| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.8544274455261938, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998938447446803e-05, | |
| "loss": 2.0078, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9957546913022665e-05, | |
| "loss": 1.8414, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9904514353459654e-05, | |
| "loss": 1.7763, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.983033183325818e-05, | |
| "loss": 1.7278, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.973506235133323e-05, | |
| "loss": 1.7139, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.96313569658781e-05, | |
| "loss": 1.6947, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.94962599008322e-05, | |
| "loss": 1.6749, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9340359581993066e-05, | |
| "loss": 1.6579, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.916378840646592e-05, | |
| "loss": 1.6499, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8966696325916515e-05, | |
| "loss": 1.6507, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8749250719225915e-05, | |
| "loss": 1.6426, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.8511636250345294e-05, | |
| "loss": 1.6288, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.825405471147153e-05, | |
| "loss": 1.6265, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.797672485167683e-05, | |
| "loss": 1.6252, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.7679882191137804e-05, | |
| "loss": 1.6218, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.7363778821121784e-05, | |
| "loss": 1.6146, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.702868318990039e-05, | |
| "loss": 1.604, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.6674879874771926e-05, | |
| "loss": 1.607, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.630266934038642e-05, | |
| "loss": 1.5972, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.591236768357833e-05, | |
| "loss": 1.5973, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.55043063649239e-05, | |
| "loss": 1.6043, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.507883192725089e-05, | |
| "loss": 1.5839, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.463630570133978e-05, | |
| "loss": 1.5923, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.417710349906658e-05, | |
| "loss": 1.5841, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.3701615294247465e-05, | |
| "loss": 1.5977, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.321024489145673e-05, | |
| "loss": 1.5954, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.270340958309888e-05, | |
| "loss": 1.5914, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.2181539795026435e-05, | |
| "loss": 1.577, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1645078721004174e-05, | |
| "loss": 1.576, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.109448194633033e-05, | |
| "loss": 1.5771, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0530217060934466e-05, | |
| "loss": 1.5825, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.9952763262280405e-05, | |
| "loss": 1.581, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9362610948411585e-05, | |
| "loss": 1.5691, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.8760261301484466e-05, | |
| "loss": 1.5795, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.8208140480771856e-05, | |
| "loss": 1.5847, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.758403340686345e-05, | |
| "loss": 1.5737, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.6949239442720976e-05, | |
| "loss": 1.5645, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.6304297682067144e-05, | |
| "loss": 1.5659, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.5649755836560106e-05, | |
| "loss": 1.5613, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.4986169770653685e-05, | |
| "loss": 1.5725, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.431410302953389e-05, | |
| "loss": 1.5619, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.363412636053269e-05, | |
| "loss": 1.5615, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.294681722842537e-05, | |
| "loss": 1.5623, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.225275932502315e-05, | |
| "loss": 1.5685, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.1552542073477555e-05, | |
| "loss": 1.566, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.084676012771753e-05, | |
| "loss": 1.5643, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.0136012867444297e-05, | |
| "loss": 1.5588, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.942090388911291e-05, | |
| "loss": 1.5598, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.8702040493332778e-05, | |
| "loss": 1.5618, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.7980033169122454e-05, | |
| "loss": 1.5602, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.7255495075456693e-05, | |
| "loss": 1.5453, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.652904152054607e-05, | |
| "loss": 1.5557, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.5801289439291388e-05, | |
| "loss": 1.5493, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5072856869356593e-05, | |
| "loss": 1.5527, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.4344362426305255e-05, | |
| "loss": 1.5564, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.3616424778246173e-05, | |
| "loss": 1.5514, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.2889662120434453e-05, | |
| "loss": 1.5427, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.216469165027406e-05, | |
| "loss": 1.5428, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.1442129043167874e-05, | |
| "loss": 1.5441, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 2.0722587929660227e-05, | |
| "loss": 1.552, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.0006679374316062e-05, | |
| "loss": 1.5555, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.9295011356779192e-05, | |
| "loss": 1.5504, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.8588188255450466e-05, | |
| "loss": 1.5452, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.7886810334224192e-05, | |
| "loss": 1.5502, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.7191473232718774e-05, | |
| "loss": 1.556, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.6502767460434588e-05, | |
| "loss": 1.5419, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.582127789526838e-05, | |
| "loss": 1.5528, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.5147583286810485e-05, | |
| "loss": 1.545, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.4482255764846225e-05, | |
| "loss": 1.5433, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.3825860353479336e-05, | |
| "loss": 1.551, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.3178954491289692e-05, | |
| "loss": 1.5424, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.2542087557933041e-05, | |
| "loss": 1.5426, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.1915800407584704e-05, | |
| "loss": 1.5555, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1300624909623463e-05, | |
| "loss": 1.5335, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.0697083496945765e-05, | |
| "loss": 1.5446, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.0105688722293643e-05, | |
| "loss": 1.5422, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.526942822973522e-06, | |
| "loss": 1.548, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.961337294335021e-06, | |
| "loss": 1.5436, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.409352472372595e-06, | |
| "loss": 1.5483, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.871457125803896e-06, | |
| "loss": 1.5348, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.348108057971728e-06, | |
| "loss": 1.5309, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 6.839749718907429e-06, | |
| "loss": 1.5343, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.3468138278852174e-06, | |
| "loss": 1.5446, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 5.8697190067880325e-06, | |
| "loss": 1.5381, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 5.408870424596238e-06, | |
| "loss": 1.5483, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.9646594533010875e-06, | |
| "loss": 1.5466, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.537463335535161e-06, | |
| "loss": 1.5397, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.127644864202104e-06, | |
| "loss": 1.5355, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.735552074377563e-06, | |
| "loss": 1.5409, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.3615179477432645e-06, | |
| "loss": 1.5383, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.0058601298048774e-06, | |
| "loss": 1.5338, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.6688806601341765e-06, | |
| "loss": 1.5384, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.350865715864278e-06, | |
| "loss": 1.5305, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0520853686560178e-06, | |
| "loss": 1.5344, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.772793355341734e-06, | |
| "loss": 1.5363, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.513226862441286e-06, | |
| "loss": 1.5468, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.273606324733284e-06, | |
| "loss": 1.5369, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.0541352380526087e-06, | |
| "loss": 1.5428, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.549999864732011e-07, | |
| "loss": 1.5338, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.763696840228456e-07, | |
| "loss": 1.5399, | |
| "step": 1000 | |
| } | |
| ], | |
| "max_steps": 1078, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.489434826550018e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |