| { | |
| "best_metric": 0.2435225546360016, | |
| "best_model_checkpoint": "finetuned-blurr-nonblur/checkpoint-112", | |
| "epoch": 10.0, | |
| "eval_steps": 100, | |
| "global_step": 140, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 54505.6171875, | |
| "learning_rate": 1.8571428571428575e-05, | |
| "loss": 0.6486, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6645569620253164, | |
| "eval_loss": 0.625545084476471, | |
| "eval_runtime": 7.9446, | |
| "eval_samples_per_second": 19.888, | |
| "eval_steps_per_second": 0.378, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 57206.48828125, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 0.552, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6772151898734177, | |
| "eval_loss": 0.5736974477767944, | |
| "eval_runtime": 7.72, | |
| "eval_samples_per_second": 20.466, | |
| "eval_steps_per_second": 0.389, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 53612.984375, | |
| "learning_rate": 1.5714285714285715e-05, | |
| "loss": 0.5206, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 51806.32421875, | |
| "learning_rate": 1.4285714285714287e-05, | |
| "loss": 0.4207, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7974683544303798, | |
| "eval_loss": 0.5174736380577087, | |
| "eval_runtime": 7.726, | |
| "eval_samples_per_second": 20.45, | |
| "eval_steps_per_second": 0.388, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "grad_norm": 56256.515625, | |
| "learning_rate": 1.2857142857142859e-05, | |
| "loss": 0.3545, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8860759493670886, | |
| "eval_loss": 0.44835516810417175, | |
| "eval_runtime": 7.7183, | |
| "eval_samples_per_second": 20.471, | |
| "eval_steps_per_second": 0.389, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "grad_norm": 70625.890625, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 0.286, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 57182.4140625, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2082, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8860759493670886, | |
| "eval_loss": 0.3621000051498413, | |
| "eval_runtime": 7.8704, | |
| "eval_samples_per_second": 20.075, | |
| "eval_steps_per_second": 0.381, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "grad_norm": 41587.15625, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.167, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.9050632911392406, | |
| "eval_loss": 0.2930063307285309, | |
| "eval_runtime": 7.906, | |
| "eval_samples_per_second": 19.985, | |
| "eval_steps_per_second": 0.379, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "grad_norm": 89662.453125, | |
| "learning_rate": 7.1428571428571436e-06, | |
| "loss": 0.176, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8860759493670886, | |
| "eval_loss": 0.3003353178501129, | |
| "eval_runtime": 7.7836, | |
| "eval_samples_per_second": 20.299, | |
| "eval_steps_per_second": 0.385, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "grad_norm": 105894.6796875, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.1271, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 73103.46875, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 0.1275, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9240506329113924, | |
| "eval_loss": 0.2435225546360016, | |
| "eval_runtime": 7.7728, | |
| "eval_samples_per_second": 20.327, | |
| "eval_steps_per_second": 0.386, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "grad_norm": 65041.953125, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 0.11, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.9050632911392406, | |
| "eval_loss": 0.25813814997673035, | |
| "eval_runtime": 7.8334, | |
| "eval_samples_per_second": 20.17, | |
| "eval_steps_per_second": 0.383, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "grad_norm": 42873.71484375, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 0.107, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 64285.99609375, | |
| "learning_rate": 0.0, | |
| "loss": 0.1009, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.9113924050632911, | |
| "eval_loss": 0.24741442501544952, | |
| "eval_runtime": 7.9653, | |
| "eval_samples_per_second": 19.836, | |
| "eval_steps_per_second": 0.377, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 140, | |
| "total_flos": 6.912285473621606e+17, | |
| "train_loss": 0.2789982352937971, | |
| "train_runtime": 718.4172, | |
| "train_samples_per_second": 12.416, | |
| "train_steps_per_second": 0.195 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 140, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 100, | |
| "total_flos": 6.912285473621606e+17, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |