xls-r-300m-nyanja-model_v1 / trainer_state.json
unza's picture
End of training
e73120a
raw
history blame
3.05 kB
{
"best_metric": 0.27719950675964355,
"best_model_checkpoint": "./xls-r-300m-nyanja-model_v1/checkpoint-1000",
"epoch": 5.0,
"global_step": 1580,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.32,
"learning_rate": 4.9000000000000005e-05,
"loss": 9.53,
"step": 100
},
{
"epoch": 0.63,
"learning_rate": 9.900000000000001e-05,
"loss": 3.1099,
"step": 200
},
{
"epoch": 0.95,
"learning_rate": 0.000149,
"loss": 2.7322,
"step": 300
},
{
"epoch": 1.27,
"learning_rate": 0.000199,
"loss": 1.2227,
"step": 400
},
{
"epoch": 1.58,
"learning_rate": 0.000249,
"loss": 0.7585,
"step": 500
},
{
"epoch": 1.58,
"eval_loss": 0.35743284225463867,
"eval_runtime": 290.0599,
"eval_samples_per_second": 5.806,
"eval_steps_per_second": 0.727,
"eval_wer": 0.9679334916864608,
"step": 500
},
{
"epoch": 1.9,
"learning_rate": 0.000299,
"loss": 0.6269,
"step": 600
},
{
"epoch": 2.22,
"learning_rate": 0.00034899999999999997,
"loss": 0.5478,
"step": 700
},
{
"epoch": 2.53,
"learning_rate": 0.00039900000000000005,
"loss": 0.5208,
"step": 800
},
{
"epoch": 2.85,
"learning_rate": 0.000449,
"loss": 0.5211,
"step": 900
},
{
"epoch": 3.16,
"learning_rate": 0.000499,
"loss": 0.4736,
"step": 1000
},
{
"epoch": 3.16,
"eval_loss": 0.27719950675964355,
"eval_runtime": 292.8589,
"eval_samples_per_second": 5.75,
"eval_steps_per_second": 0.72,
"eval_wer": 0.9073634204275535,
"step": 1000
},
{
"epoch": 3.48,
"learning_rate": 0.000549,
"loss": 0.4533,
"step": 1100
},
{
"epoch": 3.8,
"learning_rate": 0.000599,
"loss": 0.4563,
"step": 1200
},
{
"epoch": 4.11,
"learning_rate": 0.0006490000000000001,
"loss": 0.4883,
"step": 1300
},
{
"epoch": 4.43,
"learning_rate": 0.000699,
"loss": 0.465,
"step": 1400
},
{
"epoch": 4.75,
"learning_rate": 0.000749,
"loss": 0.4776,
"step": 1500
},
{
"epoch": 4.75,
"eval_loss": 0.28532925248146057,
"eval_runtime": 291.534,
"eval_samples_per_second": 5.776,
"eval_steps_per_second": 0.724,
"eval_wer": 0.9578384798099763,
"step": 1500
},
{
"epoch": 5.0,
"step": 1580,
"total_flos": 7.433221301093351e+18,
"train_loss": 1.4388394899006132,
"train_runtime": 7577.7535,
"train_samples_per_second": 3.334,
"train_steps_per_second": 0.209
}
],
"max_steps": 1580,
"num_train_epochs": 5,
"total_flos": 7.433221301093351e+18,
"trial_name": null,
"trial_params": null
}