate": 5.2395e-05,
"loss": 1.606,
"step": 1500
},
{
"epoch": 19.05,
"learning_rate": 5.589499999999999e-05,
"loss": 1.3918,
"step": 1600
},
{
"epoch": 20.24,
"learning_rate": 5.9394999999999996e-05,
"loss": 1.2684,
"step": 1700
},
{
"epoch": 21.43,
"learning_rate": 6.289499999999999e-05,
"loss": 1.193,
"step": 1800
},
{
"epoch": 22.62,
"learning_rate": 6.639499999999999e-05,
"loss": 1.152,
"step": 1900
},
{
"epoch": 23.81,
"learning_rate": 6.9895e-05,
"loss": 1.1004,
"step": 2000
},
{
"epoch": 23.81,
"eval_loss": 0.371005654335022,
"eval_runtime": 72.2015,
"eval_samples_per_second": 17.174,
"eval_steps_per_second": 17.174,
"eval_wer": 0.41910403554239173,
"step": 2000
},
{
"epoch": 25.0,
"learning_rate": 6.915965346534654e-05,
"loss": 1.0522,
"step": 2100
},
{
"epoch": 26.19,
"learning_rate": 6.829331683168317e-05,
"loss": 1.0103,
"step": 2200
},
{
"epoch": 27.38,
"learning_rate": 6.74269801980198e-05,
"loss": 0.9936,
"step": 2300
},
{
"epoch": 28.57,
"learning_rate": 6.656064356435643e-05,
"loss": 0.9501,
"step": 2400
},
{
"epoch": 29.76,
"learning_rate": 6.569430693069306e-05,
"loss": 0.9198,
"step": 2500
},
{
"epoch": 30.95,
"learning_rate": 6.48279702970297e-05,
"loss": 0.9018,
"step": 2600
},
{
"epoch": 32.14,
"learning_rate": 6.396163366336633e-05,
"loss": 0.8733,
"step": 2700
},
{
"epoch": 33.33,
"learning_rate": 6.309529702970296e-05,
"loss": 0.8516,
"step": 2800
},
{
"epoch": 34.52,
"learning_rate": 6.222896039603959e-05,
"loss": 0.8474,
"step": 2900
},
{
"epoch": 35.71,
"learning_rate": 6.136262376237622e-05,
"loss": 0.8208,
"step": 3000
},
{
"epoch": 36.9,
"learning_rate": 6.049628712871287e-05,
"loss": 0.8079,
"step": 3100
},
{
"epoch": 38.1,
"learning_rate": 5.9629950495049504e-05,
"loss": 0.7882,
"step": 3200
},
{
"epoch": 39.29,
"learning_rate": 5.8763613861386134e-05,
"loss": 0.7741,
"step": 3300
},
{
"epoch": 40.48,
"learning_rate": 5.789727722772277e-05,
"loss": 0.7725,
"step": 3400
},
{
"epoch": 41.67,
"learning_rate": 5.70309405940594e-05,
"loss": 0.7445,
"step": 3500
},
{
"epoch": 42.86,
"learning_rate": 5.616460396039603e-05,
"loss": 0.7391,
"step": 3600
},
{
"epoch": 44.05,
"learning_rate": 5.5306930693069306e-05,
"loss": 0.7449,
"step": 3700
},
{
"epoch": 45.24,
"learning_rate": 5.4440594059405936e-05,
"loss": 0.7294,
"step": 3800
},
{
"epoch": 46.43,
"learning_rate": 5.357425742574257e-05,
"loss": 0.7163,
"step": 3900
},
{
"epoch": 47.62,
"learning_rate": 5.27079207920792e-05,
"loss": 0.7002,
"step": 4000
},
{
"epoch": 47.62,
"eval_loss": 0.23418071866035461,
"eval_runtime": 71.5351,
"eval_samples_per_second": 17.334,
"eval_steps_per_second": 17.334,
"eval_wer": 0.2562014068863384,
"step": 4000
},
{
"epoch": 48.81,
"learning_rate": 5.1841584158415834e-05,
"loss": 0.6952,
"step": 4100
},
{
"epoch": 50.0,
"learning_rate": 5.097524752475247e-05,
"loss": 0.685,
"step": 4200
},
{
"epoch": 51.19,
"learning_rate": 5.01089108910891e-05,
"loss": 0.6711,
"step": 4300
},
{
"epoch": 52.38,
"learning_rate": 4.924257425742574e-05,
"loss": 0.6592,
"step": 4400
},
{
"epoch": 53.57,
"learning_rate": 4.837623762376237e-05,
"loss": 0.659,
"step": 4500
},
{
"epoch": 54.76,
"learning_rate": 4.750990099009901e-05,
"loss": 0.6529,
"step": 4600
},
{
"epoch": 55.95,
"learning_rate": 4.664356435643564e-05,
"loss": 0.6481,
"step": 4700
},
{
"epoch": 57.14,
"learning_rate": 4.577722772277227e-05,
"loss": 0.6411,
"step": 4800
},
{
"epoch": 58.33,
"learning_rate": 4.491089108910891e-05,
"loss": 0.6292,
"step": 4900
},
{
"epoch": 59.52,
"learning_rate": 4.404455445544554e-05,
"loss": 0.6158,
"step": 5000
},
{
"epoch": 60.71,
"learning_rate": 4.3178217821782176e-05,
"loss": 0.6152,
"step": 5100
},
{
"epoch": 61.9,
"learning_rate": 4.2311881188118806e-05,
"loss": 0.6095,
"step": 5200
},
{
"epoch": 63.1,
"learning_rate": 4.144554455445544e-05,
"loss": 0.5974,
"step": 5300
},
{
"epoch": 64.29,
"learning_rate": 4.057920792079207e-05,
"loss": 0.5963,
"step": 5400
},
{
"epoch": 65.48,
"learning_rate": 3.9712871287128704e-05,
"loss": 0.602,
"step": 5500
},
{
"epoch": 66.67,
"learning_rate": 3.884653465346535e-05,
"loss": 0.5889,
"step": 5600
},
{
"epoch": 67.86,
"learning_rate": 3.798886138613861e-05,
"loss": 0.5811,
"step": 5700
},
{
"epoch": 69.05,
"learning_rate": 3.7122524752475246e-05,
"loss": 0.5693,
"step": 5800
},
{
"epoch": 70.24,
"learning_rate": 3.6256188118811876e-05,
"loss": 0.5674,
"step": 5900
},
{
"epoch": 71.43,
"learning_rate": 3.5389851485148506e-05,
"loss": 0.5573,
"step": 6000
},
{
"epoch": 71.43,
"eval_loss": 0.21746528148651123,
"eval_runtime": 71.6386,
"eval_samples_per_second": 17.309,
"eval_steps_per_second": 17.309,
"eval_wer": 0.21769714920399852,
"step": 6000
},
{
"epoch": 72.62,
"learning_rate": 3.452351485148514e-05,
"loss": 0.5548,
"step": 6100
},
{
"epoch": 73.81,
"learning_rate": 3.365717821782178e-05,
"loss": 0.5465,
"step": 6200
},
{
"epoch": 75.0,
"learning_rate": 3.279084158415842e-05,
"loss": 0.5558,
"step": 6300
},
{
"epoch": 76.19,
"learning_rate": 3.192450495049505e-05,
"loss": 0.5373,
"step": 6400
},
{
"epoch": 77.38,
"learning_rate": 3.1058168316831684e-05,
"loss": 0.5469,
"step": 6500
},
{
"epoch": 78.57,
"learning_rate": 3.0191831683168314e-05,
"loss": 0.5386,
"step": 6600
},
{
"epoch": 79.76,
"learning_rate": 2.9325495049504948e-05,
"loss": 0.5402,
"step": 6700
},
{
"epoch": 80.95,
"learning_rate": 2.845915841584158e-05,
"loss": 0.5313,
"step": 6800
},
{
"epoch": 82.14,
"learning_rate": 2.7592821782178218e-05,
"loss": 0.523,
"step": 6900
},
{
"epoch": 83.33,
"learning_rate": 2.672648514851485e-05,
"loss": 0.5107,
"step": 7000
},
{
"epoch": 84.52,
"learning_rate": 2.5860148514851482e-05,
"loss": 0.5147,
"step": 7100
},
{
"epoch": 85.71,
"learning_rate": 2.4993811881188116e-05,
"loss": 0.5185,
"step": 7200
},
{
"epoch": 86.9,
"learning_rate": 2.412747524752475e-05,
"loss": 0.5066,
"step": 7300
},
{
"epoch": 88.1,
"learning_rate": 2.3261138613861386e-05,
"loss": 0.4988,
"step": 7400
},
{
"epoch": 89.29,
"learning_rate": 2.239480198019802e-05,
"loss": 0.4997,
"step": 7500
},
{
"epoch": 90.48,
"learning_rate": 2.152846534653465e-05,
"loss": 0.5079,
"step": 7600
},
{
"epoch": 91.67,
"learning_rate": 2.0662128712871283e-05,
"loss": 0.5039,
"step": 7700
},
{
"epoch": 92.86,
"learning_rate": 1.9795792079207917e-05,
"loss": 0.4924,
"step": 7800
},
{
"epoch": 94.05,
"learning_rate": 1.8929455445544554e-05,
"loss": 0.4821,
"step": 7900
},
{
"epoch": 95.24,
"learning_rate": 1.8063118811881187e-05,
"loss": 0.4799,
"step": 8000
},
{
"epoch": 95.24,
"eval_loss": 0.21092435717582703,
"eval_runtime": 72.1193,
"eval_samples_per_second": 17.194,
"eval_steps_per_second": 17.194,
"eval_wer": 0.19872269529803777,
"step": 8000
},
{
"epoch": 96.43,
"learning_rate": 1.719678217821782e-05,
"loss": 0.485,
"step": 8100
},
{
"epoch": 97.62,
"learning_rate": 1.6330445544554455e-05,
"loss": 0.4806,
"step": 8200
},
{
"epoch": 98.81,
"learning_rate": 1.5464108910891088e-05,
"loss": 0.4726,
"step": 8300
},
{
"epoch": 100.0,
"learning_rate": 1.459777227722772e-05,
"loss": 0.4649,
"step": 8400
},
{
"epoch": 101.19,
"learning_rate": 1.3731435643564355e-05,
"loss": 0.4752,
"step": 8500
},
{
"epoch": 102.38,
"learning_rate": 1.2865099009900989e-05,
"loss": 0.467,
"step": 8600
},
{
"epoch": 103.57,
"learning_rate": 1.1998762376237624e-05,
"loss": 0.4623,
"step": 8700
},
{
"epoch": 104.76,
"learning_rate": 1.1132425742574256e-05,
"loss": 0.4651,
"step": 8800
},
{
"epoch": 105.95,
"learning_rate": 1.026608910891089e-05,
"loss": 0.469,
"step": 8900
},
{
"epoch": 107.14,
"learning_rate": 9.399752475247525e-06,
"loss": 0.464,
"step": 9000
},
{
"epoch": 108.33,
"learning_rate": 8.533415841584157e-06,
"loss": 0.4614,
"step": 9100
},
{
"epoch": 109.52,
"learning_rate": 7.667079207920792e-06,
"loss": 0.4526,
"step": 9200
},
{
"epoch": 110.71,
"learning_rate": 6.800742574257426e-06,
"loss": 0.4485,
"step": 9300
},
{
"epoch": 111.9,
"learning_rate": 5.934405940594059e-06,
"loss": 0.4532,
"step": 9400
},
{
"epoch": 113.1,
"learning_rate": 5.068069306930692e-06,
"loss": 0.4417,
"step": 9500
},
{
"epoch": 114.29,
"learning_rate": 4.21039603960396e-06,
"loss": 0.4453,
"step": 9600
},
{
"epoch": 115.48,
"learning_rate": 3.344059405940594e-06,
"loss": 0.4517,
"step": 9700
},
{
"epoch": 116.67,
"learning_rate": 2.477722772277227e-06,
"loss": 0.4399,
"step": 9800
},
{
"epoch": 117.86,
"learning_rate": 1.6113861386138611e-06,
"loss": 0.4467,
"step": 9900
},
{
"epoch": 119.05,
"learning_rate": 7.45049504950495e-07,
"loss": 0.4511,
"step": 10000
},
{
"epoch": 119.05,
"eval_loss": 0.2164391726255417,
"eval_runtime": 71.6824,
"eval_samples_per_second": 17.299,
"eval_steps_per_second": 17.299,
"eval_wer": 0.19751943724546464,
"step": 10000
},
{
"epoch": 120.0,
"step": 10080,
"total_flos": 5.597929800172578e+19,
"train_loss": 1.2021882954097929,
"train_runtime": 20445.8304,
"train_samples_per_second": 15.7,
"train_steps_per_second": 0.493
}
],
"max_steps": 10080,
"num_train_epochs": 120,
"total_flos": 5.597929800172578e+19,
"trial_name": null,
"trial_params": null
}