| { | |
| "best_metric": 0.3921276926994324, | |
| "best_model_checkpoint": "./vit-base-3e-5-randaug/checkpoint-2750", | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 2750, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 4.155110836029053, | |
| "learning_rate": 2.9902226030228252e-05, | |
| "loss": 2.7863, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 4.886130332946777, | |
| "learning_rate": 2.9610178754135005e-05, | |
| "loss": 1.6815, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.7737574552683897, | |
| "eval_loss": 0.9074916839599609, | |
| "eval_runtime": 54.9169, | |
| "eval_samples_per_second": 45.796, | |
| "eval_steps_per_second": 0.728, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 4.915349960327148, | |
| "learning_rate": 2.9127665454592872e-05, | |
| "loss": 1.2844, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 4.22341775894165, | |
| "learning_rate": 2.8460976430370375e-05, | |
| "loss": 1.06, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 4.15113639831543, | |
| "learning_rate": 2.7618802992467718e-05, | |
| "loss": 0.9759, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8500994035785289, | |
| "eval_loss": 0.5867449641227722, | |
| "eval_runtime": 54.5476, | |
| "eval_samples_per_second": 46.106, | |
| "eval_steps_per_second": 0.733, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 4.799998760223389, | |
| "learning_rate": 2.661212415958624e-05, | |
| "loss": 0.9053, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 4.3418684005737305, | |
| "learning_rate": 2.5454063529829405e-05, | |
| "loss": 0.8298, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 4.906313896179199, | |
| "learning_rate": 2.4159718194531573e-05, | |
| "loss": 0.7955, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8548707753479126, | |
| "eval_loss": 0.5190673470497131, | |
| "eval_runtime": 55.308, | |
| "eval_samples_per_second": 45.473, | |
| "eval_steps_per_second": 0.723, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "grad_norm": 5.728692531585693, | |
| "learning_rate": 2.274596192458443e-05, | |
| "loss": 0.7404, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 4.377862453460693, | |
| "learning_rate": 2.12312251950283e-05, | |
| "loss": 0.7126, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 4.678811550140381, | |
| "learning_rate": 1.963525491562421e-05, | |
| "loss": 0.7056, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8755467196819086, | |
| "eval_loss": 0.45482131838798523, | |
| "eval_runtime": 55.04, | |
| "eval_samples_per_second": 45.694, | |
| "eval_steps_per_second": 0.727, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "grad_norm": 4.7580246925354, | |
| "learning_rate": 1.7978856999686182e-05, | |
| "loss": 0.6839, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "grad_norm": 4.103386878967285, | |
| "learning_rate": 1.6283625127182596e-05, | |
| "loss": 0.6455, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8854870775347913, | |
| "eval_loss": 0.42557355761528015, | |
| "eval_runtime": 55.0116, | |
| "eval_samples_per_second": 45.718, | |
| "eval_steps_per_second": 0.727, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "grad_norm": 4.85685396194458, | |
| "learning_rate": 1.4571659238094557e-05, | |
| "loss": 0.6445, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "grad_norm": 4.542230606079102, | |
| "learning_rate": 1.2865277425900725e-05, | |
| "loss": 0.5996, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "grad_norm": 3.4589812755584717, | |
| "learning_rate": 1.11867249870977e-05, | |
| "loss": 0.6249, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8846918489065606, | |
| "eval_loss": 0.411356657743454, | |
| "eval_runtime": 54.1236, | |
| "eval_samples_per_second": 46.468, | |
| "eval_steps_per_second": 0.739, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "grad_norm": 3.62654972076416, | |
| "learning_rate": 9.557884419740387e-06, | |
| "loss": 0.6076, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "grad_norm": 4.838754177093506, | |
| "learning_rate": 7.999990151614895e-06, | |
| "loss": 0.5802, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "grad_norm": 4.127110004425049, | |
| "learning_rate": 6.533351716998466e-06, | |
| "loss": 0.5742, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8874751491053678, | |
| "eval_loss": 0.4025728404521942, | |
| "eval_runtime": 54.4896, | |
| "eval_samples_per_second": 46.156, | |
| "eval_steps_per_second": 0.734, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 3.6499452590942383, | |
| "learning_rate": 5.177088990820725e-06, | |
| "loss": 0.5646, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "grad_norm": 3.194026231765747, | |
| "learning_rate": 3.948882931853924e-06, | |
| "loss": 0.5627, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 4.581117630004883, | |
| "learning_rate": 2.86474508437579e-06, | |
| "loss": 0.5782, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.890258449304175, | |
| "eval_loss": 0.3943153917789459, | |
| "eval_runtime": 54.2999, | |
| "eval_samples_per_second": 46.317, | |
| "eval_steps_per_second": 0.737, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "grad_norm": 4.3230791091918945, | |
| "learning_rate": 1.9388088432033446e-06, | |
| "loss": 0.5551, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "grad_norm": 4.666913032531738, | |
| "learning_rate": 1.1831452032772499e-06, | |
| "loss": 0.5383, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.8882703777335984, | |
| "eval_loss": 0.3928951919078827, | |
| "eval_runtime": 54.0056, | |
| "eval_samples_per_second": 46.569, | |
| "eval_steps_per_second": 0.741, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "grad_norm": 5.079577922821045, | |
| "learning_rate": 6.076053957825411e-07, | |
| "loss": 0.5563, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "grad_norm": 4.677091598510742, | |
| "learning_rate": 2.1969246228460526e-07, | |
| "loss": 0.5505, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "grad_norm": 2.785017967224121, | |
| "learning_rate": 2.4463441107965278e-08, | |
| "loss": 0.5495, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8878727634194831, | |
| "eval_loss": 0.3921276926994324, | |
| "eval_runtime": 55.0192, | |
| "eval_samples_per_second": 45.711, | |
| "eval_steps_per_second": 0.727, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 2750, | |
| "total_flos": 1.3627285673563054e+19, | |
| "train_loss": 0.8061258281360973, | |
| "train_runtime": 7150.5683, | |
| "train_samples_per_second": 24.587, | |
| "train_steps_per_second": 0.385 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 2750, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 1.3627285673563054e+19, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |