| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 2890, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 4.816294846566506, | |
| "learning_rate": 1.7301038062283738e-08, | |
| "loss": 0.8531, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 4.872596535971759, | |
| "learning_rate": 8.65051903114187e-08, | |
| "loss": 0.8583, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 3.6310650985999184, | |
| "learning_rate": 1.730103806228374e-07, | |
| "loss": 0.8626, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 3.744415952772262, | |
| "learning_rate": 2.5951557093425607e-07, | |
| "loss": 0.8594, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 4.189549391166134, | |
| "learning_rate": 3.460207612456748e-07, | |
| "loss": 0.8542, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 2.388097158161717, | |
| "learning_rate": 4.325259515570934e-07, | |
| "loss": 0.8489, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 2.131273138920553, | |
| "learning_rate": 5.190311418685121e-07, | |
| "loss": 0.8319, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 1.9741009575137782, | |
| "learning_rate": 6.055363321799308e-07, | |
| "loss": 0.8334, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 1.9300494756762427, | |
| "learning_rate": 6.920415224913496e-07, | |
| "loss": 0.8206, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.7903640743849025, | |
| "learning_rate": 7.785467128027683e-07, | |
| "loss": 0.7872, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 1.9336160245419374, | |
| "learning_rate": 8.650519031141868e-07, | |
| "loss": 0.8006, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 1.8895155835527617, | |
| "learning_rate": 9.515570934256056e-07, | |
| "loss": 0.8079, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 1.802498228874264, | |
| "learning_rate": 1.0380622837370243e-06, | |
| "loss": 0.803, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.777164779599201, | |
| "learning_rate": 1.124567474048443e-06, | |
| "loss": 0.8124, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 1.8054173138128906, | |
| "learning_rate": 1.2110726643598616e-06, | |
| "loss": 0.8183, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 1.9350493231536463, | |
| "learning_rate": 1.2975778546712802e-06, | |
| "loss": 0.8029, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 1.8950537177996267, | |
| "learning_rate": 1.3840830449826992e-06, | |
| "loss": 0.8138, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 1.8019972698305375, | |
| "learning_rate": 1.4705882352941177e-06, | |
| "loss": 0.8036, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.9102546088528127, | |
| "learning_rate": 1.5570934256055365e-06, | |
| "loss": 0.7893, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.8718974603674414, | |
| "learning_rate": 1.643598615916955e-06, | |
| "loss": 0.7821, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.8905357930834985, | |
| "learning_rate": 1.7301038062283736e-06, | |
| "loss": 0.7811, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 1.8446962030261018, | |
| "learning_rate": 1.8166089965397926e-06, | |
| "loss": 0.7687, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 1.9762667024263982, | |
| "learning_rate": 1.9031141868512112e-06, | |
| "loss": 0.7876, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 1.9200753281543292, | |
| "learning_rate": 1.9896193771626298e-06, | |
| "loss": 0.7856, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.9558456094506125, | |
| "learning_rate": 2.0761245674740485e-06, | |
| "loss": 0.7532, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 1.9761173872923619, | |
| "learning_rate": 2.1626297577854673e-06, | |
| "loss": 0.7732, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 1.9910728274837823, | |
| "learning_rate": 2.249134948096886e-06, | |
| "loss": 0.7694, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 2.1510352769592007, | |
| "learning_rate": 2.335640138408305e-06, | |
| "loss": 0.7612, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 2.0884221130022373, | |
| "learning_rate": 2.4221453287197232e-06, | |
| "loss": 0.7667, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 2.0730526163013345, | |
| "learning_rate": 2.508650519031142e-06, | |
| "loss": 0.7734, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 2.071966255411125, | |
| "learning_rate": 2.5951557093425604e-06, | |
| "loss": 0.7558, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.140324284318863, | |
| "learning_rate": 2.6816608996539796e-06, | |
| "loss": 0.7453, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 2.0794979181710866, | |
| "learning_rate": 2.7681660899653983e-06, | |
| "loss": 0.7413, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 2.079407078921554, | |
| "learning_rate": 2.8546712802768167e-06, | |
| "loss": 0.7331, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 2.0051529944042445, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 0.751, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 2.075289040593278, | |
| "learning_rate": 3.027681660899654e-06, | |
| "loss": 0.7369, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.009911706722013, | |
| "learning_rate": 3.114186851211073e-06, | |
| "loss": 0.7239, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 2.114683663985779, | |
| "learning_rate": 3.200692041522492e-06, | |
| "loss": 0.7219, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.1266461925060343, | |
| "learning_rate": 3.28719723183391e-06, | |
| "loss": 0.7163, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 2.0699467756030048, | |
| "learning_rate": 3.373702422145329e-06, | |
| "loss": 0.7246, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.9178177043552207, | |
| "learning_rate": 3.4602076124567473e-06, | |
| "loss": 0.715, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.9401101711848872, | |
| "learning_rate": 3.5467128027681665e-06, | |
| "loss": 0.6888, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 1.7637994756169328, | |
| "learning_rate": 3.6332179930795853e-06, | |
| "loss": 0.7095, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 1.8879618486444918, | |
| "learning_rate": 3.7197231833910036e-06, | |
| "loss": 0.7159, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 1.7360031299470728, | |
| "learning_rate": 3.8062283737024224e-06, | |
| "loss": 0.7111, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 1.6156910333813757, | |
| "learning_rate": 3.892733564013842e-06, | |
| "loss": 0.7136, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 1.5128961988226293, | |
| "learning_rate": 3.9792387543252595e-06, | |
| "loss": 0.7199, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 1.5314262584578413, | |
| "learning_rate": 4.065743944636678e-06, | |
| "loss": 0.7051, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.7503445297432856, | |
| "learning_rate": 4.152249134948097e-06, | |
| "loss": 0.6951, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.727205989077596, | |
| "learning_rate": 4.238754325259516e-06, | |
| "loss": 0.6967, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 1.512173137191571, | |
| "learning_rate": 4.325259515570935e-06, | |
| "loss": 0.6951, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1.5906205947939107, | |
| "learning_rate": 4.411764705882353e-06, | |
| "loss": 0.7177, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 1.4011742158056826, | |
| "learning_rate": 4.498269896193772e-06, | |
| "loss": 0.6955, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 1.3418261511629705, | |
| "learning_rate": 4.58477508650519e-06, | |
| "loss": 0.6845, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 1.5187095864301963, | |
| "learning_rate": 4.67128027681661e-06, | |
| "loss": 0.7068, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.62902883041799, | |
| "learning_rate": 4.7577854671280285e-06, | |
| "loss": 0.6937, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.3634212608530258, | |
| "learning_rate": 4.8442906574394464e-06, | |
| "loss": 0.6858, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 1.5038430309774804, | |
| "learning_rate": 4.930795847750865e-06, | |
| "loss": 0.6988, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.3823728989860273, | |
| "learning_rate": 4.999998176402445e-06, | |
| "loss": 0.6895, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.7851921485205167, | |
| "learning_rate": 4.999934350767369e-06, | |
| "loss": 0.6972, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 1.5012843187747695, | |
| "learning_rate": 4.99977934791494e-06, | |
| "loss": 0.6909, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1.4846464217080884, | |
| "learning_rate": 4.9995331734983985e-06, | |
| "loss": 0.6784, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.4416671840566515, | |
| "learning_rate": 4.999195836496179e-06, | |
| "loss": 0.6885, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.3968335334596063, | |
| "learning_rate": 4.998767349211587e-06, | |
| "loss": 0.6632, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1.4727540225441376, | |
| "learning_rate": 4.9982477272723404e-06, | |
| "loss": 0.6997, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 2.0150510241724295, | |
| "learning_rate": 4.997636989630012e-06, | |
| "loss": 0.681, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 1.5786095690969104, | |
| "learning_rate": 4.996935158559329e-06, | |
| "loss": 0.7052, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.4930115902628418, | |
| "learning_rate": 4.9961422596573674e-06, | |
| "loss": 0.683, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 1.417488194165555, | |
| "learning_rate": 4.995258321842611e-06, | |
| "loss": 0.6724, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1.5814802052424026, | |
| "learning_rate": 4.994283377353902e-06, | |
| "loss": 0.708, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 1.626327112384558, | |
| "learning_rate": 4.993217461749266e-06, | |
| "loss": 0.6769, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 1.4561720499145596, | |
| "learning_rate": 4.992060613904611e-06, | |
| "loss": 0.6827, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.4209579264748098, | |
| "learning_rate": 4.990812876012311e-06, | |
| "loss": 0.6843, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1.3963649914734755, | |
| "learning_rate": 4.989474293579669e-06, | |
| "loss": 0.678, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 1.5340840894447103, | |
| "learning_rate": 4.988044915427257e-06, | |
| "loss": 0.6753, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1.4322002446736155, | |
| "learning_rate": 4.986524793687131e-06, | |
| "loss": 0.6913, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 1.494759555714755, | |
| "learning_rate": 4.984913983800936e-06, | |
| "loss": 0.6601, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 1.5286688918221005, | |
| "learning_rate": 4.983212544517878e-06, | |
| "loss": 0.6708, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 1.52282481083145, | |
| "learning_rate": 4.9814205378925895e-06, | |
| "loss": 0.6582, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1.6753513410463947, | |
| "learning_rate": 4.979538029282856e-06, | |
| "loss": 0.679, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 1.5212718767367754, | |
| "learning_rate": 4.977565087347239e-06, | |
| "loss": 0.667, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 1.4524081871963455, | |
| "learning_rate": 4.975501784042572e-06, | |
| "loss": 0.6707, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 1.7921858867602187, | |
| "learning_rate": 4.973348194621329e-06, | |
| "loss": 0.6741, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 1.5845517579748196, | |
| "learning_rate": 4.9711043976288935e-06, | |
| "loss": 0.6791, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1.4343033765555364, | |
| "learning_rate": 4.968770474900677e-06, | |
| "loss": 0.6607, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1.5295122616615218, | |
| "learning_rate": 4.966346511559149e-06, | |
| "loss": 0.6624, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1.4896437285857649, | |
| "learning_rate": 4.963832596010723e-06, | |
| "loss": 0.6677, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 1.5044244825090314, | |
| "learning_rate": 4.961228819942538e-06, | |
| "loss": 0.6722, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.4992891216972632, | |
| "learning_rate": 4.958535278319109e-06, | |
| "loss": 0.6529, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 1.5676366345308723, | |
| "learning_rate": 4.955752069378867e-06, | |
| "loss": 0.6682, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 1.566761422297736, | |
| "learning_rate": 4.952879294630577e-06, | |
| "loss": 0.6674, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.593607408287518, | |
| "learning_rate": 4.949917058849635e-06, | |
| "loss": 0.6685, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.4655841734364505, | |
| "learning_rate": 4.946865470074241e-06, | |
| "loss": 0.6706, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.4680095746651978, | |
| "learning_rate": 4.9437246396014695e-06, | |
| "loss": 0.6503, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1.6414158311618077, | |
| "learning_rate": 4.940494681983201e-06, | |
| "loss": 0.6483, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 1.647070417024019, | |
| "learning_rate": 4.9371757150219475e-06, | |
| "loss": 0.6708, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 1.5786850858034276, | |
| "learning_rate": 4.933767859766557e-06, | |
| "loss": 0.6406, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.3967395390307664, | |
| "learning_rate": 4.930271240507797e-06, | |
| "loss": 0.6469, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 1.863770908509767, | |
| "learning_rate": 4.926685984773822e-06, | |
| "loss": 0.6721, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 1.4091042690875146, | |
| "learning_rate": 4.923012223325522e-06, | |
| "loss": 0.6418, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 1.4211964673604418, | |
| "learning_rate": 4.919250090151754e-06, | |
| "loss": 0.6384, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 1.5350233939025446, | |
| "learning_rate": 4.915399722464455e-06, | |
| "loss": 0.6365, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.5608101804229912, | |
| "learning_rate": 4.911461260693639e-06, | |
| "loss": 0.6349, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 1.471894754845902, | |
| "learning_rate": 4.907434848482268e-06, | |
| "loss": 0.6324, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.5341990166130748, | |
| "learning_rate": 4.903320632681026e-06, | |
| "loss": 0.6434, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.4680803495559631, | |
| "learning_rate": 4.899118763342952e-06, | |
| "loss": 0.6466, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 1.786833703301667, | |
| "learning_rate": 4.89482939371797e-06, | |
| "loss": 0.6599, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 1.5749569225927729, | |
| "learning_rate": 4.890452680247305e-06, | |
| "loss": 0.641, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 1.3756860207273784, | |
| "learning_rate": 4.885988782557771e-06, | |
| "loss": 0.6372, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.6093553203322744, | |
| "learning_rate": 4.88143786345595e-06, | |
| "loss": 0.6053, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1.417201052972172, | |
| "learning_rate": 4.876800088922257e-06, | |
| "loss": 0.6288, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.512172943891037, | |
| "learning_rate": 4.872075628104884e-06, | |
| "loss": 0.6352, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 1.6791678410000617, | |
| "learning_rate": 4.867264653313633e-06, | |
| "loss": 0.6473, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 1.4591202016380542, | |
| "learning_rate": 4.8623673400136295e-06, | |
| "loss": 0.6364, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.5298614351963449, | |
| "learning_rate": 4.8573838668189225e-06, | |
| "loss": 0.6295, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.5276546684651868, | |
| "learning_rate": 4.852314415485973e-06, | |
| "loss": 0.6331, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.622621476650238, | |
| "eval_runtime": 330.1558, | |
| "eval_samples_per_second": 22.78, | |
| "eval_steps_per_second": 0.357, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.564141000897205, | |
| "learning_rate": 4.847159170907022e-06, | |
| "loss": 0.5882, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 1.702572539352643, | |
| "learning_rate": 4.841918321103349e-06, | |
| "loss": 0.5574, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 1.8542783824172295, | |
| "learning_rate": 4.836592057218416e-06, | |
| "loss": 0.5321, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 1.6168277685945358, | |
| "learning_rate": 4.83118057351089e-06, | |
| "loss": 0.5301, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 1.7040979617839176, | |
| "learning_rate": 4.825684067347565e-06, | |
| "loss": 0.5369, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 1.6826187176546037, | |
| "learning_rate": 4.82010273919616e-06, | |
| "loss": 0.521, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 1.7537542015921341, | |
| "learning_rate": 4.8144367926180084e-06, | |
| "loss": 0.5362, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 1.5596851176953166, | |
| "learning_rate": 4.8086864342606345e-06, | |
| "loss": 0.5421, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 1.612414749994373, | |
| "learning_rate": 4.802851873850214e-06, | |
| "loss": 0.5542, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 1.6043297699907004, | |
| "learning_rate": 4.796933324183928e-06, | |
| "loss": 0.526, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 1.5705483698698184, | |
| "learning_rate": 4.790931001122202e-06, | |
| "loss": 0.5267, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 1.6281592397386129, | |
| "learning_rate": 4.784845123580829e-06, | |
| "loss": 0.5367, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 1.5966059820228211, | |
| "learning_rate": 4.77867591352299e-06, | |
| "loss": 0.535, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 1.8510817844041834, | |
| "learning_rate": 4.7724235959511566e-06, | |
| "loss": 0.5495, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 1.8997498768876209, | |
| "learning_rate": 4.766088398898884e-06, | |
| "loss": 0.5403, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 1.7160489659947078, | |
| "learning_rate": 4.759670553422494e-06, | |
| "loss": 0.5347, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 1.7971855635226874, | |
| "learning_rate": 4.7531702935926505e-06, | |
| "loss": 0.537, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 1.7064772479008743, | |
| "learning_rate": 4.74658785648582e-06, | |
| "loss": 0.5579, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 1.4680823102221994, | |
| "learning_rate": 4.7399234821756285e-06, | |
| "loss": 0.5276, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 2.0480120228699814, | |
| "learning_rate": 4.733177413724098e-06, | |
| "loss": 0.5417, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 4.470377211022557, | |
| "learning_rate": 4.726349897172791e-06, | |
| "loss": 0.5292, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 1.7455236825499205, | |
| "learning_rate": 4.71944118153383e-06, | |
| "loss": 0.5486, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 1.6229800058296617, | |
| "learning_rate": 4.7124515187808186e-06, | |
| "loss": 0.5403, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 1.6120491331538838, | |
| "learning_rate": 4.705381163839652e-06, | |
| "loss": 0.5212, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 1.5707477382476385, | |
| "learning_rate": 4.698230374579216e-06, | |
| "loss": 0.513, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 1.8038956520870928, | |
| "learning_rate": 4.690999411801987e-06, | |
| "loss": 0.5505, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.4977964665899992, | |
| "learning_rate": 4.683688539234515e-06, | |
| "loss": 0.5447, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 1.64524294757822, | |
| "learning_rate": 4.676298023517811e-06, | |
| "loss": 0.5252, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.8439638298190486, | |
| "learning_rate": 4.668828134197615e-06, | |
| "loss": 0.5288, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.6551302109627803, | |
| "learning_rate": 4.66127914371457e-06, | |
| "loss": 0.5042, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 1.8432213805384519, | |
| "learning_rate": 4.653651327394284e-06, | |
| "loss": 0.5075, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 1.7991942501765485, | |
| "learning_rate": 4.64594496343729e-06, | |
| "loss": 0.5234, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 1.560979934022783, | |
| "learning_rate": 4.638160332908896e-06, | |
| "loss": 0.5076, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 1.5466509824875485, | |
| "learning_rate": 4.630297719728939e-06, | |
| "loss": 0.5204, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 1.7134330421496313, | |
| "learning_rate": 4.622357410661425e-06, | |
| "loss": 0.5136, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.9274332369383436, | |
| "learning_rate": 4.6143396953040735e-06, | |
| "loss": 0.5121, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 1.4937551003793812, | |
| "learning_rate": 4.606244866077752e-06, | |
| "loss": 0.527, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 2.658551374926485, | |
| "learning_rate": 4.598073218215817e-06, | |
| "loss": 0.5458, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 2.224040870079962, | |
| "learning_rate": 4.5898250497533395e-06, | |
| "loss": 0.5436, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 2.7319048521917635, | |
| "learning_rate": 4.581500661516237e-06, | |
| "loss": 0.5313, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 1.6998409225129496, | |
| "learning_rate": 4.573100357110306e-06, | |
| "loss": 0.5173, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 1.5581752222534802, | |
| "learning_rate": 4.564624442910144e-06, | |
| "loss": 0.5301, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 1.6962356595619175, | |
| "learning_rate": 4.556073228047977e-06, | |
| "loss": 0.5313, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 1.8219089406617814, | |
| "learning_rate": 4.547447024402385e-06, | |
| "loss": 0.5335, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 1.62352160937354, | |
| "learning_rate": 4.5387461465869275e-06, | |
| "loss": 0.5459, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 1.5257648113589735, | |
| "learning_rate": 4.529970911938669e-06, | |
| "loss": 0.5111, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 2.5973864011536416, | |
| "learning_rate": 4.521121640506605e-06, | |
| "loss": 0.5352, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 2.1347665405267384, | |
| "learning_rate": 4.512198655039988e-06, | |
| "loss": 0.5208, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 1.8537528730656059, | |
| "learning_rate": 4.503202280976557e-06, | |
| "loss": 0.5064, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 2.1504491301566784, | |
| "learning_rate": 4.494132846430672e-06, | |
| "loss": 0.5186, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 14.695605966854037, | |
| "learning_rate": 4.4849906821813406e-06, | |
| "loss": 0.5349, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 2.063212121669853, | |
| "learning_rate": 4.475776121660156e-06, | |
| "loss": 0.531, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 1.834832161110646, | |
| "learning_rate": 4.466489500939138e-06, | |
| "loss": 0.5143, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 1.7633276487477196, | |
| "learning_rate": 4.457131158718478e-06, | |
| "loss": 0.5331, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 1.809450693070927, | |
| "learning_rate": 4.447701436314176e-06, | |
| "loss": 0.5283, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 1.6601361692100707, | |
| "learning_rate": 4.438200677645607e-06, | |
| "loss": 0.5371, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 1.6494075536487005, | |
| "learning_rate": 4.428629229222962e-06, | |
| "loss": 0.5334, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 1.5995037205122475, | |
| "learning_rate": 4.418987440134624e-06, | |
| "loss": 0.5171, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 1.6352796020571934, | |
| "learning_rate": 4.4092756620344246e-06, | |
| "loss": 0.5148, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 7.045247439882991, | |
| "learning_rate": 4.3994942491288275e-06, | |
| "loss": 1.7556, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 6.128928116038949, | |
| "learning_rate": 4.389643558164005e-06, | |
| "loss": 0.5763, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 36.49282493029366, | |
| "learning_rate": 4.379723948412828e-06, | |
| "loss": 0.6, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 9.579002986726392, | |
| "learning_rate": 4.369735781661763e-06, | |
| "loss": 0.6819, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 12.266502746290532, | |
| "learning_rate": 4.3596794221976755e-06, | |
| "loss": 0.6053, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 3.9630821498228874, | |
| "learning_rate": 4.349555236794546e-06, | |
| "loss": 0.6229, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 3.4771323398284566, | |
| "learning_rate": 4.339363594700093e-06, | |
| "loss": 0.585, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 2.201457058274829, | |
| "learning_rate": 4.329104867622302e-06, | |
| "loss": 0.5611, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 1.9654853072206555, | |
| "learning_rate": 4.3187794297158765e-06, | |
| "loss": 0.5648, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 4.227758915564248, | |
| "learning_rate": 4.308387657568582e-06, | |
| "loss": 0.5481, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 1.7148546445696602, | |
| "learning_rate": 4.297929930187519e-06, | |
| "loss": 0.548, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 1.7282534502834466, | |
| "learning_rate": 4.287406628985295e-06, | |
| "loss": 0.5669, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 1.5604162135037978, | |
| "learning_rate": 4.276818137766118e-06, | |
| "loss": 0.5357, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 1.7385091646846929, | |
| "learning_rate": 4.266164842711797e-06, | |
| "loss": 0.5371, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 1.8757343048736053, | |
| "learning_rate": 4.255447132367651e-06, | |
| "loss": 0.5356, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 1.5821779705858339, | |
| "learning_rate": 4.2446653976283485e-06, | |
| "loss": 0.534, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 1.7181301166792315, | |
| "learning_rate": 4.233820031723644e-06, | |
| "loss": 0.5192, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "grad_norm": 1.7620628413002903, | |
| "learning_rate": 4.22291143020404e-06, | |
| "loss": 0.527, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 1.588909360094358, | |
| "learning_rate": 4.211939990926355e-06, | |
| "loss": 0.5205, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 1.6838208492860955, | |
| "learning_rate": 4.200906114039219e-06, | |
| "loss": 0.5298, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 1.9081461502343906, | |
| "learning_rate": 4.1898102019684765e-06, | |
| "loss": 0.502, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 1.4961424837742658, | |
| "learning_rate": 4.1786526594025065e-06, | |
| "loss": 0.5372, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 1.6252683024976837, | |
| "learning_rate": 4.167433893277473e-06, | |
| "loss": 0.5269, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 1.5490581530552676, | |
| "learning_rate": 4.156154312762467e-06, | |
| "loss": 0.5285, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 1.8061470924061633, | |
| "learning_rate": 4.1448143292446e-06, | |
| "loss": 0.5503, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 1.5936238911397005, | |
| "learning_rate": 4.133414356313988e-06, | |
| "loss": 0.535, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 1.5524395421750514, | |
| "learning_rate": 4.121954809748675e-06, | |
| "loss": 0.5278, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 1.5740976269214233, | |
| "learning_rate": 4.110436107499462e-06, | |
| "loss": 0.5197, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 1.5214013327972284, | |
| "learning_rate": 4.098858669674667e-06, | |
| "loss": 0.5202, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 1.5237741216662357, | |
| "learning_rate": 4.087222918524807e-06, | |
| "loss": 0.5088, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 1.6738078012815194, | |
| "learning_rate": 4.07552927842719e-06, | |
| "loss": 0.5166, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 1.5517613536326162, | |
| "learning_rate": 4.06377817587044e-06, | |
| "loss": 0.4918, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "grad_norm": 1.431747331311476, | |
| "learning_rate": 4.051970039438945e-06, | |
| "loss": 0.5133, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 1.4767269458087147, | |
| "learning_rate": 4.0401052997972235e-06, | |
| "loss": 0.5033, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 1.5180384694427076, | |
| "learning_rate": 4.028184389674216e-06, | |
| "loss": 0.5058, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 1.5889385254960862, | |
| "learning_rate": 4.016207743847505e-06, | |
| "loss": 0.515, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "grad_norm": 1.5532809579850755, | |
| "learning_rate": 4.004175799127458e-06, | |
| "loss": 0.5252, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "grad_norm": 1.5669745733483815, | |
| "learning_rate": 3.992088994341292e-06, | |
| "loss": 0.5237, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 1.7971179594367128, | |
| "learning_rate": 3.979947770317077e-06, | |
| "loss": 0.5401, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 1.6564666981215765, | |
| "learning_rate": 3.967752569867645e-06, | |
| "loss": 0.5315, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 1.596740218503028, | |
| "learning_rate": 3.955503837774458e-06, | |
| "loss": 0.5124, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "grad_norm": 1.5508137957824288, | |
| "learning_rate": 3.943202020771367e-06, | |
| "loss": 0.5219, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 1.7093498845192874, | |
| "learning_rate": 3.9308475675283326e-06, | |
| "loss": 0.5265, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 1.5364211220041548, | |
| "learning_rate": 3.9184409286350535e-06, | |
| "loss": 0.5249, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 1.588131611670702, | |
| "learning_rate": 3.905982556584537e-06, | |
| "loss": 0.5132, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 1.4898401463528494, | |
| "learning_rate": 3.893472905756593e-06, | |
| "loss": 0.5241, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 1.655966866153354, | |
| "learning_rate": 3.880912432401265e-06, | |
| "loss": 0.5098, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 1.4977265046553758, | |
| "learning_rate": 3.868301594622185e-06, | |
| "loss": 0.5189, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "grad_norm": 1.45173341143097, | |
| "learning_rate": 3.855640852359869e-06, | |
| "loss": 0.4974, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 1.8226392308673234, | |
| "learning_rate": 3.842930667374945e-06, | |
| "loss": 0.5122, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 1.479895371145679, | |
| "learning_rate": 3.830171503231302e-06, | |
| "loss": 0.5077, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 1.8170301741742978, | |
| "learning_rate": 3.817363825279195e-06, | |
| "loss": 0.5174, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 1.6039721848461628, | |
| "learning_rate": 3.8045081006382668e-06, | |
| "loss": 0.5145, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 2.198046435353487, | |
| "learning_rate": 3.7916047981805058e-06, | |
| "loss": 0.507, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 2.538669249099628, | |
| "learning_rate": 3.778654388513159e-06, | |
| "loss": 0.4926, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "grad_norm": 1.490519700011853, | |
| "learning_rate": 3.765657343961553e-06, | |
| "loss": 0.5091, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.609088164054738, | |
| "learning_rate": 3.7526141385518794e-06, | |
| "loss": 0.51, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.5023414492607117, | |
| "eval_runtime": 329.8071, | |
| "eval_samples_per_second": 22.804, | |
| "eval_steps_per_second": 0.358, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "grad_norm": 2.3682882817547433, | |
| "learning_rate": 3.739525247993901e-06, | |
| "loss": 0.4345, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 2.198186338081821, | |
| "learning_rate": 3.7263911496636003e-06, | |
| "loss": 0.4221, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 1.9622396068630974, | |
| "learning_rate": 3.71321232258577e-06, | |
| "loss": 0.4166, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "grad_norm": 1.9405841953564362, | |
| "learning_rate": 3.6999892474165477e-06, | |
| "loss": 0.4124, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 1.6911138242049408, | |
| "learning_rate": 3.6867224064258754e-06, | |
| "loss": 0.3999, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 1.805794848826172, | |
| "learning_rate": 3.673412283479916e-06, | |
| "loss": 0.3923, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 1.7227193868132855, | |
| "learning_rate": 3.660059364023409e-06, | |
| "loss": 0.3922, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "grad_norm": 1.5633624388497611, | |
| "learning_rate": 3.6466641350619564e-06, | |
| "loss": 0.4081, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 1.6049499642389522, | |
| "learning_rate": 3.6332270851442697e-06, | |
| "loss": 0.3988, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 1.6302545860912692, | |
| "learning_rate": 3.619748704344346e-06, | |
| "loss": 0.4153, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "grad_norm": 1.893580997382411, | |
| "learning_rate": 3.6062294842435958e-06, | |
| "loss": 0.4118, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 1.591008467389523, | |
| "learning_rate": 3.5926699179129156e-06, | |
| "loss": 0.3967, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "grad_norm": 1.8812505523353902, | |
| "learning_rate": 3.579070499894703e-06, | |
| "loss": 0.4105, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 1.7312982189755863, | |
| "learning_rate": 3.5654317261848182e-06, | |
| "loss": 0.4088, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "grad_norm": 1.715794291601741, | |
| "learning_rate": 3.5517540942144985e-06, | |
| "loss": 0.3837, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 1.571536156806173, | |
| "learning_rate": 3.538038102832213e-06, | |
| "loss": 0.4047, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "grad_norm": 1.6926403850601441, | |
| "learning_rate": 3.5242842522854687e-06, | |
| "loss": 0.4097, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "grad_norm": 1.5865520678177705, | |
| "learning_rate": 3.5104930442025665e-06, | |
| "loss": 0.3988, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 1.6304875271885666, | |
| "learning_rate": 3.4966649815743055e-06, | |
| "loss": 0.4002, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "grad_norm": 1.5709135265237464, | |
| "learning_rate": 3.4828005687356382e-06, | |
| "loss": 0.4115, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 1.6424713964605273, | |
| "learning_rate": 3.4689003113472754e-06, | |
| "loss": 0.3978, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 1.8859757363715237, | |
| "learning_rate": 3.454964716377247e-06, | |
| "loss": 0.4203, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 1.637179971274298, | |
| "learning_rate": 3.440994292082408e-06, | |
| "loss": 0.3908, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 25.023011080924313, | |
| "learning_rate": 3.4269895479899023e-06, | |
| "loss": 0.4038, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 2.830979911281441, | |
| "learning_rate": 3.412950994878582e-06, | |
| "loss": 0.4047, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 1.8263955076269602, | |
| "learning_rate": 3.3988791447603763e-06, | |
| "loss": 0.3963, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 1.6584308792186484, | |
| "learning_rate": 3.3847745108616157e-06, | |
| "loss": 0.4049, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 1.868836216780196, | |
| "learning_rate": 3.3706376076043183e-06, | |
| "loss": 0.4123, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 1.704922866865318, | |
| "learning_rate": 3.356468950587424e-06, | |
| "loss": 0.3919, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 1.8511972094459495, | |
| "learning_rate": 3.3422690565679895e-06, | |
| "loss": 0.4118, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 1.5826425092029799, | |
| "learning_rate": 3.3280384434423447e-06, | |
| "loss": 0.4072, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "grad_norm": 2.8458291903299555, | |
| "learning_rate": 3.3137776302272017e-06, | |
| "loss": 0.4044, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "grad_norm": 2.696905442165675, | |
| "learning_rate": 3.2994871370407242e-06, | |
| "loss": 0.405, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 1.8719365939252683, | |
| "learning_rate": 3.2851674850835615e-06, | |
| "loss": 0.4091, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "grad_norm": 3.602956750004364, | |
| "learning_rate": 3.2708191966198344e-06, | |
| "loss": 0.3845, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 1.7107089177149095, | |
| "learning_rate": 3.2564427949580907e-06, | |
| "loss": 0.4128, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 1.782125003384775, | |
| "learning_rate": 3.2420388044322204e-06, | |
| "loss": 0.3957, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "grad_norm": 1.7157313501617861, | |
| "learning_rate": 3.2276077503823276e-06, | |
| "loss": 0.4164, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 1.6189210159528598, | |
| "learning_rate": 3.213150159135575e-06, | |
| "loss": 0.4017, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 1.7965605988243851, | |
| "learning_rate": 3.198666557986984e-06, | |
| "loss": 0.3995, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 1.6750382286037944, | |
| "learning_rate": 3.184157475180208e-06, | |
| "loss": 0.4065, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 1.7466143412529116, | |
| "learning_rate": 3.169623439888259e-06, | |
| "loss": 0.3913, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 1.7094098689820845, | |
| "learning_rate": 3.1550649821942165e-06, | |
| "loss": 0.4055, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 1.6546779771359428, | |
| "learning_rate": 3.1404826330718884e-06, | |
| "loss": 0.4072, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 1.5822130245289208, | |
| "learning_rate": 3.1258769243664457e-06, | |
| "loss": 0.396, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 1.6175930524108444, | |
| "learning_rate": 3.1112483887750285e-06, | |
| "loss": 0.4009, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 1.6572189285469252, | |
| "learning_rate": 3.0965975598273147e-06, | |
| "loss": 0.4135, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "grad_norm": 1.7351755497284653, | |
| "learning_rate": 3.0819249718660625e-06, | |
| "loss": 0.4077, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 1.722331718347925, | |
| "learning_rate": 3.0672311600276205e-06, | |
| "loss": 0.3961, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 2.8787903703341704, | |
| "learning_rate": 3.052516660222412e-06, | |
| "loss": 0.4013, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 1.7609799996300464, | |
| "learning_rate": 3.0377820091153894e-06, | |
| "loss": 0.4182, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 1.6624761502383696, | |
| "learning_rate": 3.0230277441064576e-06, | |
| "loss": 0.4174, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 1.6422107924968583, | |
| "learning_rate": 3.0082544033108792e-06, | |
| "loss": 0.3968, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 1.6979969793176557, | |
| "learning_rate": 2.993462525539646e-06, | |
| "loss": 0.3893, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 2.0089020705022027, | |
| "learning_rate": 2.9786526502798245e-06, | |
| "loss": 0.401, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 2.4328077084567035, | |
| "learning_rate": 2.9638253176748868e-06, | |
| "loss": 0.4041, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 2.1658392118895096, | |
| "learning_rate": 2.948981068505004e-06, | |
| "loss": 0.4152, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 1.9075735219009673, | |
| "learning_rate": 2.9341204441673267e-06, | |
| "loss": 0.4019, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 79.58152669185955, | |
| "learning_rate": 2.9192439866562363e-06, | |
| "loss": 0.3986, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 1.9783510669105098, | |
| "learning_rate": 2.904352238543582e-06, | |
| "loss": 0.3977, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 2.0408912753861452, | |
| "learning_rate": 2.889445742958886e-06, | |
| "loss": 0.395, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 2.1287347922745563, | |
| "learning_rate": 2.8745250435695433e-06, | |
| "loss": 0.3944, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "grad_norm": 1.9220277314654948, | |
| "learning_rate": 2.859590684560982e-06, | |
| "loss": 0.4066, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 1.7440202984118327, | |
| "learning_rate": 2.844643210616824e-06, | |
| "loss": 0.3933, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 4.515419055182269, | |
| "learning_rate": 2.8296831668990195e-06, | |
| "loss": 0.408, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 1.804786297673, | |
| "learning_rate": 2.8147110990279576e-06, | |
| "loss": 0.3895, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 1.8010185425321945, | |
| "learning_rate": 2.7997275530625715e-06, | |
| "loss": 0.4007, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 1.7762627094164511, | |
| "learning_rate": 2.7847330754804257e-06, | |
| "loss": 0.4047, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 1.713415455379404, | |
| "learning_rate": 2.7697282131577752e-06, | |
| "loss": 0.3981, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 1.8140754605201832, | |
| "learning_rate": 2.754713513349627e-06, | |
| "loss": 0.3976, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 1.925171255479977, | |
| "learning_rate": 2.7396895236697833e-06, | |
| "loss": 0.3839, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 1.6916338723474953, | |
| "learning_rate": 2.724656792070861e-06, | |
| "loss": 0.4009, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "grad_norm": 1.7468445915903061, | |
| "learning_rate": 2.7096158668243105e-06, | |
| "loss": 0.3954, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 1.7964976583564236, | |
| "learning_rate": 2.6945672965004236e-06, | |
| "loss": 0.3943, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "grad_norm": 1.6933144469510995, | |
| "learning_rate": 2.679511629948319e-06, | |
| "loss": 0.3958, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 1.7617290180955545, | |
| "learning_rate": 2.664449416275928e-06, | |
| "loss": 0.3926, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 1.6597635128518158, | |
| "learning_rate": 2.64938120482997e-06, | |
| "loss": 0.4052, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 2.1579541798227213, | |
| "learning_rate": 2.6343075451759098e-06, | |
| "loss": 0.3962, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 1.6871485325523872, | |
| "learning_rate": 2.619228987077923e-06, | |
| "loss": 0.3975, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 2.0167255639712436, | |
| "learning_rate": 2.604146080478839e-06, | |
| "loss": 0.3794, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 1.8181334973476126, | |
| "learning_rate": 2.5890593754800826e-06, | |
| "loss": 0.4061, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "grad_norm": 1.753744913485219, | |
| "learning_rate": 2.573969422321617e-06, | |
| "loss": 0.386, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 1.7053212636407111, | |
| "learning_rate": 2.558876771361869e-06, | |
| "loss": 0.3835, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 2.0262024426335903, | |
| "learning_rate": 2.5437819730576608e-06, | |
| "loss": 0.3925, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 1.7593848195979234, | |
| "learning_rate": 2.528685577944131e-06, | |
| "loss": 0.3856, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 1.7817581606319834, | |
| "learning_rate": 2.5135881366146585e-06, | |
| "loss": 0.379, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 1.9582822708483771, | |
| "learning_rate": 2.498490199700778e-06, | |
| "loss": 0.4028, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 1.737025758915202, | |
| "learning_rate": 2.4833923178520984e-06, | |
| "loss": 0.3918, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 1.8457649877934572, | |
| "learning_rate": 2.468295041716223e-06, | |
| "loss": 0.3869, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "grad_norm": 1.7058419645908771, | |
| "learning_rate": 2.4531989219186626e-06, | |
| "loss": 0.3969, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "grad_norm": 1.7424467938884878, | |
| "learning_rate": 2.4381045090427525e-06, | |
| "loss": 0.3944, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "grad_norm": 1.7229435032812541, | |
| "learning_rate": 2.4230123536095746e-06, | |
| "loss": 0.392, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 1.6940195802109876, | |
| "learning_rate": 2.4079230060578787e-06, | |
| "loss": 0.3921, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "grad_norm": 1.926203653659556, | |
| "learning_rate": 2.3928370167240037e-06, | |
| "loss": 0.3772, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 2.3907546286805443, | |
| "learning_rate": 2.3777549358218105e-06, | |
| "loss": 0.3829, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "grad_norm": 1.7061727789752485, | |
| "learning_rate": 2.3626773134226114e-06, | |
| "loss": 0.3821, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 4.31283876155316, | |
| "learning_rate": 2.3476046994351085e-06, | |
| "loss": 0.3885, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 1.815971909695708, | |
| "learning_rate": 2.332537643585337e-06, | |
| "loss": 0.4065, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 1.8273993097485424, | |
| "learning_rate": 2.3174766953966178e-06, | |
| "loss": 0.403, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 2.3245389527585347, | |
| "learning_rate": 2.302422404169516e-06, | |
| "loss": 0.3813, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "grad_norm": 2.5008928168129305, | |
| "learning_rate": 2.2873753189618016e-06, | |
| "loss": 0.3958, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 1.7982648722278374, | |
| "learning_rate": 2.272335988568429e-06, | |
| "loss": 0.3939, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "grad_norm": 1.7113353706744565, | |
| "learning_rate": 2.257304961501521e-06, | |
| "loss": 0.3831, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 1.8571633665118874, | |
| "learning_rate": 2.242282785970361e-06, | |
| "loss": 0.4036, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 1.6973636101504117, | |
| "learning_rate": 2.2272700098614022e-06, | |
| "loss": 0.3897, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 1.8121544488710108, | |
| "learning_rate": 2.2122671807182847e-06, | |
| "loss": 0.3997, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 1.7018502297212512, | |
| "learning_rate": 2.1972748457218608e-06, | |
| "loss": 0.3907, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 1.5654175282292644, | |
| "learning_rate": 2.182293551670245e-06, | |
| "loss": 0.3849, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 1.8131129781015132, | |
| "learning_rate": 2.167323844958867e-06, | |
| "loss": 0.3991, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "grad_norm": 1.6556951661263792, | |
| "learning_rate": 2.152366271560543e-06, | |
| "loss": 0.3876, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 1.6927349202503557, | |
| "learning_rate": 2.1374213770055692e-06, | |
| "loss": 0.3942, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "grad_norm": 1.8331415491875973, | |
| "learning_rate": 2.122489706361818e-06, | |
| "loss": 0.3791, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 1.8546912165375895, | |
| "learning_rate": 2.107571804214861e-06, | |
| "loss": 0.3863, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 1.8701558336845174, | |
| "learning_rate": 2.0926682146481094e-06, | |
| "loss": 0.3918, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 1.6438729463581823, | |
| "learning_rate": 2.077779481222967e-06, | |
| "loss": 0.4058, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.41721072793006897, | |
| "eval_runtime": 329.8932, | |
| "eval_samples_per_second": 22.798, | |
| "eval_steps_per_second": 0.358, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3.362012279132523, | |
| "learning_rate": 2.0629061469590046e-06, | |
| "loss": 0.3616, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "grad_norm": 2.405517474268548, | |
| "learning_rate": 2.048048754314162e-06, | |
| "loss": 0.3281, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "grad_norm": 2.252602528540252, | |
| "learning_rate": 2.0332078451649557e-06, | |
| "loss": 0.3167, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "grad_norm": 1.8994332344961664, | |
| "learning_rate": 2.0183839607867177e-06, | |
| "loss": 0.3061, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 2.0025910547043138, | |
| "learning_rate": 2.003577641833859e-06, | |
| "loss": 0.3205, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 1.8748618784934366, | |
| "learning_rate": 1.988789428320141e-06, | |
| "loss": 0.3101, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "grad_norm": 1.8182888367049812, | |
| "learning_rate": 1.9740198595989935e-06, | |
| "loss": 0.3071, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "grad_norm": 1.9091898687921862, | |
| "learning_rate": 1.959269474343832e-06, | |
| "loss": 0.308, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 1.7973778535481801, | |
| "learning_rate": 1.9445388105284157e-06, | |
| "loss": 0.3031, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "grad_norm": 1.7575417892395457, | |
| "learning_rate": 1.9298284054072278e-06, | |
| "loss": 0.3063, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 1.8133263463584137, | |
| "learning_rate": 1.9151387954958792e-06, | |
| "loss": 0.3008, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "grad_norm": 1.892423366282607, | |
| "learning_rate": 1.9004705165515403e-06, | |
| "loss": 0.33, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "grad_norm": 3.4152436955589596, | |
| "learning_rate": 1.885824103553404e-06, | |
| "loss": 0.2967, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "grad_norm": 1.766725631679472, | |
| "learning_rate": 1.8712000906831684e-06, | |
| "loss": 0.3134, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 1.7559925086788295, | |
| "learning_rate": 1.8565990113055615e-06, | |
| "loss": 0.3117, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "grad_norm": 1.7744634591819772, | |
| "learning_rate": 1.8420213979488805e-06, | |
| "loss": 0.3188, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "grad_norm": 1.7585191953453494, | |
| "learning_rate": 1.8274677822855775e-06, | |
| "loss": 0.3052, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "grad_norm": 1.8690590027859153, | |
| "learning_rate": 1.81293869511286e-06, | |
| "loss": 0.3113, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "grad_norm": 1.8668392799746427, | |
| "learning_rate": 1.7984346663333415e-06, | |
| "loss": 0.3115, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "grad_norm": 1.8555764810481752, | |
| "learning_rate": 1.783956224935705e-06, | |
| "loss": 0.3198, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "grad_norm": 1.7812617913871858, | |
| "learning_rate": 1.7695038989754179e-06, | |
| "loss": 0.3145, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "grad_norm": 1.7431068806527559, | |
| "learning_rate": 1.7550782155554664e-06, | |
| "loss": 0.2926, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "grad_norm": 1.8768744877192187, | |
| "learning_rate": 1.7406797008071351e-06, | |
| "loss": 0.3113, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "grad_norm": 1.8667603957089496, | |
| "learning_rate": 1.7263088798708206e-06, | |
| "loss": 0.3179, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "grad_norm": 2.1299587342389072, | |
| "learning_rate": 1.7119662768768694e-06, | |
| "loss": 0.3171, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "grad_norm": 1.7486090043188862, | |
| "learning_rate": 1.6976524149264707e-06, | |
| "loss": 0.3364, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "grad_norm": 2.728922445411206, | |
| "learning_rate": 1.6833678160725746e-06, | |
| "loss": 0.3104, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 1.662819449057055, | |
| "learning_rate": 1.6691130013008514e-06, | |
| "loss": 0.3128, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "grad_norm": 1.944417319707403, | |
| "learning_rate": 1.654888490510691e-06, | |
| "loss": 0.3136, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 2.543538356158386, | |
| "learning_rate": 1.6406948024962438e-06, | |
| "loss": 0.3139, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "grad_norm": 1.8386914789973714, | |
| "learning_rate": 1.6265324549274924e-06, | |
| "loss": 0.3164, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "grad_norm": 1.8542366087207915, | |
| "learning_rate": 1.6124019643313783e-06, | |
| "loss": 0.3086, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "grad_norm": 1.774546610521189, | |
| "learning_rate": 1.59830384607296e-06, | |
| "loss": 0.3128, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "grad_norm": 2.1592993916723007, | |
| "learning_rate": 1.5842386143366157e-06, | |
| "loss": 0.3102, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "grad_norm": 1.8046821102537511, | |
| "learning_rate": 1.570206782107294e-06, | |
| "loss": 0.3156, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "grad_norm": 1.873775956948585, | |
| "learning_rate": 1.5562088611518007e-06, | |
| "loss": 0.3093, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "grad_norm": 1.8924372370403861, | |
| "learning_rate": 1.542245362000134e-06, | |
| "loss": 0.3027, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "grad_norm": 1.879324682842581, | |
| "learning_rate": 1.528316793926866e-06, | |
| "loss": 0.3263, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "grad_norm": 2.1352207068371465, | |
| "learning_rate": 1.5144236649325692e-06, | |
| "loss": 0.3163, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "grad_norm": 2.164953693631762, | |
| "learning_rate": 1.5005664817252844e-06, | |
| "loss": 0.3104, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "grad_norm": 1.6787075265475386, | |
| "learning_rate": 1.486745749702047e-06, | |
| "loss": 0.3004, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "grad_norm": 2.3266338701503537, | |
| "learning_rate": 1.4729619729304486e-06, | |
| "loss": 0.3084, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "grad_norm": 2.1291831085187827, | |
| "learning_rate": 1.4592156541302542e-06, | |
| "loss": 0.3212, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "grad_norm": 1.8834065519660157, | |
| "learning_rate": 1.4455072946550688e-06, | |
| "loss": 0.3018, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "grad_norm": 1.7558438161458612, | |
| "learning_rate": 1.4318373944740485e-06, | |
| "loss": 0.3151, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "grad_norm": 1.945827671485564, | |
| "learning_rate": 1.4182064521536693e-06, | |
| "loss": 0.3376, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "grad_norm": 1.807537024750888, | |
| "learning_rate": 1.4046149648395451e-06, | |
| "loss": 0.3275, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "grad_norm": 2.0683028288702885, | |
| "learning_rate": 1.391063428238288e-06, | |
| "loss": 0.3141, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "grad_norm": 2.0081489992818566, | |
| "learning_rate": 1.3775523365994364e-06, | |
| "loss": 0.3181, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "grad_norm": 2.056354755418753, | |
| "learning_rate": 1.3640821826974272e-06, | |
| "loss": 0.3203, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "grad_norm": 1.8074911366312743, | |
| "learning_rate": 1.3506534578136184e-06, | |
| "loss": 0.3056, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "grad_norm": 1.9405362029094402, | |
| "learning_rate": 1.3372666517183813e-06, | |
| "loss": 0.3145, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "grad_norm": 1.919499046536955, | |
| "learning_rate": 1.323922252653228e-06, | |
| "loss": 0.315, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "grad_norm": 1.9377990366808608, | |
| "learning_rate": 1.3106207473130072e-06, | |
| "loss": 0.3182, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "grad_norm": 2.342962414340334, | |
| "learning_rate": 1.2973626208281568e-06, | |
| "loss": 0.3166, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "grad_norm": 2.212340995208452, | |
| "learning_rate": 1.284148356747007e-06, | |
| "loss": 0.3141, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "grad_norm": 1.8649674715590283, | |
| "learning_rate": 1.2709784370181424e-06, | |
| "loss": 0.3077, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "grad_norm": 1.8611701311779156, | |
| "learning_rate": 1.257853341972833e-06, | |
| "loss": 0.3052, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "grad_norm": 1.7753559046449237, | |
| "learning_rate": 1.244773550307506e-06, | |
| "loss": 0.3109, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "grad_norm": 2.016998664762509, | |
| "learning_rate": 1.23173953906629e-06, | |
| "loss": 0.3023, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "grad_norm": 1.8857492837767753, | |
| "learning_rate": 1.21875178362362e-06, | |
| "loss": 0.3247, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "grad_norm": 1.868969694539886, | |
| "learning_rate": 1.205810757666894e-06, | |
| "loss": 0.3038, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "grad_norm": 1.8277877381221046, | |
| "learning_rate": 1.1929169331792018e-06, | |
| "loss": 0.3089, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "grad_norm": 1.7809098288999228, | |
| "learning_rate": 1.1800707804221068e-06, | |
| "loss": 0.3127, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "grad_norm": 1.9048483124043887, | |
| "learning_rate": 1.1672727679184984e-06, | |
| "loss": 0.3227, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "grad_norm": 1.7713895884274724, | |
| "learning_rate": 1.154523362435499e-06, | |
| "loss": 0.3149, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "grad_norm": 1.8725797518727274, | |
| "learning_rate": 1.1418230289674456e-06, | |
| "loss": 0.3021, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "grad_norm": 2.122666453139922, | |
| "learning_rate": 1.1291722307189285e-06, | |
| "loss": 0.3075, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "grad_norm": 1.9298160681079481, | |
| "learning_rate": 1.1165714290878962e-06, | |
| "loss": 0.3139, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "grad_norm": 1.8940463808973838, | |
| "learning_rate": 1.1040210836488286e-06, | |
| "loss": 0.3157, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "grad_norm": 2.061247706009593, | |
| "learning_rate": 1.091521652135976e-06, | |
| "loss": 0.3196, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "grad_norm": 1.811438510725072, | |
| "learning_rate": 1.0790735904266616e-06, | |
| "loss": 0.3148, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "grad_norm": 1.805396357161127, | |
| "learning_rate": 1.0666773525246593e-06, | |
| "loss": 0.3036, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "grad_norm": 2.045401978025098, | |
| "learning_rate": 1.0543333905436325e-06, | |
| "loss": 0.3132, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "grad_norm": 1.9050403870259225, | |
| "learning_rate": 1.0420421546906458e-06, | |
| "loss": 0.3149, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "grad_norm": 1.8742997867603652, | |
| "learning_rate": 1.0298040932497436e-06, | |
| "loss": 0.3289, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "grad_norm": 1.877012236122311, | |
| "learning_rate": 1.0176196525656032e-06, | |
| "loss": 0.3183, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "grad_norm": 1.873184037966506, | |
| "learning_rate": 1.0054892770272506e-06, | |
| "loss": 0.3092, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "grad_norm": 2.0504833760417434, | |
| "learning_rate": 9.934134090518593e-07, | |
| "loss": 0.3277, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "grad_norm": 1.9884415960242454, | |
| "learning_rate": 9.813924890686107e-07, | |
| "loss": 0.313, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "grad_norm": 1.83107837229764, | |
| "learning_rate": 9.694269555026315e-07, | |
| "loss": 0.3075, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "grad_norm": 1.764340908962033, | |
| "learning_rate": 9.575172447590032e-07, | |
| "loss": 0.3193, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "grad_norm": 1.8477826895669585, | |
| "learning_rate": 9.456637912068467e-07, | |
| "loss": 0.3125, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "grad_norm": 1.6877441904666532, | |
| "learning_rate": 9.338670271634778e-07, | |
| "loss": 0.308, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "grad_norm": 1.729827667583889, | |
| "learning_rate": 9.221273828786431e-07, | |
| "loss": 0.3002, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "grad_norm": 1.8639866274721397, | |
| "learning_rate": 9.104452865188254e-07, | |
| "loss": 0.3168, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 1.8584083780028213, | |
| "learning_rate": 8.988211641516289e-07, | |
| "loss": 0.3214, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 1.8211317436181427, | |
| "learning_rate": 8.872554397302388e-07, | |
| "loss": 0.3109, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 1.8219334782565464, | |
| "learning_rate": 8.757485350779609e-07, | |
| "loss": 0.301, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "grad_norm": 1.9771362675742574, | |
| "learning_rate": 8.64300869872832e-07, | |
| "loss": 0.3048, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "grad_norm": 1.875073162601576, | |
| "learning_rate": 8.529128616323224e-07, | |
| "loss": 0.2992, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "grad_norm": 1.8832489255751728, | |
| "learning_rate": 8.415849256980976e-07, | |
| "loss": 0.3054, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "grad_norm": 10.069268574653966, | |
| "learning_rate": 8.303174752208786e-07, | |
| "loss": 0.3205, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "grad_norm": 1.8175623122049562, | |
| "learning_rate": 8.191109211453688e-07, | |
| "loss": 0.3166, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "grad_norm": 1.884935738343672, | |
| "learning_rate": 8.079656721952686e-07, | |
| "loss": 0.3199, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "grad_norm": 1.8325908870427998, | |
| "learning_rate": 7.968821348583644e-07, | |
| "loss": 0.3017, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "grad_norm": 1.8664284061304677, | |
| "learning_rate": 7.858607133717117e-07, | |
| "loss": 0.3139, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "grad_norm": 1.8379727282817748, | |
| "learning_rate": 7.749018097068808e-07, | |
| "loss": 0.3023, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "grad_norm": 1.7887589319537402, | |
| "learning_rate": 7.640058235553044e-07, | |
| "loss": 0.3019, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "grad_norm": 1.7307328404201077, | |
| "learning_rate": 7.531731523136976e-07, | |
| "loss": 0.3121, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "grad_norm": 1.8359406746662572, | |
| "learning_rate": 7.424041910695631e-07, | |
| "loss": 0.3223, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "grad_norm": 1.866144889812394, | |
| "learning_rate": 7.316993325867827e-07, | |
| "loss": 0.313, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "grad_norm": 1.8657860951686938, | |
| "learning_rate": 7.210589672912935e-07, | |
| "loss": 0.314, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "grad_norm": 1.9579670233637116, | |
| "learning_rate": 7.104834832568438e-07, | |
| "loss": 0.2997, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "grad_norm": 1.7626608239046955, | |
| "learning_rate": 6.999732661908456e-07, | |
| "loss": 0.2939, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "grad_norm": 1.8904507508265878, | |
| "learning_rate": 6.895286994203037e-07, | |
| "loss": 0.3173, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "grad_norm": 1.8375475778504446, | |
| "learning_rate": 6.79150163877835e-07, | |
| "loss": 0.2928, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "grad_norm": 2.071120288178449, | |
| "learning_rate": 6.688380380877755e-07, | |
| "loss": 0.3104, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "grad_norm": 1.8806639293010023, | |
| "learning_rate": 6.585926981523761e-07, | |
| "loss": 0.3113, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "grad_norm": 1.7866185026537371, | |
| "learning_rate": 6.484145177380824e-07, | |
| "loss": 0.3053, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "grad_norm": 1.8712955462594827, | |
| "learning_rate": 6.3830386806191e-07, | |
| "loss": 0.3054, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "grad_norm": 1.8278128484701799, | |
| "learning_rate": 6.282611178779039e-07, | |
| "loss": 0.3226, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "grad_norm": 1.7845284828008048, | |
| "learning_rate": 6.182866334636889e-07, | |
| "loss": 0.2995, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "grad_norm": 1.908811432566092, | |
| "learning_rate": 6.083807786071111e-07, | |
| "loss": 0.3222, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "grad_norm": 1.7539407257874886, | |
| "learning_rate": 5.985439145929712e-07, | |
| "loss": 0.3106, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 1.765533340506495, | |
| "learning_rate": 5.887764001898441e-07, | |
| "loss": 0.3003, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.37727871537208557, | |
| "eval_runtime": 330.5667, | |
| "eval_samples_per_second": 22.752, | |
| "eval_steps_per_second": 0.357, | |
| "step": 2312 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "grad_norm": 2.6249172724373158, | |
| "learning_rate": 5.790785916369987e-07, | |
| "loss": 0.2687, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "grad_norm": 2.0234365981193108, | |
| "learning_rate": 5.694508426314024e-07, | |
| "loss": 0.2614, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "grad_norm": 2.0671265707120643, | |
| "learning_rate": 5.598935043148216e-07, | |
| "loss": 0.2575, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "grad_norm": 2.122780230514674, | |
| "learning_rate": 5.504069252610148e-07, | |
| "loss": 0.2634, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "grad_norm": 1.943804613668437, | |
| "learning_rate": 5.409914514630205e-07, | |
| "loss": 0.2682, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "grad_norm": 2.025428441019771, | |
| "learning_rate": 5.316474263205352e-07, | |
| "loss": 0.2499, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "grad_norm": 1.8116891241805655, | |
| "learning_rate": 5.223751906273936e-07, | |
| "loss": 0.2587, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "grad_norm": 1.9094711728143186, | |
| "learning_rate": 5.131750825591353e-07, | |
| "loss": 0.2624, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "grad_norm": 1.792419723948987, | |
| "learning_rate": 5.04047437660673e-07, | |
| "loss": 0.2596, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "grad_norm": 1.8856130920393068, | |
| "learning_rate": 4.94992588834054e-07, | |
| "loss": 0.2696, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "grad_norm": 4.886142600320586, | |
| "learning_rate": 4.860108663263185e-07, | |
| "loss": 0.279, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 1.831554357960856, | |
| "learning_rate": 4.77102597717454e-07, | |
| "loss": 0.2627, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "grad_norm": 1.8619320299708513, | |
| "learning_rate": 4.682681079084492e-07, | |
| "loss": 0.2492, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "grad_norm": 2.026722052142745, | |
| "learning_rate": 4.5950771910944603e-07, | |
| "loss": 0.2795, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "grad_norm": 1.8283283261223273, | |
| "learning_rate": 4.5082175082798244e-07, | |
| "loss": 0.2513, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "grad_norm": 1.959431295136227, | |
| "learning_rate": 4.422105198573451e-07, | |
| "loss": 0.2649, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "grad_norm": 1.7723428484005033, | |
| "learning_rate": 4.3367434026501317e-07, | |
| "loss": 0.2555, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "grad_norm": 1.978790482643724, | |
| "learning_rate": 4.2521352338120157e-07, | |
| "loss": 0.2611, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "grad_norm": 1.8319604787059154, | |
| "learning_rate": 4.1682837778750956e-07, | |
| "loss": 0.2525, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "grad_norm": 2.055092070648486, | |
| "learning_rate": 4.085192093056667e-07, | |
| "loss": 0.2683, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "grad_norm": 1.8816551471726852, | |
| "learning_rate": 4.0028632098637335e-07, | |
| "loss": 0.2599, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "grad_norm": 1.8539364404128353, | |
| "learning_rate": 3.9213001309825444e-07, | |
| "loss": 0.2564, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "grad_norm": 1.802564889277574, | |
| "learning_rate": 3.8405058311690473e-07, | |
| "loss": 0.2604, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "grad_norm": 1.921786615001957, | |
| "learning_rate": 3.7604832571403823e-07, | |
| "loss": 0.2577, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "grad_norm": 1.8432967570763548, | |
| "learning_rate": 3.681235327467453e-07, | |
| "loss": 0.2482, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "grad_norm": 1.8816609521253742, | |
| "learning_rate": 3.6027649324684417e-07, | |
| "loss": 0.2598, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "grad_norm": 1.8554972854976062, | |
| "learning_rate": 3.5250749341034007e-07, | |
| "loss": 0.2626, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "grad_norm": 1.7689335842068374, | |
| "learning_rate": 3.448168165869886e-07, | |
| "loss": 0.2501, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "grad_norm": 1.9815485309472562, | |
| "learning_rate": 3.372047432699607e-07, | |
| "loss": 0.2605, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "grad_norm": 1.8170369399087385, | |
| "learning_rate": 3.2967155108561076e-07, | |
| "loss": 0.25, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "grad_norm": 2.4030034838076886, | |
| "learning_rate": 3.222175147833556e-07, | |
| "loss": 0.2651, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "grad_norm": 1.9136750515911383, | |
| "learning_rate": 3.148429062256497e-07, | |
| "loss": 0.2621, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "grad_norm": 1.91289336281638, | |
| "learning_rate": 3.075479943780693e-07, | |
| "loss": 0.2608, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "grad_norm": 1.916237195499518, | |
| "learning_rate": 3.0033304529950733e-07, | |
| "loss": 0.2591, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "grad_norm": 1.7917439220204743, | |
| "learning_rate": 2.931983221324664e-07, | |
| "loss": 0.2559, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "grad_norm": 1.8928174769958723, | |
| "learning_rate": 2.861440850934599e-07, | |
| "loss": 0.2554, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "grad_norm": 1.8697064920019573, | |
| "learning_rate": 2.7917059146352703e-07, | |
| "loss": 0.2529, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "grad_norm": 1.8149915602515383, | |
| "learning_rate": 2.722780955788448e-07, | |
| "loss": 0.2507, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "grad_norm": 1.8085679954098952, | |
| "learning_rate": 2.6546684882145097e-07, | |
| "loss": 0.2692, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "grad_norm": 1.8501785801054293, | |
| "learning_rate": 2.587370996100802e-07, | |
| "loss": 0.2635, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "grad_norm": 1.7841943537417788, | |
| "learning_rate": 2.520890933911002e-07, | |
| "loss": 0.2726, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "grad_norm": 1.8236899418180625, | |
| "learning_rate": 2.4552307262956154e-07, | |
| "loss": 0.2629, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "grad_norm": 1.8527410868403758, | |
| "learning_rate": 2.3903927680035357e-07, | |
| "loss": 0.2601, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "grad_norm": 1.9912215998190843, | |
| "learning_rate": 2.3263794237947084e-07, | |
| "loss": 0.2541, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "grad_norm": 2.06578491146987, | |
| "learning_rate": 2.2631930283538745e-07, | |
| "loss": 0.2501, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "grad_norm": 1.8706250122502321, | |
| "learning_rate": 2.200835886205438e-07, | |
| "loss": 0.2625, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "grad_norm": 1.9781271983895208, | |
| "learning_rate": 2.139310271629405e-07, | |
| "loss": 0.2666, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "grad_norm": 1.7931386269528835, | |
| "learning_rate": 2.07861842857843e-07, | |
| "loss": 0.2693, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "grad_norm": 1.892205648139564, | |
| "learning_rate": 2.0187625705959862e-07, | |
| "loss": 0.2624, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "grad_norm": 1.8830690882464614, | |
| "learning_rate": 1.9597448807356368e-07, | |
| "loss": 0.2621, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "grad_norm": 2.792218645014677, | |
| "learning_rate": 1.901567511481389e-07, | |
| "loss": 0.2601, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "grad_norm": 1.896041115050084, | |
| "learning_rate": 1.8442325846692254e-07, | |
| "loss": 0.2659, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "grad_norm": 1.8177273159535672, | |
| "learning_rate": 1.787742191409686e-07, | |
| "loss": 0.2607, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "grad_norm": 1.8209339547730556, | |
| "learning_rate": 1.7320983920116246e-07, | |
| "loss": 0.2543, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "grad_norm": 1.8208127681176174, | |
| "learning_rate": 1.6773032159070474e-07, | |
| "loss": 0.2559, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "grad_norm": 1.8887316977772421, | |
| "learning_rate": 1.6233586615771097e-07, | |
| "loss": 0.2675, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "grad_norm": 1.8402016996115687, | |
| "learning_rate": 1.570266696479214e-07, | |
| "loss": 0.2708, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "grad_norm": 1.8135826560339299, | |
| "learning_rate": 1.5180292569752675e-07, | |
| "loss": 0.2663, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "grad_norm": 1.8932652503035179, | |
| "learning_rate": 1.4666482482610517e-07, | |
| "loss": 0.2562, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "grad_norm": 1.923411155855795, | |
| "learning_rate": 1.4161255442967314e-07, | |
| "loss": 0.2535, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "grad_norm": 1.8186664689085035, | |
| "learning_rate": 1.3664629877385233e-07, | |
| "loss": 0.2536, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "grad_norm": 1.8324657945327985, | |
| "learning_rate": 1.317662389871477e-07, | |
| "loss": 0.2626, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "grad_norm": 1.825266320838444, | |
| "learning_rate": 1.2697255305434087e-07, | |
| "loss": 0.2669, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "grad_norm": 1.8907649399842439, | |
| "learning_rate": 1.22265415810002e-07, | |
| "loss": 0.2637, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "grad_norm": 1.8689930525310243, | |
| "learning_rate": 1.1764499893210879e-07, | |
| "loss": 0.2625, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "grad_norm": 1.7851298479073345, | |
| "learning_rate": 1.1311147093578778e-07, | |
| "loss": 0.2515, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "grad_norm": 1.9022961798833327, | |
| "learning_rate": 1.0866499716716855e-07, | |
| "loss": 0.2683, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "grad_norm": 1.79257289914677, | |
| "learning_rate": 1.0430573979735154e-07, | |
| "loss": 0.2654, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "grad_norm": 1.9177553807521017, | |
| "learning_rate": 1.000338578164936e-07, | |
| "loss": 0.2607, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "grad_norm": 1.9212642014661736, | |
| "learning_rate": 9.584950702801244e-08, | |
| "loss": 0.2748, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 1.8444742116297173, | |
| "learning_rate": 9.175284004289853e-08, | |
| "loss": 0.2695, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "grad_norm": 1.879187416864439, | |
| "learning_rate": 8.774400627415498e-08, | |
| "loss": 0.2648, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "grad_norm": 1.853575218259344, | |
| "learning_rate": 8.382315193134372e-08, | |
| "loss": 0.2531, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "grad_norm": 1.8274985226049987, | |
| "learning_rate": 7.999042001525598e-08, | |
| "loss": 0.2608, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "grad_norm": 1.9822786870303424, | |
| "learning_rate": 7.624595031269388e-08, | |
| "loss": 0.2739, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "grad_norm": 2.395868204461241, | |
| "learning_rate": 7.258987939137596e-08, | |
| "loss": 0.2608, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "grad_norm": 1.9124430429661154, | |
| "learning_rate": 6.902234059495222e-08, | |
| "loss": 0.2621, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "grad_norm": 1.896035807781804, | |
| "learning_rate": 6.554346403814387e-08, | |
| "loss": 0.2604, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "grad_norm": 1.881027284048797, | |
| "learning_rate": 6.215337660199632e-08, | |
| "loss": 0.2457, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "grad_norm": 1.9007017472534657, | |
| "learning_rate": 5.8852201929252005e-08, | |
| "loss": 0.254, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "grad_norm": 1.8930962106170162, | |
| "learning_rate": 5.5640060419840655e-08, | |
| "loss": 0.264, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "grad_norm": 1.7660321817481, | |
| "learning_rate": 5.2517069226488694e-08, | |
| "loss": 0.2845, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "grad_norm": 1.794271544889709, | |
| "learning_rate": 4.9483342250445934e-08, | |
| "loss": 0.2657, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "grad_norm": 1.8831928024041047, | |
| "learning_rate": 4.653899013733115e-08, | |
| "loss": 0.2622, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "grad_norm": 1.8254137007243867, | |
| "learning_rate": 4.368412027309754e-08, | |
| "loss": 0.2613, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "grad_norm": 1.921264151045264, | |
| "learning_rate": 4.0918836780115265e-08, | |
| "loss": 0.2479, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "grad_norm": 1.8886495539506756, | |
| "learning_rate": 3.8243240513374524e-08, | |
| "loss": 0.2554, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "grad_norm": 1.842345668470861, | |
| "learning_rate": 3.5657429056806523e-08, | |
| "loss": 0.261, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "grad_norm": 1.8470545043429003, | |
| "learning_rate": 3.316149671972552e-08, | |
| "loss": 0.2451, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 1.898197482257326, | |
| "learning_rate": 3.0755534533387365e-08, | |
| "loss": 0.2605, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "grad_norm": 1.7715397317439983, | |
| "learning_rate": 2.8439630247671934e-08, | |
| "loss": 0.247, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "grad_norm": 1.8519562470145452, | |
| "learning_rate": 2.6213868327880088e-08, | |
| "loss": 0.2736, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "grad_norm": 1.9175537155386098, | |
| "learning_rate": 2.4078329951655044e-08, | |
| "loss": 0.2617, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "grad_norm": 1.8142907346054091, | |
| "learning_rate": 2.2033093006021133e-08, | |
| "loss": 0.2653, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "grad_norm": 1.7543346418920545, | |
| "learning_rate": 2.007823208454218e-08, | |
| "loss": 0.2478, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "grad_norm": 1.817163533821061, | |
| "learning_rate": 1.821381848460202e-08, | |
| "loss": 0.2513, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "grad_norm": 1.9039285159704047, | |
| "learning_rate": 1.643992020480406e-08, | |
| "loss": 0.257, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "grad_norm": 1.8371177889007853, | |
| "learning_rate": 1.4756601942490789e-08, | |
| "loss": 0.2497, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "grad_norm": 1.8366707128631479, | |
| "learning_rate": 1.3163925091384532e-08, | |
| "loss": 0.2531, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "grad_norm": 2.363794421622485, | |
| "learning_rate": 1.1661947739347867e-08, | |
| "loss": 0.2682, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "grad_norm": 1.820365380182807, | |
| "learning_rate": 1.0250724666265588e-08, | |
| "loss": 0.265, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "grad_norm": 1.8970735871610287, | |
| "learning_rate": 8.930307342046307e-09, | |
| "loss": 0.2625, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "grad_norm": 1.9794621227492666, | |
| "learning_rate": 7.700743924745902e-09, | |
| "loss": 0.2743, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "grad_norm": 2.1030207809550574, | |
| "learning_rate": 6.5620792588097505e-09, | |
| "loss": 0.2625, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "grad_norm": 1.8031485505651268, | |
| "learning_rate": 5.51435487343932e-09, | |
| "loss": 0.2589, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "grad_norm": 1.862632623834731, | |
| "learning_rate": 4.557608981075323e-09, | |
| "loss": 0.2651, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "grad_norm": 1.796005099020042, | |
| "learning_rate": 3.691876476005496e-09, | |
| "loss": 0.2537, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "grad_norm": 1.7492013741327732, | |
| "learning_rate": 2.917188933091175e-09, | |
| "loss": 0.2461, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "grad_norm": 1.9157963617346956, | |
| "learning_rate": 2.233574606616273e-09, | |
| "loss": 0.2661, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 1.8994803862391993, | |
| "learning_rate": 1.6410584292561592e-09, | |
| "loss": 0.2665, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "grad_norm": 1.846063403180044, | |
| "learning_rate": 1.139662011168663e-09, | |
| "loss": 0.2503, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "grad_norm": 1.8871406230680454, | |
| "learning_rate": 7.294036392066495e-10, | |
| "loss": 0.2601, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "grad_norm": 1.7693402402676852, | |
| "learning_rate": 4.1029827624911033e-10, | |
| "loss": 0.2494, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "grad_norm": 1.8544595991462598, | |
| "learning_rate": 1.8235756065770882e-10, | |
| "loss": 0.2647, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "grad_norm": 1.9426687566855905, | |
| "learning_rate": 4.5589805850454475e-11, | |
| "loss": 0.2615, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 1.8892335398265772, | |
| "learning_rate": 0.0, | |
| "loss": 0.2508, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.375461608171463, | |
| "eval_runtime": 329.8865, | |
| "eval_samples_per_second": 22.799, | |
| "eval_steps_per_second": 0.358, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 2890, | |
| "total_flos": 2420428819660800.0, | |
| "train_loss": 0.44474206024387714, | |
| "train_runtime": 64422.034, | |
| "train_samples_per_second": 5.737, | |
| "train_steps_per_second": 0.045 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 2890, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 2420428819660800.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |