| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 8.04289544235925, | |
| "eval_steps": 500, | |
| "global_step": 6000, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.6702412868632708, | |
| "grad_norm": 23.28346824645996, | |
| "learning_rate": 9.798511296679954e-06, | |
| "loss": 0.5531, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8025477528572083, | |
| "eval_loss": 0.4079710841178894, | |
| "eval_runtime": 9.2612, | |
| "eval_samples_per_second": 322.097, | |
| "eval_steps_per_second": 20.192, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.3404825737265416, | |
| "grad_norm": 38.64032745361328, | |
| "learning_rate": 9.094595255251797e-06, | |
| "loss": 0.3957, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8276902437210083, | |
| "eval_loss": 0.42224231362342834, | |
| "eval_runtime": 9.5099, | |
| "eval_samples_per_second": 313.673, | |
| "eval_steps_per_second": 19.664, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 2.0107238605898123, | |
| "grad_norm": 34.970176696777344, | |
| "learning_rate": 8.390679213823639e-06, | |
| "loss": 0.3185, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.680965147453083, | |
| "grad_norm": 13.735482215881348, | |
| "learning_rate": 7.686763172395482e-06, | |
| "loss": 0.1887, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.819309413433075, | |
| "eval_loss": 0.49816903471946716, | |
| "eval_runtime": 9.3819, | |
| "eval_samples_per_second": 317.952, | |
| "eval_steps_per_second": 19.932, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 3.351206434316354, | |
| "grad_norm": 10.144265174865723, | |
| "learning_rate": 6.982847130967324e-06, | |
| "loss": 0.1522, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8313778042793274, | |
| "eval_loss": 0.6922349333763123, | |
| "eval_runtime": 9.2835, | |
| "eval_samples_per_second": 321.324, | |
| "eval_steps_per_second": 20.143, | |
| "step": 2984 | |
| }, | |
| { | |
| "epoch": 4.021447721179625, | |
| "grad_norm": 39.181697845458984, | |
| "learning_rate": 6.278931089539166e-06, | |
| "loss": 0.1152, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.6916890080428955, | |
| "grad_norm": 5.586261749267578, | |
| "learning_rate": 5.575015048111009e-06, | |
| "loss": 0.0669, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8253436088562012, | |
| "eval_loss": 0.8140854835510254, | |
| "eval_runtime": 9.3552, | |
| "eval_samples_per_second": 318.859, | |
| "eval_steps_per_second": 19.989, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 5.361930294906166, | |
| "grad_norm": 0.027025775983929634, | |
| "learning_rate": 4.871099006682851e-06, | |
| "loss": 0.0526, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8223265409469604, | |
| "eval_loss": 0.9023452997207642, | |
| "eval_runtime": 9.2601, | |
| "eval_samples_per_second": 322.135, | |
| "eval_steps_per_second": 20.194, | |
| "step": 4476 | |
| }, | |
| { | |
| "epoch": 6.032171581769437, | |
| "grad_norm": 0.05576400086283684, | |
| "learning_rate": 4.167182965254693e-06, | |
| "loss": 0.0417, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.702412868632708, | |
| "grad_norm": 55.70774841308594, | |
| "learning_rate": 3.4632669238265355e-06, | |
| "loss": 0.0301, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8266845345497131, | |
| "eval_loss": 0.9764488339424133, | |
| "eval_runtime": 9.3516, | |
| "eval_samples_per_second": 318.981, | |
| "eval_steps_per_second": 19.996, | |
| "step": 5222 | |
| }, | |
| { | |
| "epoch": 7.372654155495979, | |
| "grad_norm": 0.9753480553627014, | |
| "learning_rate": 2.759350882398378e-06, | |
| "loss": 0.0218, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.8320482969284058, | |
| "eval_loss": 0.9748868346214294, | |
| "eval_runtime": 9.391, | |
| "eval_samples_per_second": 317.644, | |
| "eval_steps_per_second": 19.913, | |
| "step": 5968 | |
| }, | |
| { | |
| "epoch": 8.04289544235925, | |
| "grad_norm": 65.11773681640625, | |
| "learning_rate": 2.0554348409702203e-06, | |
| "loss": 0.0196, | |
| "step": 6000 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 7460, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.5818666004425088e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.0502427338108112e-05, | |
| "per_device_train_batch_size": 16 | |
| } | |
| } | |