| { | |
| "best_metric": 0.015285464935004711, | |
| "best_model_checkpoint": "./vit-base-beans-demo-v5/checkpoint-200", | |
| "epoch": 4.0, | |
| "eval_steps": 100, | |
| "global_step": 260, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.15384615384615385, | |
| "grad_norm": 1.8352460861206055, | |
| "learning_rate": 0.00019230769230769233, | |
| "loss": 0.7932, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 2.671342611312866, | |
| "learning_rate": 0.00018461538461538463, | |
| "loss": 0.3441, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.46153846153846156, | |
| "grad_norm": 2.3329362869262695, | |
| "learning_rate": 0.00017692307692307693, | |
| "loss": 0.233, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 2.8852927684783936, | |
| "learning_rate": 0.00016923076923076923, | |
| "loss": 0.2218, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.31967252492904663, | |
| "learning_rate": 0.00016153846153846155, | |
| "loss": 0.182, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 4.441674709320068, | |
| "learning_rate": 0.00015384615384615385, | |
| "loss": 0.2335, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.0769230769230769, | |
| "grad_norm": 0.6437997817993164, | |
| "learning_rate": 0.00014615384615384615, | |
| "loss": 0.1089, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 0.1999683678150177, | |
| "learning_rate": 0.00013846153846153847, | |
| "loss": 0.1733, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.3846153846153846, | |
| "grad_norm": 0.586209237575531, | |
| "learning_rate": 0.00013076923076923077, | |
| "loss": 0.0628, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 1.335889220237732, | |
| "learning_rate": 0.0001230769230769231, | |
| "loss": 0.0935, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "eval_accuracy": 0.9924812030075187, | |
| "eval_loss": 0.056021548807621, | |
| "eval_runtime": 1.3017, | |
| "eval_samples_per_second": 102.174, | |
| "eval_steps_per_second": 13.06, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.6923076923076923, | |
| "grad_norm": 3.6746835708618164, | |
| "learning_rate": 0.00011538461538461538, | |
| "loss": 0.07, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 0.17705152928829193, | |
| "learning_rate": 0.0001076923076923077, | |
| "loss": 0.0994, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.6592533588409424, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0643, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 0.08297152072191238, | |
| "learning_rate": 9.230769230769232e-05, | |
| "loss": 0.0399, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 0.10470888763666153, | |
| "learning_rate": 8.461538461538461e-05, | |
| "loss": 0.0482, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 0.15344949066638947, | |
| "learning_rate": 7.692307692307693e-05, | |
| "loss": 0.0458, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.6153846153846154, | |
| "grad_norm": 0.10154535621404648, | |
| "learning_rate": 6.923076923076924e-05, | |
| "loss": 0.0135, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 0.22166353464126587, | |
| "learning_rate": 6.153846153846155e-05, | |
| "loss": 0.0416, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.9230769230769234, | |
| "grad_norm": 0.35676950216293335, | |
| "learning_rate": 5.384615384615385e-05, | |
| "loss": 0.0168, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "grad_norm": 0.057582877576351166, | |
| "learning_rate": 4.615384615384616e-05, | |
| "loss": 0.012, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "eval_accuracy": 1.0, | |
| "eval_loss": 0.015285464935004711, | |
| "eval_runtime": 1.272, | |
| "eval_samples_per_second": 104.56, | |
| "eval_steps_per_second": 13.365, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.230769230769231, | |
| "grad_norm": 0.05789865553379059, | |
| "learning_rate": 3.846153846153846e-05, | |
| "loss": 0.0132, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.3846153846153846, | |
| "grad_norm": 0.053986236453056335, | |
| "learning_rate": 3.0769230769230774e-05, | |
| "loss": 0.0109, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 3.5384615384615383, | |
| "grad_norm": 0.05768540874123573, | |
| "learning_rate": 2.307692307692308e-05, | |
| "loss": 0.0105, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.6923076923076925, | |
| "grad_norm": 0.05456564202904701, | |
| "learning_rate": 1.5384615384615387e-05, | |
| "loss": 0.0102, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.8461538461538463, | |
| "grad_norm": 0.05737489089369774, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.0102, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.053701262921094894, | |
| "learning_rate": 0.0, | |
| "loss": 0.0103, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 260, | |
| "total_flos": 3.205097416476426e+17, | |
| "train_loss": 0.11395030921468367, | |
| "train_runtime": 86.7335, | |
| "train_samples_per_second": 47.686, | |
| "train_steps_per_second": 2.998 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 260, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.205097416476426e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |