|
{ |
|
"best_metric": 0.4286063983911785, |
|
"best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-4/checkpoint-8552", |
|
"epoch": 4.0, |
|
"eval_steps": 500, |
|
"global_step": 8552, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.23, |
|
"grad_norm": 4.632926940917969, |
|
"learning_rate": 1.2946011611760917e-06, |
|
"loss": 0.6259, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"grad_norm": 5.080592155456543, |
|
"learning_rate": 1.2310780423745168e-06, |
|
"loss": 0.6096, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"grad_norm": 7.2639031410217285, |
|
"learning_rate": 1.1675549235729423e-06, |
|
"loss": 0.5698, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"grad_norm": 12.567307472229004, |
|
"learning_rate": 1.1040318047713676e-06, |
|
"loss": 0.5589, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 0.5995281934738159, |
|
"eval_matthews_correlation": 0.08036809130702588, |
|
"eval_runtime": 0.3414, |
|
"eval_samples_per_second": 3055.256, |
|
"eval_steps_per_second": 193.334, |
|
"step": 2138 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"grad_norm": 15.908562660217285, |
|
"learning_rate": 1.0405086859697929e-06, |
|
"loss": 0.5387, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"grad_norm": 11.577272415161133, |
|
"learning_rate": 9.769855671682182e-07, |
|
"loss": 0.5105, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"grad_norm": 36.38253402709961, |
|
"learning_rate": 9.134624483666437e-07, |
|
"loss": 0.5256, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"grad_norm": 11.872018814086914, |
|
"learning_rate": 8.499393295650689e-07, |
|
"loss": 0.511, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 0.5569708347320557, |
|
"eval_matthews_correlation": 0.39137710349170013, |
|
"eval_runtime": 0.5165, |
|
"eval_samples_per_second": 2019.532, |
|
"eval_steps_per_second": 127.794, |
|
"step": 4276 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"grad_norm": 20.970304489135742, |
|
"learning_rate": 7.864162107634943e-07, |
|
"loss": 0.475, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"grad_norm": 9.216538429260254, |
|
"learning_rate": 7.228930919619195e-07, |
|
"loss": 0.4814, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"grad_norm": 19.621673583984375, |
|
"learning_rate": 6.593699731603449e-07, |
|
"loss": 0.4871, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"grad_norm": 19.49756622314453, |
|
"learning_rate": 5.958468543587703e-07, |
|
"loss": 0.4675, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_loss": 0.5788199305534363, |
|
"eval_matthews_correlation": 0.39891405791978285, |
|
"eval_runtime": 0.3422, |
|
"eval_samples_per_second": 3047.878, |
|
"eval_steps_per_second": 192.867, |
|
"step": 6414 |
|
}, |
|
{ |
|
"epoch": 3.04, |
|
"grad_norm": 49.59273910522461, |
|
"learning_rate": 5.323237355571956e-07, |
|
"loss": 0.5038, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 3.27, |
|
"grad_norm": 34.72640609741211, |
|
"learning_rate": 4.6880061675562097e-07, |
|
"loss": 0.4434, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 3.51, |
|
"grad_norm": 55.005916595458984, |
|
"learning_rate": 4.052774979540463e-07, |
|
"loss": 0.4816, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 3.74, |
|
"grad_norm": 79.78916931152344, |
|
"learning_rate": 3.417543791524717e-07, |
|
"loss": 0.4923, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 3.98, |
|
"grad_norm": 49.529483795166016, |
|
"learning_rate": 2.78231260350897e-07, |
|
"loss": 0.4698, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_loss": 0.6026819944381714, |
|
"eval_matthews_correlation": 0.4286063983911785, |
|
"eval_runtime": 0.3402, |
|
"eval_samples_per_second": 3065.812, |
|
"eval_steps_per_second": 194.002, |
|
"step": 8552 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 10690, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 5, |
|
"save_steps": 500, |
|
"total_flos": 138507342838860.0, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": { |
|
"learning_rate": 1.3581242799776662e-06, |
|
"num_train_epochs": 5, |
|
"per_device_train_batch_size": 4, |
|
"seed": 5 |
|
} |
|
} |
|
|