| { | |
| "best_metric": 0.998068630695343, | |
| "best_model_checkpoint": "./vit-Valence/checkpoint-200", | |
| "epoch": 4.0, | |
| "eval_steps": 100, | |
| "global_step": 512, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019609375, | |
| "loss": 1.527, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001921875, | |
| "loss": 1.3365, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00018828125, | |
| "loss": 1.3176, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000184375, | |
| "loss": 1.2405, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00018046875000000002, | |
| "loss": 1.1474, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00017656250000000002, | |
| "loss": 1.1061, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00017265625, | |
| "loss": 1.1234, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.00016875, | |
| "loss": 1.1591, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00016484375, | |
| "loss": 1.083, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001609375, | |
| "loss": 1.2256, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "eval_accuracy": 0.5450980392156862, | |
| "eval_loss": 1.0935585498809814, | |
| "eval_runtime": 8.8454, | |
| "eval_samples_per_second": 57.657, | |
| "eval_steps_per_second": 7.235, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00015703125, | |
| "loss": 1.1669, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.000153125, | |
| "loss": 1.0515, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00014921875000000002, | |
| "loss": 0.9945, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00014531250000000002, | |
| "loss": 0.7937, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00014140625, | |
| "loss": 0.7721, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001375, | |
| "loss": 0.7106, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00013359375, | |
| "loss": 0.6943, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0001296875, | |
| "loss": 0.7345, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00012578125, | |
| "loss": 0.7497, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00012187500000000001, | |
| "loss": 0.7315, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "eval_accuracy": 0.5882352941176471, | |
| "eval_loss": 0.998068630695343, | |
| "eval_runtime": 9.2665, | |
| "eval_samples_per_second": 55.037, | |
| "eval_steps_per_second": 6.907, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00011796875, | |
| "loss": 0.6931, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0001140625, | |
| "loss": 0.6189, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00011015625000000001, | |
| "loss": 0.6924, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00010625000000000001, | |
| "loss": 0.7201, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00010234375, | |
| "loss": 0.7225, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.84375e-05, | |
| "loss": 0.4755, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.453125000000001e-05, | |
| "loss": 0.2789, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.062500000000001e-05, | |
| "loss": 0.3311, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.671875e-05, | |
| "loss": 0.3663, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 8.28125e-05, | |
| "loss": 0.2118, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "eval_accuracy": 0.5901960784313726, | |
| "eval_loss": 1.1650166511535645, | |
| "eval_runtime": 8.9508, | |
| "eval_samples_per_second": 56.978, | |
| "eval_steps_per_second": 7.15, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.890625000000001e-05, | |
| "loss": 0.2309, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.2993, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.109375e-05, | |
| "loss": 0.2645, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 6.71875e-05, | |
| "loss": 0.339, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 6.328125e-05, | |
| "loss": 0.2321, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 5.9375e-05, | |
| "loss": 0.2505, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 5.546875e-05, | |
| "loss": 0.3216, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.15625e-05, | |
| "loss": 0.2255, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.765625e-05, | |
| "loss": 0.2241, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.1119, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_accuracy": 0.5862745098039216, | |
| "eval_loss": 1.2863526344299316, | |
| "eval_runtime": 8.5079, | |
| "eval_samples_per_second": 59.944, | |
| "eval_steps_per_second": 7.522, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.984375e-05, | |
| "loss": 0.127, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 3.59375e-05, | |
| "loss": 0.117, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.203125e-05, | |
| "loss": 0.0675, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 2.8125000000000003e-05, | |
| "loss": 0.0978, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 2.4218750000000003e-05, | |
| "loss": 0.0846, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.0312500000000002e-05, | |
| "loss": 0.0999, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.6406250000000002e-05, | |
| "loss": 0.0756, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.25e-05, | |
| "loss": 0.1008, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 8.59375e-06, | |
| "loss": 0.0679, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 4.6875000000000004e-06, | |
| "loss": 0.1116, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "eval_accuracy": 0.5862745098039216, | |
| "eval_loss": 1.4464128017425537, | |
| "eval_runtime": 9.2975, | |
| "eval_samples_per_second": 54.853, | |
| "eval_steps_per_second": 6.884, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 7.8125e-07, | |
| "loss": 0.1127, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 512, | |
| "total_flos": 6.311117323294802e+17, | |
| "train_loss": 0.5731766858662013, | |
| "train_runtime": 408.7689, | |
| "train_samples_per_second": 19.923, | |
| "train_steps_per_second": 1.253 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 512, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "total_flos": 6.311117323294802e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |