| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9972041006523766, | |
| "eval_steps": 500, | |
| "global_step": 201, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 2.2584, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 1.676, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 1.6463, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9998688836656322e-05, | |
| "loss": 1.584, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.998820159279591e-05, | |
| "loss": 1.4092, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9967238104745695e-05, | |
| "loss": 1.7832, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.993582036030978e-05, | |
| "loss": 2.0932, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9893981312363563e-05, | |
| "loss": 2.524, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9841764844290744e-05, | |
| "loss": 1.9114, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.977922572395571e-05, | |
| "loss": 1.5524, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9706429546259592e-05, | |
| "loss": 1.6035, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9623452664340305e-05, | |
| "loss": 1.5664, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.953038210948861e-05, | |
| "loss": 1.6138, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9427315499864345e-05, | |
| "loss": 1.9132, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9314360938108427e-05, | |
| "loss": 2.2387, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9191636897958123e-05, | |
| "loss": 2.4396, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.905927209998447e-05, | |
| "loss": 2.2283, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.8917405376582144e-05, | |
| "loss": 1.8121, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.876618552635348e-05, | |
| "loss": 1.552, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.8605771158039253e-05, | |
| "loss": 1.5683, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8436330524160048e-05, | |
| "loss": 1.5337, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.8258041344542567e-05, | |
| "loss": 1.8445, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.8071090619916095e-05, | |
| "loss": 2.0517, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.7875674435774546e-05, | |
| "loss": 2.3032, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.767199775670986e-05, | |
| "loss": 2.3695, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.7460274211432463e-05, | |
| "loss": 1.611, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.7240725868704218e-05, | |
| "loss": 1.4434, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7013583004418994e-05, | |
| "loss": 1.6615, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.6779083860075032e-05, | |
| "loss": 1.5693, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.6537474392892527e-05, | |
| "loss": 1.5989, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6289008017838447e-05, | |
| "loss": 1.8186, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.603394534182925e-05, | |
| "loss": 2.0993, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.5772553890390196e-05, | |
| "loss": 1.5693, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.5505107827058038e-05, | |
| "loss": 2.1505, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.52318876658213e-05, | |
| "loss": 1.7584, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.4953179976899878e-05, | |
| "loss": 1.4229, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.4669277086172406e-05, | |
| "loss": 1.611, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.4380476768566825e-05, | |
| "loss": 1.4103, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.4087081935735565e-05, | |
| "loss": 1.4992, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.378940031834307e-05, | |
| "loss": 1.7046, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.3487744143298822e-05, | |
| "loss": 1.9403, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.3182429806274442e-05, | |
| "loss": 1.9072, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.2873777539848284e-05, | |
| "loss": 1.7842, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.2562111077625723e-05, | |
| "loss": 1.5809, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.2247757314687296e-05, | |
| "loss": 1.4814, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1931045964720882e-05, | |
| "loss": 1.4703, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.1612309214197599e-05, | |
| "loss": 1.4649, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.1291881373954066e-05, | |
| "loss": 1.6343, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0970098528546482e-05, | |
| "loss": 1.7428, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0647298183744359e-05, | |
| "loss": 1.7637, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.0323818912533561e-05, | |
| "loss": 2.1005, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4387, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.676181087466444e-06, | |
| "loss": 1.422, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.352701816255643e-06, | |
| "loss": 1.3836, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.02990147145352e-06, | |
| "loss": 1.2649, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.708118626045939e-06, | |
| "loss": 1.5681, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.387690785802403e-06, | |
| "loss": 1.8252, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.068954035279121e-06, | |
| "loss": 1.8779, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 7.752242685312709e-06, | |
| "loss": 2.0095, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.4378889223742766e-06, | |
| "loss": 1.4449, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.126222460151719e-06, | |
| "loss": 1.4738, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.8175701937255645e-06, | |
| "loss": 1.3797, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.5122558567011775e-06, | |
| "loss": 1.3042, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.210599681656933e-06, | |
| "loss": 1.5955, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.912918064264441e-06, | |
| "loss": 1.7519, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 5.619523231433177e-06, | |
| "loss": 1.3313, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.330722913827594e-06, | |
| "loss": 1.5147, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 5.046820023100129e-06, | |
| "loss": 2.0083, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 4.7681123341787e-06, | |
| "loss": 1.4886, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 4.494892172941965e-06, | |
| "loss": 1.4057, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 4.2274461096098085e-06, | |
| "loss": 1.3837, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.966054658170754e-06, | |
| "loss": 1.438, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.7109919821615546e-06, | |
| "loss": 1.3849, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.4625256071074776e-06, | |
| "loss": 1.5841, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.2209161399249677e-06, | |
| "loss": 1.4959, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 2.9864169955810085e-06, | |
| "loss": 2.3643, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.759274131295787e-06, | |
| "loss": 1.6805, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.5397257885675396e-06, | |
| "loss": 1.3571, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.328002243290138e-06, | |
| "loss": 1.4686, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.124325564225458e-06, | |
| "loss": 1.2663, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.9289093800839067e-06, | |
| "loss": 1.4305, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.7419586554574364e-06, | |
| "loss": 1.5805, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.5636694758399563e-06, | |
| "loss": 1.5783, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.3942288419607476e-06, | |
| "loss": 2.0266, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.233814473646524e-06, | |
| "loss": 1.4056, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.0825946234178575e-06, | |
| "loss": 1.3858, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 9.407279000155311e-07, | |
| "loss": 1.5584, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 8.083631020418792e-07, | |
| "loss": 1.2246, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.856390618915775e-07, | |
| "loss": 1.3859, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.726845001356573e-07, | |
| "loss": 1.5712, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.696178905113913e-07, | |
| "loss": 1.6031, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.7654733565969826e-07, | |
| "loss": 2.1996, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 2.935704537404083e-07, | |
| "loss": 1.4927, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 2.2077427604429435e-07, | |
| "loss": 1.479, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.5823515570925763e-07, | |
| "loss": 1.3484, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.0601868763643997e-07, | |
| "loss": 1.1359, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 6.417963969022389e-08, | |
| "loss": 1.5236, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.2761895254306285e-08, | |
| "loss": 1.6316, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.179840720409331e-08, | |
| "loss": 1.5152, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.3111633436779792e-09, | |
| "loss": 1.2889, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 201, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1.1940452642598912e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |