| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9928047404063205, | |
| "global_step": 56500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.955911399548533e-05, | |
| "loss": 1.0947, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.911822799097065e-05, | |
| "loss": 0.9881, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.867734198645598e-05, | |
| "loss": 0.9375, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.8236455981941316e-05, | |
| "loss": 0.9275, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.779556997742664e-05, | |
| "loss": 0.9047, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.7354683972911966e-05, | |
| "loss": 0.8813, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.6913797968397295e-05, | |
| "loss": 0.8806, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.647291196388262e-05, | |
| "loss": 0.8718, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.6032025959367945e-05, | |
| "loss": 0.8828, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.559113995485327e-05, | |
| "loss": 0.8524, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.51502539503386e-05, | |
| "loss": 0.8464, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.470936794582393e-05, | |
| "loss": 0.8411, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.426848194130926e-05, | |
| "loss": 0.839, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.382759593679459e-05, | |
| "loss": 0.838, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.338670993227991e-05, | |
| "loss": 0.8531, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.294582392776524e-05, | |
| "loss": 0.8109, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.2504937923250566e-05, | |
| "loss": 0.8364, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.2064051918735894e-05, | |
| "loss": 0.8249, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1623165914221216e-05, | |
| "loss": 0.8349, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.118227990970655e-05, | |
| "loss": 0.8136, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.074139390519188e-05, | |
| "loss": 0.8099, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.03005079006772e-05, | |
| "loss": 0.8054, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.985962189616253e-05, | |
| "loss": 0.8195, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.941873589164786e-05, | |
| "loss": 0.8646, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.897784988713318e-05, | |
| "loss": 0.8038, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.853696388261851e-05, | |
| "loss": 0.7897, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.8096077878103843e-05, | |
| "loss": 0.7991, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.7655191873589165e-05, | |
| "loss": 0.7834, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.7214305869074494e-05, | |
| "loss": 0.8035, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.677341986455982e-05, | |
| "loss": 0.7965, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.633253386004515e-05, | |
| "loss": 0.8022, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.589164785553047e-05, | |
| "loss": 0.7959, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.54507618510158e-05, | |
| "loss": 0.8167, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.500987584650113e-05, | |
| "loss": 0.8003, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.456898984198646e-05, | |
| "loss": 0.8143, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4128103837471786e-05, | |
| "loss": 0.797, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.3687217832957114e-05, | |
| "loss": 0.7655, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.324633182844244e-05, | |
| "loss": 0.7821, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2805445823927765e-05, | |
| "loss": 0.7721, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.236455981941309e-05, | |
| "loss": 0.835, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.192367381489842e-05, | |
| "loss": 0.7945, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.148278781038374e-05, | |
| "loss": 0.8042, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.104190180586908e-05, | |
| "loss": 0.7661, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.060101580135441e-05, | |
| "loss": 0.7981, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.0160129796839732e-05, | |
| "loss": 0.7588, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.9719243792325057e-05, | |
| "loss": 0.837, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.9278357787810385e-05, | |
| "loss": 0.7686, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.883747178329571e-05, | |
| "loss": 0.7711, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.839658577878104e-05, | |
| "loss": 0.7785, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.795569977426637e-05, | |
| "loss": 0.7949, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7514813769751696e-05, | |
| "loss": 0.7737, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.7073927765237024e-05, | |
| "loss": 0.7719, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.663304176072235e-05, | |
| "loss": 0.7666, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.6192155756207674e-05, | |
| "loss": 0.7678, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.5751269751693003e-05, | |
| "loss": 0.8034, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.5310383747178328e-05, | |
| "loss": 0.748, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6928090453147888, | |
| "eval_loss": 0.7356609106063843, | |
| "eval_runtime": 352.934, | |
| "eval_samples_per_second": 66.945, | |
| "eval_steps_per_second": 11.158, | |
| "step": 28352 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.486949774266366e-05, | |
| "loss": 0.7526, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.4428611738148985e-05, | |
| "loss": 0.6793, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.398772573363431e-05, | |
| "loss": 0.7199, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.3546839729119642e-05, | |
| "loss": 0.6832, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.3105953724604967e-05, | |
| "loss": 0.6787, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.2665067720090295e-05, | |
| "loss": 0.6795, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 2.222418171557562e-05, | |
| "loss": 0.6592, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.178329571106095e-05, | |
| "loss": 0.6692, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.1342409706546277e-05, | |
| "loss": 0.6884, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.0901523702031602e-05, | |
| "loss": 0.6915, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.046063769751693e-05, | |
| "loss": 0.659, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.001975169300226e-05, | |
| "loss": 0.6741, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.9578865688487584e-05, | |
| "loss": 0.6489, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.9137979683972913e-05, | |
| "loss": 0.6568, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.869709367945824e-05, | |
| "loss": 0.6982, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.8256207674943566e-05, | |
| "loss": 0.6673, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.7815321670428895e-05, | |
| "loss": 0.675, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.737443566591422e-05, | |
| "loss": 0.6711, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.693354966139955e-05, | |
| "loss": 0.6643, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.6492663656884877e-05, | |
| "loss": 0.6788, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.6051777652370205e-05, | |
| "loss": 0.6976, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.561089164785553e-05, | |
| "loss": 0.6543, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5170005643340859e-05, | |
| "loss": 0.6651, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.4729119638826185e-05, | |
| "loss": 0.6512, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4288233634311512e-05, | |
| "loss": 0.6417, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.3847347629796839e-05, | |
| "loss": 0.662, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.3406461625282169e-05, | |
| "loss": 0.6715, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.2965575620767496e-05, | |
| "loss": 0.6519, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.2524689616252821e-05, | |
| "loss": 0.627, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.208380361173815e-05, | |
| "loss": 0.6357, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.1642917607223478e-05, | |
| "loss": 0.6612, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.1202031602708805e-05, | |
| "loss": 0.688, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.0761145598194131e-05, | |
| "loss": 0.6556, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.0320259593679458e-05, | |
| "loss": 0.6514, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.879373589164786e-06, | |
| "loss": 0.6476, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.438487584650113e-06, | |
| "loss": 0.6519, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.997601580135442e-06, | |
| "loss": 0.612, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.556715575620767e-06, | |
| "loss": 0.6314, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.115829571106095e-06, | |
| "loss": 0.6279, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.674943566591422e-06, | |
| "loss": 0.6498, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7.23405756207675e-06, | |
| "loss": 0.6608, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.793171557562076e-06, | |
| "loss": 0.645, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.352285553047405e-06, | |
| "loss": 0.6612, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.9113995485327315e-06, | |
| "loss": 0.6263, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.470513544018059e-06, | |
| "loss": 0.6374, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.029627539503387e-06, | |
| "loss": 0.6385, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.5887415349887135e-06, | |
| "loss": 0.6362, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.147855530474041e-06, | |
| "loss": 0.6342, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.7069695259593683e-06, | |
| "loss": 0.6714, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.2660835214446955e-06, | |
| "loss": 0.6355, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.8251975169300226e-06, | |
| "loss": 0.6151, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.38431151241535e-06, | |
| "loss": 0.6461, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.943425507900677e-06, | |
| "loss": 0.6533, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.5025395033860046e-06, | |
| "loss": 0.6376, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0616534988713318e-06, | |
| "loss": 0.5976, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.207674943566591e-07, | |
| "loss": 0.6318, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.7988148984198645e-07, | |
| "loss": 0.6471, | |
| "step": 56500 | |
| } | |
| ], | |
| "max_steps": 56704, | |
| "num_train_epochs": 2, | |
| "total_flos": 7.589219989624102e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |