| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.2681564245810056, | |
| "eval_steps": 500, | |
| "global_step": 60, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.5526, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.6607, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.372, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.4931, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.5619, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-06, | |
| "loss": 1.718, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.5e-06, | |
| "loss": 1.4011, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.6706, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.5431, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-06, | |
| "loss": 1.6866, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.999971592166969e-06, | |
| "loss": 1.3485, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.999886369313478e-06, | |
| "loss": 1.6039, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.999744333376324e-06, | |
| "loss": 1.6092, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.999545487583456e-06, | |
| "loss": 1.4041, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.999289836453894e-06, | |
| "loss": 1.2342, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.998977385797635e-06, | |
| "loss": 1.5679, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.998608142715515e-06, | |
| "loss": 1.5122, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.998182115599052e-06, | |
| "loss": 1.7621, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.99769931413025e-06, | |
| "loss": 1.4677, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.997159749281385e-06, | |
| "loss": 1.4017, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.996563433314751e-06, | |
| "loss": 1.5041, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9959103797823835e-06, | |
| "loss": 1.7749, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.995200603525752e-06, | |
| "loss": 1.5217, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.99443412067542e-06, | |
| "loss": 1.5351, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.993610948650681e-06, | |
| "loss": 1.6509, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992731106159161e-06, | |
| "loss": 1.3234, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9917946131963965e-06, | |
| "loss": 1.5587, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.990801491045376e-06, | |
| "loss": 1.5141, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.989751762276056e-06, | |
| "loss": 1.3626, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.988645450744855e-06, | |
| "loss": 1.413, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.987482581594101e-06, | |
| "loss": 1.5925, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.98626318125147e-06, | |
| "loss": 1.6128, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.984987277429377e-06, | |
| "loss": 1.5369, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.983654899124355e-06, | |
| "loss": 1.4611, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.982266076616386e-06, | |
| "loss": 1.3973, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9808208414682215e-06, | |
| "loss": 1.6418, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.979319226524661e-06, | |
| "loss": 1.5317, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9777612659118034e-06, | |
| "loss": 1.3446, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.976146995036279e-06, | |
| "loss": 1.6594, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.974476450584438e-06, | |
| "loss": 1.3303, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.972749670521517e-06, | |
| "loss": 1.391, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.970966694090781e-06, | |
| "loss": 1.5491, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.969127561812626e-06, | |
| "loss": 1.7895, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.967232315483664e-06, | |
| "loss": 1.3842, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.965280998175767e-06, | |
| "loss": 1.3015, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.963273654235092e-06, | |
| "loss": 1.5276, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.961210329281072e-06, | |
| "loss": 1.3836, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.959091070205381e-06, | |
| "loss": 1.3379, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9569159251708635e-06, | |
| "loss": 1.385, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.954684943610446e-06, | |
| "loss": 1.4117, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.95239817622601e-06, | |
| "loss": 1.5684, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.95005567498724e-06, | |
| "loss": 1.4568, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.947657493130443e-06, | |
| "loss": 1.4549, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.945203685157339e-06, | |
| "loss": 1.4898, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.942694306833823e-06, | |
| "loss": 1.3103, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.940129415188693e-06, | |
| "loss": 1.5988, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.937509068512362e-06, | |
| "loss": 1.607, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.934833326355525e-06, | |
| "loss": 1.5436, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.932102249527812e-06, | |
| "loss": 1.5437, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.929315900096403e-06, | |
| "loss": 1.2806, | |
| "step": 60 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 669, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 20, | |
| "total_flos": 4.243516003713024e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |