| { | |
| "dataset_name": "Mu-Math", | |
| "group_index": 4, | |
| "prompt_group_file": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/prompt_groups/Mu-Math/group_04.json", | |
| "output_dir": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/outputs/Mu-Math/group_04", | |
| "checkpoint_root": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/outputs/Mu-Math/group_04/checkpoints", | |
| "generated_at": "2025-11-06T12:12:13Z", | |
| "train_loss": 0.050706694109572306, | |
| "metrics": { | |
| "train_runtime": 30947.3969, | |
| "train_samples_per_second": 1.861, | |
| "train_steps_per_second": 0.058, | |
| "total_flos": 1.5308141101056e+18, | |
| "train_loss": 0.050706694109572306, | |
| "epoch": 450.0 | |
| }, | |
| "trainer_state": [ | |
| { | |
| "loss": 1.8699, | |
| "grad_norm": 11.692859649658203, | |
| "learning_rate": 0.0, | |
| "epoch": 0.32, | |
| "step": 1 | |
| }, | |
| { | |
| "loss": 1.8216, | |
| "grad_norm": 7.065145969390869, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "epoch": 1.32, | |
| "step": 5 | |
| }, | |
| { | |
| "loss": 1.6433, | |
| "grad_norm": 3.5841238498687744, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "epoch": 2.64, | |
| "step": 10 | |
| }, | |
| { | |
| "loss": 1.2063, | |
| "grad_norm": 1.5809892416000366, | |
| "learning_rate": 2.5925925925925925e-05, | |
| "epoch": 3.96, | |
| "step": 15 | |
| }, | |
| { | |
| "loss": 0.9937, | |
| "grad_norm": 1.4060782194137573, | |
| "learning_rate": 3.518518518518519e-05, | |
| "epoch": 5.0, | |
| "step": 20 | |
| }, | |
| { | |
| "loss": 0.77, | |
| "grad_norm": 0.5133189558982849, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "epoch": 6.32, | |
| "step": 25 | |
| }, | |
| { | |
| "loss": 0.6743, | |
| "grad_norm": 0.7827394604682922, | |
| "learning_rate": 5.370370370370371e-05, | |
| "epoch": 7.64, | |
| "step": 30 | |
| }, | |
| { | |
| "loss": 0.5614, | |
| "grad_norm": 0.5059296488761902, | |
| "learning_rate": 6.296296296296296e-05, | |
| "epoch": 8.96, | |
| "step": 35 | |
| }, | |
| { | |
| "loss": 0.4787, | |
| "grad_norm": 1.9664890766143799, | |
| "learning_rate": 7.222222222222222e-05, | |
| "epoch": 10.0, | |
| "step": 40 | |
| }, | |
| { | |
| "loss": 0.3254, | |
| "grad_norm": 0.5983086824417114, | |
| "learning_rate": 8.148148148148148e-05, | |
| "epoch": 11.32, | |
| "step": 45 | |
| }, | |
| { | |
| "loss": 0.2289, | |
| "grad_norm": 0.5863825082778931, | |
| "learning_rate": 9.074074074074075e-05, | |
| "epoch": 12.64, | |
| "step": 50 | |
| }, | |
| { | |
| "loss": 0.1611, | |
| "grad_norm": 1.312117099761963, | |
| "learning_rate": 0.0001, | |
| "epoch": 13.96, | |
| "step": 55 | |
| }, | |
| { | |
| "loss": 0.1012, | |
| "grad_norm": 1.723562240600586, | |
| "learning_rate": 9.971363115693013e-05, | |
| "epoch": 15.0, | |
| "step": 60 | |
| }, | |
| { | |
| "loss": 0.0705, | |
| "grad_norm": 0.610700249671936, | |
| "learning_rate": 9.942726231386026e-05, | |
| "epoch": 16.32, | |
| "step": 65 | |
| }, | |
| { | |
| "loss": 0.0533, | |
| "grad_norm": 0.44096142053604126, | |
| "learning_rate": 9.914089347079038e-05, | |
| "epoch": 17.64, | |
| "step": 70 | |
| }, | |
| { | |
| "loss": 0.0446, | |
| "grad_norm": 0.4199007749557495, | |
| "learning_rate": 9.885452462772051e-05, | |
| "epoch": 18.96, | |
| "step": 75 | |
| }, | |
| { | |
| "loss": 0.0383, | |
| "grad_norm": 1.3046945333480835, | |
| "learning_rate": 9.856815578465064e-05, | |
| "epoch": 20.0, | |
| "step": 80 | |
| }, | |
| { | |
| "loss": 0.0349, | |
| "grad_norm": 0.20499350130558014, | |
| "learning_rate": 9.828178694158075e-05, | |
| "epoch": 21.32, | |
| "step": 85 | |
| }, | |
| { | |
| "loss": 0.0347, | |
| "grad_norm": 0.16132423281669617, | |
| "learning_rate": 9.799541809851088e-05, | |
| "epoch": 22.64, | |
| "step": 90 | |
| }, | |
| { | |
| "loss": 0.0344, | |
| "grad_norm": 0.1392243653535843, | |
| "learning_rate": 9.7709049255441e-05, | |
| "epoch": 23.96, | |
| "step": 95 | |
| }, | |
| { | |
| "loss": 0.0411, | |
| "grad_norm": 0.2796006500720978, | |
| "learning_rate": 9.742268041237114e-05, | |
| "epoch": 25.0, | |
| "step": 100 | |
| }, | |
| { | |
| "loss": 0.0299, | |
| "grad_norm": 0.1289726346731186, | |
| "learning_rate": 9.713631156930127e-05, | |
| "epoch": 26.32, | |
| "step": 105 | |
| }, | |
| { | |
| "loss": 0.0329, | |
| "grad_norm": 0.13365666568279266, | |
| "learning_rate": 9.68499427262314e-05, | |
| "epoch": 27.64, | |
| "step": 110 | |
| }, | |
| { | |
| "loss": 0.0291, | |
| "grad_norm": 0.16668930649757385, | |
| "learning_rate": 9.656357388316152e-05, | |
| "epoch": 28.96, | |
| "step": 115 | |
| }, | |
| { | |
| "loss": 0.0297, | |
| "grad_norm": 0.34922805428504944, | |
| "learning_rate": 9.627720504009165e-05, | |
| "epoch": 30.0, | |
| "step": 120 | |
| }, | |
| { | |
| "loss": 0.0308, | |
| "grad_norm": 0.11170455068349838, | |
| "learning_rate": 9.599083619702178e-05, | |
| "epoch": 31.32, | |
| "step": 125 | |
| }, | |
| { | |
| "loss": 0.0312, | |
| "grad_norm": 0.11236865818500519, | |
| "learning_rate": 9.57044673539519e-05, | |
| "epoch": 32.64, | |
| "step": 130 | |
| }, | |
| { | |
| "loss": 0.028, | |
| "grad_norm": 0.10385998338460922, | |
| "learning_rate": 9.541809851088203e-05, | |
| "epoch": 33.96, | |
| "step": 135 | |
| }, | |
| { | |
| "loss": 0.0289, | |
| "grad_norm": 0.20068448781967163, | |
| "learning_rate": 9.513172966781214e-05, | |
| "epoch": 35.0, | |
| "step": 140 | |
| }, | |
| { | |
| "loss": 0.0321, | |
| "grad_norm": 0.08476591855287552, | |
| "learning_rate": 9.484536082474227e-05, | |
| "epoch": 36.32, | |
| "step": 145 | |
| }, | |
| { | |
| "loss": 0.0293, | |
| "grad_norm": 0.08026140183210373, | |
| "learning_rate": 9.45589919816724e-05, | |
| "epoch": 37.64, | |
| "step": 150 | |
| }, | |
| { | |
| "loss": 0.0307, | |
| "grad_norm": 0.07616273313760757, | |
| "learning_rate": 9.427262313860252e-05, | |
| "epoch": 38.96, | |
| "step": 155 | |
| }, | |
| { | |
| "loss": 0.0313, | |
| "grad_norm": 0.19227686524391174, | |
| "learning_rate": 9.398625429553265e-05, | |
| "epoch": 40.0, | |
| "step": 160 | |
| }, | |
| { | |
| "loss": 0.0307, | |
| "grad_norm": 0.09652470797300339, | |
| "learning_rate": 9.369988545246277e-05, | |
| "epoch": 41.32, | |
| "step": 165 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.09458985924720764, | |
| "learning_rate": 9.34135166093929e-05, | |
| "epoch": 42.64, | |
| "step": 170 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.06568342447280884, | |
| "learning_rate": 9.312714776632303e-05, | |
| "epoch": 43.96, | |
| "step": 175 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.12499319016933441, | |
| "learning_rate": 9.284077892325315e-05, | |
| "epoch": 45.0, | |
| "step": 180 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.07067073881626129, | |
| "learning_rate": 9.255441008018328e-05, | |
| "epoch": 46.32, | |
| "step": 185 | |
| }, | |
| { | |
| "loss": 0.0292, | |
| "grad_norm": 0.10531776398420334, | |
| "learning_rate": 9.22680412371134e-05, | |
| "epoch": 47.64, | |
| "step": 190 | |
| }, | |
| { | |
| "loss": 0.0285, | |
| "grad_norm": 0.07698921859264374, | |
| "learning_rate": 9.198167239404353e-05, | |
| "epoch": 48.96, | |
| "step": 195 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.3016670048236847, | |
| "learning_rate": 9.169530355097366e-05, | |
| "epoch": 50.0, | |
| "step": 200 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.06784846633672714, | |
| "learning_rate": 9.140893470790379e-05, | |
| "epoch": 51.32, | |
| "step": 205 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.07265879958868027, | |
| "learning_rate": 9.112256586483391e-05, | |
| "epoch": 52.64, | |
| "step": 210 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.07025367021560669, | |
| "learning_rate": 9.083619702176404e-05, | |
| "epoch": 53.96, | |
| "step": 215 | |
| }, | |
| { | |
| "loss": 0.0294, | |
| "grad_norm": 0.17506320774555206, | |
| "learning_rate": 9.054982817869416e-05, | |
| "epoch": 55.0, | |
| "step": 220 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.06594470888376236, | |
| "learning_rate": 9.026345933562429e-05, | |
| "epoch": 56.32, | |
| "step": 225 | |
| }, | |
| { | |
| "loss": 0.029, | |
| "grad_norm": 0.06064048409461975, | |
| "learning_rate": 8.997709049255442e-05, | |
| "epoch": 57.64, | |
| "step": 230 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.08256722241640091, | |
| "learning_rate": 8.969072164948454e-05, | |
| "epoch": 58.96, | |
| "step": 235 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.193592369556427, | |
| "learning_rate": 8.940435280641467e-05, | |
| "epoch": 60.0, | |
| "step": 240 | |
| }, | |
| { | |
| "loss": 0.029, | |
| "grad_norm": 0.054831646382808685, | |
| "learning_rate": 8.91179839633448e-05, | |
| "epoch": 61.32, | |
| "step": 245 | |
| }, | |
| { | |
| "loss": 0.0293, | |
| "grad_norm": 0.05805569887161255, | |
| "learning_rate": 8.883161512027491e-05, | |
| "epoch": 62.64, | |
| "step": 250 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.07056920975446701, | |
| "learning_rate": 8.854524627720504e-05, | |
| "epoch": 63.96, | |
| "step": 255 | |
| }, | |
| { | |
| "loss": 0.0316, | |
| "grad_norm": 0.20423397421836853, | |
| "learning_rate": 8.825887743413516e-05, | |
| "epoch": 65.0, | |
| "step": 260 | |
| }, | |
| { | |
| "loss": 0.0289, | |
| "grad_norm": 0.0642482191324234, | |
| "learning_rate": 8.797250859106529e-05, | |
| "epoch": 66.32, | |
| "step": 265 | |
| }, | |
| { | |
| "loss": 0.0292, | |
| "grad_norm": 0.06649542599916458, | |
| "learning_rate": 8.768613974799542e-05, | |
| "epoch": 67.64, | |
| "step": 270 | |
| }, | |
| { | |
| "loss": 0.0278, | |
| "grad_norm": 0.08042517304420471, | |
| "learning_rate": 8.739977090492554e-05, | |
| "epoch": 68.96, | |
| "step": 275 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.1491345465183258, | |
| "learning_rate": 8.711340206185567e-05, | |
| "epoch": 70.0, | |
| "step": 280 | |
| }, | |
| { | |
| "loss": 0.0285, | |
| "grad_norm": 0.061604879796504974, | |
| "learning_rate": 8.682703321878581e-05, | |
| "epoch": 71.32, | |
| "step": 285 | |
| }, | |
| { | |
| "loss": 0.0304, | |
| "grad_norm": 0.4991450011730194, | |
| "learning_rate": 8.654066437571594e-05, | |
| "epoch": 72.64, | |
| "step": 290 | |
| }, | |
| { | |
| "loss": 0.0312, | |
| "grad_norm": 0.3048219084739685, | |
| "learning_rate": 8.625429553264606e-05, | |
| "epoch": 73.96, | |
| "step": 295 | |
| }, | |
| { | |
| "loss": 0.0394, | |
| "grad_norm": 2.4087955951690674, | |
| "learning_rate": 8.596792668957619e-05, | |
| "epoch": 75.0, | |
| "step": 300 | |
| }, | |
| { | |
| "loss": 0.031, | |
| "grad_norm": 0.1915508359670639, | |
| "learning_rate": 8.56815578465063e-05, | |
| "epoch": 76.32, | |
| "step": 305 | |
| }, | |
| { | |
| "loss": 0.0291, | |
| "grad_norm": 0.11591818183660507, | |
| "learning_rate": 8.539518900343643e-05, | |
| "epoch": 77.64, | |
| "step": 310 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.3887242376804352, | |
| "learning_rate": 8.510882016036655e-05, | |
| "epoch": 78.96, | |
| "step": 315 | |
| }, | |
| { | |
| "loss": 0.0307, | |
| "grad_norm": 0.23419131338596344, | |
| "learning_rate": 8.482245131729668e-05, | |
| "epoch": 80.0, | |
| "step": 320 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.10408481955528259, | |
| "learning_rate": 8.453608247422681e-05, | |
| "epoch": 81.32, | |
| "step": 325 | |
| }, | |
| { | |
| "loss": 0.031, | |
| "grad_norm": 0.21175497770309448, | |
| "learning_rate": 8.424971363115693e-05, | |
| "epoch": 82.64, | |
| "step": 330 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.052649106830358505, | |
| "learning_rate": 8.396334478808706e-05, | |
| "epoch": 83.96, | |
| "step": 335 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.1649101972579956, | |
| "learning_rate": 8.367697594501719e-05, | |
| "epoch": 85.0, | |
| "step": 340 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.04991982877254486, | |
| "learning_rate": 8.339060710194731e-05, | |
| "epoch": 86.32, | |
| "step": 345 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.05327003449201584, | |
| "learning_rate": 8.310423825887744e-05, | |
| "epoch": 87.64, | |
| "step": 350 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.052796024829149246, | |
| "learning_rate": 8.281786941580757e-05, | |
| "epoch": 88.96, | |
| "step": 355 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.1174449771642685, | |
| "learning_rate": 8.253150057273768e-05, | |
| "epoch": 90.0, | |
| "step": 360 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.049136094748973846, | |
| "learning_rate": 8.224513172966782e-05, | |
| "epoch": 91.32, | |
| "step": 365 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.05051523074507713, | |
| "learning_rate": 8.195876288659795e-05, | |
| "epoch": 92.64, | |
| "step": 370 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.04851191118359566, | |
| "learning_rate": 8.167239404352807e-05, | |
| "epoch": 93.96, | |
| "step": 375 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.11616070568561554, | |
| "learning_rate": 8.13860252004582e-05, | |
| "epoch": 95.0, | |
| "step": 380 | |
| }, | |
| { | |
| "loss": 0.0289, | |
| "grad_norm": 0.04650884494185448, | |
| "learning_rate": 8.109965635738833e-05, | |
| "epoch": 96.32, | |
| "step": 385 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.047667283564805984, | |
| "learning_rate": 8.081328751431845e-05, | |
| "epoch": 97.64, | |
| "step": 390 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.04733484238386154, | |
| "learning_rate": 8.052691867124858e-05, | |
| "epoch": 98.96, | |
| "step": 395 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.19330866634845734, | |
| "learning_rate": 8.02405498281787e-05, | |
| "epoch": 100.0, | |
| "step": 400 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.05003388971090317, | |
| "learning_rate": 7.995418098510883e-05, | |
| "epoch": 101.32, | |
| "step": 405 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.041838694363832474, | |
| "learning_rate": 7.966781214203894e-05, | |
| "epoch": 102.64, | |
| "step": 410 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.05561203137040138, | |
| "learning_rate": 7.938144329896907e-05, | |
| "epoch": 103.96, | |
| "step": 415 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.11651888489723206, | |
| "learning_rate": 7.90950744558992e-05, | |
| "epoch": 105.0, | |
| "step": 420 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.04522060975432396, | |
| "learning_rate": 7.880870561282932e-05, | |
| "epoch": 106.32, | |
| "step": 425 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.0425175316631794, | |
| "learning_rate": 7.852233676975945e-05, | |
| "epoch": 107.64, | |
| "step": 430 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.04209710657596588, | |
| "learning_rate": 7.823596792668958e-05, | |
| "epoch": 108.96, | |
| "step": 435 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.14562757313251495, | |
| "learning_rate": 7.79495990836197e-05, | |
| "epoch": 110.0, | |
| "step": 440 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.03748113662004471, | |
| "learning_rate": 7.766323024054983e-05, | |
| "epoch": 111.32, | |
| "step": 445 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.040174227207899094, | |
| "learning_rate": 7.737686139747996e-05, | |
| "epoch": 112.64, | |
| "step": 450 | |
| }, | |
| { | |
| "loss": 0.029, | |
| "grad_norm": 0.040747471153736115, | |
| "learning_rate": 7.709049255441008e-05, | |
| "epoch": 113.96, | |
| "step": 455 | |
| }, | |
| { | |
| "loss": 0.029, | |
| "grad_norm": 0.1542213410139084, | |
| "learning_rate": 7.680412371134021e-05, | |
| "epoch": 115.0, | |
| "step": 460 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.042015790939331055, | |
| "learning_rate": 7.651775486827034e-05, | |
| "epoch": 116.32, | |
| "step": 465 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.04656674340367317, | |
| "learning_rate": 7.623138602520046e-05, | |
| "epoch": 117.64, | |
| "step": 470 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.04147423803806305, | |
| "learning_rate": 7.594501718213059e-05, | |
| "epoch": 118.96, | |
| "step": 475 | |
| }, | |
| { | |
| "loss": 0.0273, | |
| "grad_norm": 0.10380031913518906, | |
| "learning_rate": 7.565864833906071e-05, | |
| "epoch": 120.0, | |
| "step": 480 | |
| }, | |
| { | |
| "loss": 0.0296, | |
| "grad_norm": 0.036668192595243454, | |
| "learning_rate": 7.537227949599084e-05, | |
| "epoch": 121.32, | |
| "step": 485 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.03682679682970047, | |
| "learning_rate": 7.508591065292097e-05, | |
| "epoch": 122.64, | |
| "step": 490 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.035644371062517166, | |
| "learning_rate": 7.47995418098511e-05, | |
| "epoch": 123.96, | |
| "step": 495 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.14637862145900726, | |
| "learning_rate": 7.451317296678122e-05, | |
| "epoch": 125.0, | |
| "step": 500 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.058375801891088486, | |
| "learning_rate": 7.422680412371135e-05, | |
| "epoch": 126.32, | |
| "step": 505 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.04191552475094795, | |
| "learning_rate": 7.394043528064147e-05, | |
| "epoch": 127.64, | |
| "step": 510 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.03145256265997887, | |
| "learning_rate": 7.36540664375716e-05, | |
| "epoch": 128.96, | |
| "step": 515 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.1512438803911209, | |
| "learning_rate": 7.336769759450171e-05, | |
| "epoch": 130.0, | |
| "step": 520 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.033739957958459854, | |
| "learning_rate": 7.308132875143184e-05, | |
| "epoch": 131.32, | |
| "step": 525 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.037058182060718536, | |
| "learning_rate": 7.279495990836197e-05, | |
| "epoch": 132.64, | |
| "step": 530 | |
| }, | |
| { | |
| "loss": 0.0286, | |
| "grad_norm": 0.03618552163243294, | |
| "learning_rate": 7.250859106529209e-05, | |
| "epoch": 133.96, | |
| "step": 535 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.10497462004423141, | |
| "learning_rate": 7.222222222222222e-05, | |
| "epoch": 135.0, | |
| "step": 540 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.03696698695421219, | |
| "learning_rate": 7.193585337915235e-05, | |
| "epoch": 136.32, | |
| "step": 545 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.029829595237970352, | |
| "learning_rate": 7.164948453608247e-05, | |
| "epoch": 137.64, | |
| "step": 550 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.03659842535853386, | |
| "learning_rate": 7.136311569301261e-05, | |
| "epoch": 138.96, | |
| "step": 555 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.11393796652555466, | |
| "learning_rate": 7.107674684994274e-05, | |
| "epoch": 140.0, | |
| "step": 560 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.03492063283920288, | |
| "learning_rate": 7.079037800687286e-05, | |
| "epoch": 141.32, | |
| "step": 565 | |
| }, | |
| { | |
| "loss": 0.0279, | |
| "grad_norm": 0.03462018072605133, | |
| "learning_rate": 7.050400916380299e-05, | |
| "epoch": 142.64, | |
| "step": 570 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.04572908580303192, | |
| "learning_rate": 7.02176403207331e-05, | |
| "epoch": 143.96, | |
| "step": 575 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.06328148394823074, | |
| "learning_rate": 6.993127147766323e-05, | |
| "epoch": 145.0, | |
| "step": 580 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.03475407883524895, | |
| "learning_rate": 6.964490263459336e-05, | |
| "epoch": 146.32, | |
| "step": 585 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02716582454741001, | |
| "learning_rate": 6.935853379152348e-05, | |
| "epoch": 147.64, | |
| "step": 590 | |
| }, | |
| { | |
| "loss": 0.0288, | |
| "grad_norm": 0.043476201593875885, | |
| "learning_rate": 6.907216494845361e-05, | |
| "epoch": 148.96, | |
| "step": 595 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.12141745537519455, | |
| "learning_rate": 6.878579610538374e-05, | |
| "epoch": 150.0, | |
| "step": 600 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.03419503942131996, | |
| "learning_rate": 6.849942726231386e-05, | |
| "epoch": 151.32, | |
| "step": 605 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.047660406678915024, | |
| "learning_rate": 6.821305841924399e-05, | |
| "epoch": 152.64, | |
| "step": 610 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.04545063525438309, | |
| "learning_rate": 6.792668957617412e-05, | |
| "epoch": 153.96, | |
| "step": 615 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.18920128047466278, | |
| "learning_rate": 6.764032073310424e-05, | |
| "epoch": 155.0, | |
| "step": 620 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.044060833752155304, | |
| "learning_rate": 6.735395189003437e-05, | |
| "epoch": 156.32, | |
| "step": 625 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.031964316964149475, | |
| "learning_rate": 6.706758304696448e-05, | |
| "epoch": 157.64, | |
| "step": 630 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.04160606116056442, | |
| "learning_rate": 6.678121420389462e-05, | |
| "epoch": 158.96, | |
| "step": 635 | |
| }, | |
| { | |
| "loss": 0.031, | |
| "grad_norm": 0.18344326317310333, | |
| "learning_rate": 6.649484536082475e-05, | |
| "epoch": 160.0, | |
| "step": 640 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.034284207969903946, | |
| "learning_rate": 6.620847651775487e-05, | |
| "epoch": 161.32, | |
| "step": 645 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.040628831833601, | |
| "learning_rate": 6.5922107674685e-05, | |
| "epoch": 162.64, | |
| "step": 650 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.0370308980345726, | |
| "learning_rate": 6.563573883161513e-05, | |
| "epoch": 163.96, | |
| "step": 655 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.09529199451208115, | |
| "learning_rate": 6.534936998854525e-05, | |
| "epoch": 165.0, | |
| "step": 660 | |
| }, | |
| { | |
| "loss": 0.0281, | |
| "grad_norm": 0.041172780096530914, | |
| "learning_rate": 6.506300114547538e-05, | |
| "epoch": 166.32, | |
| "step": 665 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.03005264140665531, | |
| "learning_rate": 6.477663230240551e-05, | |
| "epoch": 167.64, | |
| "step": 670 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.03522620350122452, | |
| "learning_rate": 6.449026345933563e-05, | |
| "epoch": 168.96, | |
| "step": 675 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.0983290746808052, | |
| "learning_rate": 6.420389461626576e-05, | |
| "epoch": 170.0, | |
| "step": 680 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.037480439990758896, | |
| "learning_rate": 6.391752577319587e-05, | |
| "epoch": 171.32, | |
| "step": 685 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.028597865253686905, | |
| "learning_rate": 6.3631156930126e-05, | |
| "epoch": 172.64, | |
| "step": 690 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.03485028073191643, | |
| "learning_rate": 6.334478808705613e-05, | |
| "epoch": 173.96, | |
| "step": 695 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.09555519372224808, | |
| "learning_rate": 6.305841924398625e-05, | |
| "epoch": 175.0, | |
| "step": 700 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.038160115480422974, | |
| "learning_rate": 6.277205040091638e-05, | |
| "epoch": 176.32, | |
| "step": 705 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.03564232960343361, | |
| "learning_rate": 6.24856815578465e-05, | |
| "epoch": 177.64, | |
| "step": 710 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.04173216223716736, | |
| "learning_rate": 6.219931271477663e-05, | |
| "epoch": 178.96, | |
| "step": 715 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.09208651632070541, | |
| "learning_rate": 6.191294387170676e-05, | |
| "epoch": 180.0, | |
| "step": 720 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.03402038663625717, | |
| "learning_rate": 6.162657502863689e-05, | |
| "epoch": 181.32, | |
| "step": 725 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.029249833896756172, | |
| "learning_rate": 6.134020618556701e-05, | |
| "epoch": 182.64, | |
| "step": 730 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.04164959862828255, | |
| "learning_rate": 6.105383734249714e-05, | |
| "epoch": 183.96, | |
| "step": 735 | |
| }, | |
| { | |
| "loss": 0.0292, | |
| "grad_norm": 0.10854028165340424, | |
| "learning_rate": 6.076746849942726e-05, | |
| "epoch": 185.0, | |
| "step": 740 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.034494444727897644, | |
| "learning_rate": 6.0481099656357384e-05, | |
| "epoch": 186.32, | |
| "step": 745 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.03181561455130577, | |
| "learning_rate": 6.019473081328752e-05, | |
| "epoch": 187.64, | |
| "step": 750 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.03192061558365822, | |
| "learning_rate": 5.9908361970217644e-05, | |
| "epoch": 188.96, | |
| "step": 755 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.08837488293647766, | |
| "learning_rate": 5.962199312714777e-05, | |
| "epoch": 190.0, | |
| "step": 760 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.030650176107883453, | |
| "learning_rate": 5.93356242840779e-05, | |
| "epoch": 191.32, | |
| "step": 765 | |
| }, | |
| { | |
| "loss": 0.0273, | |
| "grad_norm": 0.03677394241094589, | |
| "learning_rate": 5.904925544100802e-05, | |
| "epoch": 192.64, | |
| "step": 770 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.032607655972242355, | |
| "learning_rate": 5.876288659793815e-05, | |
| "epoch": 193.96, | |
| "step": 775 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.10432080179452896, | |
| "learning_rate": 5.8476517754868276e-05, | |
| "epoch": 195.0, | |
| "step": 780 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.03283718228340149, | |
| "learning_rate": 5.81901489117984e-05, | |
| "epoch": 196.32, | |
| "step": 785 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.029840612784028053, | |
| "learning_rate": 5.790378006872853e-05, | |
| "epoch": 197.64, | |
| "step": 790 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.0333683080971241, | |
| "learning_rate": 5.761741122565865e-05, | |
| "epoch": 198.96, | |
| "step": 795 | |
| }, | |
| { | |
| "loss": 0.0281, | |
| "grad_norm": 0.12805673480033875, | |
| "learning_rate": 5.7331042382588775e-05, | |
| "epoch": 200.0, | |
| "step": 800 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.03851447254419327, | |
| "learning_rate": 5.70446735395189e-05, | |
| "epoch": 201.32, | |
| "step": 805 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.03337559476494789, | |
| "learning_rate": 5.675830469644903e-05, | |
| "epoch": 202.64, | |
| "step": 810 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.03541240841150284, | |
| "learning_rate": 5.6471935853379155e-05, | |
| "epoch": 203.96, | |
| "step": 815 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.10263597220182419, | |
| "learning_rate": 5.618556701030928e-05, | |
| "epoch": 205.0, | |
| "step": 820 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.03297778591513634, | |
| "learning_rate": 5.589919816723941e-05, | |
| "epoch": 206.32, | |
| "step": 825 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.03233632445335388, | |
| "learning_rate": 5.5612829324169534e-05, | |
| "epoch": 207.64, | |
| "step": 830 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.04101260006427765, | |
| "learning_rate": 5.532646048109966e-05, | |
| "epoch": 208.96, | |
| "step": 835 | |
| }, | |
| { | |
| "loss": 0.0287, | |
| "grad_norm": 0.11795070767402649, | |
| "learning_rate": 5.504009163802979e-05, | |
| "epoch": 210.0, | |
| "step": 840 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.03282361850142479, | |
| "learning_rate": 5.4753722794959914e-05, | |
| "epoch": 211.32, | |
| "step": 845 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.030075691640377045, | |
| "learning_rate": 5.4467353951890033e-05, | |
| "epoch": 212.64, | |
| "step": 850 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.032813820987939835, | |
| "learning_rate": 5.418098510882016e-05, | |
| "epoch": 213.96, | |
| "step": 855 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.11462122201919556, | |
| "learning_rate": 5.3894616265750286e-05, | |
| "epoch": 215.0, | |
| "step": 860 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.027200449258089066, | |
| "learning_rate": 5.360824742268041e-05, | |
| "epoch": 216.32, | |
| "step": 865 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.03272516280412674, | |
| "learning_rate": 5.332187857961054e-05, | |
| "epoch": 217.64, | |
| "step": 870 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.036671776324510574, | |
| "learning_rate": 5.3035509736540666e-05, | |
| "epoch": 218.96, | |
| "step": 875 | |
| }, | |
| { | |
| "loss": 0.0294, | |
| "grad_norm": 0.21629348397254944, | |
| "learning_rate": 5.274914089347079e-05, | |
| "epoch": 220.0, | |
| "step": 880 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.03564731404185295, | |
| "learning_rate": 5.246277205040092e-05, | |
| "epoch": 221.32, | |
| "step": 885 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.0319940410554409, | |
| "learning_rate": 5.2176403207331045e-05, | |
| "epoch": 222.64, | |
| "step": 890 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.033614080399274826, | |
| "learning_rate": 5.189003436426118e-05, | |
| "epoch": 223.96, | |
| "step": 895 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.10172655433416367, | |
| "learning_rate": 5.1603665521191305e-05, | |
| "epoch": 225.0, | |
| "step": 900 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.03494977578520775, | |
| "learning_rate": 5.131729667812142e-05, | |
| "epoch": 226.32, | |
| "step": 905 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.029910268262028694, | |
| "learning_rate": 5.1030927835051544e-05, | |
| "epoch": 227.64, | |
| "step": 910 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.02917870134115219, | |
| "learning_rate": 5.074455899198167e-05, | |
| "epoch": 228.96, | |
| "step": 915 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.11739416420459747, | |
| "learning_rate": 5.04581901489118e-05, | |
| "epoch": 230.0, | |
| "step": 920 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.03160247951745987, | |
| "learning_rate": 5.0171821305841924e-05, | |
| "epoch": 231.32, | |
| "step": 925 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.028692517429590225, | |
| "learning_rate": 4.988545246277205e-05, | |
| "epoch": 232.64, | |
| "step": 930 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.03336671367287636, | |
| "learning_rate": 4.9599083619702184e-05, | |
| "epoch": 233.96, | |
| "step": 935 | |
| }, | |
| { | |
| "loss": 0.023, | |
| "grad_norm": 0.09917111694812775, | |
| "learning_rate": 4.931271477663231e-05, | |
| "epoch": 235.0, | |
| "step": 940 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.03005743958055973, | |
| "learning_rate": 4.902634593356243e-05, | |
| "epoch": 236.32, | |
| "step": 945 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.031525760889053345, | |
| "learning_rate": 4.8739977090492556e-05, | |
| "epoch": 237.64, | |
| "step": 950 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.040231119841337204, | |
| "learning_rate": 4.845360824742268e-05, | |
| "epoch": 238.96, | |
| "step": 955 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.12149109691381454, | |
| "learning_rate": 4.816723940435281e-05, | |
| "epoch": 240.0, | |
| "step": 960 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.03200831636786461, | |
| "learning_rate": 4.7880870561282936e-05, | |
| "epoch": 241.32, | |
| "step": 965 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.033598385751247406, | |
| "learning_rate": 4.7594501718213055e-05, | |
| "epoch": 242.64, | |
| "step": 970 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.03309132531285286, | |
| "learning_rate": 4.730813287514318e-05, | |
| "epoch": 243.96, | |
| "step": 975 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.1105155199766159, | |
| "learning_rate": 4.7021764032073315e-05, | |
| "epoch": 245.0, | |
| "step": 980 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.029582051560282707, | |
| "learning_rate": 4.673539518900344e-05, | |
| "epoch": 246.32, | |
| "step": 985 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.02913469821214676, | |
| "learning_rate": 4.644902634593357e-05, | |
| "epoch": 247.64, | |
| "step": 990 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.030919795855879784, | |
| "learning_rate": 4.6162657502863694e-05, | |
| "epoch": 248.96, | |
| "step": 995 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.10500980913639069, | |
| "learning_rate": 4.5876288659793814e-05, | |
| "epoch": 250.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.030042476952075958, | |
| "learning_rate": 4.558991981672394e-05, | |
| "epoch": 251.32, | |
| "step": 1005 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.035833071917295456, | |
| "learning_rate": 4.530355097365407e-05, | |
| "epoch": 252.64, | |
| "step": 1010 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.027148740366101265, | |
| "learning_rate": 4.5017182130584194e-05, | |
| "epoch": 253.96, | |
| "step": 1015 | |
| }, | |
| { | |
| "loss": 0.0227, | |
| "grad_norm": 0.05335044860839844, | |
| "learning_rate": 4.473081328751432e-05, | |
| "epoch": 255.0, | |
| "step": 1020 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.026993224397301674, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "epoch": 256.32, | |
| "step": 1025 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.026406891644001007, | |
| "learning_rate": 4.415807560137457e-05, | |
| "epoch": 257.64, | |
| "step": 1030 | |
| }, | |
| { | |
| "loss": 0.0293, | |
| "grad_norm": 0.02821189910173416, | |
| "learning_rate": 4.38717067583047e-05, | |
| "epoch": 258.96, | |
| "step": 1035 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.1354018896818161, | |
| "learning_rate": 4.3585337915234826e-05, | |
| "epoch": 260.0, | |
| "step": 1040 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.029364751651883125, | |
| "learning_rate": 4.329896907216495e-05, | |
| "epoch": 261.32, | |
| "step": 1045 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.029653150588274002, | |
| "learning_rate": 4.301260022909508e-05, | |
| "epoch": 262.64, | |
| "step": 1050 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.028902199119329453, | |
| "learning_rate": 4.27262313860252e-05, | |
| "epoch": 263.96, | |
| "step": 1055 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.08469967544078827, | |
| "learning_rate": 4.2439862542955325e-05, | |
| "epoch": 265.0, | |
| "step": 1060 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.027035946026444435, | |
| "learning_rate": 4.215349369988545e-05, | |
| "epoch": 266.32, | |
| "step": 1065 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.030749531462788582, | |
| "learning_rate": 4.1867124856815585e-05, | |
| "epoch": 267.64, | |
| "step": 1070 | |
| }, | |
| { | |
| "loss": 0.0228, | |
| "grad_norm": 0.02827923186123371, | |
| "learning_rate": 4.158075601374571e-05, | |
| "epoch": 268.96, | |
| "step": 1075 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.06870191544294357, | |
| "learning_rate": 4.129438717067583e-05, | |
| "epoch": 270.0, | |
| "step": 1080 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.030150286853313446, | |
| "learning_rate": 4.100801832760596e-05, | |
| "epoch": 271.32, | |
| "step": 1085 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.028299298137426376, | |
| "learning_rate": 4.0721649484536084e-05, | |
| "epoch": 272.64, | |
| "step": 1090 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.029368359595537186, | |
| "learning_rate": 4.043528064146621e-05, | |
| "epoch": 273.96, | |
| "step": 1095 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.07552497833967209, | |
| "learning_rate": 4.014891179839634e-05, | |
| "epoch": 275.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.03882739692926407, | |
| "learning_rate": 3.9862542955326463e-05, | |
| "epoch": 276.32, | |
| "step": 1105 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.02948675863444805, | |
| "learning_rate": 3.957617411225659e-05, | |
| "epoch": 277.64, | |
| "step": 1110 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.03421991318464279, | |
| "learning_rate": 3.9289805269186716e-05, | |
| "epoch": 278.96, | |
| "step": 1115 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.15604497492313385, | |
| "learning_rate": 3.900343642611684e-05, | |
| "epoch": 280.0, | |
| "step": 1120 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.031152017414569855, | |
| "learning_rate": 3.871706758304697e-05, | |
| "epoch": 281.32, | |
| "step": 1125 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.026346053928136826, | |
| "learning_rate": 3.8430698739977096e-05, | |
| "epoch": 282.64, | |
| "step": 1130 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.035875216126441956, | |
| "learning_rate": 3.8144329896907216e-05, | |
| "epoch": 283.96, | |
| "step": 1135 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.11470595002174377, | |
| "learning_rate": 3.785796105383734e-05, | |
| "epoch": 285.0, | |
| "step": 1140 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.030024651437997818, | |
| "learning_rate": 3.757159221076747e-05, | |
| "epoch": 286.32, | |
| "step": 1145 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.027808278799057007, | |
| "learning_rate": 3.7285223367697595e-05, | |
| "epoch": 287.64, | |
| "step": 1150 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.026050791144371033, | |
| "learning_rate": 3.699885452462772e-05, | |
| "epoch": 288.96, | |
| "step": 1155 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.07873599231243134, | |
| "learning_rate": 3.671248568155785e-05, | |
| "epoch": 290.0, | |
| "step": 1160 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.025554237887263298, | |
| "learning_rate": 3.6426116838487974e-05, | |
| "epoch": 291.32, | |
| "step": 1165 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.02678663283586502, | |
| "learning_rate": 3.61397479954181e-05, | |
| "epoch": 292.64, | |
| "step": 1170 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.025933323428034782, | |
| "learning_rate": 3.585337915234823e-05, | |
| "epoch": 293.96, | |
| "step": 1175 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.07074404507875443, | |
| "learning_rate": 3.5567010309278354e-05, | |
| "epoch": 295.0, | |
| "step": 1180 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.028462108224630356, | |
| "learning_rate": 3.528064146620848e-05, | |
| "epoch": 296.32, | |
| "step": 1185 | |
| }, | |
| { | |
| "loss": 0.0286, | |
| "grad_norm": 0.03762601688504219, | |
| "learning_rate": 3.49942726231386e-05, | |
| "epoch": 297.64, | |
| "step": 1190 | |
| }, | |
| { | |
| "loss": 0.0227, | |
| "grad_norm": 0.037543535232543945, | |
| "learning_rate": 3.4707903780068726e-05, | |
| "epoch": 298.96, | |
| "step": 1195 | |
| }, | |
| { | |
| "loss": 0.0293, | |
| "grad_norm": 0.12059560418128967, | |
| "learning_rate": 3.442153493699885e-05, | |
| "epoch": 300.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.031975675374269485, | |
| "learning_rate": 3.4135166093928986e-05, | |
| "epoch": 301.32, | |
| "step": 1205 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.0250416100025177, | |
| "learning_rate": 3.384879725085911e-05, | |
| "epoch": 302.64, | |
| "step": 1210 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.027184171602129936, | |
| "learning_rate": 3.356242840778923e-05, | |
| "epoch": 303.96, | |
| "step": 1215 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.10560937970876694, | |
| "learning_rate": 3.327605956471936e-05, | |
| "epoch": 305.0, | |
| "step": 1220 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.028360804542899132, | |
| "learning_rate": 3.2989690721649485e-05, | |
| "epoch": 306.32, | |
| "step": 1225 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.02822004072368145, | |
| "learning_rate": 3.270332187857961e-05, | |
| "epoch": 307.64, | |
| "step": 1230 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.02768874540925026, | |
| "learning_rate": 3.241695303550974e-05, | |
| "epoch": 308.96, | |
| "step": 1235 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.07531046867370605, | |
| "learning_rate": 3.2130584192439865e-05, | |
| "epoch": 310.0, | |
| "step": 1240 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02392926998436451, | |
| "learning_rate": 3.184421534936999e-05, | |
| "epoch": 311.32, | |
| "step": 1245 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.032872848212718964, | |
| "learning_rate": 3.155784650630012e-05, | |
| "epoch": 312.64, | |
| "step": 1250 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.028521280735731125, | |
| "learning_rate": 3.1271477663230244e-05, | |
| "epoch": 313.96, | |
| "step": 1255 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.07936836779117584, | |
| "learning_rate": 3.098510882016037e-05, | |
| "epoch": 315.0, | |
| "step": 1260 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02847984805703163, | |
| "learning_rate": 3.06987399770905e-05, | |
| "epoch": 316.32, | |
| "step": 1265 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.02831309661269188, | |
| "learning_rate": 3.0412371134020617e-05, | |
| "epoch": 317.64, | |
| "step": 1270 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.027881620451807976, | |
| "learning_rate": 3.0126002290950743e-05, | |
| "epoch": 318.96, | |
| "step": 1275 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.09501548111438751, | |
| "learning_rate": 2.983963344788087e-05, | |
| "epoch": 320.0, | |
| "step": 1280 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.02640927955508232, | |
| "learning_rate": 2.9553264604811e-05, | |
| "epoch": 321.32, | |
| "step": 1285 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.031759195029735565, | |
| "learning_rate": 2.9266895761741126e-05, | |
| "epoch": 322.64, | |
| "step": 1290 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.028879089280962944, | |
| "learning_rate": 2.8980526918671253e-05, | |
| "epoch": 323.96, | |
| "step": 1295 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.11872436851263046, | |
| "learning_rate": 2.8694158075601372e-05, | |
| "epoch": 325.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.02618909813463688, | |
| "learning_rate": 2.8407789232531502e-05, | |
| "epoch": 326.32, | |
| "step": 1305 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.030796971172094345, | |
| "learning_rate": 2.812142038946163e-05, | |
| "epoch": 327.64, | |
| "step": 1310 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.027313588187098503, | |
| "learning_rate": 2.7835051546391755e-05, | |
| "epoch": 328.96, | |
| "step": 1315 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.12215082347393036, | |
| "learning_rate": 2.754868270332188e-05, | |
| "epoch": 330.0, | |
| "step": 1320 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.02814553491771221, | |
| "learning_rate": 2.7262313860252005e-05, | |
| "epoch": 331.32, | |
| "step": 1325 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.027494393289089203, | |
| "learning_rate": 2.697594501718213e-05, | |
| "epoch": 332.64, | |
| "step": 1330 | |
| }, | |
| { | |
| "loss": 0.028, | |
| "grad_norm": 0.029484819620847702, | |
| "learning_rate": 2.6689576174112258e-05, | |
| "epoch": 333.96, | |
| "step": 1335 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.07523085922002792, | |
| "learning_rate": 2.6403207331042384e-05, | |
| "epoch": 335.0, | |
| "step": 1340 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02781621366739273, | |
| "learning_rate": 2.611683848797251e-05, | |
| "epoch": 336.32, | |
| "step": 1345 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.03604988008737564, | |
| "learning_rate": 2.5830469644902637e-05, | |
| "epoch": 337.64, | |
| "step": 1350 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.028637420386075974, | |
| "learning_rate": 2.554410080183276e-05, | |
| "epoch": 338.96, | |
| "step": 1355 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.07377194613218307, | |
| "learning_rate": 2.5257731958762887e-05, | |
| "epoch": 340.0, | |
| "step": 1360 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.02640937641263008, | |
| "learning_rate": 2.4971363115693013e-05, | |
| "epoch": 341.32, | |
| "step": 1365 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.035442180931568146, | |
| "learning_rate": 2.468499427262314e-05, | |
| "epoch": 342.64, | |
| "step": 1370 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.028201589360833168, | |
| "learning_rate": 2.4398625429553266e-05, | |
| "epoch": 343.96, | |
| "step": 1375 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.0743686780333519, | |
| "learning_rate": 2.4112256586483393e-05, | |
| "epoch": 345.0, | |
| "step": 1380 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.030458997935056686, | |
| "learning_rate": 2.3825887743413516e-05, | |
| "epoch": 346.32, | |
| "step": 1385 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.03134138882160187, | |
| "learning_rate": 2.3539518900343642e-05, | |
| "epoch": 347.64, | |
| "step": 1390 | |
| }, | |
| { | |
| "loss": 0.0226, | |
| "grad_norm": 0.027610359713435173, | |
| "learning_rate": 2.3253150057273772e-05, | |
| "epoch": 348.96, | |
| "step": 1395 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.08298386633396149, | |
| "learning_rate": 2.2966781214203895e-05, | |
| "epoch": 350.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.02439303509891033, | |
| "learning_rate": 2.268041237113402e-05, | |
| "epoch": 351.32, | |
| "step": 1405 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.030637478455901146, | |
| "learning_rate": 2.2394043528064148e-05, | |
| "epoch": 352.64, | |
| "step": 1410 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.029427411034703255, | |
| "learning_rate": 2.210767468499427e-05, | |
| "epoch": 353.96, | |
| "step": 1415 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.07798276096582413, | |
| "learning_rate": 2.18213058419244e-05, | |
| "epoch": 355.0, | |
| "step": 1420 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.028970472514629364, | |
| "learning_rate": 2.1534936998854528e-05, | |
| "epoch": 356.32, | |
| "step": 1425 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.02799619734287262, | |
| "learning_rate": 2.124856815578465e-05, | |
| "epoch": 357.64, | |
| "step": 1430 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.027115605771541595, | |
| "learning_rate": 2.0962199312714777e-05, | |
| "epoch": 358.96, | |
| "step": 1435 | |
| }, | |
| { | |
| "loss": 0.0224, | |
| "grad_norm": 0.061676882207393646, | |
| "learning_rate": 2.0675830469644904e-05, | |
| "epoch": 360.0, | |
| "step": 1440 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.028663355857133865, | |
| "learning_rate": 2.038946162657503e-05, | |
| "epoch": 361.32, | |
| "step": 1445 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.025531966239213943, | |
| "learning_rate": 2.0103092783505157e-05, | |
| "epoch": 362.64, | |
| "step": 1450 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.02323620207607746, | |
| "learning_rate": 1.981672394043528e-05, | |
| "epoch": 363.96, | |
| "step": 1455 | |
| }, | |
| { | |
| "loss": 0.0221, | |
| "grad_norm": 0.05577632412314415, | |
| "learning_rate": 1.9530355097365406e-05, | |
| "epoch": 365.0, | |
| "step": 1460 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.02796008065342903, | |
| "learning_rate": 1.9243986254295536e-05, | |
| "epoch": 366.32, | |
| "step": 1465 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.024931611493229866, | |
| "learning_rate": 1.895761741122566e-05, | |
| "epoch": 367.64, | |
| "step": 1470 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.0304824598133564, | |
| "learning_rate": 1.8671248568155786e-05, | |
| "epoch": 368.96, | |
| "step": 1475 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.08039630204439163, | |
| "learning_rate": 1.8384879725085912e-05, | |
| "epoch": 370.0, | |
| "step": 1480 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.028660640120506287, | |
| "learning_rate": 1.809851088201604e-05, | |
| "epoch": 371.32, | |
| "step": 1485 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.02648288756608963, | |
| "learning_rate": 1.7812142038946165e-05, | |
| "epoch": 372.64, | |
| "step": 1490 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.02419593371450901, | |
| "learning_rate": 1.7525773195876288e-05, | |
| "epoch": 373.96, | |
| "step": 1495 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.0976887196302414, | |
| "learning_rate": 1.7239404352806415e-05, | |
| "epoch": 375.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.030413702130317688, | |
| "learning_rate": 1.695303550973654e-05, | |
| "epoch": 376.32, | |
| "step": 1505 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.026200460270047188, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "epoch": 377.64, | |
| "step": 1510 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.02892214059829712, | |
| "learning_rate": 1.6380297823596794e-05, | |
| "epoch": 378.96, | |
| "step": 1515 | |
| }, | |
| { | |
| "loss": 0.0225, | |
| "grad_norm": 0.05756858363747597, | |
| "learning_rate": 1.609392898052692e-05, | |
| "epoch": 380.0, | |
| "step": 1520 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.034238461405038834, | |
| "learning_rate": 1.5807560137457044e-05, | |
| "epoch": 381.32, | |
| "step": 1525 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.026337947696447372, | |
| "learning_rate": 1.5521191294387173e-05, | |
| "epoch": 382.64, | |
| "step": 1530 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.026005607098340988, | |
| "learning_rate": 1.5234822451317298e-05, | |
| "epoch": 383.96, | |
| "step": 1535 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.11806559562683105, | |
| "learning_rate": 1.4948453608247423e-05, | |
| "epoch": 385.0, | |
| "step": 1540 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.02969103865325451, | |
| "learning_rate": 1.466208476517755e-05, | |
| "epoch": 386.32, | |
| "step": 1545 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.025757048279047012, | |
| "learning_rate": 1.4375715922107674e-05, | |
| "epoch": 387.64, | |
| "step": 1550 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.028338925912976265, | |
| "learning_rate": 1.40893470790378e-05, | |
| "epoch": 388.96, | |
| "step": 1555 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.11359255015850067, | |
| "learning_rate": 1.3802978235967929e-05, | |
| "epoch": 390.0, | |
| "step": 1560 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.024127336218953133, | |
| "learning_rate": 1.3516609392898052e-05, | |
| "epoch": 391.32, | |
| "step": 1565 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.025410732254385948, | |
| "learning_rate": 1.323024054982818e-05, | |
| "epoch": 392.64, | |
| "step": 1570 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.028184223920106888, | |
| "learning_rate": 1.2943871706758307e-05, | |
| "epoch": 393.96, | |
| "step": 1575 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.09448428452014923, | |
| "learning_rate": 1.2657502863688431e-05, | |
| "epoch": 395.0, | |
| "step": 1580 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.030733933672308922, | |
| "learning_rate": 1.2371134020618558e-05, | |
| "epoch": 396.32, | |
| "step": 1585 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.027529794722795486, | |
| "learning_rate": 1.2084765177548683e-05, | |
| "epoch": 397.64, | |
| "step": 1590 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.03000178188085556, | |
| "learning_rate": 1.1798396334478809e-05, | |
| "epoch": 398.96, | |
| "step": 1595 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.0866531953215599, | |
| "learning_rate": 1.1512027491408934e-05, | |
| "epoch": 400.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.02824021689593792, | |
| "learning_rate": 1.1225658648339062e-05, | |
| "epoch": 401.32, | |
| "step": 1605 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.024802206084132195, | |
| "learning_rate": 1.0939289805269187e-05, | |
| "epoch": 402.64, | |
| "step": 1610 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.03183116391301155, | |
| "learning_rate": 1.0652920962199313e-05, | |
| "epoch": 403.96, | |
| "step": 1615 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.08067861944437027, | |
| "learning_rate": 1.036655211912944e-05, | |
| "epoch": 405.0, | |
| "step": 1620 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02937748283147812, | |
| "learning_rate": 1.0080183276059566e-05, | |
| "epoch": 406.32, | |
| "step": 1625 | |
| }, | |
| { | |
| "loss": 0.0231, | |
| "grad_norm": 0.025826288387179375, | |
| "learning_rate": 9.793814432989691e-06, | |
| "epoch": 407.64, | |
| "step": 1630 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.02986573800444603, | |
| "learning_rate": 9.507445589919818e-06, | |
| "epoch": 408.96, | |
| "step": 1635 | |
| }, | |
| { | |
| "loss": 0.0231, | |
| "grad_norm": 0.07292810082435608, | |
| "learning_rate": 9.221076746849944e-06, | |
| "epoch": 410.0, | |
| "step": 1640 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.023759571835398674, | |
| "learning_rate": 8.934707903780069e-06, | |
| "epoch": 411.32, | |
| "step": 1645 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.025804780423641205, | |
| "learning_rate": 8.648339060710195e-06, | |
| "epoch": 412.64, | |
| "step": 1650 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.02864696830511093, | |
| "learning_rate": 8.36197021764032e-06, | |
| "epoch": 413.96, | |
| "step": 1655 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.09891517460346222, | |
| "learning_rate": 8.075601374570448e-06, | |
| "epoch": 415.0, | |
| "step": 1660 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.023955030366778374, | |
| "learning_rate": 7.789232531500573e-06, | |
| "epoch": 416.32, | |
| "step": 1665 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.03313145041465759, | |
| "learning_rate": 7.502863688430699e-06, | |
| "epoch": 417.64, | |
| "step": 1670 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.025648707523941994, | |
| "learning_rate": 7.216494845360824e-06, | |
| "epoch": 418.96, | |
| "step": 1675 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.0878920704126358, | |
| "learning_rate": 6.930126002290952e-06, | |
| "epoch": 420.0, | |
| "step": 1680 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.025199044495821, | |
| "learning_rate": 6.643757159221077e-06, | |
| "epoch": 421.32, | |
| "step": 1685 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.028068676590919495, | |
| "learning_rate": 6.357388316151203e-06, | |
| "epoch": 422.64, | |
| "step": 1690 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.027017563581466675, | |
| "learning_rate": 6.071019473081329e-06, | |
| "epoch": 423.96, | |
| "step": 1695 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.08914289623498917, | |
| "learning_rate": 5.784650630011455e-06, | |
| "epoch": 425.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.029386265203356743, | |
| "learning_rate": 5.498281786941581e-06, | |
| "epoch": 426.32, | |
| "step": 1705 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.027373168617486954, | |
| "learning_rate": 5.211912943871707e-06, | |
| "epoch": 427.64, | |
| "step": 1710 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.027136381715536118, | |
| "learning_rate": 4.925544100801833e-06, | |
| "epoch": 428.96, | |
| "step": 1715 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.06883777678012848, | |
| "learning_rate": 4.639175257731959e-06, | |
| "epoch": 430.0, | |
| "step": 1720 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.02670559100806713, | |
| "learning_rate": 4.352806414662085e-06, | |
| "epoch": 431.32, | |
| "step": 1725 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.026326850056648254, | |
| "learning_rate": 4.066437571592211e-06, | |
| "epoch": 432.64, | |
| "step": 1730 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.031075820326805115, | |
| "learning_rate": 3.7800687285223365e-06, | |
| "epoch": 433.96, | |
| "step": 1735 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.10043574869632721, | |
| "learning_rate": 3.493699885452463e-06, | |
| "epoch": 435.0, | |
| "step": 1740 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.029350019991397858, | |
| "learning_rate": 3.2073310423825886e-06, | |
| "epoch": 436.32, | |
| "step": 1745 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.026561835780739784, | |
| "learning_rate": 2.920962199312715e-06, | |
| "epoch": 437.64, | |
| "step": 1750 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.02692883089184761, | |
| "learning_rate": 2.6345933562428407e-06, | |
| "epoch": 438.96, | |
| "step": 1755 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.14109838008880615, | |
| "learning_rate": 2.3482245131729668e-06, | |
| "epoch": 440.0, | |
| "step": 1760 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.026362616568803787, | |
| "learning_rate": 2.061855670103093e-06, | |
| "epoch": 441.32, | |
| "step": 1765 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.026044240221381187, | |
| "learning_rate": 1.7754868270332189e-06, | |
| "epoch": 442.64, | |
| "step": 1770 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02880522608757019, | |
| "learning_rate": 1.4891179839633447e-06, | |
| "epoch": 443.96, | |
| "step": 1775 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.07317947596311569, | |
| "learning_rate": 1.202749140893471e-06, | |
| "epoch": 445.0, | |
| "step": 1780 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.0290087778121233, | |
| "learning_rate": 9.163802978235968e-07, | |
| "epoch": 446.32, | |
| "step": 1785 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.02604903280735016, | |
| "learning_rate": 6.300114547537229e-07, | |
| "epoch": 447.64, | |
| "step": 1790 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.029062993824481964, | |
| "learning_rate": 3.436426116838488e-07, | |
| "epoch": 448.96, | |
| "step": 1795 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.10057289898395538, | |
| "learning_rate": 5.72737686139748e-08, | |
| "epoch": 450.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "train_runtime": 30947.3969, | |
| "train_samples_per_second": 1.861, | |
| "train_steps_per_second": 0.058, | |
| "total_flos": 1.5308141101056e+18, | |
| "train_loss": 0.050706694109572306, | |
| "epoch": 450.0, | |
| "step": 1800 | |
| } | |
| ], | |
| "training_args": { | |
| "output_dir": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/outputs/Mu-Math/group_04/checkpoints", | |
| "overwrite_output_dir": false, | |
| "do_train": false, | |
| "do_eval": false, | |
| "do_predict": false, | |
| "eval_strategy": "no", | |
| "prediction_loss_only": false, | |
| "per_device_train_batch_size": 2, | |
| "per_device_eval_batch_size": 8, | |
| "per_gpu_train_batch_size": null, | |
| "per_gpu_eval_batch_size": null, | |
| "gradient_accumulation_steps": 16, | |
| "eval_accumulation_steps": null, | |
| "eval_delay": 0, | |
| "torch_empty_cache_steps": null, | |
| "learning_rate": 0.0001, | |
| "weight_decay": 0.01, | |
| "adam_beta1": 0.9, | |
| "adam_beta2": 0.999, | |
| "adam_epsilon": 1e-08, | |
| "max_grad_norm": 1.0, | |
| "num_train_epochs": 12, | |
| "max_steps": 1800, | |
| "lr_scheduler_type": "linear", | |
| "lr_scheduler_kwargs": {}, | |
| "warmup_ratio": 0.03, | |
| "warmup_steps": 0, | |
| "log_level": "passive", | |
| "log_level_replica": "warning", | |
| "log_on_each_node": true, | |
| "logging_dir": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/logs/Mu-Math/group_04", | |
| "logging_strategy": "steps", | |
| "logging_first_step": true, | |
| "logging_steps": 5, | |
| "logging_nan_inf_filter": true, | |
| "save_strategy": "steps", | |
| "save_steps": 300, | |
| "save_total_limit": 6, | |
| "save_safetensors": true, | |
| "save_on_each_node": false, | |
| "save_only_model": false, | |
| "restore_callback_states_from_checkpoint": false, | |
| "no_cuda": false, | |
| "use_cpu": false, | |
| "use_mps_device": false, | |
| "seed": 42, | |
| "data_seed": null, | |
| "jit_mode_eval": false, | |
| "bf16": true, | |
| "fp16": false, | |
| "fp16_opt_level": "O1", | |
| "half_precision_backend": "auto", | |
| "bf16_full_eval": false, | |
| "fp16_full_eval": false, | |
| "tf32": null, | |
| "local_rank": 0, | |
| "ddp_backend": null, | |
| "tpu_num_cores": null, | |
| "tpu_metrics_debug": false, | |
| "debug": [], | |
| "dataloader_drop_last": false, | |
| "eval_steps": null, | |
| "dataloader_num_workers": 0, | |
| "dataloader_prefetch_factor": null, | |
| "past_index": -1, | |
| "run_name": null, | |
| "disable_tqdm": false, | |
| "remove_unused_columns": true, | |
| "label_names": null, | |
| "load_best_model_at_end": false, | |
| "metric_for_best_model": null, | |
| "greater_is_better": null, | |
| "ignore_data_skip": false, | |
| "fsdp": [], | |
| "fsdp_min_num_params": 0, | |
| "fsdp_config": { | |
| "min_num_params": 0, | |
| "xla": false, | |
| "xla_fsdp_v2": false, | |
| "xla_fsdp_grad_ckpt": false | |
| }, | |
| "fsdp_transformer_layer_cls_to_wrap": null, | |
| "accelerator_config": { | |
| "split_batches": false, | |
| "dispatch_batches": null, | |
| "even_batches": true, | |
| "use_seedable_sampler": true, | |
| "non_blocking": false, | |
| "gradient_accumulation_kwargs": null | |
| }, | |
| "parallelism_config": null, | |
| "deepspeed": null, | |
| "label_smoothing_factor": 0.0, | |
| "optim": "adamw_torch", | |
| "optim_args": null, | |
| "adafactor": false, | |
| "group_by_length": false, | |
| "length_column_name": "length", | |
| "report_to": [], | |
| "project": "huggingface", | |
| "trackio_space_id": "trackio", | |
| "ddp_find_unused_parameters": null, | |
| "ddp_bucket_cap_mb": null, | |
| "ddp_broadcast_buffers": null, | |
| "dataloader_pin_memory": true, | |
| "dataloader_persistent_workers": false, | |
| "skip_memory_metrics": true, | |
| "use_legacy_prediction_loop": false, | |
| "push_to_hub": false, | |
| "resume_from_checkpoint": null, | |
| "hub_model_id": null, | |
| "hub_strategy": "every_save", | |
| "hub_token": "<HUB_TOKEN>", | |
| "hub_private_repo": null, | |
| "hub_always_push": false, | |
| "hub_revision": null, | |
| "gradient_checkpointing": true, | |
| "gradient_checkpointing_kwargs": null, | |
| "include_inputs_for_metrics": false, | |
| "include_for_metrics": [], | |
| "eval_do_concat_batches": true, | |
| "fp16_backend": "auto", | |
| "push_to_hub_model_id": null, | |
| "push_to_hub_organization": null, | |
| "push_to_hub_token": "<PUSH_TO_HUB_TOKEN>", | |
| "mp_parameters": "", | |
| "auto_find_batch_size": false, | |
| "full_determinism": false, | |
| "torchdynamo": null, | |
| "ray_scope": "last", | |
| "ddp_timeout": 1800, | |
| "torch_compile": false, | |
| "torch_compile_backend": null, | |
| "torch_compile_mode": null, | |
| "include_tokens_per_second": false, | |
| "include_num_input_tokens_seen": "no", | |
| "neftune_noise_alpha": null, | |
| "optim_target_modules": null, | |
| "batch_eval_metrics": false, | |
| "eval_on_start": false, | |
| "use_liger_kernel": false, | |
| "liger_kernel_config": null, | |
| "eval_use_gather_object": false, | |
| "average_tokens_across_devices": true | |
| }, | |
| "lora_config": { | |
| "r": 64, | |
| "alpha": 128, | |
| "dropout": 0.05, | |
| "target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ] | |
| }, | |
| "effective_batch_size": 32, | |
| "world_size": 1, | |
| "git_commit": "" | |
| } |