|
{ |
|
"best_metric": 0.4472995698451996, |
|
"best_model_checkpoint": "AlexWang99/byt5_add_1k/checkpoint-60", |
|
"epoch": 30.0, |
|
"eval_steps": 500, |
|
"global_step": 60, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 0.6308889389038086, |
|
"eval_runtime": 10.5961, |
|
"eval_samples_per_second": 943.743, |
|
"eval_steps_per_second": 1.227, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_loss": 0.6276265978813171, |
|
"eval_runtime": 10.6791, |
|
"eval_samples_per_second": 936.411, |
|
"eval_steps_per_second": 1.217, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_loss": 0.6149245500564575, |
|
"eval_runtime": 10.8446, |
|
"eval_samples_per_second": 922.121, |
|
"eval_steps_per_second": 1.199, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_loss": 0.6245794296264648, |
|
"eval_runtime": 10.6658, |
|
"eval_samples_per_second": 937.573, |
|
"eval_steps_per_second": 1.219, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_loss": 0.6215338110923767, |
|
"eval_runtime": 10.8289, |
|
"eval_samples_per_second": 923.456, |
|
"eval_steps_per_second": 1.2, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_loss": 0.6247537136077881, |
|
"eval_runtime": 10.7084, |
|
"eval_samples_per_second": 933.843, |
|
"eval_steps_per_second": 1.214, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"eval_loss": 0.6244410872459412, |
|
"eval_runtime": 10.8139, |
|
"eval_samples_per_second": 924.732, |
|
"eval_steps_per_second": 1.202, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_loss": 0.6034420728683472, |
|
"eval_runtime": 10.7264, |
|
"eval_samples_per_second": 932.277, |
|
"eval_steps_per_second": 1.212, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"eval_loss": 0.585522472858429, |
|
"eval_runtime": 10.8073, |
|
"eval_samples_per_second": 925.3, |
|
"eval_steps_per_second": 1.203, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_loss": 0.5661865472793579, |
|
"eval_runtime": 10.713, |
|
"eval_samples_per_second": 933.448, |
|
"eval_steps_per_second": 1.213, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 11.0, |
|
"eval_loss": 0.5532593727111816, |
|
"eval_runtime": 10.858, |
|
"eval_samples_per_second": 920.977, |
|
"eval_steps_per_second": 1.197, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 12.0, |
|
"eval_loss": 0.5615524053573608, |
|
"eval_runtime": 10.7654, |
|
"eval_samples_per_second": 928.899, |
|
"eval_steps_per_second": 1.208, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 13.0, |
|
"eval_loss": 0.5644332766532898, |
|
"eval_runtime": 10.8917, |
|
"eval_samples_per_second": 918.128, |
|
"eval_steps_per_second": 1.194, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 14.0, |
|
"eval_loss": 0.5305653214454651, |
|
"eval_runtime": 10.768, |
|
"eval_samples_per_second": 928.681, |
|
"eval_steps_per_second": 1.207, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 15.0, |
|
"eval_loss": 0.4982789158821106, |
|
"eval_runtime": 10.828, |
|
"eval_samples_per_second": 923.53, |
|
"eval_steps_per_second": 1.201, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 16.0, |
|
"eval_loss": 0.49113744497299194, |
|
"eval_runtime": 10.7309, |
|
"eval_samples_per_second": 931.888, |
|
"eval_steps_per_second": 1.211, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 17.0, |
|
"eval_loss": 0.5076947808265686, |
|
"eval_runtime": 10.8747, |
|
"eval_samples_per_second": 919.562, |
|
"eval_steps_per_second": 1.195, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 18.0, |
|
"eval_loss": 0.5289499163627625, |
|
"eval_runtime": 10.7947, |
|
"eval_samples_per_second": 926.376, |
|
"eval_steps_per_second": 1.204, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 19.0, |
|
"eval_loss": 0.5172982215881348, |
|
"eval_runtime": 10.7036, |
|
"eval_samples_per_second": 934.267, |
|
"eval_steps_per_second": 1.215, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 20.0, |
|
"eval_loss": 0.4997369050979614, |
|
"eval_runtime": 10.8464, |
|
"eval_samples_per_second": 921.968, |
|
"eval_steps_per_second": 1.199, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 21.0, |
|
"eval_loss": 0.4779321253299713, |
|
"eval_runtime": 10.704, |
|
"eval_samples_per_second": 934.226, |
|
"eval_steps_per_second": 1.214, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 22.0, |
|
"eval_loss": 0.4699294865131378, |
|
"eval_runtime": 10.847, |
|
"eval_samples_per_second": 921.918, |
|
"eval_steps_per_second": 1.198, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 23.0, |
|
"eval_loss": 0.47573691606521606, |
|
"eval_runtime": 10.773, |
|
"eval_samples_per_second": 928.246, |
|
"eval_steps_per_second": 1.207, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 24.0, |
|
"eval_loss": 0.4945220649242401, |
|
"eval_runtime": 10.9053, |
|
"eval_samples_per_second": 916.987, |
|
"eval_steps_per_second": 1.192, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 25.0, |
|
"eval_loss": 0.5078331828117371, |
|
"eval_runtime": 10.7019, |
|
"eval_samples_per_second": 934.413, |
|
"eval_steps_per_second": 1.215, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 26.0, |
|
"eval_loss": 0.49670201539993286, |
|
"eval_runtime": 10.7461, |
|
"eval_samples_per_second": 930.57, |
|
"eval_steps_per_second": 1.21, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 27.0, |
|
"eval_loss": 0.4805784523487091, |
|
"eval_runtime": 10.7106, |
|
"eval_samples_per_second": 933.651, |
|
"eval_steps_per_second": 1.214, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 28.0, |
|
"eval_loss": 0.4583061635494232, |
|
"eval_runtime": 10.7342, |
|
"eval_samples_per_second": 931.602, |
|
"eval_steps_per_second": 1.211, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 29.0, |
|
"eval_loss": 0.4492928683757782, |
|
"eval_runtime": 10.8064, |
|
"eval_samples_per_second": 925.38, |
|
"eval_steps_per_second": 1.203, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 30.0, |
|
"eval_loss": 0.4472995698451996, |
|
"eval_runtime": 10.7236, |
|
"eval_samples_per_second": 932.526, |
|
"eval_steps_per_second": 1.212, |
|
"step": 60 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 160, |
|
"num_train_epochs": 80, |
|
"save_steps": 500, |
|
"total_flos": 861327728640000.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|