| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9986232216613126, | |
| "eval_steps": 500, | |
| "global_step": 272, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999332998034515e-05, | |
| "loss": 1.9596, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.99833293993636e-05, | |
| "loss": 1.9397, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.99333731792395e-05, | |
| "loss": 1.8704, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.985029789966671e-05, | |
| "loss": 1.7903, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9734380543606932e-05, | |
| "loss": 1.6992, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9586007593450098e-05, | |
| "loss": 1.6678, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9405673742435677e-05, | |
| "loss": 1.6189, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9193980245285967e-05, | |
| "loss": 1.6122, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8951632913550625e-05, | |
| "loss": 1.5687, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8679439762346186e-05, | |
| "loss": 1.5472, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8378308316336585e-05, | |
| "loss": 1.5302, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8049242583936923e-05, | |
| "loss": 1.5045, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.769333970982879e-05, | |
| "loss": 1.49, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7311786316948112e-05, | |
| "loss": 1.4782, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6905854550141717e-05, | |
| "loss": 1.4596, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.647689783468362e-05, | |
| "loss": 1.4509, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6026346363792565e-05, | |
| "loss": 1.4295, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5555702330196024e-05, | |
| "loss": 1.4301, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5066534917639195e-05, | |
| "loss": 1.4297, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4560475069037895e-05, | |
| "loss": 1.4161, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.403921004871895e-05, | |
| "loss": 1.4144, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.350447781687826e-05, | |
| "loss": 1.4051, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.2958061235012707e-05, | |
| "loss": 1.4061, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2401782121645767e-05, | |
| "loss": 1.3968, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1837495178165706e-05, | |
| "loss": 1.3946, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.126708180502834e-05, | |
| "loss": 1.382, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0692443828941918e-05, | |
| "loss": 1.3844, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0115497161948409e-05, | |
| "loss": 1.3857, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.538165413542607e-06, | |
| "loss": 1.3812, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.962373477126983e-06, | |
| "loss": 1.3657, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.39004111218587e-06, | |
| "loss": 1.3757, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.823076543576718e-06, | |
| "loss": 1.3713, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.263370099279173e-06, | |
| "loss": 1.3653, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.712787907814542e-06, | |
| "loss": 1.3749, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.173165676349103e-06, | |
| "loss": 1.3718, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.646302570225919e-06, | |
| "loss": 1.3728, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.133955214331439e-06, | |
| "loss": 1.3672, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.637831836297103e-06, | |
| "loss": 1.366, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.1595865710632366e-06, | |
| "loss": 1.3708, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.700813945794425e-06, | |
| "loss": 1.37, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.2630435635344283e-06, | |
| "loss": 1.3679, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.847735003325868e-06, | |
| "loss": 1.3671, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.456272953798361e-06, | |
| "loss": 1.3677, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.0899625964503113e-06, | |
| "loss": 1.3675, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7500252540169782e-06, | |
| "loss": 1.3647, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4375943184337871e-06, | |
| "loss": 1.371, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.1537114719714482e-06, | |
| "loss": 1.3622, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.993232141421415e-07, | |
| "loss": 1.3596, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.752777059564431e-07, | |
| "loss": 1.369, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.823219420526182e-07, | |
| "loss": 1.3621, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.2109926012677484e-07, | |
| "loss": 1.3649, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.921471959676957e-07, | |
| "loss": 1.3646, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.589569124794918e-08, | |
| "loss": 1.3552, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.266566004670013e-08, | |
| "loss": 1.3669, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.667918883627607e-09, | |
| "loss": 1.3681, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.0922234058380127, | |
| "eval_runtime": 2386.6161, | |
| "eval_samples_per_second": 6.466, | |
| "eval_steps_per_second": 0.101, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 272, | |
| "total_flos": 7631468079611904.0, | |
| "train_loss": 1.4444872824584736, | |
| "train_runtime": 72477.8005, | |
| "train_samples_per_second": 1.924, | |
| "train_steps_per_second": 0.004 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 272, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 7631468079611904.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |