| { | |
| "best_metric": 0.05549783632159233, | |
| "best_model_checkpoint": "./vit-base-beans/checkpoint-160", | |
| "epoch": 3.6363636363636362, | |
| "eval_steps": 40, | |
| "global_step": 160, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.8493722081184387, | |
| "learning_rate": 0.00018863636363636364, | |
| "loss": 0.444, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.272053599357605, | |
| "learning_rate": 0.00017727272727272728, | |
| "loss": 0.151, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.1838751882314682, | |
| "learning_rate": 0.00016590909090909094, | |
| "loss": 0.117, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 4.570974349975586, | |
| "learning_rate": 0.00015454545454545454, | |
| "loss": 0.2073, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "eval_accuracy": 0.975, | |
| "eval_loss": 0.09584420919418335, | |
| "eval_runtime": 57.723, | |
| "eval_samples_per_second": 3.465, | |
| "eval_steps_per_second": 0.433, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 0.50724858045578, | |
| "learning_rate": 0.0001431818181818182, | |
| "loss": 0.0636, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 0.09627090394496918, | |
| "learning_rate": 0.0001318181818181818, | |
| "loss": 0.0651, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 0.08125822246074677, | |
| "learning_rate": 0.00012045454545454546, | |
| "loss": 0.0298, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 0.0715242475271225, | |
| "learning_rate": 0.00010909090909090909, | |
| "loss": 0.0342, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "eval_accuracy": 0.965, | |
| "eval_loss": 0.09033375978469849, | |
| "eval_runtime": 3.0566, | |
| "eval_samples_per_second": 65.433, | |
| "eval_steps_per_second": 8.179, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.06209810450673103, | |
| "learning_rate": 9.772727272727274e-05, | |
| "loss": 0.0213, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 0.056180860847234726, | |
| "learning_rate": 8.636363636363637e-05, | |
| "loss": 0.0081, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.05113891884684563, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0073, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 0.05209748074412346, | |
| "learning_rate": 6.363636363636364e-05, | |
| "loss": 0.0139, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "eval_accuracy": 0.965, | |
| "eval_loss": 0.09859169274568558, | |
| "eval_runtime": 2.9938, | |
| "eval_samples_per_second": 66.805, | |
| "eval_steps_per_second": 8.351, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "grad_norm": 0.06674375385046005, | |
| "learning_rate": 5.2272727272727274e-05, | |
| "loss": 0.0068, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "grad_norm": 0.04334596171975136, | |
| "learning_rate": 4.0909090909090915e-05, | |
| "loss": 0.0068, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "grad_norm": 0.04224753752350807, | |
| "learning_rate": 2.954545454545455e-05, | |
| "loss": 0.0057, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 0.048290498554706573, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 0.0069, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "eval_accuracy": 0.97, | |
| "eval_loss": 0.05549783632159233, | |
| "eval_runtime": 3.0091, | |
| "eval_samples_per_second": 66.466, | |
| "eval_steps_per_second": 8.308, | |
| "step": 160 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 176, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 40, | |
| "total_flos": 1.9721711356913664e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |