| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.9998415967052114, | |
| "global_step": 3156, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.374999999999999e-05, | |
| "loss": 1.8139, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018749999999999998, | |
| "loss": 1.7876, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00028125, | |
| "loss": 1.8662, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00037499999999999995, | |
| "loss": 1.8047, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00046875, | |
| "loss": 1.9482, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005625, | |
| "loss": 1.9624, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005999986347533043, | |
| "loss": 1.8496, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005999902916240525, | |
| "loss": 1.8962, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005999743640466132, | |
| "loss": 1.9655, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005999508524236727, | |
| "loss": 1.9308, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000599919757349659, | |
| "loss": 2.0406, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0005998810796107273, | |
| "loss": 1.8524, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0005998348201847405, | |
| "loss": 1.9677, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0005997809802412437, | |
| "loss": 1.963, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0005997195611414353, | |
| "loss": 1.7783, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000599650564438132, | |
| "loss": 1.9602, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0005995739918757302, | |
| "loss": 1.941, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000599489845390161, | |
| "loss": 1.908, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0005993981271088423, | |
| "loss": 1.8707, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0005992988393506244, | |
| "loss": 1.9848, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0005991919846257313, | |
| "loss": 1.9445, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0005990775656356976, | |
| "loss": 1.8844, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0005989555852732997, | |
| "loss": 1.9902, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0005988260466224834, | |
| "loss": 1.8382, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0005986889529582852, | |
| "loss": 2.0264, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00059854430774675, | |
| "loss": 1.9612, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000598392114644843, | |
| "loss": 2.0919, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0005982323775003579, | |
| "loss": 1.9855, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0005980651003518189, | |
| "loss": 1.8857, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0005978902874283793, | |
| "loss": 1.972, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0005977079431497138, | |
| "loss": 1.8702, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0005975180721259074, | |
| "loss": 1.886, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0005973206791573386, | |
| "loss": 1.9137, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0005971157692345581, | |
| "loss": 1.9225, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0005969033475381623, | |
| "loss": 1.8239, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000596683419438663, | |
| "loss": 1.9465, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0005964559904963512, | |
| "loss": 1.8873, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0005962210664611563, | |
| "loss": 1.9938, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000595978653272501, | |
| "loss": 1.9666, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0005957287570591513, | |
| "loss": 1.8345, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000595471384139061, | |
| "loss": 1.983, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000595206541019213, | |
| "loss": 1.9706, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0005949342343954535, | |
| "loss": 1.9829, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0005946544711523236, | |
| "loss": 1.8883, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0005943672583628848, | |
| "loss": 2.0109, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0005940726032885408, | |
| "loss": 1.9654, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0005937705133788529, | |
| "loss": 1.8733, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0005934609962713525, | |
| "loss": 1.9419, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0005931440597913479, | |
| "loss": 1.9287, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0005928197119517257, | |
| "loss": 1.9507, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0005924879609527498, | |
| "loss": 1.9224, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005921488151818521, | |
| "loss": 1.9124, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000591802283213422, | |
| "loss": 1.9667, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005914483738085887, | |
| "loss": 1.955, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005910870959150005, | |
| "loss": 1.949, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005907184586665977, | |
| "loss": 1.7752, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005903424713833823, | |
| "loss": 1.8766, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005899591435711821, | |
| "loss": 2.0573, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005895684849214106, | |
| "loss": 1.9837, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005891705053108217, | |
| "loss": 2.0551, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005887652148012601, | |
| "loss": 2.0254, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.000588352623639407, | |
| "loss": 2.0329, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005879327422565209, | |
| "loss": 1.8451, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000587505581268174, | |
| "loss": 1.9613, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005870711514739836, | |
| "loss": 1.9022, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005866294638573397, | |
| "loss": 2.0333, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005861805295851261, | |
| "loss": 1.8614, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005857243600074394, | |
| "loss": 1.937, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005852609666573011, | |
