| { | |
| "best_metric": 0.7700450584385239, | |
| "best_model_checkpoint": "./checkpoints/clip-10240-supervised-multilingual/checkpoint-60", | |
| "epoch": 0.8346374543557642, | |
| "global_step": 1600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-06, | |
| "loss": 4.2235, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.454, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.9628, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.6604, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3e-05, | |
| "loss": 1.4726, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9670329670329673e-05, | |
| "loss": 1.3632, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.934065934065934e-05, | |
| "loss": 1.2782, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.9010989010989013e-05, | |
| "loss": 1.2024, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.8681318681318685e-05, | |
| "loss": 1.1547, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.8351648351648353e-05, | |
| "loss": 1.1092, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.802197802197802e-05, | |
| "loss": 1.0757, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.7692307692307694e-05, | |
| "loss": 1.044, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.7362637362637362e-05, | |
| "loss": 1.0206, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.7032967032967034e-05, | |
| "loss": 0.9943, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.6703296703296702e-05, | |
| "loss": 0.976, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.6373626373626374e-05, | |
| "loss": 0.955, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.6043956043956046e-05, | |
| "loss": 0.9377, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.5714285714285714e-05, | |
| "loss": 0.9052, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.5384615384615386e-05, | |
| "loss": 0.9012, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.5054945054945058e-05, | |
| "loss": 0.8827, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.4725274725274723e-05, | |
| "loss": 0.8755, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.4395604395604395e-05, | |
| "loss": 0.8625, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.4065934065934067e-05, | |
| "loss": 0.8507, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.3736263736263735e-05, | |
| "loss": 0.8301, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.3406593406593407e-05, | |
| "loss": 0.8237, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.307692307692308e-05, | |
| "loss": 0.8128, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.2747252747252748e-05, | |
| "loss": 0.8115, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.241758241758242e-05, | |
| "loss": 0.7954, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.208791208791209e-05, | |
| "loss": 0.7855, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.175824175824176e-05, | |
| "loss": 0.7718, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.1428571428571428e-05, | |
| "loss": 0.7685, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.10989010989011e-05, | |
| "loss": 0.7562, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.076923076923077e-05, | |
| "loss": 0.752, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.043956043956044e-05, | |
| "loss": 0.7468, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.010989010989011e-05, | |
| "loss": 0.7395, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.978021978021978e-05, | |
| "loss": 0.7314, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9450549450549452e-05, | |
| "loss": 0.7256, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.912087912087912e-05, | |
| "loss": 0.7159, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8791208791208793e-05, | |
| "loss": 0.7139, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8461538461538465e-05, | |
| "loss": 0.7042, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.813186813186813e-05, | |
| "loss": 0.7033, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.78021978021978e-05, | |
| "loss": 0.6961, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.7472527472527473e-05, | |
| "loss": 0.6908, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 0.6893, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6813186813186814e-05, | |
| "loss": 0.6789, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.6483516483516486e-05, | |
| "loss": 0.6742, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6153846153846154e-05, | |
| "loss": 0.6702, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.5824175824175826e-05, | |
| "loss": 0.6681, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.5494505494505498e-05, | |
| "loss": 0.6618, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5164835164835164e-05, | |
| "loss": 0.6572, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4835164835164836e-05, | |
| "loss": 0.6571, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.4505494505494506e-05, | |
| "loss": 0.6512, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4175824175824177e-05, | |
| "loss": 0.6533, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.3846153846153847e-05, | |
| "loss": 0.6432, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.3516483516483517e-05, | |
| "loss": 0.642, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.3186813186813187e-05, | |
| "loss": 0.6336, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.2857142857142857e-05, | |
| "loss": 0.6356, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.2527472527472529e-05, | |
| "loss": 0.6291, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.2197802197802198e-05, | |
| "loss": 0.6312, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.1868131868131868e-05, | |
| "loss": 0.6281, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.153846153846154e-05, | |
| "loss": 0.6251, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.120879120879121e-05, | |
| "loss": 0.6237, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.087912087912088e-05, | |
| "loss": 0.614, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.054945054945055e-05, | |
| "loss": 0.6199, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.021978021978022e-05, | |
| "loss": 0.6127, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.89010989010989e-06, | |
| "loss": 0.6153, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.56043956043956e-06, | |
| "loss": 0.6075, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 0.6046, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.9010989010989e-06, | |
| "loss": 0.6106, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.6083, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.241758241758243e-06, | |
| "loss": 0.5984, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.912087912087913e-06, | |
| "loss": 0.5974, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.582417582417582e-06, | |
| "loss": 0.5987, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.252747252747253e-06, | |
| "loss": 0.5972, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 0.5891, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.5934065934065935e-06, | |
| "loss": 0.5924, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.2637362637362645e-06, | |
| "loss": 0.5881, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.934065934065934e-06, | |
| "loss": 0.5886, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.604395604395605e-06, | |
| "loss": 0.5889, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.274725274725275e-06, | |
| "loss": 0.5904, | |
| "step": 1600 | |
| } | |
| ], | |
| "max_steps": 1920, | |
| "num_train_epochs": 2, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |