VERSIL91's picture
Training in progress, step 5, checkpoint
b9d8858 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.03508771929824561,
"eval_steps": 2,
"global_step": 5,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.007017543859649123,
"grad_norm": NaN,
"learning_rate": 1e-05,
"loss": 16.576,
"step": 1
},
{
"epoch": 0.007017543859649123,
"eval_loss": NaN,
"eval_runtime": 56.4868,
"eval_samples_per_second": 4.249,
"eval_steps_per_second": 2.124,
"step": 1
},
{
"epoch": 0.014035087719298246,
"grad_norm": NaN,
"learning_rate": 2e-05,
"loss": 17.6386,
"step": 2
},
{
"epoch": 0.014035087719298246,
"eval_loss": NaN,
"eval_runtime": 5.9101,
"eval_samples_per_second": 40.608,
"eval_steps_per_second": 20.304,
"step": 2
},
{
"epoch": 0.021052631578947368,
"grad_norm": NaN,
"learning_rate": 3e-05,
"loss": 5.3622,
"step": 3
},
{
"epoch": 0.028070175438596492,
"grad_norm": NaN,
"learning_rate": 4e-05,
"loss": 9.3633,
"step": 4
},
{
"epoch": 0.028070175438596492,
"eval_loss": NaN,
"eval_runtime": 7.9437,
"eval_samples_per_second": 30.213,
"eval_steps_per_second": 15.106,
"step": 4
},
{
"epoch": 0.03508771929824561,
"grad_norm": NaN,
"learning_rate": 5e-05,
"loss": 11.4757,
"step": 5
}
],
"logging_steps": 1,
"max_steps": 5,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 2,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3699128496291840.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}