| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 39, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5431, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2e-05, | |
| "loss": 0.5217, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9963974885425267e-05, | |
| "loss": 0.4643, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9856159103477085e-05, | |
| "loss": 0.4449, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.967732946933499e-05, | |
| "loss": 0.4399, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.9428774454610845e-05, | |
| "loss": 0.4314, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.911228490388136e-05, | |
| "loss": 0.44, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8730141131611882e-05, | |
| "loss": 0.422, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.8285096492438424e-05, | |
| "loss": 0.4005, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.7780357543184396e-05, | |
| "loss": 0.4117, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.7219560939545246e-05, | |
| "loss": 0.3954, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6606747233900816e-05, | |
| "loss": 0.3866, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.594633176304287e-05, | |
| "loss": 0.3906, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.5243072835572319e-05, | |
| "loss": 0.3271, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.4502037448176734e-05, | |
| "loss": 0.3647, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.3728564777803089e-05, | |
| "loss": 0.3168, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.2928227712765504e-05, | |
| "loss": 0.3172, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.2106792699957264e-05, | |
| "loss": 0.3152, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.1270178197468788e-05, | |
| "loss": 0.3392, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.0424412031961485e-05, | |
| "loss": 0.3301, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.57558796803852e-06, | |
| "loss": 0.2919, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.729821802531213e-06, | |
| "loss": 0.2873, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.89320730004274e-06, | |
| "loss": 0.3345, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.071772287234497e-06, | |
| "loss": 0.3276, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.2714352221969155e-06, | |
| "loss": 0.2866, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.497962551823266e-06, | |
| "loss": 0.2947, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 4.756927164427685e-06, | |
| "loss": 0.2738, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 4.053668236957135e-06, | |
| "loss": 0.2464, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.3932527660991877e-06, | |
| "loss": 0.2749, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 2.780439060454756e-06, | |
| "loss": 0.2664, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.2196424568156073e-06, | |
| "loss": 0.2852, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.7149035075615795e-06, | |
| "loss": 0.2876, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1.2698588683881185e-06, | |
| "loss": 0.2691, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 8.87715096118642e-07, | |
| "loss": 0.2448, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.71225545389158e-07, | |
| "loss": 0.2784, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.226705306650113e-07, | |
| "loss": 0.2722, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.4384089652291544e-07, | |
| "loss": 0.2862, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.602511457473479e-08, | |
| "loss": 0.2631, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.2251, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 39, | |
| "total_flos": 2.4951731691454464e+16, | |
| "train_loss": 0.3409684506746439, | |
| "train_runtime": 1410.2975, | |
| "train_samples_per_second": 3.54, | |
| "train_steps_per_second": 0.028 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.2879543900489807, | |
| "eval_runtime": 10.2141, | |
| "eval_samples_per_second": 40.728, | |
| "eval_steps_per_second": 1.273, | |
| "step": 39 | |
| } | |
| ], | |
| "max_steps": 39, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.4951731691454464e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |