| { | |
| "best_metric": 0.5750909447669983, | |
| "best_model_checkpoint": "finetuned-mango-types/checkpoint-396", | |
| "epoch": 20.0, | |
| "eval_steps": 100, | |
| "global_step": 440, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 60231.65234375, | |
| "learning_rate": 1.9545454545454546e-05, | |
| "loss": 2.0767, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 64952.8828125, | |
| "learning_rate": 1.9090909090909094e-05, | |
| "loss": 1.9926, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.38333333333333336, | |
| "eval_loss": 1.9525998830795288, | |
| "eval_runtime": 3.0528, | |
| "eval_samples_per_second": 78.616, | |
| "eval_steps_per_second": 1.31, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 80774.6875, | |
| "learning_rate": 1.8636363636363638e-05, | |
| "loss": 1.8898, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 85769.2890625, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 1.7976, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6083333333333333, | |
| "eval_loss": 1.7500048875808716, | |
| "eval_runtime": 3.0095, | |
| "eval_samples_per_second": 79.749, | |
| "eval_steps_per_second": 1.329, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 89406.2109375, | |
| "learning_rate": 1.772727272727273e-05, | |
| "loss": 1.684, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 100806.59375, | |
| "learning_rate": 1.7272727272727274e-05, | |
| "loss": 1.5678, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7583333333333333, | |
| "eval_loss": 1.502477765083313, | |
| "eval_runtime": 3.0462, | |
| "eval_samples_per_second": 78.787, | |
| "eval_steps_per_second": 1.313, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "grad_norm": 100292.0, | |
| "learning_rate": 1.681818181818182e-05, | |
| "loss": 1.4796, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 116212.53125, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 1.3907, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9, | |
| "eval_loss": 1.280394196510315, | |
| "eval_runtime": 3.0736, | |
| "eval_samples_per_second": 78.084, | |
| "eval_steps_per_second": 1.301, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "grad_norm": 110518.28125, | |
| "learning_rate": 1.590909090909091e-05, | |
| "loss": 1.245, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 108781.6015625, | |
| "learning_rate": 1.5454545454545454e-05, | |
| "loss": 1.1536, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 215031.46875, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 1.0873, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9041666666666667, | |
| "eval_loss": 1.1004973649978638, | |
| "eval_runtime": 3.2979, | |
| "eval_samples_per_second": 72.773, | |
| "eval_steps_per_second": 1.213, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "grad_norm": 134639.890625, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 1.0071, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "grad_norm": 167596.140625, | |
| "learning_rate": 1.4090909090909092e-05, | |
| "loss": 0.9511, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8875, | |
| "eval_loss": 1.0129581689834595, | |
| "eval_runtime": 3.0931, | |
| "eval_samples_per_second": 77.592, | |
| "eval_steps_per_second": 1.293, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "grad_norm": 161001.609375, | |
| "learning_rate": 1.3636363636363637e-05, | |
| "loss": 0.9019, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "grad_norm": 152811.796875, | |
| "learning_rate": 1.3181818181818183e-05, | |
| "loss": 0.8476, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8833333333333333, | |
| "eval_loss": 0.9424096941947937, | |
| "eval_runtime": 3.1124, | |
| "eval_samples_per_second": 77.112, | |
| "eval_steps_per_second": 1.285, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 165279.65625, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 0.7811, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "grad_norm": 132651.1875, | |
| "learning_rate": 1.2272727272727274e-05, | |
| "loss": 0.7511, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9041666666666667, | |
| "eval_loss": 0.8324654698371887, | |
| "eval_runtime": 3.1226, | |
| "eval_samples_per_second": 76.86, | |
| "eval_steps_per_second": 1.281, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "grad_norm": 133902.6875, | |
| "learning_rate": 1.181818181818182e-05, | |
| "loss": 0.7406, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "grad_norm": 122836.96875, | |
| "learning_rate": 1.1363636363636366e-05, | |
| "loss": 0.6985, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.9083333333333333, | |
| "eval_loss": 0.7894275784492493, | |
| "eval_runtime": 3.163, | |
| "eval_samples_per_second": 75.877, | |
| "eval_steps_per_second": 1.265, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "grad_norm": 134674.1875, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 0.6442, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "grad_norm": 170883.515625, | |
| "learning_rate": 1.0454545454545455e-05, | |
| "loss": 0.6472, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "grad_norm": 244726.8125, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6515, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.8791666666666667, | |
| "eval_loss": 0.8052219748497009, | |
| "eval_runtime": 3.1419, | |
| "eval_samples_per_second": 76.386, | |
| "eval_steps_per_second": 1.273, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "grad_norm": 184393.625, | |
| "learning_rate": 9.545454545454547e-06, | |
| "loss": 0.6142, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "grad_norm": 132595.8125, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 0.5775, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.8791666666666667, | |
| "eval_loss": 0.7600470781326294, | |
| "eval_runtime": 3.2066, | |
| "eval_samples_per_second": 74.845, | |
| "eval_steps_per_second": 1.247, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "grad_norm": 131146.15625, | |
| "learning_rate": 8.636363636363637e-06, | |
| "loss": 0.5393, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "grad_norm": 191549.59375, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 0.5458, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.925, | |
| "eval_loss": 0.6684209108352661, | |
| "eval_runtime": 3.1291, | |
| "eval_samples_per_second": 76.699, | |
| "eval_steps_per_second": 1.278, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "grad_norm": 97074.78125, | |
| "learning_rate": 7.727272727272727e-06, | |
| "loss": 0.5246, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 12.73, | |
| "grad_norm": 125002.2109375, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 0.5331, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.8916666666666667, | |
| "eval_loss": 0.7147873044013977, | |
| "eval_runtime": 3.1553, | |
| "eval_samples_per_second": 76.061, | |
| "eval_steps_per_second": 1.268, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 13.18, | |
| "grad_norm": 191931.578125, | |
| "learning_rate": 6.818181818181818e-06, | |
| "loss": 0.4916, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 13.64, | |
| "grad_norm": 200110.078125, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 0.4823, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.9125, | |
| "eval_loss": 0.6849376559257507, | |
| "eval_runtime": 3.143, | |
| "eval_samples_per_second": 76.361, | |
| "eval_steps_per_second": 1.273, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "grad_norm": 147656.59375, | |
| "learning_rate": 5.90909090909091e-06, | |
| "loss": 0.4584, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "grad_norm": 149433.0625, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 0.4763, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 169986.890625, | |
| "learning_rate": 5e-06, | |
| "loss": 0.4579, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.9166666666666666, | |
| "eval_loss": 0.6413583755493164, | |
| "eval_runtime": 3.1496, | |
| "eval_samples_per_second": 76.2, | |
| "eval_steps_per_second": 1.27, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "grad_norm": 128090.9921875, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 0.4544, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 15.91, | |
| "grad_norm": 114307.015625, | |
| "learning_rate": 4.0909090909090915e-06, | |
| "loss": 0.4435, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.8833333333333333, | |
| "eval_loss": 0.6557222008705139, | |
| "eval_runtime": 3.3456, | |
| "eval_samples_per_second": 71.737, | |
| "eval_steps_per_second": 1.196, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "grad_norm": 153500.765625, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 0.4625, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 16.82, | |
| "grad_norm": 231700.671875, | |
| "learning_rate": 3.181818181818182e-06, | |
| "loss": 0.4411, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.9083333333333333, | |
| "eval_loss": 0.5968121886253357, | |
| "eval_runtime": 3.1534, | |
| "eval_samples_per_second": 76.108, | |
| "eval_steps_per_second": 1.268, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 17.27, | |
| "grad_norm": 209887.0625, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 0.4593, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "grad_norm": 160558.59375, | |
| "learning_rate": 2.2727272727272728e-06, | |
| "loss": 0.453, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.9291666666666667, | |
| "eval_loss": 0.5750909447669983, | |
| "eval_runtime": 3.2046, | |
| "eval_samples_per_second": 74.892, | |
| "eval_steps_per_second": 1.248, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "grad_norm": 147230.578125, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 0.4292, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 18.64, | |
| "grad_norm": 208387.1875, | |
| "learning_rate": 1.3636363636363636e-06, | |
| "loss": 0.445, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.9083333333333333, | |
| "eval_loss": 0.6034529209136963, | |
| "eval_runtime": 3.1393, | |
| "eval_samples_per_second": 76.45, | |
| "eval_steps_per_second": 1.274, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "grad_norm": 142534.578125, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 0.4364, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 19.55, | |
| "grad_norm": 190416.796875, | |
| "learning_rate": 4.5454545454545457e-07, | |
| "loss": 0.4115, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "grad_norm": 414293.15625, | |
| "learning_rate": 0.0, | |
| "loss": 0.4357, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.9083333333333333, | |
| "eval_loss": 0.6010429263114929, | |
| "eval_runtime": 3.1505, | |
| "eval_samples_per_second": 76.178, | |
| "eval_steps_per_second": 1.27, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 440, | |
| "total_flos": 2.1078954658234368e+18, | |
| "train_loss": 0.8308919093825601, | |
| "train_runtime": 956.1106, | |
| "train_samples_per_second": 28.449, | |
| "train_steps_per_second": 0.46 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 440, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 100, | |
| "total_flos": 2.1078954658234368e+18, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |