| { | |
| "best_metric": 0.016295962035655975, | |
| "best_model_checkpoint": "./outputs/checkpoint-160", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 160, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 3.219058036804199, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 0.4776, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.2438404560089111, | |
| "learning_rate": 1.7500000000000002e-05, | |
| "loss": 0.1885, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.8410812616348267, | |
| "learning_rate": 1.6250000000000002e-05, | |
| "loss": 0.0948, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.994, | |
| "eval_loss": 0.038248706609010696, | |
| "eval_runtime": 27.2538, | |
| "eval_samples_per_second": 18.346, | |
| "eval_steps_per_second": 0.294, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 0.9565186500549316, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.0511, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 1.1085598468780518, | |
| "learning_rate": 1.375e-05, | |
| "loss": 0.045, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 1.4812817573547363, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.045, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.996, | |
| "eval_loss": 0.020908161997795105, | |
| "eval_runtime": 4.4414, | |
| "eval_samples_per_second": 112.577, | |
| "eval_steps_per_second": 1.801, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 2.213463068008423, | |
| "learning_rate": 1.125e-05, | |
| "loss": 0.0408, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 1.0551693439483643, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0311, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "grad_norm": 1.5963162183761597, | |
| "learning_rate": 8.750000000000001e-06, | |
| "loss": 0.0421, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.996, | |
| "eval_loss": 0.017463386058807373, | |
| "eval_runtime": 4.5818, | |
| "eval_samples_per_second": 109.128, | |
| "eval_steps_per_second": 1.746, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 0.7329137921333313, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.0219, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "grad_norm": 1.8856861591339111, | |
| "learning_rate": 6.25e-06, | |
| "loss": 0.0257, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 0.23129290342330933, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0223, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.996, | |
| "eval_loss": 0.016901519149541855, | |
| "eval_runtime": 4.3629, | |
| "eval_samples_per_second": 114.603, | |
| "eval_steps_per_second": 1.834, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "grad_norm": 1.9281848669052124, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.0396, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "grad_norm": 0.3660925030708313, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0267, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "grad_norm": 0.5336365699768066, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.0304, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 3.7920114994049072, | |
| "learning_rate": 0.0, | |
| "loss": 0.025, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.996, | |
| "eval_loss": 0.016295962035655975, | |
| "eval_runtime": 4.1331, | |
| "eval_samples_per_second": 120.976, | |
| "eval_steps_per_second": 1.936, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 160, | |
| "total_flos": 7.7491989614592e+17, | |
| "train_loss": 0.07546138893812895, | |
| "train_runtime": 293.9975, | |
| "train_samples_per_second": 34.014, | |
| "train_steps_per_second": 0.544 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 160, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 7.7491989614592e+17, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |