{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.659568854733151, "global_step": 215000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 0.00014999965456413008, "loss": 0.6484, "step": 500 }, { "epoch": 0.02, "learning_rate": 0.0001499986182599296, "loss": 0.5436, "step": 1000 }, { "epoch": 0.03, "learning_rate": 0.00014999689109762647, "loss": 0.5149, "step": 1500 }, { "epoch": 0.03, "learning_rate": 0.00014999447309426713, "loss": 0.4996, "step": 2000 }, { "epoch": 0.04, "learning_rate": 0.00014999136427371623, "loss": 0.4888, "step": 2500 }, { "epoch": 0.05, "learning_rate": 0.0001499875646666566, "loss": 0.4797, "step": 3000 }, { "epoch": 0.06, "learning_rate": 0.00014998307431058887, "loss": 0.4711, "step": 3500 }, { "epoch": 0.07, "learning_rate": 0.00014997789324983099, "loss": 0.463, "step": 4000 }, { "epoch": 0.08, "learning_rate": 0.00014997202153551792, "loss": 0.4545, "step": 4500 }, { "epoch": 0.09, "learning_rate": 0.0001499654592256012, "loss": 0.4466, "step": 5000 }, { "epoch": 0.09, "learning_rate": 0.0001499582063848481, "loss": 0.4398, "step": 5500 }, { "epoch": 0.1, "learning_rate": 0.00014995026308484122, "loss": 0.4337, "step": 6000 }, { "epoch": 0.11, "learning_rate": 0.00014994162940397778, "loss": 0.4286, "step": 6500 }, { "epoch": 0.12, "learning_rate": 0.00014993230542746872, "loss": 0.4242, "step": 7000 }, { "epoch": 0.13, "learning_rate": 0.00014992229124733788, "loss": 0.4203, "step": 7500 }, { "epoch": 0.14, "learning_rate": 0.0001499115869624212, "loss": 0.4173, "step": 8000 }, { "epoch": 0.14, "learning_rate": 0.00014990019267836565, "loss": 0.4141, "step": 8500 }, { "epoch": 0.15, "learning_rate": 0.00014988810850762824, "loss": 0.4116, "step": 9000 }, { "epoch": 0.16, "learning_rate": 0.00014987533456947482, "loss": 0.4095, "step": 9500 }, { "epoch": 0.17, "learning_rate": 0.000149861870989979, "loss": 0.4075, "step": 10000 }, { "epoch": 0.18, "learning_rate": 0.0001498477758876907, "loss": 0.4057, "step": 10500 }, { "epoch": 0.19, "learning_rate": 0.00014983293618814342, "loss": 0.4038, "step": 11000 }, { "epoch": 0.2, "learning_rate": 0.00014981740726570863, "loss": 0.4021, "step": 11500 }, { "epoch": 0.2, "learning_rate": 0.00014980118927365057, "loss": 0.4005, "step": 12000 }, { "epoch": 0.21, "learning_rate": 0.00014978428237203426, "loss": 0.3991, "step": 12500 }, { "epoch": 0.22, "learning_rate": 0.00014976668672772396, "loss": 0.3977, "step": 13000 }, { "epoch": 0.23, "learning_rate": 0.00014974843976988143, "loss": 0.3968, "step": 13500 }, { "epoch": 0.24, "learning_rate": 0.00014972946854455738, "loss": 0.3953, "step": 14000 }, { "epoch": 0.25, "learning_rate": 0.00014970980911752974, "loss": 0.3945, "step": 14500 }, { "epoch": 0.26, "learning_rate": 0.0001496894616828291, "loss": 0.3931, "step": 15000 }, { "epoch": 0.26, "learning_rate": 0.0001496684264412764, "loss": 0.3921, "step": 15500 }, { "epoch": 0.27, "learning_rate": 0.00014964670360048104, "loss": 0.3909, "step": 16000 }, { "epoch": 0.28, "learning_rate": 0.00014962429337483864, "loss": 0.3903, "step": 16500 }, { "epoch": 0.29, "learning_rate": 0.00014960119598552907, "loss": 0.3892, "step": 17000 }, { "epoch": 0.3, "learning_rate": 0.00014957745991457127, "loss": 0.3885, "step": 17500 }, { "epoch": 0.31, "learning_rate": 0.00014955303988623036, "loss": 0.3877, "step": 18000 }, { "epoch": 0.31, "learning_rate": 0.00014952788514615529, "loss": 0.3871, "step": 18500 }, { "epoch": 0.32, "learning_rate": 0.0001495020441939234, "loss": 0.3858, "step": 19000 }, { "epoch": 0.33, "learning_rate": 0.0001494755172845744, "loss": 0.3854, "step": 19500 }, { "epoch": 0.34, "learning_rate": 0.00014944830467991823, "loss": 0.3847, "step": 20000 }, { "epoch": 0.35, "learning_rate": 0.00014942046312846905, "loss": 0.3839, "step": 20500 }, { "epoch": 0.36, "learning_rate": 0.0001493918813157191, "loss": 0.3837, "step": 21000 }, { "epoch": 0.37, "learning_rate": 0.0001493626146331152, "loss": 0.383, "step": 21500 }, { "epoch": 0.37, "learning_rate": 0.00014933266336950772, "loss": 0.3821, "step": 22000 }, { "epoch": 0.38, "learning_rate": 0.00014930208977431846, "loss": 0.382, "step": 22500 }, { "epoch": 0.39, "learning_rate": 0.00014927077160993785, "loss": 0.381, "step": 23000 }, { "epoch": 0.4, "learning_rate": 0.0001492387697710072, "loss": 0.3804, "step": 23500 }, { "epoch": 0.41, "learning_rate": 0.0001492060845733717, "loss": 0.38, "step": 24000 }, { "epoch": 0.42, "learning_rate": 0.00014917278375754084, "loss": 0.3793, "step": 24500 }, { "epoch": 0.43, "learning_rate": 0.00014913880296235724, "loss": 0.3791, "step": 25000 }, { "epoch": 0.43, "learning_rate": 0.00014910407237991193, "loss": 0.3787, "step": 25500 }, { "epoch": 0.44, "learning_rate": 0.00014906865976817066, "loss": 0.3779, "step": 26000 }, { "epoch": 0.45, "learning_rate": 0.00014903256547664165, "loss": 0.3778, "step": 26500 }, { "epoch": 0.46, "learning_rate": 0.0001489958640925119, "loss": 0.377, "step": 27000 }, { "epoch": 0.47, "learning_rate": 0.00014895840887839445, "loss": 0.3768, "step": 27500 }, { "epoch": 0.48, "learning_rate": 0.000148920273072621, "loss": 0.3761, "step": 28000 }, { "epoch": 0.49, "learning_rate": 0.00014888145705157658, "loss": 0.3759, "step": 28500 }, { "epoch": 0.49, "learning_rate": 0.0001488420408682806, "loss": 0.3755, "step": 29000 }, { "epoch": 0.5, "learning_rate": 0.00014880186693119052, "loss": 0.3749, "step": 29500 }, { "epoch": 0.51, "learning_rate": 0.00014876101394745023, "loss": 0.3747, "step": 30000 }, { "epoch": 0.52, "learning_rate": 0.00014871948232026222, "loss": 0.3741, "step": 30500 }, { "epoch": 0.53, "learning_rate": 0.00014867727245952692, "loss": 0.3739, "step": 31000 }, { "epoch": 0.54, "learning_rate": 0.00014863447123337502, "loss": 0.3736, "step": 31500 }, { "epoch": 0.54, "learning_rate": 0.00014859090751637814, "loss": 0.3733, "step": 32000 }, { "epoch": 0.55, "learning_rate": 0.00014854666683481534, "loss": 0.3728, "step": 32500 }, { "epoch": 0.56, "learning_rate": 0.0001485017496253243, "loss": 0.3725, "step": 33000 }, { "epoch": 0.57, "learning_rate": 0.00014845624819224238, "loss": 0.3722, "step": 33500 }, { "epoch": 0.58, "learning_rate": 0.00014840998061432743, "loss": 0.3719, "step": 34000 }, { "epoch": 0.59, "learning_rate": 0.00014836313241655734, "loss": 0.3717, "step": 34500 }, { "epoch": 0.6, "learning_rate": 0.00014831551629313194, "loss": 0.3711, "step": 35000 }, { "epoch": 0.6, "learning_rate": 0.00014826722592314076, "loss": 0.3709, "step": 35500 }, { "epoch": 0.61, "learning_rate": 0.00014821826178319038, "loss": 0.3709, "step": 36000 }, { "epoch": 0.62, "learning_rate": 0.0001481687243030059, "loss": 0.3703, "step": 36500 }, { "epoch": 0.63, "learning_rate": 0.00014811841542465117, "loss": 0.3698, "step": 37000 }, { "epoch": 0.64, "learning_rate": 0.00014806743424503673, "loss": 0.3697, "step": 37500 }, { "epoch": 0.65, "learning_rate": 0.0001480157812673262, "loss": 0.3692, "step": 38000 }, { "epoch": 0.66, "learning_rate": 0.0001479634570013137, "loss": 0.369, "step": 38500 }, { "epoch": 0.66, "learning_rate": 0.00014791046196341849, "loss": 0.3689, "step": 39000 }, { "epoch": 0.67, "learning_rate": 0.0001478567966766803, "loss": 0.3685, "step": 39500 }, { "epoch": 0.68, "learning_rate": 0.0001478024616707538, "loss": 0.3683, "step": 40000 }, { "epoch": 0.69, "learning_rate": 0.00014774756815776608, "loss": 0.368, "step": 40500 }, { "epoch": 0.7, "learning_rate": 0.00014769189666559521, "loss": 0.3679, "step": 41000 }, { "epoch": 0.71, "learning_rate": 0.00014763555708173254, "loss": 0.3676, "step": 41500 }, { "epoch": 0.71, "learning_rate": 0.0001475785499622271, "loss": 0.3669, "step": 42000 }, { "epoch": 0.72, "learning_rate": 0.00014752087586971605, "loss": 0.3673, "step": 42500 }, { "epoch": 0.73, "learning_rate": 0.0001474626527191014, "loss": 0.3668, "step": 43000 }, { "epoch": 0.74, "learning_rate": 0.00014740364772589337, "loss": 0.3666, "step": 43500 }, { "epoch": 0.75, "learning_rate": 0.00014734397748589426, "loss": 0.3665, "step": 44000 }, { "epoch": 0.76, "learning_rate": 0.0001472836425880252, "loss": 0.366, "step": 44500 }, { "epoch": 0.77, "learning_rate": 0.00014722264362776731, "loss": 0.3657, "step": 45000 }, { "epoch": 0.77, "learning_rate": 0.0001471609812071557, "loss": 0.3657, "step": 45500 }, { "epoch": 0.78, "learning_rate": 0.00014709865593477356, "loss": 0.3653, "step": 46000 }, { "epoch": 0.79, "learning_rate": 0.00014703566842574615, "loss": 0.3652, "step": 46500 }, { "epoch": 0.8, "learning_rate": 0.0001469721472598581, "loss": 0.3645, "step": 47000 }, { "epoch": 0.81, "learning_rate": 0.00014690783847039622, "loss": 0.3647, "step": 47500 }, { "epoch": 0.82, "learning_rate": 0.00014684286932758046, "loss": 0.3643, "step": 48000 }, { "epoch": 0.83, "learning_rate": 0.00014677724047262996, "loss": 0.3644, "step": 48500 }, { "epoch": 0.83, "learning_rate": 0.0001467110857864263, "loss": 0.3641, "step": 49000 }, { "epoch": 0.84, "learning_rate": 0.00014664414077306508, "loss": 0.3636, "step": 49500 }, { "epoch": 0.85, "learning_rate": 0.0001465765380089398, "loss": 0.3635, "step": 50000 }, { "epoch": 0.86, "learning_rate": 0.00014650841533628042, "loss": 0.3635, "step": 50500 }, { "epoch": 0.87, "learning_rate": 0.00014643950039089107, "loss": 0.3631, "step": 51000 }, { "epoch": 0.88, "learning_rate": 0.0001463699297144556, "loss": 0.3628, "step": 51500 }, { "epoch": 0.89, "learning_rate": 0.00014629970399360846, "loss": 0.363, "step": 52000 }, { "epoch": 0.89, "learning_rate": 0.00014622882392144917, "loss": 0.3624, "step": 52500 }, { "epoch": 0.9, "learning_rate": 0.00014615729019753547, "loss": 0.3623, "step": 53000 }, { "epoch": 0.91, "learning_rate": 0.00014608510352787625, "loss": 0.3623, "step": 53500 }, { "epoch": 0.92, "learning_rate": 0.0001460122646249249, "loss": 0.3621, "step": 54000 }, { "epoch": 0.93, "learning_rate": 0.00014593892183813694, "loss": 0.3617, "step": 54500 }, { "epoch": 0.94, "learning_rate": 0.00014586478193255307, "loss": 0.3617, "step": 55000 }, { "epoch": 0.94, "learning_rate": 0.00014578999196816224, "loss": 0.3616, "step": 55500 }, { "epoch": 0.95, "learning_rate": 0.00014571455268311119, "loss": 0.361, "step": 56000 }, { "epoch": 0.96, "learning_rate": 0.00014563861764445836, "loss": 0.3613, "step": 56500 }, { "epoch": 0.97, "learning_rate": 0.00014556188325305048, "loss": 0.3607, "step": 57000 }, { "epoch": 0.98, "learning_rate": 0.00014548450179232327, "loss": 0.3606, "step": 57500 }, { "epoch": 0.99, "learning_rate": 0.00014540647402600055, "loss": 0.3607, "step": 58000 }, { "epoch": 1.0, "learning_rate": 0.00014532780072418487, "loss": 0.3602, "step": 58500 }, { "epoch": 1.0, "learning_rate": 0.00014524864194242183, "loss": 0.3607, "step": 59000 }, { "epoch": 1.01, "learning_rate": 0.00014516868119257174, "loss": 0.3599, "step": 59500 }, { "epoch": 1.02, "learning_rate": 0.00014508823910339927, "loss": 0.3599, "step": 60000 }, { "epoch": 1.03, "learning_rate": 0.00014500699405591802, "loss": 0.3599, "step": 60500 }, { "epoch": 1.04, "learning_rate": 0.00014492510741564985, "loss": 0.3595, "step": 61000 }, { "epoch": 1.05, "learning_rate": 0.00014484257999078278, "loss": 0.3593, "step": 61500 }, { "epoch": 1.06, "learning_rate": 0.00014475941259582925, "loss": 0.3595, "step": 62000 }, { "epoch": 1.06, "learning_rate": 0.0001446756060516178, "loss": 0.359, "step": 62500 }, { "epoch": 1.07, "learning_rate": 0.00014459116118528524, "loss": 0.3589, "step": 63000 }, { "epoch": 1.08, "learning_rate": 0.00014450607883026827, "loss": 0.3586, "step": 63500 }, { "epoch": 1.09, "learning_rate": 0.00014442053189911774, "loss": 0.3587, "step": 64000 }, { "epoch": 1.1, "learning_rate": 0.00014433417836295833, "loss": 0.3583, "step": 64500 }, { "epoch": 1.11, "learning_rate": 0.00014424718987443127, "loss": 0.3583, "step": 65000 }, { "epoch": 1.11, "learning_rate": 0.00014415956729207775, "loss": 0.3582, "step": 65500 }, { "epoch": 1.12, "learning_rate": 0.00014407131148069734, "loss": 0.3579, "step": 66000 }, { "epoch": 1.13, "learning_rate": 0.00014398260171818916, "loss": 0.3578, "step": 66500 }, { "epoch": 1.14, "learning_rate": 0.00014389308333022545, "loss": 0.3576, "step": 67000 }, { "epoch": 1.15, "learning_rate": 0.00014380311527004513, "loss": 0.3577, "step": 67500 }, { "epoch": 1.16, "learning_rate": 0.00014371233783246623, "loss": 0.3579, "step": 68000 }, { "epoch": 1.17, "learning_rate": 0.00014362093157983466, "loss": 0.3572, "step": 68500 }, { "epoch": 1.17, "learning_rate": 0.0001435288974142932, "loss": 0.3575, "step": 69000 }, { "epoch": 1.18, "learning_rate": 0.00014343623624418192, "loss": 0.3571, "step": 69500 }, { "epoch": 1.19, "learning_rate": 0.00014334294898402916, "loss": 0.3567, "step": 70000 }, { "epoch": 1.2, "learning_rate": 0.0001432490365545425, "loss": 0.3567, "step": 70500 }, { "epoch": 1.21, "learning_rate": 0.00014315449988259974, "loss": 0.3566, "step": 71000 }, { "epoch": 1.22, "learning_rate": 0.0001430593399012396, "loss": 0.3564, "step": 71500 }, { "epoch": 1.23, "learning_rate": 0.00014296374973485408, "loss": 0.3561, "step": 72000 }, { "epoch": 1.23, "learning_rate": 0.0001428675406248976, "loss": 0.356, "step": 72500 }, { "epoch": 1.24, "learning_rate": 0.00014277051885498018, "loss": 0.3562, "step": 73000 }, { "epoch": 1.25, "learning_rate": 0.00014267287756538312, "loss": 0.3557, "step": 73500 }, { "epoch": 1.26, "learning_rate": 0.00014257461771978656, "loss": 0.3557, "step": 74000 }, { "epoch": 1.27, "learning_rate": 0.0001424757402879755, "loss": 0.3556, "step": 74500 }, { "epoch": 1.28, "learning_rate": 0.0001423762462458303, "loss": 0.3552, "step": 75000 }, { "epoch": 1.29, "learning_rate": 0.00014227613657531694, "loss": 0.3552, "step": 75500 }, { "epoch": 1.29, "learning_rate": 0.00014217541226447747, "loss": 0.3551, "step": 76000 }, { "epoch": 1.3, "learning_rate": 0.00014207427759508945, "loss": 0.3552, "step": 76500 }, { "epoch": 1.31, "learning_rate": 0.000141972328216269, "loss": 0.355, "step": 77000 }, { "epoch": 1.32, "learning_rate": 0.0001418697671955883, "loss": 0.3546, "step": 77500 }, { "epoch": 1.33, "learning_rate": 0.00014176659554528318, "loss": 0.3546, "step": 78000 }, { "epoch": 1.34, "learning_rate": 0.00014166302245385192, "loss": 0.3544, "step": 78500 }, { "epoch": 1.34, "learning_rate": 0.00014155863382125022, "loss": 0.3545, "step": 79000 }, { "epoch": 1.35, "learning_rate": 0.00014145363762978493, "loss": 0.3544, "step": 79500 }, { "epoch": 1.36, "learning_rate": 0.00014134803491572606, "loss": 0.3542, "step": 80000 }, { "epoch": 1.37, "learning_rate": 0.00014124203974129223, "loss": 0.3539, "step": 80500 }, { "epoch": 1.38, "learning_rate": 0.00014113522832260466, "loss": 0.3541, "step": 81000 }, { "epoch": 1.39, "learning_rate": 0.00014102781352389485, "loss": 0.3537, "step": 81500 }, { "epoch": 1.4, "learning_rate": 0.00014091979640530345, "loss": 0.3535, "step": 82000 }, { "epoch": 1.4, "learning_rate": 0.00014081139586900036, "loss": 0.3533, "step": 82500 }, { "epoch": 1.41, "learning_rate": 0.0001407021785141258, "loss": 0.3535, "step": 83000 }, { "epoch": 1.42, "learning_rate": 0.00014059236205325574, "loss": 0.3536, "step": 83500 }, { "epoch": 1.43, "learning_rate": 0.00014048194757023435, "loss": 0.353, "step": 84000 }, { "epoch": 1.44, "learning_rate": 0.00014037115877265014, "loss": 0.353, "step": 84500 }, { "epoch": 1.45, "learning_rate": 0.00014025955271103284, "loss": 0.3529, "step": 85000 }, { "epoch": 1.46, "learning_rate": 0.00014014735191195807, "loss": 0.3528, "step": 85500 }, { "epoch": 1.46, "learning_rate": 0.00014003478366336457, "loss": 0.3527, "step": 86000 }, { "epoch": 1.47, "learning_rate": 0.00013992253468820287, "loss": 0.3559, "step": 86500 }, { "epoch": 1.48, "learning_rate": 0.00013980856343455598, "loss": 0.3566, "step": 87000 }, { "epoch": 1.49, "learning_rate": 0.0001396940018945337, "loss": 0.3554, "step": 87500 }, { "epoch": 1.5, "learning_rate": 0.00013957885119881218, "loss": 0.3552, "step": 88000 }, { "epoch": 1.51, "learning_rate": 0.00013946311248388228, "loss": 0.3548, "step": 88500 }, { "epoch": 1.51, "learning_rate": 0.00013934678689203836, "loss": 0.3547, "step": 89000 }, { "epoch": 1.52, "learning_rate": 0.0001392298755713671, "loss": 0.3547, "step": 89500 }, { "epoch": 1.53, "learning_rate": 0.00013911237967573605, "loss": 0.3541, "step": 90000 }, { "epoch": 1.54, "learning_rate": 0.0001389943003647822, "loss": 0.3539, "step": 90500 }, { "epoch": 1.55, "learning_rate": 0.00013887563880390073, "loss": 0.3536, "step": 91000 }, { "epoch": 1.56, "learning_rate": 0.0001387563961642333, "loss": 0.3537, "step": 91500 }, { "epoch": 1.57, "learning_rate": 0.00013863657362265668, "loss": 0.3549, "step": 92000 }, { "epoch": 1.57, "learning_rate": 0.0001385161723617709, "loss": 0.3539, "step": 92500 }, { "epoch": 1.58, "learning_rate": 0.00013839519356988788, "loss": 0.3535, "step": 93000 }, { "epoch": 1.59, "learning_rate": 0.0001382736384410194, "loss": 0.3537, "step": 93500 }, { "epoch": 1.6, "learning_rate": 0.00013815150817486554, "loss": 0.3527, "step": 94000 }, { "epoch": 1.61, "learning_rate": 0.0001380288039768027, "loss": 0.3529, "step": 94500 }, { "epoch": 1.62, "learning_rate": 0.00013790552705787178, "loss": 0.353, "step": 95000 }, { "epoch": 1.63, "learning_rate": 0.0001377816786347662, "loss": 0.3526, "step": 95500 }, { "epoch": 1.63, "learning_rate": 0.00013765725992981992, "loss": 0.3527, "step": 96000 }, { "epoch": 1.64, "learning_rate": 0.00013753227217099532, "loss": 0.3528, "step": 96500 }, { "epoch": 1.65, "learning_rate": 0.0001374067165918711, "loss": 0.3524, "step": 97000 }, { "epoch": 1.66, "learning_rate": 0.00013728059443163013, "loss": 0.352, "step": 97500 }, { "epoch": 1.67, "learning_rate": 0.00013715390693504724, "loss": 0.3519, "step": 98000 }, { "epoch": 1.68, "learning_rate": 0.00013702665535247687, "loss": 0.3519, "step": 98500 }, { "epoch": 1.69, "learning_rate": 0.00013689884093984076, "loss": 0.3519, "step": 99000 }, { "epoch": 1.69, "learning_rate": 0.00013677072227018223, "loss": 0.3526, "step": 99500 }, { "epoch": 1.7, "learning_rate": 0.00013664178710672204, "loss": 0.3517, "step": 100000 }, { "epoch": 1.71, "learning_rate": 0.00013651229291169033, "loss": 0.3516, "step": 100500 }, { "epoch": 1.72, "learning_rate": 0.00013638250162282539, "loss": 0.3516, "step": 101000 }, { "epoch": 1.73, "learning_rate": 0.00013625189431597695, "loss": 0.3515, "step": 101500 }, { "epoch": 1.74, "learning_rate": 0.00013612073182564145, "loss": 0.3512, "step": 102000 }, { "epoch": 1.74, "learning_rate": 0.00013599007137178395, "loss": 0.3527, "step": 102500 }, { "epoch": 1.75, "learning_rate": 0.0001358578068190425, "loss": 0.352, "step": 103000 }, { "epoch": 1.76, "learning_rate": 0.00013572499097229843, "loss": 0.3512, "step": 103500 }, { "epoch": 1.77, "learning_rate": 0.00013559162514239058, "loss": 0.3509, "step": 104000 }, { "epoch": 1.78, "learning_rate": 0.0001354577106455858, "loss": 0.3504, "step": 104500 }, { "epoch": 1.79, "learning_rate": 0.00013532324880356615, "loss": 0.3507, "step": 105000 }, { "epoch": 1.8, "learning_rate": 0.0001351882409434157, "loss": 0.3504, "step": 105500 }, { "epoch": 1.8, "learning_rate": 0.00013505268839760753, "loss": 0.3501, "step": 106000 }, { "epoch": 1.81, "learning_rate": 0.00013491659250399054, "loss": 0.3503, "step": 106500 }, { "epoch": 1.82, "learning_rate": 0.00013477995460577628, "loss": 0.3505, "step": 107000 }, { "epoch": 1.83, "learning_rate": 0.00013464277605152562, "loss": 0.3502, "step": 107500 }, { "epoch": 1.84, "learning_rate": 0.0001345050581951355, "loss": 0.3503, "step": 108000 }, { "epoch": 1.85, "learning_rate": 0.00013436680239582559, "loss": 0.35, "step": 108500 }, { "epoch": 1.86, "learning_rate": 0.00013422801001812477, "loss": 0.3506, "step": 109000 }, { "epoch": 1.86, "learning_rate": 0.00013408868243185783, "loss": 0.3504, "step": 109500 }, { "epoch": 1.87, "learning_rate": 0.00013394882101213176, "loss": 0.3497, "step": 110000 }, { "epoch": 1.88, "learning_rate": 0.00013380842713932233, "loss": 0.3495, "step": 110500 }, { "epoch": 1.89, "learning_rate": 0.00013366750219906035, "loss": 0.3499, "step": 111000 }, { "epoch": 1.9, "learning_rate": 0.00013352604758221816, "loss": 0.3494, "step": 111500 }, { "epoch": 1.91, "learning_rate": 0.00013338406468489563, "loss": 0.3499, "step": 112000 }, { "epoch": 1.91, "learning_rate": 0.0001332415549084067, "loss": 0.3496, "step": 112500 }, { "epoch": 1.92, "learning_rate": 0.0001330985196592653, "loss": 0.349, "step": 113000 }, { "epoch": 1.93, "learning_rate": 0.00013295496034917162, "loss": 0.3491, "step": 113500 }, { "epoch": 1.94, "learning_rate": 0.00013281087839499807, "loss": 0.3489, "step": 114000 }, { "epoch": 1.95, "learning_rate": 0.00013266627521877535, "loss": 0.3485, "step": 114500 }, { "epoch": 1.96, "learning_rate": 0.0001325211522476785, "loss": 0.3487, "step": 115000 }, { "epoch": 1.97, "learning_rate": 0.00013237551091401257, "loss": 0.349, "step": 115500 }, { "epoch": 1.97, "learning_rate": 0.0001322299383160388, "loss": 0.3494, "step": 116000 }, { "epoch": 1.98, "learning_rate": 0.0001320832666336504, "loss": 0.3493, "step": 116500 }, { "epoch": 1.99, "learning_rate": 0.00013193637579394162, "loss": 0.3491, "step": 117000 }, { "epoch": 2.0, "learning_rate": 0.00013178867850631044, "loss": 0.3497, "step": 117500 }, { "epoch": 2.01, "learning_rate": 0.00013164047008533037, "loss": 0.3489, "step": 118000 }, { "epoch": 2.02, "learning_rate": 0.00013149175199375862, "loss": 0.3489, "step": 118500 }, { "epoch": 2.03, "learning_rate": 0.00013134252569938276, "loss": 0.3479, "step": 119000 }, { "epoch": 2.03, "learning_rate": 0.0001311927926750061, "loss": 0.3482, "step": 119500 }, { "epoch": 2.04, "learning_rate": 0.00013104255439843306, "loss": 0.3479, "step": 120000 }, { "epoch": 2.05, "learning_rate": 0.00013089181235245484, "loss": 0.3476, "step": 120500 }, { "epoch": 2.06, "learning_rate": 0.00013074056802483454, "loss": 0.348, "step": 121000 }, { "epoch": 2.07, "learning_rate": 0.00013058882290829265, "loss": 0.3481, "step": 121500 }, { "epoch": 2.08, "learning_rate": 0.00013043688348660256, "loss": 0.3482, "step": 122000 }, { "epoch": 2.09, "learning_rate": 0.00013028414228420884, "loss": 0.3474, "step": 122500 }, { "epoch": 2.09, "learning_rate": 0.0001301309047976313, "loss": 0.3478, "step": 123000 }, { "epoch": 2.1, "learning_rate": 0.00012997717253926207, "loss": 0.3479, "step": 123500 }, { "epoch": 2.11, "learning_rate": 0.00012982294702637649, "loss": 0.3473, "step": 124000 }, { "epoch": 2.12, "learning_rate": 0.0001296682297811181, "loss": 0.3472, "step": 124500 }, { "epoch": 2.13, "learning_rate": 0.00012951302233048365, "loss": 0.3475, "step": 125000 }, { "epoch": 2.14, "learning_rate": 0.00012935732620630798, "loss": 0.3473, "step": 125500 }, { "epoch": 2.14, "learning_rate": 0.00012920145579691002, "loss": 0.3473, "step": 126000 }, { "epoch": 2.15, "learning_rate": 0.00012904478791008278, "loss": 0.3476, "step": 126500 }, { "epoch": 2.16, "learning_rate": 0.00012888763597099904, "loss": 0.3474, "step": 127000 }, { "epoch": 2.17, "learning_rate": 0.000128730001530685, "loss": 0.3473, "step": 127500 }, { "epoch": 2.18, "learning_rate": 0.00012857188614492896, "loss": 0.3471, "step": 128000 }, { "epoch": 2.19, "learning_rate": 0.0001284136090411938, "loss": 0.3471, "step": 128500 }, { "epoch": 2.2, "learning_rate": 0.00012825453740496415, "loss": 0.3467, "step": 129000 }, { "epoch": 2.2, "learning_rate": 0.00012809498951593147, "loss": 0.3467, "step": 129500 }, { "epoch": 2.21, "learning_rate": 0.000127934966948769, "loss": 0.3466, "step": 130000 }, { "epoch": 2.22, "learning_rate": 0.00012777447128283487, "loss": 0.3464, "step": 130500 }, { "epoch": 2.23, "learning_rate": 0.0001276135041021565, "loss": 0.3462, "step": 131000 }, { "epoch": 2.24, "learning_rate": 0.00012745239033755645, "loss": 0.3467, "step": 131500 }, { "epoch": 2.25, "learning_rate": 0.0001272908101084899, "loss": 0.3464, "step": 132000 }, { "epoch": 2.26, "learning_rate": 0.0001271284397979516, "loss": 0.3463, "step": 132500 }, { "epoch": 2.26, "learning_rate": 0.000126965604348739, "loss": 0.3463, "step": 133000 }, { "epoch": 2.27, "learning_rate": 0.00012680230536797228, "loss": 0.3461, "step": 133500 }, { "epoch": 2.28, "learning_rate": 0.0001266385444673464, "loss": 0.3458, "step": 134000 }, { "epoch": 2.29, "learning_rate": 0.00012647432326311537, "loss": 0.3459, "step": 134500 }, { "epoch": 2.3, "learning_rate": 0.00012630964337607617, "loss": 0.3459, "step": 135000 }, { "epoch": 2.31, "learning_rate": 0.00012614483716050272, "loss": 0.3456, "step": 135500 }, { "epoch": 2.31, "learning_rate": 0.0001259792456975559, "loss": 0.3459, "step": 136000 }, { "epoch": 2.32, "learning_rate": 0.00012581320043801664, "loss": 0.3456, "step": 136500 }, { "epoch": 2.33, "learning_rate": 0.00012564670302068462, "loss": 0.3457, "step": 137000 }, { "epoch": 2.34, "learning_rate": 0.00012547975508882204, "loss": 0.3452, "step": 137500 }, { "epoch": 2.35, "learning_rate": 0.00012531235829013772, "loss": 0.3452, "step": 138000 }, { "epoch": 2.36, "learning_rate": 0.00012514451427677039, "loss": 0.3451, "step": 138500 }, { "epoch": 2.37, "learning_rate": 0.0001249762247052727, "loss": 0.3446, "step": 139000 }, { "epoch": 2.37, "learning_rate": 0.00012480749123659483, "loss": 0.3445, "step": 139500 }, { "epoch": 2.38, "learning_rate": 0.00012463865432770735, "loss": 0.3455, "step": 140000 }, { "epoch": 2.39, "learning_rate": 0.00012446903894448242, "loss": 0.345, "step": 140500 }, { "epoch": 2.4, "learning_rate": 0.00012429898466979599, "loss": 0.3447, "step": 141000 }, { "epoch": 2.41, "learning_rate": 0.0001241284931820151, "loss": 0.3445, "step": 141500 }, { "epoch": 2.42, "learning_rate": 0.00012395756616382189, "loss": 0.3446, "step": 142000 }, { "epoch": 2.43, "learning_rate": 0.0001237865484557726, "loss": 0.3444, "step": 142500 }, { "epoch": 2.43, "learning_rate": 0.00012361475630459222, "loss": 0.3441, "step": 143000 }, { "epoch": 2.44, "learning_rate": 0.0001234428785670676, "loss": 0.3443, "step": 143500 }, { "epoch": 2.45, "learning_rate": 0.00012327022805140274, "loss": 0.3438, "step": 144000 }, { "epoch": 2.46, "learning_rate": 0.00012309715047605592, "loss": 0.3437, "step": 144500 }, { "epoch": 2.47, "learning_rate": 0.00012292364754923295, "loss": 0.3444, "step": 145000 }, { "epoch": 2.48, "learning_rate": 0.0001227497209833376, "loss": 0.3441, "step": 145500 }, { "epoch": 2.49, "learning_rate": 0.00012257572161186733, "loss": 0.3439, "step": 146000 }, { "epoch": 2.49, "learning_rate": 0.00012240095376042992, "loss": 0.3436, "step": 146500 }, { "epoch": 2.5, "learning_rate": 0.00012222576742869685, "loss": 0.3434, "step": 147000 }, { "epoch": 2.51, "learning_rate": 0.00012205016434568649, "loss": 0.3434, "step": 147500 }, { "epoch": 2.52, "learning_rate": 0.00012187414624453037, "loss": 0.3433, "step": 148000 }, { "epoch": 2.53, "learning_rate": 0.00012169771486245609, "loss": 0.3429, "step": 148500 }, { "epoch": 2.54, "learning_rate": 0.00012152087194077019, "loss": 0.3429, "step": 149000 }, { "epoch": 2.54, "learning_rate": 0.00012134361922484087, "loss": 0.3435, "step": 149500 }, { "epoch": 2.55, "learning_rate": 0.0001211659584640809, "loss": 0.3427, "step": 150000 }, { "epoch": 2.56, "learning_rate": 0.00012098789141193026, "loss": 0.3423, "step": 150500 }, { "epoch": 2.57, "learning_rate": 0.00012080941982583888, "loss": 0.3425, "step": 151000 }, { "epoch": 2.58, "learning_rate": 0.00012063054546724922, "loss": 0.3425, "step": 151500 }, { "epoch": 2.59, "learning_rate": 0.00012045127010157905, "loss": 0.3426, "step": 152000 }, { "epoch": 2.6, "learning_rate": 0.00012027195524467175, "loss": 0.3422, "step": 152500 }, { "epoch": 2.6, "learning_rate": 0.00012009188397006361, "loss": 0.342, "step": 153000 }, { "epoch": 2.61, "learning_rate": 0.00011991141700474647, "loss": 0.3422, "step": 153500 }, { "epoch": 2.62, "learning_rate": 0.00011973055612985635, "loss": 0.3417, "step": 154000 }, { "epoch": 2.63, "learning_rate": 0.00011954930313041712, "loss": 0.3416, "step": 154500 }, { "epoch": 2.64, "learning_rate": 0.00011936802347035692, "loss": 0.3417, "step": 155000 }, { "epoch": 2.65, "learning_rate": 0.00011918599236764778, "loss": 0.3415, "step": 155500 }, { "epoch": 2.66, "learning_rate": 0.00011900357451501401, "loss": 0.3417, "step": 156000 }, { "epoch": 2.66, "learning_rate": 0.00011882077171284612, "loss": 0.3421, "step": 156500 }, { "epoch": 2.67, "learning_rate": 0.00011863795251840704, "loss": 0.3414, "step": 157000 }, { "epoch": 2.68, "learning_rate": 0.00011845512101771738, "loss": 0.3419, "step": 157500 }, { "epoch": 2.69, "learning_rate": 0.00011827117647893832, "loss": 0.3414, "step": 158000 }, { "epoch": 2.7, "learning_rate": 0.0001180868542190981, "loss": 0.3423, "step": 158500 }, { "epoch": 2.71, "learning_rate": 0.00011790215605738305, "loss": 0.3417, "step": 159000 }, { "epoch": 2.71, "learning_rate": 0.00011771708381668941, "loss": 0.3412, "step": 159500 }, { "epoch": 2.72, "learning_rate": 0.00011753163932360551, "loss": 0.3408, "step": 160000 }, { "epoch": 2.73, "learning_rate": 0.00011734582440839364, "loss": 0.3411, "step": 160500 }, { "epoch": 2.74, "learning_rate": 0.00011715964090497195, "loss": 0.3406, "step": 161000 }, { "epoch": 2.75, "learning_rate": 0.00011697309065089649, "loss": 0.3412, "step": 161500 }, { "epoch": 2.76, "learning_rate": 0.00011678654968062415, "loss": 0.3411, "step": 162000 }, { "epoch": 2.77, "learning_rate": 0.00011659964709277792, "loss": 0.3414, "step": 162500 }, { "epoch": 2.77, "learning_rate": 0.00011641200908936181, "loss": 0.3414, "step": 163000 }, { "epoch": 2.78, "learning_rate": 0.00011622401171411614, "loss": 0.3412, "step": 163500 }, { "epoch": 2.79, "learning_rate": 0.00011603565682249906, "loss": 0.3411, "step": 164000 }, { "epoch": 2.8, "learning_rate": 0.00011584732404830404, "loss": 0.3408, "step": 164500 }, { "epoch": 2.81, "learning_rate": 0.0001156586388892709, "loss": 0.3405, "step": 165000 }, { "epoch": 2.82, "learning_rate": 0.00011546922402047642, "loss": 0.3403, "step": 165500 }, { "epoch": 2.83, "learning_rate": 0.00011527945908476035, "loss": 0.3409, "step": 166000 }, { "epoch": 2.83, "learning_rate": 0.000115089345955026, "loss": 0.3406, "step": 166500 }, { "epoch": 2.84, "learning_rate": 0.00011489926777088424, "loss": 0.3406, "step": 167000 }, { "epoch": 2.85, "learning_rate": 0.00011470846457254845, "loss": 0.3403, "step": 167500 }, { "epoch": 2.86, "learning_rate": 0.00011451731881567971, "loss": 0.3402, "step": 168000 }, { "epoch": 2.87, "learning_rate": 0.00011432583238680949, "loss": 0.3405, "step": 168500 }, { "epoch": 2.88, "learning_rate": 0.00011413400717583146, "loss": 0.3401, "step": 169000 }, { "epoch": 2.89, "learning_rate": 0.00011394184507598311, "loss": 0.3398, "step": 169500 }, { "epoch": 2.89, "learning_rate": 0.00011374934798382673, "loss": 0.3396, "step": 170000 }, { "epoch": 2.9, "learning_rate": 0.00011355651779923088, "loss": 0.3393, "step": 170500 }, { "epoch": 2.91, "learning_rate": 0.00011336335642535168, "loss": 0.3391, "step": 171000 }, { "epoch": 2.92, "learning_rate": 0.00011316986576861393, "loss": 0.3394, "step": 171500 }, { "epoch": 2.93, "learning_rate": 0.00011297604773869221, "loss": 0.3395, "step": 172000 }, { "epoch": 2.94, "learning_rate": 0.00011278229285900902, "loss": 0.3397, "step": 172500 }, { "epoch": 2.94, "learning_rate": 0.00011258782646982266, "loss": 0.3396, "step": 173000 }, { "epoch": 2.95, "learning_rate": 0.0001123930384519453, "loss": 0.3392, "step": 173500 }, { "epoch": 2.96, "learning_rate": 0.00011219793072785603, "loss": 0.3394, "step": 174000 }, { "epoch": 2.97, "learning_rate": 0.00011200250522318928, "loss": 0.3391, "step": 174500 }, { "epoch": 2.98, "learning_rate": 0.00011180676386671593, "loss": 0.3389, "step": 175000 }, { "epoch": 2.99, "learning_rate": 0.00011161070859032411, "loss": 0.3387, "step": 175500 }, { "epoch": 3.0, "learning_rate": 0.00011141434132900025, "loss": 0.3385, "step": 176000 }, { "epoch": 3.0, "learning_rate": 0.00011121805768356343, "loss": 0.339, "step": 176500 }, { "epoch": 3.01, "learning_rate": 0.00011102107288390457, "loss": 0.3386, "step": 177000 }, { "epoch": 3.02, "learning_rate": 0.00011082378191878007, "loss": 0.3383, "step": 177500 }, { "epoch": 3.03, "learning_rate": 0.000110626582228053, "loss": 0.3385, "step": 178000 }, { "epoch": 3.04, "learning_rate": 0.00011042908147320174, "loss": 0.3391, "step": 178500 }, { "epoch": 3.05, "learning_rate": 0.00011023128159568977, "loss": 0.339, "step": 179000 }, { "epoch": 3.06, "learning_rate": 0.00011003278725476702, "loss": 0.3386, "step": 179500 }, { "epoch": 3.06, "learning_rate": 0.00010983399650235882, "loss": 0.3384, "step": 180000 }, { "epoch": 3.07, "learning_rate": 0.00010963491130044959, "loss": 0.3385, "step": 180500 }, { "epoch": 3.08, "learning_rate": 0.00010943553361392991, "loss": 0.339, "step": 181000 }, { "epoch": 3.09, "learning_rate": 0.00010923586541057708, "loss": 0.3388, "step": 181500 }, { "epoch": 3.1, "learning_rate": 0.00010903590866103563, "loss": 0.3386, "step": 182000 }, { "epoch": 3.11, "learning_rate": 0.00010883606611012889, "loss": 0.3387, "step": 182500 }, { "epoch": 3.11, "learning_rate": 0.00010863553875873429, "loss": 0.3383, "step": 183000 }, { "epoch": 3.12, "learning_rate": 0.00010843472878613276, "loss": 0.3387, "step": 183500 }, { "epoch": 3.13, "learning_rate": 0.0001082336381742377, "loss": 0.3381, "step": 184000 }, { "epoch": 3.14, "learning_rate": 0.00010803226890773223, "loss": 0.3381, "step": 184500 }, { "epoch": 3.15, "learning_rate": 0.00010783062297404974, "loss": 0.338, "step": 185000 }, { "epoch": 3.16, "learning_rate": 0.00010762870236335414, "loss": 0.3379, "step": 185500 }, { "epoch": 3.17, "learning_rate": 0.0001074265090685204, "loss": 0.338, "step": 186000 }, { "epoch": 3.17, "learning_rate": 0.00010722404508511468, "loss": 0.3377, "step": 186500 }, { "epoch": 3.18, "learning_rate": 0.00010702131241137474, "loss": 0.3376, "step": 187000 }, { "epoch": 3.19, "learning_rate": 0.00010681831304819025, "loss": 0.3378, "step": 187500 }, { "epoch": 3.2, "learning_rate": 0.00010661504899908298, "loss": 0.3377, "step": 188000 }, { "epoch": 3.21, "learning_rate": 0.00010641192958446453, "loss": 0.3384, "step": 188500 }, { "epoch": 3.22, "learning_rate": 0.00010620814270384223, "loss": 0.3378, "step": 189000 }, { "epoch": 3.23, "learning_rate": 0.00010600450550732924, "loss": 0.3378, "step": 189500 }, { "epoch": 3.23, "learning_rate": 0.00010580020382426679, "loss": 0.3376, "step": 190000 }, { "epoch": 3.24, "learning_rate": 0.00010559564750360856, "loss": 0.3375, "step": 190500 }, { "epoch": 3.25, "learning_rate": 0.00010539083856424286, "loss": 0.3372, "step": 191000 }, { "epoch": 3.26, "learning_rate": 0.00010518618939537595, "loss": 0.337, "step": 191500 }, { "epoch": 3.27, "learning_rate": 0.00010498088178033872, "loss": 0.3367, "step": 192000 }, { "epoch": 3.28, "learning_rate": 0.0001047753276140835, "loss": 0.3368, "step": 192500 }, { "epoch": 3.29, "learning_rate": 0.00010456952892534694, "loss": 0.3367, "step": 193000 }, { "epoch": 3.29, "learning_rate": 0.00010436348774527898, "loss": 0.3364, "step": 193500 }, { "epoch": 3.3, "learning_rate": 0.00010415720610742296, "loss": 0.3366, "step": 194000 }, { "epoch": 3.31, "learning_rate": 0.00010395068604769528, "loss": 0.3365, "step": 194500 }, { "epoch": 3.32, "learning_rate": 0.00010374392960436561, "loss": 0.3369, "step": 195000 }, { "epoch": 3.33, "learning_rate": 0.00010353693881803657, "loss": 0.3364, "step": 195500 }, { "epoch": 3.34, "learning_rate": 0.00010332971573162363, "loss": 0.3363, "step": 196000 }, { "epoch": 3.34, "learning_rate": 0.00010312226239033498, "loss": 0.336, "step": 196500 }, { "epoch": 3.35, "learning_rate": 0.0001029145808416514, "loss": 0.336, "step": 197000 }, { "epoch": 3.36, "learning_rate": 0.00010270667313530591, "loss": 0.3357, "step": 197500 }, { "epoch": 3.37, "learning_rate": 0.0001024985413232636, "loss": 0.3357, "step": 198000 }, { "epoch": 3.38, "learning_rate": 0.0001022901874597015, "loss": 0.3354, "step": 198500 }, { "epoch": 3.39, "learning_rate": 0.00010208161360098805, "loss": 0.3355, "step": 199000 }, { "epoch": 3.4, "learning_rate": 0.00010187282180566307, "loss": 0.3354, "step": 199500 }, { "epoch": 3.4, "learning_rate": 0.00010166506882009926, "loss": 0.3356, "step": 200000 }, { "epoch": 3.41, "learning_rate": 0.00010145626726491797, "loss": 0.3363, "step": 200500 }, { "epoch": 3.42, "learning_rate": 0.00010124683571819103, "loss": 0.3363, "step": 201000 }, { "epoch": 3.43, "learning_rate": 0.00010103719447377321, "loss": 0.3363, "step": 201500 }, { "epoch": 3.44, "learning_rate": 0.00010082734560073903, "loss": 0.3358, "step": 202000 }, { "epoch": 3.45, "learning_rate": 0.00010061729117021209, "loss": 0.336, "step": 202500 }, { "epoch": 3.46, "learning_rate": 0.00010040703325534484, "loss": 0.336, "step": 203000 }, { "epoch": 3.46, "learning_rate": 0.00010019657393129802, "loss": 0.3359, "step": 203500 }, { "epoch": 3.47, "learning_rate": 9.998591527522016e-05, "loss": 0.3357, "step": 204000 }, { "epoch": 3.48, "learning_rate": 9.977505936622718e-05, "loss": 0.3356, "step": 204500 }, { "epoch": 3.49, "learning_rate": 9.956400828538178e-05, "loss": 0.3358, "step": 205000 }, { "epoch": 3.5, "learning_rate": 9.935318679532955e-05, "loss": 0.3353, "step": 205500 }, { "epoch": 3.51, "learning_rate": 9.914217506040476e-05, "loss": 0.3356, "step": 206000 }, { "epoch": 3.51, "learning_rate": 9.893055172104751e-05, "loss": 0.3354, "step": 206500 }, { "epoch": 3.52, "learning_rate": 9.871874154478737e-05, "loss": 0.3352, "step": 207000 }, { "epoch": 3.53, "learning_rate": 9.850674662210523e-05, "loss": 0.3349, "step": 207500 }, { "epoch": 3.54, "learning_rate": 9.829456904530539e-05, "loss": 0.3349, "step": 208000 }, { "epoch": 3.55, "learning_rate": 9.808263580357464e-05, "loss": 0.3351, "step": 208500 }, { "epoch": 3.56, "learning_rate": 9.787052480668757e-05, "loss": 0.3352, "step": 209000 }, { "epoch": 3.57, "learning_rate": 9.76578125405694e-05, "loss": 0.3355, "step": 209500 }, { "epoch": 3.57, "learning_rate": 9.744535194458411e-05, "loss": 0.3354, "step": 210000 }, { "epoch": 3.58, "learning_rate": 9.723229357085885e-05, "loss": 0.3354, "step": 210500 }, { "epoch": 3.59, "learning_rate": 9.701906512135297e-05, "loss": 0.3355, "step": 211000 }, { "epoch": 3.6, "learning_rate": 9.680566870054522e-05, "loss": 0.3348, "step": 211500 }, { "epoch": 3.61, "learning_rate": 9.659210641457217e-05, "loss": 0.3349, "step": 212000 }, { "epoch": 3.62, "learning_rate": 9.637838037120726e-05, "loss": 0.3346, "step": 212500 }, { "epoch": 3.63, "learning_rate": 9.616449267984031e-05, "loss": 0.3346, "step": 213000 }, { "epoch": 3.63, "learning_rate": 9.59504454514565e-05, "loss": 0.3343, "step": 213500 }, { "epoch": 3.64, "learning_rate": 9.573666936362591e-05, "loss": 0.3341, "step": 214000 }, { "epoch": 3.65, "learning_rate": 9.5522309708951e-05, "loss": 0.3344, "step": 214500 }, { "epoch": 3.66, "learning_rate": 9.53077968553459e-05, "loss": 0.3339, "step": 215000 } ], "max_steps": 500000, "num_train_epochs": 9, "total_flos": 0.0, "trial_name": null, "trial_params": null }