Jzuluaga's picture
updating the repo with the fine-tuned model
03154fa
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 21.231422505307854,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.06,
"eval_loss": 3.1697003841400146,
"eval_runtime": 96.1702,
"eval_samples_per_second": 29.344,
"eval_steps_per_second": 2.454,
"eval_wer": 1.0,
"step": 500
},
{
"epoch": 2.12,
"learning_rate": 9.98e-05,
"loss": 3.1489,
"step": 1000
},
{
"epoch": 2.12,
"eval_loss": 1.418416142463684,
"eval_runtime": 83.1708,
"eval_samples_per_second": 33.93,
"eval_steps_per_second": 2.838,
"eval_wer": 0.5677877428998506,
"step": 1000
},
{
"epoch": 3.18,
"eval_loss": 0.8497655987739563,
"eval_runtime": 83.5791,
"eval_samples_per_second": 33.764,
"eval_steps_per_second": 2.824,
"eval_wer": 0.3365844544095665,
"step": 1500
},
{
"epoch": 4.25,
"learning_rate": 8.89111111111111e-05,
"loss": 0.8499,
"step": 2000
},
{
"epoch": 4.25,
"eval_loss": 0.8088647127151489,
"eval_runtime": 83.4386,
"eval_samples_per_second": 33.821,
"eval_steps_per_second": 2.828,
"eval_wer": 0.2755231689088191,
"step": 2000
},
{
"epoch": 5.31,
"eval_loss": 0.7338947057723999,
"eval_runtime": 83.1453,
"eval_samples_per_second": 33.941,
"eval_steps_per_second": 2.838,
"eval_wer": 0.29633781763826605,
"step": 2500
},
{
"epoch": 6.37,
"learning_rate": 7.780000000000001e-05,
"loss": 0.5901,
"step": 3000
},
{
"epoch": 6.37,
"eval_loss": 0.6375774145126343,
"eval_runtime": 83.3907,
"eval_samples_per_second": 33.841,
"eval_steps_per_second": 2.83,
"eval_wer": 0.24024663677130045,
"step": 3000
},
{
"epoch": 7.43,
"eval_loss": 0.688980221748352,
"eval_runtime": 85.6719,
"eval_samples_per_second": 32.94,
"eval_steps_per_second": 2.755,
"eval_wer": 0.2335575485799701,
"step": 3500
},
{
"epoch": 8.49,
"learning_rate": 6.66888888888889e-05,
"loss": 0.4724,
"step": 4000
},
{
"epoch": 8.49,
"eval_loss": 0.6844218969345093,
"eval_runtime": 88.5555,
"eval_samples_per_second": 31.867,
"eval_steps_per_second": 2.665,
"eval_wer": 0.22402840059790732,
"step": 4000
},
{
"epoch": 9.55,
"eval_loss": 0.6900164484977722,
"eval_runtime": 87.8731,
"eval_samples_per_second": 32.114,
"eval_steps_per_second": 2.686,
"eval_wer": 0.22223467862481316,
"step": 4500
},
{
"epoch": 10.62,
"learning_rate": 5.5577777777777784e-05,
"loss": 0.3981,
"step": 5000
},
{
"epoch": 10.62,
"eval_loss": 0.7050736546516418,
"eval_runtime": 88.2857,
"eval_samples_per_second": 31.964,
"eval_steps_per_second": 2.673,
"eval_wer": 0.21229446935724963,
"step": 5000
},
{
"epoch": 11.68,
"eval_loss": 0.6670752167701721,
"eval_runtime": 88.0544,
"eval_samples_per_second": 32.048,
"eval_steps_per_second": 2.68,
"eval_wer": 0.2095291479820628,
"step": 5500
},
{
"epoch": 12.74,
"learning_rate": 4.4466666666666666e-05,
"loss": 0.3436,
"step": 6000
},
{
"epoch": 12.74,
"eval_loss": 0.7424847483634949,
"eval_runtime": 89.6214,
"eval_samples_per_second": 31.488,
"eval_steps_per_second": 2.633,
"eval_wer": 0.20489536621823617,
"step": 6000
},
{
"epoch": 13.8,
"eval_loss": 0.7134818434715271,
"eval_runtime": 96.8293,
"eval_samples_per_second": 29.144,
"eval_steps_per_second": 2.437,
"eval_wer": 0.19943946188340808,
"step": 6500
},
{
"epoch": 14.86,
"learning_rate": 3.3355555555555554e-05,
"loss": 0.2925,
"step": 7000
},
{
"epoch": 14.86,
"eval_loss": 0.7350390553474426,
"eval_runtime": 94.8583,
"eval_samples_per_second": 29.75,
"eval_steps_per_second": 2.488,
"eval_wer": 0.20123318385650224,
"step": 7000
},
{
"epoch": 15.92,
"eval_loss": 0.785513699054718,
"eval_runtime": 92.4034,
"eval_samples_per_second": 30.54,
"eval_steps_per_second": 2.554,
"eval_wer": 0.1945067264573991,
"step": 7500
},
{
"epoch": 16.99,
"learning_rate": 2.225555555555556e-05,
"loss": 0.2525,
"step": 8000
},
{
"epoch": 16.99,
"eval_loss": 0.7933125495910645,
"eval_runtime": 92.9055,
"eval_samples_per_second": 30.375,
"eval_steps_per_second": 2.54,
"eval_wer": 0.19461883408071748,
"step": 8000
},
{
"epoch": 18.05,
"eval_loss": 0.801628589630127,
"eval_runtime": 92.1414,
"eval_samples_per_second": 30.627,
"eval_steps_per_second": 2.561,
"eval_wer": 0.19151718983557547,
"step": 8500
},
{
"epoch": 19.11,
"learning_rate": 1.1155555555555556e-05,
"loss": 0.2285,
"step": 9000
},
{
"epoch": 19.11,
"eval_loss": 0.8283743262290955,
"eval_runtime": 94.3902,
"eval_samples_per_second": 29.897,
"eval_steps_per_second": 2.5,
"eval_wer": 0.1906576980568012,
"step": 9000
},
{
"epoch": 20.17,
"eval_loss": 0.8274713754653931,
"eval_runtime": 89.6678,
"eval_samples_per_second": 31.472,
"eval_steps_per_second": 2.632,
"eval_wer": 0.19024663677130044,
"step": 9500
},
{
"epoch": 21.23,
"learning_rate": 4.444444444444445e-08,
"loss": 0.2025,
"step": 10000
},
{
"epoch": 21.23,
"eval_loss": 0.84695965051651,
"eval_runtime": 89.481,
"eval_samples_per_second": 31.537,
"eval_steps_per_second": 2.637,
"eval_wer": 0.18976083707025412,
"step": 10000
},
{
"epoch": 21.23,
"step": 10000,
"total_flos": 2.495144460541511e+19,
"train_loss": 0.6778903991699219,
"train_runtime": 11598.0436,
"train_samples_per_second": 20.693,
"train_steps_per_second": 0.862
}
],
"max_steps": 10000,
"num_train_epochs": 22,
"total_flos": 2.495144460541511e+19,
"trial_name": null,
"trial_params": null
}