| "loss": 1.8581, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005847903612503668, | |
| "loss": 1.9423, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000584312555684629, | |
| "loss": 1.9807, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005838275620401171, | |
| "loss": 1.9902, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0005833353925785922, | |
| "loss": 1.8792, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0005828360597432358, | |
| "loss": 1.9335, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005823295761583368, | |
| "loss": 1.9518, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0005818159546289711, | |
| "loss": 2.0066, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005812952081406786, | |
| "loss": 2.0164, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0005807673498591347, | |
| "loss": 2.0275, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0005802323931298169, | |
| "loss": 1.983, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005796903514776685, | |
| "loss": 1.9721, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0005791412386067555, | |
| "loss": 1.8981, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.000578585068399921, | |
| "loss": 2.073, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0005780218549184339, | |
| "loss": 1.9388, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.000577451612401633, | |
| "loss": 2.0487, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0005768743552665678, | |
| "loss": 2.0537, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0005762900981076332, | |
| "loss": 1.9692, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000575698855696201, | |
| "loss": 1.9719, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0005751006429802467, | |
| "loss": 1.9929, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0005744954750839706, | |
| "loss": 2.0479, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0005738833673074165, | |
| "loss": 1.8216, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0005732643351260843, | |
| "loss": 1.8762, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0005726383941905386, | |
| "loss": 1.9418, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0005720055603260138, | |
| "loss": 2.0108, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0005713658495320128, | |
| "loss": 1.9163, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.000570719277981904, | |
| "loss": 1.9502, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0005700658620225108, | |
| "loss": 2.1221, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0005694056181736996, | |
| "loss": 1.9281, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0005687385631279615, | |
| "loss": 2.0308, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0005680647137499902, | |
| "loss": 1.9564, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0005673840870762561, | |
| "loss": 1.9229, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0005666967003145753, | |
| "loss": 2.0006, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0005660025708436743, | |
| "loss": 2.0517, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0005653017162127511, | |
| "loss": 1.9226, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0005645941541410314, | |
| "loss": 1.9131, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00056387990251732, | |
| "loss": 2.0067, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0005631589793995495, | |
| "loss": 1.7818, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0005624314030143236, | |
| "loss": 1.9343, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0005616971917564549, | |
| "loss": 1.9357, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.000560956364188502, | |
| "loss": 1.8905, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0005602089390402986, | |
| "loss": 1.9669, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0005594549352084807, | |
| "loss": 1.9771, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0005586943717560083, | |
| "loss": 1.9627, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0005579272679116841, | |
| "loss": 1.9329, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.000557153643069667, | |
| "loss": 1.9314, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0005563735167889817, | |
| "loss": 1.8892, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.0005555869087930242, | |
| "loss": 2.0062, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0005547938389690635, | |
| "loss": 1.9624, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0005539943273677384, | |
| "loss": 2.0077, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0005531883942025507, | |
| "loss": 2.0083, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0005523760598493544, | |
| "loss": 2.0212, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0005515573448458399, | |
| "loss": 1.9307, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0005507322698910156, | |
| "loss": 1.7577, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.000549900855844684, | |
| "loss": 1.9654, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0005490631237269147, | |
| "loss": 1.7376, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0005482190947175124, | |
| "loss": 1.9602, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0005473687901554821, | |
| "loss": 1.9872, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0005465122315384889, | |
| "loss": 1.9741, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0005456494405223152, | |
| "loss": 1.9905, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0005447804389203129, | |
| "loss": 1.8507, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.0005439052487028517, | |
| "loss": 1.987, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0005430238919967637, | |
| "loss": 1.9029, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0005421363910847844, | |
| "loss": 1.9159, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0005412427684049888, | |
| "loss": 1.914, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0005403430465502244, | |
| "loss": 1.9955, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.00053943724826754, | |
| "loss": 1.9325, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0005385253964576104, | |
| "loss": 1.8784, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.000537607514174158, | |
| "loss": 1.8416, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0005366836246233687, | |
| "loss": 1.9717, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0005357537511633068, | |
| "loss": 1.8637, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0005348179173033233, | |
| "loss": 1.804, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.000533876146703462, | |
| "loss": 1.9949, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0005329284631738609, | |
| "loss": 1.9731, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0005319748906741511, | |
| "loss": 1.8443, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0005310154533128501, | |
| "loss": 1.8656, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0005300501753467525, | |
| "loss": 1.9067, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0005290790811803176, | |
| "loss": 1.9235, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0005281021953650507, | |
| "loss": 2.0954, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0005271195425988842, | |
| "loss": 1.9317, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.000526131147725552, | |
| "loss": 1.9577, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0005251370357339616, | |
| "loss": 1.8044, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0005241372317575629, | |
| "loss": 2.0009, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0005231317610737121, | |
| "loss": 1.875, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0005221206491030327, | |
| "loss": 1.9521, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0005211039214087733, | |
| "loss": 2.0129, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0005200816036961606, | |
| "loss": 1.9002, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0005190537218117502, | |
| "loss": 1.9327, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0005180203017427728, | |
| "loss": 1.9045, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 2.0005805492401123, | |
| "eval_runtime": 92.7966, | |
| "eval_samples_per_second": 138.022, | |
| "eval_steps_per_second": 8.632, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0005169813696164772, | |
| "loss": 2.0109, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0005159369516994695, | |
| "loss": 1.7846, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0005148870743970495, | |
| "loss": 1.8012, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.000513831764252543, | |
| "loss": 1.9483, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0005127710479466303, | |
| "loss": 1.9163, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0005117049522966723, | |
| "loss": 1.8544, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0005106335042560317, | |
| "loss": 1.7812, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0005095567309133926, | |
| "loss": 1.8897, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0005084746594920745, | |
| "loss": 1.8461, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.000507387317349345, | |
| "loss": 1.8396, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0005062947319757274, | |
| "loss": 1.8366, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0005051969309943062, | |
| "loss": 1.8389, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0005040939421600284, | |
| "loss": 1.7767, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0005029857933590021, | |
| "loss": 1.8001, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0005018725126077912, | |
| "loss": 1.8145, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0005007541280527068, | |
| "loss": 1.8317, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0004996306679690967, | |
| "loss": 1.8302, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0004985021607606293, | |
| "loss": 1.9682, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0004973686349585758, | |
| "loss": 1.841, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0004962301192210897, | |
| "loss": 1.8658, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.000495086642332481, | |
| "loss": 1.8247, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0004939382332024895, | |
| "loss": 1.8037, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.000492784920865553, | |
| "loss": 1.9416, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0004916267344800742, | |
| "loss": 1.8733, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0004904637033276827, | |
| "loss": 1.8063, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.000489295856812495, | |
| "loss": 1.9041, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0004881232244603716, | |
| "loss": 1.8182, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0004869458359181693, | |
| "loss": 1.7881, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00048576372095299273, | |
| "loss": 1.7624, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00048457690945144147, | |
| "loss": 1.9506, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00048338543141885444, | |
| "loss": 1.8866, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.000482189316978551, | |
| "loss": 1.8682, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0004809885963710698, | |
| "loss": 1.8928, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00047978329995340403, | |
| "loss": 1.8434, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.000478573458198234, | |
| "loss": 1.8086, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00047735910169315645, | |
| "loss": 1.9049, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0004761402611399116, | |
| "loss": 1.7393, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0004749169673536068, | |
| "loss": 1.7668, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0004736892512619375, | |
| "loss": 1.8751, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0004724571439044048, | |
| "loss": 1.7533, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0004712206764315317, | |
| "loss": 1.7789, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00046997988010407483, | |
| "loss": 1.9063, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0004687347862922342, | |
| "loss": 1.7936, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00046748542647486, | |
| "loss": 1.784, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0004662318322386573, | |
| "loss": 1.8396, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0004649740352773868, | |
| "loss": 1.8145, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00046371206739106383, | |
| "loss": 1.7536, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00046244596048515443, | |
| "loss": 1.7931, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00046117574656976855, | |
| "loss": 1.8734, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0004599014577588509, | |
| "loss": 1.8342, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00045862312626936884, | |
| "loss": 1.8973, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00045734078442049826, | |
| "loss": 1.7734, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00045605446463280576, | |
| "loss": 1.7644, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00045476419942742994, | |
| "loss": 1.8709, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.0004534700214252584, | |
| "loss": 1.7612, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0004521719633461034, | |
| "loss": 1.9362, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0004508700580078742, | |
| "loss": 1.8653, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0004495643383257482, | |
| "loss": 1.767, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0004482548373113377, | |
| "loss": 1.829, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00044694158807185636, | |
| "loss": 1.8861, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0004456246238092812, | |
| "loss": 1.7582, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0004443039778195137, | |
| "loss": 1.8504, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0004429796834915381, | |
| "loss": 1.8637, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00044165177430657685, | |
| "loss": 1.8534, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00044032028383724446, | |
| "loss": 1.7965, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00043898524574669873, | |
| "loss": 1.8788, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00043764669378778913, | |
| "loss": 1.8786, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0004363046618022041, | |
| "loss": 1.7909, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0004349591837196154, | |
| "loss": 1.8828, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0004336102935568196, | |
| "loss": 1.9064, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0004322580254168791, | |
| "loss": 1.7983, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00043090241348825876, | |
| "loss": 1.8791, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.0004295434920439628, | |
| "loss": 1.8287, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00042818129544066713, | |
| "loss": 1.8337, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.0004268158581178515, | |
| "loss": 1.8558, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00042544721459692846, | |
| "loss": 1.8345, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.0004240753994803708, | |
| "loss": 1.8341, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00042270044745083645, | |
| "loss": 1.8161, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00042132239327029173, | |
| "loss": 1.6921, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00041994127177913247, | |
| "loss": 1.7944, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00041855711789530345, | |
| "loss": 1.8489, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00041716996661341507, | |
| "loss": 1.8568, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00041577985300385913, | |
| "loss": 1.883, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0004143868122119217, | |
| "loss": 1.9306, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00041299087945689496, | |
| "loss": 1.8549, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0004115920900311864, | |
| "loss": 1.7132, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.000410190479299427, | |
| "loss": 1.8957, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0004087860826975767, | |
| "loss": 1.7544, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00040737893573202877, | |
| "loss": 1.8321, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00040596907397871205, | |
| "loss": 1.8146, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00040455653308219127, | |
| "loss": 1.7742, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0004031413487547665, | |
| "loss": 1.8268, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0004017235567755694, | |
| "loss": 1.8419, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0004003031929896594, | |
| "loss": 1.839, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00039888029330711704, | |
| "loss": 1.8722, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.0003974548937021363, | |
| "loss": 1.9201, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.00039602703021211484, | |
| "loss": 1.9658, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.0003945967389367431, | |
| "loss": 1.8658, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00039316405603709157, | |
| "loss": 1.8716, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0003917290177346964, | |
| "loss": 1.7848, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0003902916603106439, | |
| "loss": 2.0245, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0003888520201046532, | |
| "loss": 1.7693, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0003874101335141571, | |
| "loss": 1.8287, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00038596603699338276, | |
| "loss": 1.8367, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0003845197670524289, | |
| "loss": 1.9042, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0003830713602563437, | |
| "loss": 1.7802, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00038162085322419973, | |
| "loss": 1.7705, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0003801682826281687, | |
| "loss": 1.8283, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.0003787136851925937, | |
| "loss": 1.9271, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00037725709769306114, | |
| "loss": 1.8433, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0003757985569554707, | |
| "loss": 1.8718, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00037433809985510486, | |
| "loss": 1.842, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00037287576331569564, | |
| "loss": 1.9048, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0003714115843084918, | |
| "loss": 1.8235, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00036994559985132415, | |
| "loss": 1.8576, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00036847784700766933, | |
| "loss": 1.836, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0003670083628857128, | |
| "loss": 1.8414, | |
| "step": 1370 |