312 lines
7.3 KiB
JSON
312 lines
7.3 KiB
JSON
{
|
|
"best_global_step": 4220,
|
|
"best_metric": 4.378113269805908,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 18.95734597156398,
|
|
"eval_steps": 500,
|
|
"global_step": 8000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.002369668246445498,
|
|
"grad_norm": 20.34552001953125,
|
|
"learning_rate": 0.0,
|
|
"loss": 10.3309,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 1.0193040370941162,
|
|
"learning_rate": 8.42e-05,
|
|
"loss": 5.762,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_loss": 4.781050205230713,
|
|
"eval_runtime": 21.3156,
|
|
"eval_samples_per_second": 1292.106,
|
|
"eval_steps_per_second": 5.067,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"grad_norm": 0.6101345419883728,
|
|
"learning_rate": 9.566919191919192e-05,
|
|
"loss": 4.6298,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"eval_loss": 4.602915287017822,
|
|
"eval_runtime": 22.4738,
|
|
"eval_samples_per_second": 1225.514,
|
|
"eval_steps_per_second": 4.806,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"grad_norm": 0.5042136907577515,
|
|
"learning_rate": 9.03409090909091e-05,
|
|
"loss": 4.5095,
|
|
"step": 1266
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 4.522061347961426,
|
|
"eval_runtime": 31.9704,
|
|
"eval_samples_per_second": 861.483,
|
|
"eval_steps_per_second": 3.378,
|
|
"step": 1266
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"grad_norm": 0.5082475543022156,
|
|
"learning_rate": 8.501262626262628e-05,
|
|
"loss": 4.4388,
|
|
"step": 1688
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"eval_loss": 4.47515869140625,
|
|
"eval_runtime": 31.848,
|
|
"eval_samples_per_second": 864.796,
|
|
"eval_steps_per_second": 3.391,
|
|
"step": 1688
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.5105661749839783,
|
|
"learning_rate": 7.968434343434343e-05,
|
|
"loss": 4.3851,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"eval_loss": 4.442058563232422,
|
|
"eval_runtime": 32.0508,
|
|
"eval_samples_per_second": 859.323,
|
|
"eval_steps_per_second": 3.37,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"grad_norm": 0.5460625886917114,
|
|
"learning_rate": 7.435606060606061e-05,
|
|
"loss": 4.3391,
|
|
"step": 2532
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"eval_loss": 4.4180145263671875,
|
|
"eval_runtime": 30.8915,
|
|
"eval_samples_per_second": 891.572,
|
|
"eval_steps_per_second": 3.496,
|
|
"step": 2532
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"grad_norm": 0.5497872233390808,
|
|
"learning_rate": 6.902777777777779e-05,
|
|
"loss": 4.2975,
|
|
"step": 2954
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"eval_loss": 4.39995813369751,
|
|
"eval_runtime": 32.8624,
|
|
"eval_samples_per_second": 838.1,
|
|
"eval_steps_per_second": 3.286,
|
|
"step": 2954
|
|
},
|
|
{
|
|
"epoch": 8.0,
|
|
"grad_norm": 0.5838669538497925,
|
|
"learning_rate": 6.369949494949495e-05,
|
|
"loss": 4.258,
|
|
"step": 3376
|
|
},
|
|
{
|
|
"epoch": 8.0,
|
|
"eval_loss": 4.387345790863037,
|
|
"eval_runtime": 30.4389,
|
|
"eval_samples_per_second": 904.829,
|
|
"eval_steps_per_second": 3.548,
|
|
"step": 3376
|
|
},
|
|
{
|
|
"epoch": 9.0,
|
|
"grad_norm": 0.6011605262756348,
|
|
"learning_rate": 5.837121212121213e-05,
|
|
"loss": 4.2193,
|
|
"step": 3798
|
|
},
|
|
{
|
|
"epoch": 9.0,
|
|
"eval_loss": 4.380561351776123,
|
|
"eval_runtime": 31.2627,
|
|
"eval_samples_per_second": 880.986,
|
|
"eval_steps_per_second": 3.455,
|
|
"step": 3798
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"grad_norm": 0.6363890171051025,
|
|
"learning_rate": 5.30429292929293e-05,
|
|
"loss": 4.1807,
|
|
"step": 4220
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"eval_loss": 4.378113269805908,
|
|
"eval_runtime": 31.7839,
|
|
"eval_samples_per_second": 866.54,
|
|
"eval_steps_per_second": 3.398,
|
|
"step": 4220
|
|
},
|
|
{
|
|
"epoch": 11.0,
|
|
"grad_norm": 0.663550078868866,
|
|
"learning_rate": 4.771464646464647e-05,
|
|
"loss": 4.1418,
|
|
"step": 4642
|
|
},
|
|
{
|
|
"epoch": 11.0,
|
|
"eval_loss": 4.379039764404297,
|
|
"eval_runtime": 29.9854,
|
|
"eval_samples_per_second": 918.515,
|
|
"eval_steps_per_second": 3.602,
|
|
"step": 4642
|
|
},
|
|
{
|
|
"epoch": 12.0,
|
|
"grad_norm": 0.744036078453064,
|
|
"learning_rate": 4.238636363636364e-05,
|
|
"loss": 4.1029,
|
|
"step": 5064
|
|
},
|
|
{
|
|
"epoch": 12.0,
|
|
"eval_loss": 4.383896350860596,
|
|
"eval_runtime": 25.563,
|
|
"eval_samples_per_second": 1077.418,
|
|
"eval_steps_per_second": 4.225,
|
|
"step": 5064
|
|
},
|
|
{
|
|
"epoch": 13.0,
|
|
"grad_norm": 0.8011893630027771,
|
|
"learning_rate": 3.705808080808081e-05,
|
|
"loss": 4.0643,
|
|
"step": 5486
|
|
},
|
|
{
|
|
"epoch": 13.0,
|
|
"eval_loss": 4.391408920288086,
|
|
"eval_runtime": 30.7558,
|
|
"eval_samples_per_second": 895.506,
|
|
"eval_steps_per_second": 3.512,
|
|
"step": 5486
|
|
},
|
|
{
|
|
"epoch": 14.0,
|
|
"grad_norm": 0.8544487357139587,
|
|
"learning_rate": 3.172979797979798e-05,
|
|
"loss": 4.0259,
|
|
"step": 5908
|
|
},
|
|
{
|
|
"epoch": 14.0,
|
|
"eval_loss": 4.40255880355835,
|
|
"eval_runtime": 31.2899,
|
|
"eval_samples_per_second": 880.219,
|
|
"eval_steps_per_second": 3.452,
|
|
"step": 5908
|
|
},
|
|
{
|
|
"epoch": 15.0,
|
|
"grad_norm": 0.9503664970397949,
|
|
"learning_rate": 2.6401515151515155e-05,
|
|
"loss": 3.9896,
|
|
"step": 6330
|
|
},
|
|
{
|
|
"epoch": 15.0,
|
|
"eval_loss": 4.4131035804748535,
|
|
"eval_runtime": 21.2356,
|
|
"eval_samples_per_second": 1296.974,
|
|
"eval_steps_per_second": 5.086,
|
|
"step": 6330
|
|
},
|
|
{
|
|
"epoch": 16.0,
|
|
"grad_norm": 1.0018138885498047,
|
|
"learning_rate": 2.1073232323232324e-05,
|
|
"loss": 3.9551,
|
|
"step": 6752
|
|
},
|
|
{
|
|
"epoch": 16.0,
|
|
"eval_loss": 4.427628040313721,
|
|
"eval_runtime": 21.2661,
|
|
"eval_samples_per_second": 1295.115,
|
|
"eval_steps_per_second": 5.079,
|
|
"step": 6752
|
|
},
|
|
{
|
|
"epoch": 17.0,
|
|
"grad_norm": 1.0533849000930786,
|
|
"learning_rate": 1.5744949494949496e-05,
|
|
"loss": 3.9231,
|
|
"step": 7174
|
|
},
|
|
{
|
|
"epoch": 17.0,
|
|
"eval_loss": 4.4376349449157715,
|
|
"eval_runtime": 21.2661,
|
|
"eval_samples_per_second": 1295.114,
|
|
"eval_steps_per_second": 5.079,
|
|
"step": 7174
|
|
},
|
|
{
|
|
"epoch": 18.0,
|
|
"grad_norm": 1.074864149093628,
|
|
"learning_rate": 1.0416666666666668e-05,
|
|
"loss": 3.8953,
|
|
"step": 7596
|
|
},
|
|
{
|
|
"epoch": 18.0,
|
|
"eval_loss": 4.450303077697754,
|
|
"eval_runtime": 21.2275,
|
|
"eval_samples_per_second": 1297.471,
|
|
"eval_steps_per_second": 5.088,
|
|
"step": 7596
|
|
}
|
|
],
|
|
"logging_steps": 500,
|
|
"max_steps": 8440,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 20,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1.33770938056704e+17,
|
|
"train_batch_size": 256,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|