| { | |
| "dataset_name": "Mu-Math", | |
| "group_index": 0, | |
| "prompt_group_file": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/prompt_groups/Mu-Math/group_00.json", | |
| "output_dir": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/outputs/Mu-Math/group_00", | |
| "checkpoint_root": "/hkfs/work/workspace/scratch/tum_fmp0582-dndworkspace/自己训练lora/train_lora/outputs/Mu-Math/group_00/checkpoints", | |
| "generated_at": "2025-11-06T12:11:32Z", | |
| "train_loss": 0.04942113804734415, | |
| "metrics": { | |
| "train_runtime": 30905.7388, | |
| "train_samples_per_second": 1.864, | |
| "train_steps_per_second": 0.058, | |
| "total_flos": 1.5308141101056e+18, | |
| "train_loss": 0.04942113804734415, | |
| "epoch": 450.0 | |
| }, | |
| "trainer_state": [ | |
| { | |
| "loss": 1.6173, | |
| "grad_norm": 9.742866516113281, | |
| "learning_rate": 0.0, | |
| "epoch": 0.32, | |
| "step": 1 | |
| }, | |
| { | |
| "loss": 1.8185, | |
| "grad_norm": 6.767038345336914, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "epoch": 1.32, | |
| "step": 5 | |
| }, | |
| { | |
| "loss": 1.4859, | |
| "grad_norm": 3.1260478496551514, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "epoch": 2.64, | |
| "step": 10 | |
| }, | |
| { | |
| "loss": 1.1518, | |
| "grad_norm": 1.7969225645065308, | |
| "learning_rate": 2.5925925925925925e-05, | |
| "epoch": 3.96, | |
| "step": 15 | |
| }, | |
| { | |
| "loss": 0.9266, | |
| "grad_norm": 1.1876064538955688, | |
| "learning_rate": 3.518518518518519e-05, | |
| "epoch": 5.0, | |
| "step": 20 | |
| }, | |
| { | |
| "loss": 0.7755, | |
| "grad_norm": 0.5301205515861511, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "epoch": 6.32, | |
| "step": 25 | |
| }, | |
| { | |
| "loss": 0.653, | |
| "grad_norm": 0.4444913864135742, | |
| "learning_rate": 5.370370370370371e-05, | |
| "epoch": 7.64, | |
| "step": 30 | |
| }, | |
| { | |
| "loss": 0.5831, | |
| "grad_norm": 0.4524156153202057, | |
| "learning_rate": 6.296296296296296e-05, | |
| "epoch": 8.96, | |
| "step": 35 | |
| }, | |
| { | |
| "loss": 0.4939, | |
| "grad_norm": 1.3137646913528442, | |
| "learning_rate": 7.222222222222222e-05, | |
| "epoch": 10.0, | |
| "step": 40 | |
| }, | |
| { | |
| "loss": 0.3608, | |
| "grad_norm": 0.5690760612487793, | |
| "learning_rate": 8.148148148148148e-05, | |
| "epoch": 11.32, | |
| "step": 45 | |
| }, | |
| { | |
| "loss": 0.2369, | |
| "grad_norm": 0.8014142513275146, | |
| "learning_rate": 9.074074074074075e-05, | |
| "epoch": 12.64, | |
| "step": 50 | |
| }, | |
| { | |
| "loss": 0.1779, | |
| "grad_norm": 0.8277022242546082, | |
| "learning_rate": 0.0001, | |
| "epoch": 13.96, | |
| "step": 55 | |
| }, | |
| { | |
| "loss": 0.0976, | |
| "grad_norm": 1.2155102491378784, | |
| "learning_rate": 9.971363115693013e-05, | |
| "epoch": 15.0, | |
| "step": 60 | |
| }, | |
| { | |
| "loss": 0.0712, | |
| "grad_norm": 0.5272746086120605, | |
| "learning_rate": 9.942726231386026e-05, | |
| "epoch": 16.32, | |
| "step": 65 | |
| }, | |
| { | |
| "loss": 0.0563, | |
| "grad_norm": 0.3631560802459717, | |
| "learning_rate": 9.914089347079038e-05, | |
| "epoch": 17.64, | |
| "step": 70 | |
| }, | |
| { | |
| "loss": 0.0494, | |
| "grad_norm": 0.4824874699115753, | |
| "learning_rate": 9.885452462772051e-05, | |
| "epoch": 18.96, | |
| "step": 75 | |
| }, | |
| { | |
| "loss": 0.0384, | |
| "grad_norm": 0.38946136832237244, | |
| "learning_rate": 9.856815578465064e-05, | |
| "epoch": 20.0, | |
| "step": 80 | |
| }, | |
| { | |
| "loss": 0.0367, | |
| "grad_norm": 0.5088748931884766, | |
| "learning_rate": 9.828178694158075e-05, | |
| "epoch": 21.32, | |
| "step": 85 | |
| }, | |
| { | |
| "loss": 0.0365, | |
| "grad_norm": 0.4311531186103821, | |
| "learning_rate": 9.799541809851088e-05, | |
| "epoch": 22.64, | |
| "step": 90 | |
| }, | |
| { | |
| "loss": 0.0374, | |
| "grad_norm": 0.5566326975822449, | |
| "learning_rate": 9.7709049255441e-05, | |
| "epoch": 23.96, | |
| "step": 95 | |
| }, | |
| { | |
| "loss": 0.0421, | |
| "grad_norm": 1.2089989185333252, | |
| "learning_rate": 9.742268041237114e-05, | |
| "epoch": 25.0, | |
| "step": 100 | |
| }, | |
| { | |
| "loss": 0.0337, | |
| "grad_norm": 0.17220769822597504, | |
| "learning_rate": 9.713631156930127e-05, | |
| "epoch": 26.32, | |
| "step": 105 | |
| }, | |
| { | |
| "loss": 0.0318, | |
| "grad_norm": 0.1414509117603302, | |
| "learning_rate": 9.68499427262314e-05, | |
| "epoch": 27.64, | |
| "step": 110 | |
| }, | |
| { | |
| "loss": 0.0324, | |
| "grad_norm": 0.202336847782135, | |
| "learning_rate": 9.656357388316152e-05, | |
| "epoch": 28.96, | |
| "step": 115 | |
| }, | |
| { | |
| "loss": 0.0356, | |
| "grad_norm": 0.6799734234809875, | |
| "learning_rate": 9.627720504009165e-05, | |
| "epoch": 30.0, | |
| "step": 120 | |
| }, | |
| { | |
| "loss": 0.0324, | |
| "grad_norm": 0.12131154537200928, | |
| "learning_rate": 9.599083619702178e-05, | |
| "epoch": 31.32, | |
| "step": 125 | |
| }, | |
| { | |
| "loss": 0.0326, | |
| "grad_norm": 0.1356758326292038, | |
| "learning_rate": 9.57044673539519e-05, | |
| "epoch": 32.64, | |
| "step": 130 | |
| }, | |
| { | |
| "loss": 0.0312, | |
| "grad_norm": 0.15141494572162628, | |
| "learning_rate": 9.541809851088203e-05, | |
| "epoch": 33.96, | |
| "step": 135 | |
| }, | |
| { | |
| "loss": 0.0341, | |
| "grad_norm": 0.3755043148994446, | |
| "learning_rate": 9.513172966781214e-05, | |
| "epoch": 35.0, | |
| "step": 140 | |
| }, | |
| { | |
| "loss": 0.0302, | |
| "grad_norm": 0.11444094032049179, | |
| "learning_rate": 9.484536082474227e-05, | |
| "epoch": 36.32, | |
| "step": 145 | |
| }, | |
| { | |
| "loss": 0.0343, | |
| "grad_norm": 0.0892108827829361, | |
| "learning_rate": 9.45589919816724e-05, | |
| "epoch": 37.64, | |
| "step": 150 | |
| }, | |
| { | |
| "loss": 0.0305, | |
| "grad_norm": 0.09671330451965332, | |
| "learning_rate": 9.427262313860252e-05, | |
| "epoch": 38.96, | |
| "step": 155 | |
| }, | |
| { | |
| "loss": 0.0305, | |
| "grad_norm": 0.258009135723114, | |
| "learning_rate": 9.398625429553265e-05, | |
| "epoch": 40.0, | |
| "step": 160 | |
| }, | |
| { | |
| "loss": 0.0297, | |
| "grad_norm": 0.0754060447216034, | |
| "learning_rate": 9.369988545246277e-05, | |
| "epoch": 41.32, | |
| "step": 165 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.0693526491522789, | |
| "learning_rate": 9.34135166093929e-05, | |
| "epoch": 42.64, | |
| "step": 170 | |
| }, | |
| { | |
| "loss": 0.0296, | |
| "grad_norm": 0.08370771259069443, | |
| "learning_rate": 9.312714776632303e-05, | |
| "epoch": 43.96, | |
| "step": 175 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.18955636024475098, | |
| "learning_rate": 9.284077892325315e-05, | |
| "epoch": 45.0, | |
| "step": 180 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.07864312082529068, | |
| "learning_rate": 9.255441008018328e-05, | |
| "epoch": 46.32, | |
| "step": 185 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.06608983129262924, | |
| "learning_rate": 9.22680412371134e-05, | |
| "epoch": 47.64, | |
| "step": 190 | |
| }, | |
| { | |
| "loss": 0.0329, | |
| "grad_norm": 0.07156423479318619, | |
| "learning_rate": 9.198167239404353e-05, | |
| "epoch": 48.96, | |
| "step": 195 | |
| }, | |
| { | |
| "loss": 0.03, | |
| "grad_norm": 0.18776871263980865, | |
| "learning_rate": 9.169530355097366e-05, | |
| "epoch": 50.0, | |
| "step": 200 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.08733785152435303, | |
| "learning_rate": 9.140893470790379e-05, | |
| "epoch": 51.32, | |
| "step": 205 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.05940118059515953, | |
| "learning_rate": 9.112256586483391e-05, | |
| "epoch": 52.64, | |
| "step": 210 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.06646186858415604, | |
| "learning_rate": 9.083619702176404e-05, | |
| "epoch": 53.96, | |
| "step": 215 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.1825263947248459, | |
| "learning_rate": 9.054982817869416e-05, | |
| "epoch": 55.0, | |
| "step": 220 | |
| }, | |
| { | |
| "loss": 0.0291, | |
| "grad_norm": 0.062162525951862335, | |
| "learning_rate": 9.026345933562429e-05, | |
| "epoch": 56.32, | |
| "step": 225 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.0541149340569973, | |
| "learning_rate": 8.997709049255442e-05, | |
| "epoch": 57.64, | |
| "step": 230 | |
| }, | |
| { | |
| "loss": 0.0281, | |
| "grad_norm": 0.05463428795337677, | |
| "learning_rate": 8.969072164948454e-05, | |
| "epoch": 58.96, | |
| "step": 235 | |
| }, | |
| { | |
| "loss": 0.0284, | |
| "grad_norm": 0.17292532324790955, | |
| "learning_rate": 8.940435280641467e-05, | |
| "epoch": 60.0, | |
| "step": 240 | |
| }, | |
| { | |
| "loss": 0.0302, | |
| "grad_norm": 0.06024179607629776, | |
| "learning_rate": 8.91179839633448e-05, | |
| "epoch": 61.32, | |
| "step": 245 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.05445247143507004, | |
| "learning_rate": 8.883161512027491e-05, | |
| "epoch": 62.64, | |
| "step": 250 | |
| }, | |
| { | |
| "loss": 0.028, | |
| "grad_norm": 0.050848882645368576, | |
| "learning_rate": 8.854524627720504e-05, | |
| "epoch": 63.96, | |
| "step": 255 | |
| }, | |
| { | |
| "loss": 0.0295, | |
| "grad_norm": 0.17552168667316437, | |
| "learning_rate": 8.825887743413516e-05, | |
| "epoch": 65.0, | |
| "step": 260 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.049610335379838943, | |
| "learning_rate": 8.797250859106529e-05, | |
| "epoch": 66.32, | |
| "step": 265 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.0633014366030693, | |
| "learning_rate": 8.768613974799542e-05, | |
| "epoch": 67.64, | |
| "step": 270 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.05042235553264618, | |
| "learning_rate": 8.739977090492554e-05, | |
| "epoch": 68.96, | |
| "step": 275 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.10001793503761292, | |
| "learning_rate": 8.711340206185567e-05, | |
| "epoch": 70.0, | |
| "step": 280 | |
| }, | |
| { | |
| "loss": 0.0295, | |
| "grad_norm": 0.04931037500500679, | |
| "learning_rate": 8.682703321878581e-05, | |
| "epoch": 71.32, | |
| "step": 285 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.049116216599941254, | |
| "learning_rate": 8.654066437571594e-05, | |
| "epoch": 72.64, | |
| "step": 290 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.060398977249860764, | |
| "learning_rate": 8.625429553264606e-05, | |
| "epoch": 73.96, | |
| "step": 295 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.13518516719341278, | |
| "learning_rate": 8.596792668957619e-05, | |
| "epoch": 75.0, | |
| "step": 300 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.054935161024332047, | |
| "learning_rate": 8.56815578465063e-05, | |
| "epoch": 76.32, | |
| "step": 305 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.053364258259534836, | |
| "learning_rate": 8.539518900343643e-05, | |
| "epoch": 77.64, | |
| "step": 310 | |
| }, | |
| { | |
| "loss": 0.0296, | |
| "grad_norm": 0.062308233231306076, | |
| "learning_rate": 8.510882016036655e-05, | |
| "epoch": 78.96, | |
| "step": 315 | |
| }, | |
| { | |
| "loss": 0.0278, | |
| "grad_norm": 0.11498274654150009, | |
| "learning_rate": 8.482245131729668e-05, | |
| "epoch": 80.0, | |
| "step": 320 | |
| }, | |
| { | |
| "loss": 0.0287, | |
| "grad_norm": 0.04860518500208855, | |
| "learning_rate": 8.453608247422681e-05, | |
| "epoch": 81.32, | |
| "step": 325 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.047843027859926224, | |
| "learning_rate": 8.424971363115693e-05, | |
| "epoch": 82.64, | |
| "step": 330 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.052064020186662674, | |
| "learning_rate": 8.396334478808706e-05, | |
| "epoch": 83.96, | |
| "step": 335 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.08848381042480469, | |
| "learning_rate": 8.367697594501719e-05, | |
| "epoch": 85.0, | |
| "step": 340 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.047083113342523575, | |
| "learning_rate": 8.339060710194731e-05, | |
| "epoch": 86.32, | |
| "step": 345 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.04805876687169075, | |
| "learning_rate": 8.310423825887744e-05, | |
| "epoch": 87.64, | |
| "step": 350 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.05759628862142563, | |
| "learning_rate": 8.281786941580757e-05, | |
| "epoch": 88.96, | |
| "step": 355 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.1436251699924469, | |
| "learning_rate": 8.253150057273768e-05, | |
| "epoch": 90.0, | |
| "step": 360 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.05638458579778671, | |
| "learning_rate": 8.224513172966782e-05, | |
| "epoch": 91.32, | |
| "step": 365 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.062409110367298126, | |
| "learning_rate": 8.195876288659795e-05, | |
| "epoch": 92.64, | |
| "step": 370 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.04246116802096367, | |
| "learning_rate": 8.167239404352807e-05, | |
| "epoch": 93.96, | |
| "step": 375 | |
| }, | |
| { | |
| "loss": 0.0232, | |
| "grad_norm": 0.15006567537784576, | |
| "learning_rate": 8.13860252004582e-05, | |
| "epoch": 95.0, | |
| "step": 380 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.04267548397183418, | |
| "learning_rate": 8.109965635738833e-05, | |
| "epoch": 96.32, | |
| "step": 385 | |
| }, | |
| { | |
| "loss": 0.0283, | |
| "grad_norm": 0.05433656647801399, | |
| "learning_rate": 8.081328751431845e-05, | |
| "epoch": 97.64, | |
| "step": 390 | |
| }, | |
| { | |
| "loss": 0.0262, | |
| "grad_norm": 0.04750007763504982, | |
| "learning_rate": 8.052691867124858e-05, | |
| "epoch": 98.96, | |
| "step": 395 | |
| }, | |
| { | |
| "loss": 0.0305, | |
| "grad_norm": 0.24594779312610626, | |
| "learning_rate": 8.02405498281787e-05, | |
| "epoch": 100.0, | |
| "step": 400 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.05753310024738312, | |
| "learning_rate": 7.995418098510883e-05, | |
| "epoch": 101.32, | |
| "step": 405 | |
| }, | |
| { | |
| "loss": 0.0289, | |
| "grad_norm": 0.04496898129582405, | |
| "learning_rate": 7.966781214203894e-05, | |
| "epoch": 102.64, | |
| "step": 410 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.04672890901565552, | |
| "learning_rate": 7.938144329896907e-05, | |
| "epoch": 103.96, | |
| "step": 415 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.13345938920974731, | |
| "learning_rate": 7.90950744558992e-05, | |
| "epoch": 105.0, | |
| "step": 420 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.04086393490433693, | |
| "learning_rate": 7.880870561282932e-05, | |
| "epoch": 106.32, | |
| "step": 425 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.044156208634376526, | |
| "learning_rate": 7.852233676975945e-05, | |
| "epoch": 107.64, | |
| "step": 430 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.04240730404853821, | |
| "learning_rate": 7.823596792668958e-05, | |
| "epoch": 108.96, | |
| "step": 435 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.10477716475725174, | |
| "learning_rate": 7.79495990836197e-05, | |
| "epoch": 110.0, | |
| "step": 440 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.04656030237674713, | |
| "learning_rate": 7.766323024054983e-05, | |
| "epoch": 111.32, | |
| "step": 445 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.04922478273510933, | |
| "learning_rate": 7.737686139747996e-05, | |
| "epoch": 112.64, | |
| "step": 450 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.04471348226070404, | |
| "learning_rate": 7.709049255441008e-05, | |
| "epoch": 113.96, | |
| "step": 455 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.16744805872440338, | |
| "learning_rate": 7.680412371134021e-05, | |
| "epoch": 115.0, | |
| "step": 460 | |
| }, | |
| { | |
| "loss": 0.0281, | |
| "grad_norm": 0.04978503659367561, | |
| "learning_rate": 7.651775486827034e-05, | |
| "epoch": 116.32, | |
| "step": 465 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.041166212409734726, | |
| "learning_rate": 7.623138602520046e-05, | |
| "epoch": 117.64, | |
| "step": 470 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.04655733332037926, | |
| "learning_rate": 7.594501718213059e-05, | |
| "epoch": 118.96, | |
| "step": 475 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.11510545760393143, | |
| "learning_rate": 7.565864833906071e-05, | |
| "epoch": 120.0, | |
| "step": 480 | |
| }, | |
| { | |
| "loss": 0.0281, | |
| "grad_norm": 0.039519622921943665, | |
| "learning_rate": 7.537227949599084e-05, | |
| "epoch": 121.32, | |
| "step": 485 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.037291865795850754, | |
| "learning_rate": 7.508591065292097e-05, | |
| "epoch": 122.64, | |
| "step": 490 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.055157747119665146, | |
| "learning_rate": 7.47995418098511e-05, | |
| "epoch": 123.96, | |
| "step": 495 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.09586930274963379, | |
| "learning_rate": 7.451317296678122e-05, | |
| "epoch": 125.0, | |
| "step": 500 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.04437963292002678, | |
| "learning_rate": 7.422680412371135e-05, | |
| "epoch": 126.32, | |
| "step": 505 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.03398558869957924, | |
| "learning_rate": 7.394043528064147e-05, | |
| "epoch": 127.64, | |
| "step": 510 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.039411090314388275, | |
| "learning_rate": 7.36540664375716e-05, | |
| "epoch": 128.96, | |
| "step": 515 | |
| }, | |
| { | |
| "loss": 0.0288, | |
| "grad_norm": 0.22975099086761475, | |
| "learning_rate": 7.336769759450171e-05, | |
| "epoch": 130.0, | |
| "step": 520 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.043207693845033646, | |
| "learning_rate": 7.308132875143184e-05, | |
| "epoch": 131.32, | |
| "step": 525 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.03736112266778946, | |
| "learning_rate": 7.279495990836197e-05, | |
| "epoch": 132.64, | |
| "step": 530 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.033578503876924515, | |
| "learning_rate": 7.250859106529209e-05, | |
| "epoch": 133.96, | |
| "step": 535 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.16706548631191254, | |
| "learning_rate": 7.222222222222222e-05, | |
| "epoch": 135.0, | |
| "step": 540 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.036494217813014984, | |
| "learning_rate": 7.193585337915235e-05, | |
| "epoch": 136.32, | |
| "step": 545 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.03684375807642937, | |
| "learning_rate": 7.164948453608247e-05, | |
| "epoch": 137.64, | |
| "step": 550 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.04394157975912094, | |
| "learning_rate": 7.136311569301261e-05, | |
| "epoch": 138.96, | |
| "step": 555 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.1644020974636078, | |
| "learning_rate": 7.107674684994274e-05, | |
| "epoch": 140.0, | |
| "step": 560 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.0372483991086483, | |
| "learning_rate": 7.079037800687286e-05, | |
| "epoch": 141.32, | |
| "step": 565 | |
| }, | |
| { | |
| "loss": 0.0279, | |
| "grad_norm": 0.04550137743353844, | |
| "learning_rate": 7.050400916380299e-05, | |
| "epoch": 142.64, | |
| "step": 570 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.04950069263577461, | |
| "learning_rate": 7.02176403207331e-05, | |
| "epoch": 143.96, | |
| "step": 575 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.10132873803377151, | |
| "learning_rate": 6.993127147766323e-05, | |
| "epoch": 145.0, | |
| "step": 580 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.03352991119027138, | |
| "learning_rate": 6.964490263459336e-05, | |
| "epoch": 146.32, | |
| "step": 585 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.031004367396235466, | |
| "learning_rate": 6.935853379152348e-05, | |
| "epoch": 147.64, | |
| "step": 590 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.03967953845858574, | |
| "learning_rate": 6.907216494845361e-05, | |
| "epoch": 148.96, | |
| "step": 595 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.09593302011489868, | |
| "learning_rate": 6.878579610538374e-05, | |
| "epoch": 150.0, | |
| "step": 600 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.03218228742480278, | |
| "learning_rate": 6.849942726231386e-05, | |
| "epoch": 151.32, | |
| "step": 605 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.034557685256004333, | |
| "learning_rate": 6.821305841924399e-05, | |
| "epoch": 152.64, | |
| "step": 610 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.03427565097808838, | |
| "learning_rate": 6.792668957617412e-05, | |
| "epoch": 153.96, | |
| "step": 615 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.15740765631198883, | |
| "learning_rate": 6.764032073310424e-05, | |
| "epoch": 155.0, | |
| "step": 620 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.03638317063450813, | |
| "learning_rate": 6.735395189003437e-05, | |
| "epoch": 156.32, | |
| "step": 625 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.03340018168091774, | |
| "learning_rate": 6.706758304696448e-05, | |
| "epoch": 157.64, | |
| "step": 630 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.04529392719268799, | |
| "learning_rate": 6.678121420389462e-05, | |
| "epoch": 158.96, | |
| "step": 635 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.10604052245616913, | |
| "learning_rate": 6.649484536082475e-05, | |
| "epoch": 160.0, | |
| "step": 640 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.03510366007685661, | |
| "learning_rate": 6.620847651775487e-05, | |
| "epoch": 161.32, | |
| "step": 645 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.041160210967063904, | |
| "learning_rate": 6.5922107674685e-05, | |
| "epoch": 162.64, | |
| "step": 650 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.03276336193084717, | |
| "learning_rate": 6.563573883161513e-05, | |
| "epoch": 163.96, | |
| "step": 655 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.07443840801715851, | |
| "learning_rate": 6.534936998854525e-05, | |
| "epoch": 165.0, | |
| "step": 660 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.03302273154258728, | |
| "learning_rate": 6.506300114547538e-05, | |
| "epoch": 166.32, | |
| "step": 665 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.03181939944624901, | |
| "learning_rate": 6.477663230240551e-05, | |
| "epoch": 167.64, | |
| "step": 670 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.03427578881382942, | |
| "learning_rate": 6.449026345933563e-05, | |
| "epoch": 168.96, | |
| "step": 675 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.09033139050006866, | |
| "learning_rate": 6.420389461626576e-05, | |
| "epoch": 170.0, | |
| "step": 680 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.027491431683301926, | |
| "learning_rate": 6.391752577319587e-05, | |
| "epoch": 171.32, | |
| "step": 685 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.028740208595991135, | |
| "learning_rate": 6.3631156930126e-05, | |
| "epoch": 172.64, | |
| "step": 690 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.030295995995402336, | |
| "learning_rate": 6.334478808705613e-05, | |
| "epoch": 173.96, | |
| "step": 695 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.06706272065639496, | |
| "learning_rate": 6.305841924398625e-05, | |
| "epoch": 175.0, | |
| "step": 700 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.03938639909029007, | |
| "learning_rate": 6.277205040091638e-05, | |
| "epoch": 176.32, | |
| "step": 705 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.058989059180021286, | |
| "learning_rate": 6.24856815578465e-05, | |
| "epoch": 177.64, | |
| "step": 710 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.03354703262448311, | |
| "learning_rate": 6.219931271477663e-05, | |
| "epoch": 178.96, | |
| "step": 715 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.11492135375738144, | |
| "learning_rate": 6.191294387170676e-05, | |
| "epoch": 180.0, | |
| "step": 720 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.039485782384872437, | |
| "learning_rate": 6.162657502863689e-05, | |
| "epoch": 181.32, | |
| "step": 725 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.030940620228648186, | |
| "learning_rate": 6.134020618556701e-05, | |
| "epoch": 182.64, | |
| "step": 730 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.03291434049606323, | |
| "learning_rate": 6.105383734249714e-05, | |
| "epoch": 183.96, | |
| "step": 735 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.07285232841968536, | |
| "learning_rate": 6.076746849942726e-05, | |
| "epoch": 185.0, | |
| "step": 740 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.03124931827187538, | |
| "learning_rate": 6.0481099656357384e-05, | |
| "epoch": 186.32, | |
| "step": 745 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.03053557686507702, | |
| "learning_rate": 6.019473081328752e-05, | |
| "epoch": 187.64, | |
| "step": 750 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.032770149409770966, | |
| "learning_rate": 5.9908361970217644e-05, | |
| "epoch": 188.96, | |
| "step": 755 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.10088206827640533, | |
| "learning_rate": 5.962199312714777e-05, | |
| "epoch": 190.0, | |
| "step": 760 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.031048694625496864, | |
| "learning_rate": 5.93356242840779e-05, | |
| "epoch": 191.32, | |
| "step": 765 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.03410292789340019, | |
| "learning_rate": 5.904925544100802e-05, | |
| "epoch": 192.64, | |
| "step": 770 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.03622349351644516, | |
| "learning_rate": 5.876288659793815e-05, | |
| "epoch": 193.96, | |
| "step": 775 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.14193035662174225, | |
| "learning_rate": 5.8476517754868276e-05, | |
| "epoch": 195.0, | |
| "step": 780 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.042379122227430344, | |
| "learning_rate": 5.81901489117984e-05, | |
| "epoch": 196.32, | |
| "step": 785 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02714547887444496, | |
| "learning_rate": 5.790378006872853e-05, | |
| "epoch": 197.64, | |
| "step": 790 | |
| }, | |
| { | |
| "loss": 0.0268, | |
| "grad_norm": 0.025107022374868393, | |
| "learning_rate": 5.761741122565865e-05, | |
| "epoch": 198.96, | |
| "step": 795 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.09233396500349045, | |
| "learning_rate": 5.7331042382588775e-05, | |
| "epoch": 200.0, | |
| "step": 800 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.03396382927894592, | |
| "learning_rate": 5.70446735395189e-05, | |
| "epoch": 201.32, | |
| "step": 805 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.026768192648887634, | |
| "learning_rate": 5.675830469644903e-05, | |
| "epoch": 202.64, | |
| "step": 810 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.03228536620736122, | |
| "learning_rate": 5.6471935853379155e-05, | |
| "epoch": 203.96, | |
| "step": 815 | |
| }, | |
| { | |
| "loss": 0.0232, | |
| "grad_norm": 0.09274106472730637, | |
| "learning_rate": 5.618556701030928e-05, | |
| "epoch": 205.0, | |
| "step": 820 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.029722807928919792, | |
| "learning_rate": 5.589919816723941e-05, | |
| "epoch": 206.32, | |
| "step": 825 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.03271517902612686, | |
| "learning_rate": 5.5612829324169534e-05, | |
| "epoch": 207.64, | |
| "step": 830 | |
| }, | |
| { | |
| "loss": 0.0275, | |
| "grad_norm": 0.0331767238676548, | |
| "learning_rate": 5.532646048109966e-05, | |
| "epoch": 208.96, | |
| "step": 835 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.10759365558624268, | |
| "learning_rate": 5.504009163802979e-05, | |
| "epoch": 210.0, | |
| "step": 840 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.03202605992555618, | |
| "learning_rate": 5.4753722794959914e-05, | |
| "epoch": 211.32, | |
| "step": 845 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.030170518904924393, | |
| "learning_rate": 5.4467353951890033e-05, | |
| "epoch": 212.64, | |
| "step": 850 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.03028750605881214, | |
| "learning_rate": 5.418098510882016e-05, | |
| "epoch": 213.96, | |
| "step": 855 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.0875907689332962, | |
| "learning_rate": 5.3894616265750286e-05, | |
| "epoch": 215.0, | |
| "step": 860 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.032579585909843445, | |
| "learning_rate": 5.360824742268041e-05, | |
| "epoch": 216.32, | |
| "step": 865 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.03846624866127968, | |
| "learning_rate": 5.332187857961054e-05, | |
| "epoch": 217.64, | |
| "step": 870 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.032183196395635605, | |
| "learning_rate": 5.3035509736540666e-05, | |
| "epoch": 218.96, | |
| "step": 875 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.07549957185983658, | |
| "learning_rate": 5.274914089347079e-05, | |
| "epoch": 220.0, | |
| "step": 880 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.03417748212814331, | |
| "learning_rate": 5.246277205040092e-05, | |
| "epoch": 221.32, | |
| "step": 885 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.026088347658514977, | |
| "learning_rate": 5.2176403207331045e-05, | |
| "epoch": 222.64, | |
| "step": 890 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.03899826109409332, | |
| "learning_rate": 5.189003436426118e-05, | |
| "epoch": 223.96, | |
| "step": 895 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.10932710766792297, | |
| "learning_rate": 5.1603665521191305e-05, | |
| "epoch": 225.0, | |
| "step": 900 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.026801714673638344, | |
| "learning_rate": 5.131729667812142e-05, | |
| "epoch": 226.32, | |
| "step": 905 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.032709065824747086, | |
| "learning_rate": 5.1030927835051544e-05, | |
| "epoch": 227.64, | |
| "step": 910 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.034237563610076904, | |
| "learning_rate": 5.074455899198167e-05, | |
| "epoch": 228.96, | |
| "step": 915 | |
| }, | |
| { | |
| "loss": 0.0298, | |
| "grad_norm": 0.12856259942054749, | |
| "learning_rate": 5.04581901489118e-05, | |
| "epoch": 230.0, | |
| "step": 920 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.03215862438082695, | |
| "learning_rate": 5.0171821305841924e-05, | |
| "epoch": 231.32, | |
| "step": 925 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.028518445789813995, | |
| "learning_rate": 4.988545246277205e-05, | |
| "epoch": 232.64, | |
| "step": 930 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.033663809299468994, | |
| "learning_rate": 4.9599083619702184e-05, | |
| "epoch": 233.96, | |
| "step": 935 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.07673133164644241, | |
| "learning_rate": 4.931271477663231e-05, | |
| "epoch": 235.0, | |
| "step": 940 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.0316704586148262, | |
| "learning_rate": 4.902634593356243e-05, | |
| "epoch": 236.32, | |
| "step": 945 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.03111797384917736, | |
| "learning_rate": 4.8739977090492556e-05, | |
| "epoch": 237.64, | |
| "step": 950 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.035214830189943314, | |
| "learning_rate": 4.845360824742268e-05, | |
| "epoch": 238.96, | |
| "step": 955 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.10965805500745773, | |
| "learning_rate": 4.816723940435281e-05, | |
| "epoch": 240.0, | |
| "step": 960 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.034535009413957596, | |
| "learning_rate": 4.7880870561282936e-05, | |
| "epoch": 241.32, | |
| "step": 965 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.03475532680749893, | |
| "learning_rate": 4.7594501718213055e-05, | |
| "epoch": 242.64, | |
| "step": 970 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.029789257794618607, | |
| "learning_rate": 4.730813287514318e-05, | |
| "epoch": 243.96, | |
| "step": 975 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.10357765108346939, | |
| "learning_rate": 4.7021764032073315e-05, | |
| "epoch": 245.0, | |
| "step": 980 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.027163034304976463, | |
| "learning_rate": 4.673539518900344e-05, | |
| "epoch": 246.32, | |
| "step": 985 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.03338053822517395, | |
| "learning_rate": 4.644902634593357e-05, | |
| "epoch": 247.64, | |
| "step": 990 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.03142627701163292, | |
| "learning_rate": 4.6162657502863694e-05, | |
| "epoch": 248.96, | |
| "step": 995 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.08924303948879242, | |
| "learning_rate": 4.5876288659793814e-05, | |
| "epoch": 250.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.033579908311367035, | |
| "learning_rate": 4.558991981672394e-05, | |
| "epoch": 251.32, | |
| "step": 1005 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.030911900103092194, | |
| "learning_rate": 4.530355097365407e-05, | |
| "epoch": 252.64, | |
| "step": 1010 | |
| }, | |
| { | |
| "loss": 0.0228, | |
| "grad_norm": 0.03175857290625572, | |
| "learning_rate": 4.5017182130584194e-05, | |
| "epoch": 253.96, | |
| "step": 1015 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.07784884423017502, | |
| "learning_rate": 4.473081328751432e-05, | |
| "epoch": 255.0, | |
| "step": 1020 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.026503948494791985, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "epoch": 256.32, | |
| "step": 1025 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.02679341658949852, | |
| "learning_rate": 4.415807560137457e-05, | |
| "epoch": 257.64, | |
| "step": 1030 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.029903169721364975, | |
| "learning_rate": 4.38717067583047e-05, | |
| "epoch": 258.96, | |
| "step": 1035 | |
| }, | |
| { | |
| "loss": 0.0271, | |
| "grad_norm": 0.11588968336582184, | |
| "learning_rate": 4.3585337915234826e-05, | |
| "epoch": 260.0, | |
| "step": 1040 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.024833589792251587, | |
| "learning_rate": 4.329896907216495e-05, | |
| "epoch": 261.32, | |
| "step": 1045 | |
| }, | |
| { | |
| "loss": 0.0246, | |
| "grad_norm": 0.026744505390524864, | |
| "learning_rate": 4.301260022909508e-05, | |
| "epoch": 262.64, | |
| "step": 1050 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.02916337549686432, | |
| "learning_rate": 4.27262313860252e-05, | |
| "epoch": 263.96, | |
| "step": 1055 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.0802888572216034, | |
| "learning_rate": 4.2439862542955325e-05, | |
| "epoch": 265.0, | |
| "step": 1060 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.024433529004454613, | |
| "learning_rate": 4.215349369988545e-05, | |
| "epoch": 266.32, | |
| "step": 1065 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.030277010053396225, | |
| "learning_rate": 4.1867124856815585e-05, | |
| "epoch": 267.64, | |
| "step": 1070 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.025442861020565033, | |
| "learning_rate": 4.158075601374571e-05, | |
| "epoch": 268.96, | |
| "step": 1075 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.09739919006824493, | |
| "learning_rate": 4.129438717067583e-05, | |
| "epoch": 270.0, | |
| "step": 1080 | |
| }, | |
| { | |
| "loss": 0.0277, | |
| "grad_norm": 0.02988714724779129, | |
| "learning_rate": 4.100801832760596e-05, | |
| "epoch": 271.32, | |
| "step": 1085 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.03097035363316536, | |
| "learning_rate": 4.0721649484536084e-05, | |
| "epoch": 272.64, | |
| "step": 1090 | |
| }, | |
| { | |
| "loss": 0.0285, | |
| "grad_norm": 0.027229204773902893, | |
| "learning_rate": 4.043528064146621e-05, | |
| "epoch": 273.96, | |
| "step": 1095 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.14006158709526062, | |
| "learning_rate": 4.014891179839634e-05, | |
| "epoch": 275.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "loss": 0.0274, | |
| "grad_norm": 0.03101869486272335, | |
| "learning_rate": 3.9862542955326463e-05, | |
| "epoch": 276.32, | |
| "step": 1105 | |
| }, | |
| { | |
| "loss": 0.0259, | |
| "grad_norm": 0.032572630792856216, | |
| "learning_rate": 3.957617411225659e-05, | |
| "epoch": 277.64, | |
| "step": 1110 | |
| }, | |
| { | |
| "loss": 0.0225, | |
| "grad_norm": 0.024676939472556114, | |
| "learning_rate": 3.9289805269186716e-05, | |
| "epoch": 278.96, | |
| "step": 1115 | |
| }, | |
| { | |
| "loss": 0.0261, | |
| "grad_norm": 0.09308966249227524, | |
| "learning_rate": 3.900343642611684e-05, | |
| "epoch": 280.0, | |
| "step": 1120 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.034002844244241714, | |
| "learning_rate": 3.871706758304697e-05, | |
| "epoch": 281.32, | |
| "step": 1125 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.028817787766456604, | |
| "learning_rate": 3.8430698739977096e-05, | |
| "epoch": 282.64, | |
| "step": 1130 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.026497583836317062, | |
| "learning_rate": 3.8144329896907216e-05, | |
| "epoch": 283.96, | |
| "step": 1135 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.09492871910333633, | |
| "learning_rate": 3.785796105383734e-05, | |
| "epoch": 285.0, | |
| "step": 1140 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.028076212853193283, | |
| "learning_rate": 3.757159221076747e-05, | |
| "epoch": 286.32, | |
| "step": 1145 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02799973078072071, | |
| "learning_rate": 3.7285223367697595e-05, | |
| "epoch": 287.64, | |
| "step": 1150 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.02943967841565609, | |
| "learning_rate": 3.699885452462772e-05, | |
| "epoch": 288.96, | |
| "step": 1155 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.10187557339668274, | |
| "learning_rate": 3.671248568155785e-05, | |
| "epoch": 290.0, | |
| "step": 1160 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.03237089514732361, | |
| "learning_rate": 3.6426116838487974e-05, | |
| "epoch": 291.32, | |
| "step": 1165 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.028723599389195442, | |
| "learning_rate": 3.61397479954181e-05, | |
| "epoch": 292.64, | |
| "step": 1170 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.03231768310070038, | |
| "learning_rate": 3.585337915234823e-05, | |
| "epoch": 293.96, | |
| "step": 1175 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.06909634917974472, | |
| "learning_rate": 3.5567010309278354e-05, | |
| "epoch": 295.0, | |
| "step": 1180 | |
| }, | |
| { | |
| "loss": 0.0288, | |
| "grad_norm": 0.028529174625873566, | |
| "learning_rate": 3.528064146620848e-05, | |
| "epoch": 296.32, | |
| "step": 1185 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.02624385990202427, | |
| "learning_rate": 3.49942726231386e-05, | |
| "epoch": 297.64, | |
| "step": 1190 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.02835160307586193, | |
| "learning_rate": 3.4707903780068726e-05, | |
| "epoch": 298.96, | |
| "step": 1195 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.0636526346206665, | |
| "learning_rate": 3.442153493699885e-05, | |
| "epoch": 300.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.027017105370759964, | |
| "learning_rate": 3.4135166093928986e-05, | |
| "epoch": 301.32, | |
| "step": 1205 | |
| }, | |
| { | |
| "loss": 0.0232, | |
| "grad_norm": 0.03256632387638092, | |
| "learning_rate": 3.384879725085911e-05, | |
| "epoch": 302.64, | |
| "step": 1210 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.025508297607302666, | |
| "learning_rate": 3.356242840778923e-05, | |
| "epoch": 303.96, | |
| "step": 1215 | |
| }, | |
| { | |
| "loss": 0.0225, | |
| "grad_norm": 0.05160168185830116, | |
| "learning_rate": 3.327605956471936e-05, | |
| "epoch": 305.0, | |
| "step": 1220 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.030119648203253746, | |
| "learning_rate": 3.2989690721649485e-05, | |
| "epoch": 306.32, | |
| "step": 1225 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.035157229751348495, | |
| "learning_rate": 3.270332187857961e-05, | |
| "epoch": 307.64, | |
| "step": 1230 | |
| }, | |
| { | |
| "loss": 0.0231, | |
| "grad_norm": 0.022623248398303986, | |
| "learning_rate": 3.241695303550974e-05, | |
| "epoch": 308.96, | |
| "step": 1235 | |
| }, | |
| { | |
| "loss": 0.0278, | |
| "grad_norm": 0.09421593695878983, | |
| "learning_rate": 3.2130584192439865e-05, | |
| "epoch": 310.0, | |
| "step": 1240 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.029251769185066223, | |
| "learning_rate": 3.184421534936999e-05, | |
| "epoch": 311.32, | |
| "step": 1245 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.03382476419210434, | |
| "learning_rate": 3.155784650630012e-05, | |
| "epoch": 312.64, | |
| "step": 1250 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.027744019404053688, | |
| "learning_rate": 3.1271477663230244e-05, | |
| "epoch": 313.96, | |
| "step": 1255 | |
| }, | |
| { | |
| "loss": 0.0276, | |
| "grad_norm": 0.09388919919729233, | |
| "learning_rate": 3.098510882016037e-05, | |
| "epoch": 315.0, | |
| "step": 1260 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.025077903643250465, | |
| "learning_rate": 3.06987399770905e-05, | |
| "epoch": 316.32, | |
| "step": 1265 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.030428890138864517, | |
| "learning_rate": 3.0412371134020617e-05, | |
| "epoch": 317.64, | |
| "step": 1270 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.026998048648238182, | |
| "learning_rate": 3.0126002290950743e-05, | |
| "epoch": 318.96, | |
| "step": 1275 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.09231777489185333, | |
| "learning_rate": 2.983963344788087e-05, | |
| "epoch": 320.0, | |
| "step": 1280 | |
| }, | |
| { | |
| "loss": 0.0239, | |
| "grad_norm": 0.02951677329838276, | |
| "learning_rate": 2.9553264604811e-05, | |
| "epoch": 321.32, | |
| "step": 1285 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.030611136928200722, | |
| "learning_rate": 2.9266895761741126e-05, | |
| "epoch": 322.64, | |
| "step": 1290 | |
| }, | |
| { | |
| "loss": 0.0272, | |
| "grad_norm": 0.029771588742733, | |
| "learning_rate": 2.8980526918671253e-05, | |
| "epoch": 323.96, | |
| "step": 1295 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.08960679918527603, | |
| "learning_rate": 2.8694158075601372e-05, | |
| "epoch": 325.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.02549550123512745, | |
| "learning_rate": 2.8407789232531502e-05, | |
| "epoch": 326.32, | |
| "step": 1305 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.03180180862545967, | |
| "learning_rate": 2.812142038946163e-05, | |
| "epoch": 327.64, | |
| "step": 1310 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.032844457775354385, | |
| "learning_rate": 2.7835051546391755e-05, | |
| "epoch": 328.96, | |
| "step": 1315 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.10202164947986603, | |
| "learning_rate": 2.754868270332188e-05, | |
| "epoch": 330.0, | |
| "step": 1320 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.028862839564681053, | |
| "learning_rate": 2.7262313860252005e-05, | |
| "epoch": 331.32, | |
| "step": 1325 | |
| }, | |
| { | |
| "loss": 0.0247, | |
| "grad_norm": 0.030968671664595604, | |
| "learning_rate": 2.697594501718213e-05, | |
| "epoch": 332.64, | |
| "step": 1330 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.0348350964486599, | |
| "learning_rate": 2.6689576174112258e-05, | |
| "epoch": 333.96, | |
| "step": 1335 | |
| }, | |
| { | |
| "loss": 0.0229, | |
| "grad_norm": 0.04857328534126282, | |
| "learning_rate": 2.6403207331042384e-05, | |
| "epoch": 335.0, | |
| "step": 1340 | |
| }, | |
| { | |
| "loss": 0.0255, | |
| "grad_norm": 0.02903318591415882, | |
| "learning_rate": 2.611683848797251e-05, | |
| "epoch": 336.32, | |
| "step": 1345 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.02943980135023594, | |
| "learning_rate": 2.5830469644902637e-05, | |
| "epoch": 337.64, | |
| "step": 1350 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02523176558315754, | |
| "learning_rate": 2.554410080183276e-05, | |
| "epoch": 338.96, | |
| "step": 1355 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.11107093840837479, | |
| "learning_rate": 2.5257731958762887e-05, | |
| "epoch": 340.0, | |
| "step": 1360 | |
| }, | |
| { | |
| "loss": 0.0263, | |
| "grad_norm": 0.033267661929130554, | |
| "learning_rate": 2.4971363115693013e-05, | |
| "epoch": 341.32, | |
| "step": 1365 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.029709186404943466, | |
| "learning_rate": 2.468499427262314e-05, | |
| "epoch": 342.64, | |
| "step": 1370 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.026531100273132324, | |
| "learning_rate": 2.4398625429553266e-05, | |
| "epoch": 343.96, | |
| "step": 1375 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.11106812953948975, | |
| "learning_rate": 2.4112256586483393e-05, | |
| "epoch": 345.0, | |
| "step": 1380 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.028001897037029266, | |
| "learning_rate": 2.3825887743413516e-05, | |
| "epoch": 346.32, | |
| "step": 1385 | |
| }, | |
| { | |
| "loss": 0.0237, | |
| "grad_norm": 0.0313006266951561, | |
| "learning_rate": 2.3539518900343642e-05, | |
| "epoch": 347.64, | |
| "step": 1390 | |
| }, | |
| { | |
| "loss": 0.0232, | |
| "grad_norm": 0.03095093183219433, | |
| "learning_rate": 2.3253150057273772e-05, | |
| "epoch": 348.96, | |
| "step": 1395 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.06695447117090225, | |
| "learning_rate": 2.2966781214203895e-05, | |
| "epoch": 350.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02904387377202511, | |
| "learning_rate": 2.268041237113402e-05, | |
| "epoch": 351.32, | |
| "step": 1405 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.02876129560172558, | |
| "learning_rate": 2.2394043528064148e-05, | |
| "epoch": 352.64, | |
| "step": 1410 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.026445671916007996, | |
| "learning_rate": 2.210767468499427e-05, | |
| "epoch": 353.96, | |
| "step": 1415 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.0666656345129013, | |
| "learning_rate": 2.18213058419244e-05, | |
| "epoch": 355.0, | |
| "step": 1420 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.03694348409771919, | |
| "learning_rate": 2.1534936998854528e-05, | |
| "epoch": 356.32, | |
| "step": 1425 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.026420310139656067, | |
| "learning_rate": 2.124856815578465e-05, | |
| "epoch": 357.64, | |
| "step": 1430 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.027356242761015892, | |
| "learning_rate": 2.0962199312714777e-05, | |
| "epoch": 358.96, | |
| "step": 1435 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.08381753414869308, | |
| "learning_rate": 2.0675830469644904e-05, | |
| "epoch": 360.0, | |
| "step": 1440 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02863556146621704, | |
| "learning_rate": 2.038946162657503e-05, | |
| "epoch": 361.32, | |
| "step": 1445 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.031583286821842194, | |
| "learning_rate": 2.0103092783505157e-05, | |
| "epoch": 362.64, | |
| "step": 1450 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.024988096207380295, | |
| "learning_rate": 1.981672394043528e-05, | |
| "epoch": 363.96, | |
| "step": 1455 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.09980056434869766, | |
| "learning_rate": 1.9530355097365406e-05, | |
| "epoch": 365.0, | |
| "step": 1460 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02719324827194214, | |
| "learning_rate": 1.9243986254295536e-05, | |
| "epoch": 366.32, | |
| "step": 1465 | |
| }, | |
| { | |
| "loss": 0.0229, | |
| "grad_norm": 0.021068066358566284, | |
| "learning_rate": 1.895761741122566e-05, | |
| "epoch": 367.64, | |
| "step": 1470 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.027141617611050606, | |
| "learning_rate": 1.8671248568155786e-05, | |
| "epoch": 368.96, | |
| "step": 1475 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.10978229343891144, | |
| "learning_rate": 1.8384879725085912e-05, | |
| "epoch": 370.0, | |
| "step": 1480 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.030765032395720482, | |
| "learning_rate": 1.809851088201604e-05, | |
| "epoch": 371.32, | |
| "step": 1485 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.026865428313612938, | |
| "learning_rate": 1.7812142038946165e-05, | |
| "epoch": 372.64, | |
| "step": 1490 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.024347811937332153, | |
| "learning_rate": 1.7525773195876288e-05, | |
| "epoch": 373.96, | |
| "step": 1495 | |
| }, | |
| { | |
| "loss": 0.0265, | |
| "grad_norm": 0.0818612203001976, | |
| "learning_rate": 1.7239404352806415e-05, | |
| "epoch": 375.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.030260901898145676, | |
| "learning_rate": 1.695303550973654e-05, | |
| "epoch": 376.32, | |
| "step": 1505 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.028559433296322823, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "epoch": 377.64, | |
| "step": 1510 | |
| }, | |
| { | |
| "loss": 0.023, | |
| "grad_norm": 0.029224252328276634, | |
| "learning_rate": 1.6380297823596794e-05, | |
| "epoch": 378.96, | |
| "step": 1515 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.1068309098482132, | |
| "learning_rate": 1.609392898052692e-05, | |
| "epoch": 380.0, | |
| "step": 1520 | |
| }, | |
| { | |
| "loss": 0.0241, | |
| "grad_norm": 0.030471889302134514, | |
| "learning_rate": 1.5807560137457044e-05, | |
| "epoch": 381.32, | |
| "step": 1525 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.02888798713684082, | |
| "learning_rate": 1.5521191294387173e-05, | |
| "epoch": 382.64, | |
| "step": 1530 | |
| }, | |
| { | |
| "loss": 0.0256, | |
| "grad_norm": 0.031952738761901855, | |
| "learning_rate": 1.5234822451317298e-05, | |
| "epoch": 383.96, | |
| "step": 1535 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.1276218146085739, | |
| "learning_rate": 1.4948453608247423e-05, | |
| "epoch": 385.0, | |
| "step": 1540 | |
| }, | |
| { | |
| "loss": 0.027, | |
| "grad_norm": 0.03086724691092968, | |
| "learning_rate": 1.466208476517755e-05, | |
| "epoch": 386.32, | |
| "step": 1545 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.03255283832550049, | |
| "learning_rate": 1.4375715922107674e-05, | |
| "epoch": 387.64, | |
| "step": 1550 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.042661767452955246, | |
| "learning_rate": 1.40893470790378e-05, | |
| "epoch": 388.96, | |
| "step": 1555 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.10741709172725677, | |
| "learning_rate": 1.3802978235967929e-05, | |
| "epoch": 390.0, | |
| "step": 1560 | |
| }, | |
| { | |
| "loss": 0.0267, | |
| "grad_norm": 0.02813476137816906, | |
| "learning_rate": 1.3516609392898052e-05, | |
| "epoch": 391.32, | |
| "step": 1565 | |
| }, | |
| { | |
| "loss": 0.025, | |
| "grad_norm": 0.03153588995337486, | |
| "learning_rate": 1.323024054982818e-05, | |
| "epoch": 392.64, | |
| "step": 1570 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.030935829505324364, | |
| "learning_rate": 1.2943871706758307e-05, | |
| "epoch": 393.96, | |
| "step": 1575 | |
| }, | |
| { | |
| "loss": 0.0266, | |
| "grad_norm": 0.1351139396429062, | |
| "learning_rate": 1.2657502863688431e-05, | |
| "epoch": 395.0, | |
| "step": 1580 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.02933109737932682, | |
| "learning_rate": 1.2371134020618558e-05, | |
| "epoch": 396.32, | |
| "step": 1585 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.024457110092043877, | |
| "learning_rate": 1.2084765177548683e-05, | |
| "epoch": 397.64, | |
| "step": 1590 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.02621588297188282, | |
| "learning_rate": 1.1798396334478809e-05, | |
| "epoch": 398.96, | |
| "step": 1595 | |
| }, | |
| { | |
| "loss": 0.0264, | |
| "grad_norm": 0.1131487563252449, | |
| "learning_rate": 1.1512027491408934e-05, | |
| "epoch": 400.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.030307544395327568, | |
| "learning_rate": 1.1225658648339062e-05, | |
| "epoch": 401.32, | |
| "step": 1605 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.030524153262376785, | |
| "learning_rate": 1.0939289805269187e-05, | |
| "epoch": 402.64, | |
| "step": 1610 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.027147499844431877, | |
| "learning_rate": 1.0652920962199313e-05, | |
| "epoch": 403.96, | |
| "step": 1615 | |
| }, | |
| { | |
| "loss": 0.0297, | |
| "grad_norm": 0.13672912120819092, | |
| "learning_rate": 1.036655211912944e-05, | |
| "epoch": 405.0, | |
| "step": 1620 | |
| }, | |
| { | |
| "loss": 0.0254, | |
| "grad_norm": 0.03520699590444565, | |
| "learning_rate": 1.0080183276059566e-05, | |
| "epoch": 406.32, | |
| "step": 1625 | |
| }, | |
| { | |
| "loss": 0.0223, | |
| "grad_norm": 0.024485955014824867, | |
| "learning_rate": 9.793814432989691e-06, | |
| "epoch": 407.64, | |
| "step": 1630 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.028952937573194504, | |
| "learning_rate": 9.507445589919818e-06, | |
| "epoch": 408.96, | |
| "step": 1635 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.14274798333644867, | |
| "learning_rate": 9.221076746849944e-06, | |
| "epoch": 410.0, | |
| "step": 1640 | |
| }, | |
| { | |
| "loss": 0.0238, | |
| "grad_norm": 0.024746006354689598, | |
| "learning_rate": 8.934707903780069e-06, | |
| "epoch": 411.32, | |
| "step": 1645 | |
| }, | |
| { | |
| "loss": 0.0251, | |
| "grad_norm": 0.03118201531469822, | |
| "learning_rate": 8.648339060710195e-06, | |
| "epoch": 412.64, | |
| "step": 1650 | |
| }, | |
| { | |
| "loss": 0.0245, | |
| "grad_norm": 0.02892090007662773, | |
| "learning_rate": 8.36197021764032e-06, | |
| "epoch": 413.96, | |
| "step": 1655 | |
| }, | |
| { | |
| "loss": 0.0234, | |
| "grad_norm": 0.09150885790586472, | |
| "learning_rate": 8.075601374570448e-06, | |
| "epoch": 415.0, | |
| "step": 1660 | |
| }, | |
| { | |
| "loss": 0.0269, | |
| "grad_norm": 0.027218345552682877, | |
| "learning_rate": 7.789232531500573e-06, | |
| "epoch": 416.32, | |
| "step": 1665 | |
| }, | |
| { | |
| "loss": 0.0233, | |
| "grad_norm": 0.030960749834775925, | |
| "learning_rate": 7.502863688430699e-06, | |
| "epoch": 417.64, | |
| "step": 1670 | |
| }, | |
| { | |
| "loss": 0.022, | |
| "grad_norm": 0.028846222907304764, | |
| "learning_rate": 7.216494845360824e-06, | |
| "epoch": 418.96, | |
| "step": 1675 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.10136660933494568, | |
| "learning_rate": 6.930126002290952e-06, | |
| "epoch": 420.0, | |
| "step": 1680 | |
| }, | |
| { | |
| "loss": 0.0222, | |
| "grad_norm": 0.0227106474339962, | |
| "learning_rate": 6.643757159221077e-06, | |
| "epoch": 421.32, | |
| "step": 1685 | |
| }, | |
| { | |
| "loss": 0.0252, | |
| "grad_norm": 0.029144780710339546, | |
| "learning_rate": 6.357388316151203e-06, | |
| "epoch": 422.64, | |
| "step": 1690 | |
| }, | |
| { | |
| "loss": 0.022, | |
| "grad_norm": 0.026727862656116486, | |
| "learning_rate": 6.071019473081329e-06, | |
| "epoch": 423.96, | |
| "step": 1695 | |
| }, | |
| { | |
| "loss": 0.0257, | |
| "grad_norm": 0.10660794377326965, | |
| "learning_rate": 5.784650630011455e-06, | |
| "epoch": 425.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "loss": 0.0244, | |
| "grad_norm": 0.03457332029938698, | |
| "learning_rate": 5.498281786941581e-06, | |
| "epoch": 426.32, | |
| "step": 1705 | |
| }, | |
| { | |
| "loss": 0.0227, | |
| "grad_norm": 0.026608219370245934, | |
| "learning_rate": 5.211912943871707e-06, | |
| "epoch": 427.64, | |
| "step": 1710 | |
| }, | |
| { | |
| "loss": 0.0243, | |
| "grad_norm": 0.02732602320611477, | |
| "learning_rate": 4.925544100801833e-06, | |
| "epoch": 428.96, | |
| "step": 1715 | |
| }, | |
| { | |
| "loss": 0.0273, | |
| "grad_norm": 0.15714120864868164, | |
| "learning_rate": 4.639175257731959e-06, | |
| "epoch": 430.0, | |
| "step": 1720 | |
| }, | |
| { | |
| "loss": 0.0228, | |
| "grad_norm": 0.028533920645713806, | |
| "learning_rate": 4.352806414662085e-06, | |
| "epoch": 431.32, | |
| "step": 1725 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.030768312513828278, | |
| "learning_rate": 4.066437571592211e-06, | |
| "epoch": 432.64, | |
| "step": 1730 | |
| }, | |
| { | |
| "loss": 0.0249, | |
| "grad_norm": 0.028397932648658752, | |
| "learning_rate": 3.7800687285223365e-06, | |
| "epoch": 433.96, | |
| "step": 1735 | |
| }, | |
| { | |
| "loss": 0.024, | |
| "grad_norm": 0.09805476665496826, | |
| "learning_rate": 3.493699885452463e-06, | |
| "epoch": 435.0, | |
| "step": 1740 | |
| }, | |
| { | |
| "loss": 0.0258, | |
| "grad_norm": 0.028490280732512474, | |
| "learning_rate": 3.2073310423825886e-06, | |
| "epoch": 436.32, | |
| "step": 1745 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.030706044286489487, | |
| "learning_rate": 2.920962199312715e-06, | |
| "epoch": 437.64, | |
| "step": 1750 | |
| }, | |
| { | |
| "loss": 0.0231, | |
| "grad_norm": 0.03271190822124481, | |
| "learning_rate": 2.6345933562428407e-06, | |
| "epoch": 438.96, | |
| "step": 1755 | |
| }, | |
| { | |
| "loss": 0.0253, | |
| "grad_norm": 0.11334367841482162, | |
| "learning_rate": 2.3482245131729668e-06, | |
| "epoch": 440.0, | |
| "step": 1760 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "grad_norm": 0.026566533371806145, | |
| "learning_rate": 2.061855670103093e-06, | |
| "epoch": 441.32, | |
| "step": 1765 | |
| }, | |
| { | |
| "loss": 0.023, | |
| "grad_norm": 0.03375476971268654, | |
| "learning_rate": 1.7754868270332189e-06, | |
| "epoch": 442.64, | |
| "step": 1770 | |
| }, | |
| { | |
| "loss": 0.0242, | |
| "grad_norm": 0.027957361191511154, | |
| "learning_rate": 1.4891179839633447e-06, | |
| "epoch": 443.96, | |
| "step": 1775 | |
| }, | |
| { | |
| "loss": 0.0282, | |
| "grad_norm": 0.1172029972076416, | |
| "learning_rate": 1.202749140893471e-06, | |
| "epoch": 445.0, | |
| "step": 1780 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "grad_norm": 0.02673257142305374, | |
| "learning_rate": 9.163802978235968e-07, | |
| "epoch": 446.32, | |
| "step": 1785 | |
| }, | |
| { | |
| "loss": 0.0236, | |
| "grad_norm": 0.02195633016526699, | |
| "learning_rate": 6.300114547537229e-07, | |
| "epoch": 447.64, | |
| "step": 1790 | |
| }, | |
| { | |
| "loss": 0.026, | |
| "grad_norm": 0.02480531483888626, | |
| "learning_rate": 3.436426116838488e-07, | |
| "epoch": 448.96, | |
| "step": 1795 | |
| }, | |
| { | |
| "loss": 0.029, | |
| "grad_norm": 0.11198581010103226, | |
| "learning_rate": 5.72737686139748e-08, | |
| "epoch": 450.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "train_runtime": 30905.7388, | |
| "train_samples_per_second": 1.864, | |
| "train_steps_per_second": 0.058, | |
| "total_flos": 1.5308141101056e+18, | |
| "train_loss": 0.04942113804734415, | |
| "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_00/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_00", | |
| "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": "" | |
| } |