| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 435, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013562298850574713, | |
| "loss": 4.7314, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013404597701149426, | |
| "loss": 3.8299, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001324689655172414, | |
| "loss": 3.7012, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0001308919540229885, | |
| "loss": 2.6889, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00012931494252873562, | |
| "loss": 2.8895, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00012773793103448275, | |
| "loss": 3.0844, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00012616091954022987, | |
| "loss": 2.2248, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000124583908045977, | |
| "loss": 1.9715, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00012300689655172413, | |
| "loss": 1.6414, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00012142988505747127, | |
| "loss": 2.3638, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001198528735632184, | |
| "loss": 1.8585, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00011827586206896552, | |
| "loss": 1.7226, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00011669885057471265, | |
| "loss": 1.6046, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00011512183908045976, | |
| "loss": 1.151, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00011354482758620689, | |
| "loss": 0.9814, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00011196781609195402, | |
| "loss": 1.0576, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00011039080459770115, | |
| "loss": 1.496, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00010881379310344828, | |
| "loss": 0.9055, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0001072367816091954, | |
| "loss": 1.7868, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00010565977011494252, | |
| "loss": 2.0508, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00010408275862068965, | |
| "loss": 1.1797, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00010250574712643678, | |
| "loss": 0.7991, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001009287356321839, | |
| "loss": 1.1201, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.935172413793103e-05, | |
| "loss": 1.4825, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.777471264367816e-05, | |
| "loss": 1.0949, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.619770114942529e-05, | |
| "loss": 1.762, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.462068965517242e-05, | |
| "loss": 1.2538, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.304367816091953e-05, | |
| "loss": 0.6892, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.146666666666666e-05, | |
| "loss": 0.8873, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.988965517241379e-05, | |
| "loss": 1.1749, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.831264367816092e-05, | |
| "loss": 0.3432, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.673563218390805e-05, | |
| "loss": 0.8749, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.515862068965518e-05, | |
| "loss": 0.8816, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.35816091954023e-05, | |
| "loss": 1.0882, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.200459770114943e-05, | |
| "loss": 1.2177, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.042758620689655e-05, | |
| "loss": 0.7566, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.885057471264368e-05, | |
| "loss": 0.5707, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.72735632183908e-05, | |
| "loss": 0.8469, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.569655172413794e-05, | |
| "loss": 0.312, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.411954022988506e-05, | |
| "loss": 0.6802, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.254252873563219e-05, | |
| "loss": 1.0342, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.096551724137932e-05, | |
| "loss": 0.5038, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 6.938850574712645e-05, | |
| "loss": 0.6356, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.781149425287356e-05, | |
| "loss": 1.1177, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.62344827586207e-05, | |
| "loss": 0.5872, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.465747126436781e-05, | |
| "loss": 0.7614, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 6.308045977011494e-05, | |
| "loss": 0.7013, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.150344827586207e-05, | |
| "loss": 0.6858, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.99264367816092e-05, | |
| "loss": 0.5427, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.834942528735632e-05, | |
| "loss": 0.6479, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.6772413793103445e-05, | |
| "loss": 0.4818, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.519540229885057e-05, | |
| "loss": 0.7337, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.36183908045977e-05, | |
| "loss": 0.7244, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.2041379310344824e-05, | |
| "loss": 0.4943, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.046436781609195e-05, | |
| "loss": 0.5786, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.888735632183908e-05, | |
| "loss": 0.5664, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.731034482758621e-05, | |
| "loss": 0.7201, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.573333333333333e-05, | |
| "loss": 0.6205, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.415632183908046e-05, | |
| "loss": 0.5324, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.257931034482759e-05, | |
| "loss": 0.4303, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.100229885057472e-05, | |
| "loss": 0.5163, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.942528735632184e-05, | |
| "loss": 0.4597, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.784827586206897e-05, | |
| "loss": 0.2826, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.6271264367816096e-05, | |
| "loss": 0.3543, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.4694252873563225e-05, | |
| "loss": 0.4478, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.311724137931035e-05, | |
| "loss": 0.2724, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.154022988505747e-05, | |
| "loss": 0.4695, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.99632183908046e-05, | |
| "loss": 0.438, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.8386206896551722e-05, | |
| "loss": 0.3359, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.680919540229885e-05, | |
| "loss": 0.4893, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.5232183908045976e-05, | |
| "loss": 0.4421, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.3655172413793105e-05, | |
| "loss": 0.4793, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.207816091954023e-05, | |
| "loss": 0.4306, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.050114942528736e-05, | |
| "loss": 0.4226, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8924137931034484e-05, | |
| "loss": 0.4516, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.7347126436781612e-05, | |
| "loss": 0.3301, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.5770114942528734e-05, | |
| "loss": 0.323, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4193103448275861e-05, | |
| "loss": 0.4705, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.2616091954022988e-05, | |
| "loss": 0.2607, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.1039080459770115e-05, | |
| "loss": 0.6484, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.462068965517242e-06, | |
| "loss": 0.3808, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 7.885057471264367e-06, | |
| "loss": 0.4006, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.308045977011494e-06, | |
| "loss": 0.5295, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.731034482758621e-06, | |
| "loss": 0.3334, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.154022988505747e-06, | |
| "loss": 0.3786, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.5770114942528735e-06, | |
| "loss": 0.4675, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4202, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 435, | |
| "total_flos": 807969604239360.0, | |
| "train_loss": 1.0010668743615863, | |
| "train_runtime": 452.5068, | |
| "train_samples_per_second": 0.961, | |
| "train_steps_per_second": 0.961 | |
| } | |
| ], | |
| "max_steps": 435, | |
| "num_train_epochs": 1, | |
| "total_flos": 807969604239360.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |