diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,8421 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.1765497760024717, + "eval_steps": 500, + "global_step": 6000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 53.40031433105469, + "learning_rate": 5e-06, + "loss": 12.357, + "step": 5 + }, + { + "epoch": 0.0, + "grad_norm": 46.18193817138672, + "learning_rate": 1e-05, + "loss": 11.8094, + "step": 10 + }, + { + "epoch": 0.0, + "grad_norm": 40.68670654296875, + "learning_rate": 1.5e-05, + "loss": 10.7979, + "step": 15 + }, + { + "epoch": 0.0, + "grad_norm": 22.82342529296875, + "learning_rate": 2e-05, + "loss": 9.6584, + "step": 20 + }, + { + "epoch": 0.0, + "grad_norm": 12.88599967956543, + "learning_rate": 2.5e-05, + "loss": 9.0296, + "step": 25 + }, + { + "epoch": 0.0, + "grad_norm": 7.4407782554626465, + "learning_rate": 3e-05, + "loss": 8.862, + "step": 30 + }, + { + "epoch": 0.0, + "grad_norm": 6.075540065765381, + "learning_rate": 3.5e-05, + "loss": 8.8179, + "step": 35 + }, + { + "epoch": 0.0, + "grad_norm": 5.882711887359619, + "learning_rate": 4e-05, + "loss": 8.7705, + "step": 40 + }, + { + "epoch": 0.0, + "grad_norm": 6.74395751953125, + "learning_rate": 4.5e-05, + "loss": 8.6586, + "step": 45 + }, + { + "epoch": 0.0, + "grad_norm": 5.318438529968262, + "learning_rate": 5e-05, + "loss": 8.4054, + "step": 50 + }, + { + "epoch": 0.0, + "grad_norm": 6.7087883949279785, + "learning_rate": 5.500000000000001e-05, + "loss": 8.1832, + "step": 55 + }, + { + "epoch": 0.0, + "grad_norm": 6.021890640258789, + "learning_rate": 6e-05, + "loss": 8.1139, + "step": 60 + }, + { + "epoch": 0.0, + "grad_norm": 6.515568256378174, + "learning_rate": 6.500000000000001e-05, + "loss": 7.8704, + "step": 65 + }, + { + "epoch": 0.0, + "grad_norm": 4.975854873657227, + "learning_rate": 7e-05, + "loss": 7.781, + "step": 70 + }, + { + "epoch": 0.0, + "grad_norm": 6.243017196655273, + "learning_rate": 7.500000000000001e-05, + "loss": 7.6151, + "step": 75 + }, + { + "epoch": 0.0, + "grad_norm": 4.646777629852295, + "learning_rate": 8e-05, + "loss": 7.3846, + "step": 80 + }, + { + "epoch": 0.0, + "grad_norm": 4.6621503829956055, + "learning_rate": 8.5e-05, + "loss": 7.1813, + "step": 85 + }, + { + "epoch": 0.0, + "grad_norm": 7.049370288848877, + "learning_rate": 9e-05, + "loss": 7.0247, + "step": 90 + }, + { + "epoch": 0.0, + "grad_norm": 5.0783185958862305, + "learning_rate": 9.5e-05, + "loss": 6.896, + "step": 95 + }, + { + "epoch": 0.0, + "grad_norm": 4.685661792755127, + "learning_rate": 0.0001, + "loss": 6.6327, + "step": 100 + }, + { + "epoch": 0.0, + "grad_norm": 4.423750400543213, + "learning_rate": 9.999999462732701e-05, + "loss": 6.5843, + "step": 105 + }, + { + "epoch": 0.0, + "grad_norm": 3.9485859870910645, + "learning_rate": 9.999997850930916e-05, + "loss": 6.4402, + "step": 110 + }, + { + "epoch": 0.0, + "grad_norm": 3.3882434368133545, + "learning_rate": 9.999995164594991e-05, + "loss": 6.3868, + "step": 115 + }, + { + "epoch": 0.0, + "grad_norm": 4.029066562652588, + "learning_rate": 9.999991403725507e-05, + "loss": 6.2127, + "step": 120 + }, + { + "epoch": 0.0, + "grad_norm": 3.4108099937438965, + "learning_rate": 9.99998656832327e-05, + "loss": 6.2279, + "step": 125 + }, + { + "epoch": 0.0, + "grad_norm": 3.042518377304077, + "learning_rate": 9.999980658389317e-05, + "loss": 6.0734, + "step": 130 + }, + { + "epoch": 0.0, + "grad_norm": 8.128089904785156, + "learning_rate": 9.999973673924921e-05, + "loss": 6.1032, + "step": 135 + }, + { + "epoch": 0.0, + "grad_norm": 4.333005905151367, + "learning_rate": 9.999965614931583e-05, + "loss": 6.1, + "step": 140 + }, + { + "epoch": 0.0, + "grad_norm": 3.101750135421753, + "learning_rate": 9.999956481411035e-05, + "loss": 6.019, + "step": 145 + }, + { + "epoch": 0.0, + "grad_norm": 3.08514404296875, + "learning_rate": 9.999946273365238e-05, + "loss": 6.0091, + "step": 150 + }, + { + "epoch": 0.0, + "grad_norm": 4.458160400390625, + "learning_rate": 9.999934990796387e-05, + "loss": 5.9941, + "step": 155 + }, + { + "epoch": 0.0, + "grad_norm": 3.456221342086792, + "learning_rate": 9.999922633706906e-05, + "loss": 5.7836, + "step": 160 + }, + { + "epoch": 0.0, + "grad_norm": 3.13814377784729, + "learning_rate": 9.999909202099453e-05, + "loss": 5.79, + "step": 165 + }, + { + "epoch": 0.01, + "grad_norm": 3.343902349472046, + "learning_rate": 9.999894695976912e-05, + "loss": 5.7679, + "step": 170 + }, + { + "epoch": 0.01, + "grad_norm": 2.463756561279297, + "learning_rate": 9.9998791153424e-05, + "loss": 5.6633, + "step": 175 + }, + { + "epoch": 0.01, + "grad_norm": 2.939295768737793, + "learning_rate": 9.999862460199267e-05, + "loss": 5.6737, + "step": 180 + }, + { + "epoch": 0.01, + "grad_norm": 3.0761544704437256, + "learning_rate": 9.999844730551092e-05, + "loss": 5.6778, + "step": 185 + }, + { + "epoch": 0.01, + "grad_norm": 4.1005778312683105, + "learning_rate": 9.999825926401686e-05, + "loss": 5.5748, + "step": 190 + }, + { + "epoch": 0.01, + "grad_norm": 3.084240436553955, + "learning_rate": 9.999806047755088e-05, + "loss": 5.6615, + "step": 195 + }, + { + "epoch": 0.01, + "grad_norm": 3.453418016433716, + "learning_rate": 9.999785094615573e-05, + "loss": 5.4704, + "step": 200 + }, + { + "epoch": 0.01, + "grad_norm": 2.6560404300689697, + "learning_rate": 9.99976306698764e-05, + "loss": 5.4657, + "step": 205 + }, + { + "epoch": 0.01, + "grad_norm": 2.5087716579437256, + "learning_rate": 9.999739964876026e-05, + "loss": 5.4744, + "step": 210 + }, + { + "epoch": 0.01, + "grad_norm": 2.5395045280456543, + "learning_rate": 9.999715788285695e-05, + "loss": 5.3517, + "step": 215 + }, + { + "epoch": 0.01, + "grad_norm": 3.417387008666992, + "learning_rate": 9.999690537221844e-05, + "loss": 5.3933, + "step": 220 + }, + { + "epoch": 0.01, + "grad_norm": 2.5052287578582764, + "learning_rate": 9.999664211689897e-05, + "loss": 5.4842, + "step": 225 + }, + { + "epoch": 0.01, + "grad_norm": 2.3961122035980225, + "learning_rate": 9.999636811695513e-05, + "loss": 5.3338, + "step": 230 + }, + { + "epoch": 0.01, + "grad_norm": 4.5207929611206055, + "learning_rate": 9.99960833724458e-05, + "loss": 5.3798, + "step": 235 + }, + { + "epoch": 0.01, + "grad_norm": 2.4632210731506348, + "learning_rate": 9.999578788343218e-05, + "loss": 5.222, + "step": 240 + }, + { + "epoch": 0.01, + "grad_norm": 2.3988687992095947, + "learning_rate": 9.999548164997778e-05, + "loss": 5.376, + "step": 245 + }, + { + "epoch": 0.01, + "grad_norm": 2.5891101360321045, + "learning_rate": 9.999516467214839e-05, + "loss": 5.2554, + "step": 250 + }, + { + "epoch": 0.01, + "grad_norm": 2.408022880554199, + "learning_rate": 9.999483695001214e-05, + "loss": 5.1993, + "step": 255 + }, + { + "epoch": 0.01, + "grad_norm": 2.2588911056518555, + "learning_rate": 9.999449848363946e-05, + "loss": 5.0835, + "step": 260 + }, + { + "epoch": 0.01, + "grad_norm": 2.498990774154663, + "learning_rate": 9.999414927310309e-05, + "loss": 5.1574, + "step": 265 + }, + { + "epoch": 0.01, + "grad_norm": 2.3929696083068848, + "learning_rate": 9.99937893184781e-05, + "loss": 5.0843, + "step": 270 + }, + { + "epoch": 0.01, + "grad_norm": 3.079639196395874, + "learning_rate": 9.99934186198418e-05, + "loss": 5.1971, + "step": 275 + }, + { + "epoch": 0.01, + "grad_norm": 2.3947091102600098, + "learning_rate": 9.999303717727389e-05, + "loss": 5.0761, + "step": 280 + }, + { + "epoch": 0.01, + "grad_norm": 2.9632606506347656, + "learning_rate": 9.999264499085633e-05, + "loss": 5.2332, + "step": 285 + }, + { + "epoch": 0.01, + "grad_norm": 2.2455055713653564, + "learning_rate": 9.999224206067341e-05, + "loss": 5.1196, + "step": 290 + }, + { + "epoch": 0.01, + "grad_norm": 2.0398213863372803, + "learning_rate": 9.999182838681171e-05, + "loss": 5.0213, + "step": 295 + }, + { + "epoch": 0.01, + "grad_norm": 2.003916025161743, + "learning_rate": 9.999140396936016e-05, + "loss": 4.9621, + "step": 300 + }, + { + "epoch": 0.01, + "grad_norm": 2.212294340133667, + "learning_rate": 9.999096880840994e-05, + "loss": 4.9991, + "step": 305 + }, + { + "epoch": 0.01, + "grad_norm": 2.6768558025360107, + "learning_rate": 9.999052290405459e-05, + "loss": 4.9915, + "step": 310 + }, + { + "epoch": 0.01, + "grad_norm": 2.460846185684204, + "learning_rate": 9.999006625638994e-05, + "loss": 4.9958, + "step": 315 + }, + { + "epoch": 0.01, + "grad_norm": 2.033607244491577, + "learning_rate": 9.998959886551411e-05, + "loss": 4.9831, + "step": 320 + }, + { + "epoch": 0.01, + "grad_norm": 1.9576119184494019, + "learning_rate": 9.998912073152755e-05, + "loss": 4.9127, + "step": 325 + }, + { + "epoch": 0.01, + "grad_norm": 17.875638961791992, + "learning_rate": 9.998863185453302e-05, + "loss": 4.8488, + "step": 330 + }, + { + "epoch": 0.01, + "grad_norm": 2.9045534133911133, + "learning_rate": 9.998813223463557e-05, + "loss": 5.0234, + "step": 335 + }, + { + "epoch": 0.01, + "grad_norm": 1.881321907043457, + "learning_rate": 9.998762187194258e-05, + "loss": 4.9818, + "step": 340 + }, + { + "epoch": 0.01, + "grad_norm": 2.7776927947998047, + "learning_rate": 9.998710076656373e-05, + "loss": 4.9621, + "step": 345 + }, + { + "epoch": 0.01, + "grad_norm": 3.389038324356079, + "learning_rate": 9.998656891861102e-05, + "loss": 4.9083, + "step": 350 + }, + { + "epoch": 0.01, + "grad_norm": 2.7731683254241943, + "learning_rate": 9.998602632819874e-05, + "loss": 4.8998, + "step": 355 + }, + { + "epoch": 0.01, + "grad_norm": 2.8221945762634277, + "learning_rate": 9.998547299544348e-05, + "loss": 4.8623, + "step": 360 + }, + { + "epoch": 0.01, + "grad_norm": 2.151465892791748, + "learning_rate": 9.99849089204642e-05, + "loss": 4.8068, + "step": 365 + }, + { + "epoch": 0.01, + "grad_norm": 1.9444559812545776, + "learning_rate": 9.998433410338206e-05, + "loss": 4.7842, + "step": 370 + }, + { + "epoch": 0.01, + "grad_norm": 3.131049633026123, + "learning_rate": 9.998374854432064e-05, + "loss": 4.6842, + "step": 375 + }, + { + "epoch": 0.01, + "grad_norm": 2.0474250316619873, + "learning_rate": 9.998315224340577e-05, + "loss": 4.8366, + "step": 380 + }, + { + "epoch": 0.01, + "grad_norm": 1.8289575576782227, + "learning_rate": 9.99825452007656e-05, + "loss": 4.7341, + "step": 385 + }, + { + "epoch": 0.01, + "grad_norm": 4.700809478759766, + "learning_rate": 9.998192741653057e-05, + "loss": 4.8011, + "step": 390 + }, + { + "epoch": 0.01, + "grad_norm": 2.144601583480835, + "learning_rate": 9.998129889083346e-05, + "loss": 4.8091, + "step": 395 + }, + { + "epoch": 0.01, + "grad_norm": 2.0928916931152344, + "learning_rate": 9.998065962380936e-05, + "loss": 4.7802, + "step": 400 + }, + { + "epoch": 0.01, + "grad_norm": 2.3940982818603516, + "learning_rate": 9.998000961559562e-05, + "loss": 4.6233, + "step": 405 + }, + { + "epoch": 0.01, + "grad_norm": 1.9397621154785156, + "learning_rate": 9.997934886633195e-05, + "loss": 4.7236, + "step": 410 + }, + { + "epoch": 0.01, + "grad_norm": 2.2297585010528564, + "learning_rate": 9.997867737616035e-05, + "loss": 4.6891, + "step": 415 + }, + { + "epoch": 0.01, + "grad_norm": 3.5420126914978027, + "learning_rate": 9.997799514522513e-05, + "loss": 4.7119, + "step": 420 + }, + { + "epoch": 0.01, + "grad_norm": 2.504870891571045, + "learning_rate": 9.99773021736729e-05, + "loss": 4.6529, + "step": 425 + }, + { + "epoch": 0.01, + "grad_norm": 2.6012051105499268, + "learning_rate": 9.99765984616526e-05, + "loss": 4.749, + "step": 430 + }, + { + "epoch": 0.01, + "grad_norm": 3.6821699142456055, + "learning_rate": 9.997588400931542e-05, + "loss": 4.7926, + "step": 435 + }, + { + "epoch": 0.01, + "grad_norm": 2.571035146713257, + "learning_rate": 9.997515881681496e-05, + "loss": 4.6978, + "step": 440 + }, + { + "epoch": 0.01, + "grad_norm": 2.228336811065674, + "learning_rate": 9.9974422884307e-05, + "loss": 4.663, + "step": 445 + }, + { + "epoch": 0.01, + "grad_norm": 1.9010071754455566, + "learning_rate": 9.997367621194977e-05, + "loss": 4.6081, + "step": 450 + }, + { + "epoch": 0.01, + "grad_norm": 2.0545012950897217, + "learning_rate": 9.99729187999037e-05, + "loss": 4.5983, + "step": 455 + }, + { + "epoch": 0.01, + "grad_norm": 2.011831045150757, + "learning_rate": 9.997215064833155e-05, + "loss": 4.6558, + "step": 460 + }, + { + "epoch": 0.01, + "grad_norm": 1.9707472324371338, + "learning_rate": 9.99713717573984e-05, + "loss": 4.5787, + "step": 465 + }, + { + "epoch": 0.01, + "grad_norm": 4.238839149475098, + "learning_rate": 9.997058212727169e-05, + "loss": 4.6272, + "step": 470 + }, + { + "epoch": 0.01, + "grad_norm": 2.0530169010162354, + "learning_rate": 9.996978175812105e-05, + "loss": 4.5355, + "step": 475 + }, + { + "epoch": 0.01, + "grad_norm": 2.0808329582214355, + "learning_rate": 9.996897065011852e-05, + "loss": 4.7705, + "step": 480 + }, + { + "epoch": 0.01, + "grad_norm": 2.6587131023406982, + "learning_rate": 9.996814880343842e-05, + "loss": 4.6555, + "step": 485 + }, + { + "epoch": 0.01, + "grad_norm": 1.9901516437530518, + "learning_rate": 9.996731621825736e-05, + "loss": 4.512, + "step": 490 + }, + { + "epoch": 0.01, + "grad_norm": 1.9074862003326416, + "learning_rate": 9.996647289475426e-05, + "loss": 4.5713, + "step": 495 + }, + { + "epoch": 0.01, + "grad_norm": 2.065422296524048, + "learning_rate": 9.996561883311036e-05, + "loss": 4.5273, + "step": 500 + }, + { + "epoch": 0.01, + "grad_norm": 2.092353105545044, + "learning_rate": 9.996475403350921e-05, + "loss": 4.4907, + "step": 505 + }, + { + "epoch": 0.02, + "grad_norm": 2.095177173614502, + "learning_rate": 9.996387849613667e-05, + "loss": 4.5194, + "step": 510 + }, + { + "epoch": 0.02, + "grad_norm": 1.6454585790634155, + "learning_rate": 9.996299222118086e-05, + "loss": 4.5851, + "step": 515 + }, + { + "epoch": 0.02, + "grad_norm": 1.7797366380691528, + "learning_rate": 9.99620952088323e-05, + "loss": 4.5143, + "step": 520 + }, + { + "epoch": 0.02, + "grad_norm": 1.713352918624878, + "learning_rate": 9.996118745928373e-05, + "loss": 4.5963, + "step": 525 + }, + { + "epoch": 0.02, + "grad_norm": 1.757695198059082, + "learning_rate": 9.996026897273024e-05, + "loss": 4.4696, + "step": 530 + }, + { + "epoch": 0.02, + "grad_norm": 1.9573163986206055, + "learning_rate": 9.995933974936922e-05, + "loss": 4.496, + "step": 535 + }, + { + "epoch": 0.02, + "grad_norm": 1.871819257736206, + "learning_rate": 9.995839978940036e-05, + "loss": 4.481, + "step": 540 + }, + { + "epoch": 0.02, + "grad_norm": 1.8936526775360107, + "learning_rate": 9.995744909302567e-05, + "loss": 4.5889, + "step": 545 + }, + { + "epoch": 0.02, + "grad_norm": 1.9036074876785278, + "learning_rate": 9.995648766044947e-05, + "loss": 4.49, + "step": 550 + }, + { + "epoch": 0.02, + "grad_norm": 1.78731369972229, + "learning_rate": 9.995551549187836e-05, + "loss": 4.581, + "step": 555 + }, + { + "epoch": 0.02, + "grad_norm": 1.716098427772522, + "learning_rate": 9.995453258752127e-05, + "loss": 4.3321, + "step": 560 + }, + { + "epoch": 0.02, + "grad_norm": 1.601149559020996, + "learning_rate": 9.995353894758945e-05, + "loss": 4.4168, + "step": 565 + }, + { + "epoch": 0.02, + "grad_norm": 1.7256990671157837, + "learning_rate": 9.995253457229643e-05, + "loss": 4.4181, + "step": 570 + }, + { + "epoch": 0.02, + "grad_norm": 1.7807209491729736, + "learning_rate": 9.995151946185804e-05, + "loss": 4.5075, + "step": 575 + }, + { + "epoch": 0.02, + "grad_norm": 1.679442048072815, + "learning_rate": 9.995049361649247e-05, + "loss": 4.3712, + "step": 580 + }, + { + "epoch": 0.02, + "grad_norm": 1.6432149410247803, + "learning_rate": 9.994945703642013e-05, + "loss": 4.5308, + "step": 585 + }, + { + "epoch": 0.02, + "grad_norm": 1.7374026775360107, + "learning_rate": 9.994840972186386e-05, + "loss": 4.5147, + "step": 590 + }, + { + "epoch": 0.02, + "grad_norm": 1.7408748865127563, + "learning_rate": 9.994735167304866e-05, + "loss": 4.448, + "step": 595 + }, + { + "epoch": 0.02, + "grad_norm": 1.7771662473678589, + "learning_rate": 9.994628289020195e-05, + "loss": 4.3363, + "step": 600 + }, + { + "epoch": 0.02, + "grad_norm": 2.2652812004089355, + "learning_rate": 9.994520337355344e-05, + "loss": 4.3429, + "step": 605 + }, + { + "epoch": 0.02, + "grad_norm": 2.032334327697754, + "learning_rate": 9.994411312333507e-05, + "loss": 4.4761, + "step": 610 + }, + { + "epoch": 0.02, + "grad_norm": 1.5236808061599731, + "learning_rate": 9.99430121397812e-05, + "loss": 4.5198, + "step": 615 + }, + { + "epoch": 0.02, + "grad_norm": 1.5445153713226318, + "learning_rate": 9.994190042312839e-05, + "loss": 4.4431, + "step": 620 + }, + { + "epoch": 0.02, + "grad_norm": 1.7421549558639526, + "learning_rate": 9.994077797361559e-05, + "loss": 4.4322, + "step": 625 + }, + { + "epoch": 0.02, + "grad_norm": 1.5315024852752686, + "learning_rate": 9.9939644791484e-05, + "loss": 4.3377, + "step": 630 + }, + { + "epoch": 0.02, + "grad_norm": 1.887389063835144, + "learning_rate": 9.993850087697716e-05, + "loss": 4.3453, + "step": 635 + }, + { + "epoch": 0.02, + "grad_norm": 1.8790143728256226, + "learning_rate": 9.99373462303409e-05, + "loss": 4.3521, + "step": 640 + }, + { + "epoch": 0.02, + "grad_norm": 1.5290522575378418, + "learning_rate": 9.993618085182337e-05, + "loss": 4.3287, + "step": 645 + }, + { + "epoch": 0.02, + "grad_norm": 1.6216731071472168, + "learning_rate": 9.993500474167501e-05, + "loss": 4.3345, + "step": 650 + }, + { + "epoch": 0.02, + "grad_norm": 1.5201752185821533, + "learning_rate": 9.99338179001486e-05, + "loss": 4.284, + "step": 655 + }, + { + "epoch": 0.02, + "grad_norm": 1.5362247228622437, + "learning_rate": 9.993262032749914e-05, + "loss": 4.2759, + "step": 660 + }, + { + "epoch": 0.02, + "grad_norm": 1.6463065147399902, + "learning_rate": 9.993141202398407e-05, + "loss": 4.2841, + "step": 665 + }, + { + "epoch": 0.02, + "grad_norm": 1.6376757621765137, + "learning_rate": 9.993019298986301e-05, + "loss": 4.3057, + "step": 670 + }, + { + "epoch": 0.02, + "grad_norm": 1.470133662223816, + "learning_rate": 9.992896322539796e-05, + "loss": 4.2641, + "step": 675 + }, + { + "epoch": 0.02, + "grad_norm": 1.5886218547821045, + "learning_rate": 9.992772273085319e-05, + "loss": 4.2911, + "step": 680 + }, + { + "epoch": 0.02, + "grad_norm": 1.6224122047424316, + "learning_rate": 9.99264715064953e-05, + "loss": 4.4176, + "step": 685 + }, + { + "epoch": 0.02, + "grad_norm": 1.470272421836853, + "learning_rate": 9.992520955259319e-05, + "loss": 4.2921, + "step": 690 + }, + { + "epoch": 0.02, + "grad_norm": 1.9054683446884155, + "learning_rate": 9.992393686941808e-05, + "loss": 4.2424, + "step": 695 + }, + { + "epoch": 0.02, + "grad_norm": 1.6701070070266724, + "learning_rate": 9.992265345724344e-05, + "loss": 4.3357, + "step": 700 + }, + { + "epoch": 0.02, + "grad_norm": 1.4918732643127441, + "learning_rate": 9.99213593163451e-05, + "loss": 4.3432, + "step": 705 + }, + { + "epoch": 0.02, + "grad_norm": 1.5541293621063232, + "learning_rate": 9.992005444700121e-05, + "loss": 4.2593, + "step": 710 + }, + { + "epoch": 0.02, + "grad_norm": 1.5747699737548828, + "learning_rate": 9.991873884949214e-05, + "loss": 4.3348, + "step": 715 + }, + { + "epoch": 0.02, + "grad_norm": 1.4049266576766968, + "learning_rate": 9.991741252410067e-05, + "loss": 4.2806, + "step": 720 + }, + { + "epoch": 0.02, + "grad_norm": 1.4820458889007568, + "learning_rate": 9.99160754711118e-05, + "loss": 4.2543, + "step": 725 + }, + { + "epoch": 0.02, + "grad_norm": 1.8397589921951294, + "learning_rate": 9.99147276908129e-05, + "loss": 4.2046, + "step": 730 + }, + { + "epoch": 0.02, + "grad_norm": 1.5264321565628052, + "learning_rate": 9.991336918349361e-05, + "loss": 4.3166, + "step": 735 + }, + { + "epoch": 0.02, + "grad_norm": 1.7536901235580444, + "learning_rate": 9.991199994944586e-05, + "loss": 4.1905, + "step": 740 + }, + { + "epoch": 0.02, + "grad_norm": 1.8334250450134277, + "learning_rate": 9.991061998896393e-05, + "loss": 4.1802, + "step": 745 + }, + { + "epoch": 0.02, + "grad_norm": 1.6241623163223267, + "learning_rate": 9.99092293023444e-05, + "loss": 4.1424, + "step": 750 + }, + { + "epoch": 0.02, + "grad_norm": 1.6951171159744263, + "learning_rate": 9.990782788988611e-05, + "loss": 4.2014, + "step": 755 + }, + { + "epoch": 0.02, + "grad_norm": 1.4351670742034912, + "learning_rate": 9.990641575189022e-05, + "loss": 4.1817, + "step": 760 + }, + { + "epoch": 0.02, + "grad_norm": 1.409759283065796, + "learning_rate": 9.990499288866024e-05, + "loss": 4.1793, + "step": 765 + }, + { + "epoch": 0.02, + "grad_norm": 1.4067161083221436, + "learning_rate": 9.990355930050194e-05, + "loss": 4.2829, + "step": 770 + }, + { + "epoch": 0.02, + "grad_norm": 1.9757593870162964, + "learning_rate": 9.990211498772341e-05, + "loss": 4.2151, + "step": 775 + }, + { + "epoch": 0.02, + "grad_norm": 1.7121752500534058, + "learning_rate": 9.990065995063506e-05, + "loss": 4.2459, + "step": 780 + }, + { + "epoch": 0.02, + "grad_norm": 1.8657753467559814, + "learning_rate": 9.989919418954956e-05, + "loss": 4.1444, + "step": 785 + }, + { + "epoch": 0.02, + "grad_norm": 1.5653979778289795, + "learning_rate": 9.989771770478191e-05, + "loss": 4.1272, + "step": 790 + }, + { + "epoch": 0.02, + "grad_norm": 1.483831763267517, + "learning_rate": 9.989623049664944e-05, + "loss": 4.305, + "step": 795 + }, + { + "epoch": 0.02, + "grad_norm": 1.47762131690979, + "learning_rate": 9.989473256547175e-05, + "loss": 4.178, + "step": 800 + }, + { + "epoch": 0.02, + "grad_norm": 2.0902953147888184, + "learning_rate": 9.989322391157076e-05, + "loss": 4.3145, + "step": 805 + }, + { + "epoch": 0.02, + "grad_norm": 1.4770091772079468, + "learning_rate": 9.989170453527068e-05, + "loss": 4.1518, + "step": 810 + }, + { + "epoch": 0.02, + "grad_norm": 1.5169522762298584, + "learning_rate": 9.989017443689805e-05, + "loss": 4.1888, + "step": 815 + }, + { + "epoch": 0.02, + "grad_norm": 1.5774534940719604, + "learning_rate": 9.988863361678169e-05, + "loss": 4.2305, + "step": 820 + }, + { + "epoch": 0.02, + "grad_norm": 1.6230158805847168, + "learning_rate": 9.988708207525272e-05, + "loss": 4.0553, + "step": 825 + }, + { + "epoch": 0.02, + "grad_norm": 1.6451687812805176, + "learning_rate": 9.98855198126446e-05, + "loss": 4.051, + "step": 830 + }, + { + "epoch": 0.02, + "grad_norm": 1.4685438871383667, + "learning_rate": 9.988394682929305e-05, + "loss": 4.0169, + "step": 835 + }, + { + "epoch": 0.02, + "grad_norm": 1.3612847328186035, + "learning_rate": 9.988236312553614e-05, + "loss": 4.1332, + "step": 840 + }, + { + "epoch": 0.02, + "grad_norm": 1.6016806364059448, + "learning_rate": 9.988076870171419e-05, + "loss": 4.0773, + "step": 845 + }, + { + "epoch": 0.03, + "grad_norm": 1.3579212427139282, + "learning_rate": 9.987916355816988e-05, + "loss": 4.1718, + "step": 850 + }, + { + "epoch": 0.03, + "grad_norm": 1.3493858575820923, + "learning_rate": 9.987754769524814e-05, + "loss": 4.1674, + "step": 855 + }, + { + "epoch": 0.03, + "grad_norm": 1.4003750085830688, + "learning_rate": 9.987592111329626e-05, + "loss": 4.1567, + "step": 860 + }, + { + "epoch": 0.03, + "grad_norm": 1.7907814979553223, + "learning_rate": 9.987428381266377e-05, + "loss": 4.1239, + "step": 865 + }, + { + "epoch": 0.03, + "grad_norm": 1.6452932357788086, + "learning_rate": 9.987263579370257e-05, + "loss": 4.1446, + "step": 870 + }, + { + "epoch": 0.03, + "grad_norm": 1.5121257305145264, + "learning_rate": 9.987097705676681e-05, + "loss": 4.1587, + "step": 875 + }, + { + "epoch": 0.03, + "grad_norm": 1.3372843265533447, + "learning_rate": 9.986930760221296e-05, + "loss": 4.1626, + "step": 880 + }, + { + "epoch": 0.03, + "grad_norm": 1.2996832132339478, + "learning_rate": 9.986762743039983e-05, + "loss": 4.082, + "step": 885 + }, + { + "epoch": 0.03, + "grad_norm": 1.6313203573226929, + "learning_rate": 9.986593654168846e-05, + "loss": 4.2007, + "step": 890 + }, + { + "epoch": 0.03, + "grad_norm": 1.4239163398742676, + "learning_rate": 9.986423493644225e-05, + "loss": 4.213, + "step": 895 + }, + { + "epoch": 0.03, + "grad_norm": 1.465247631072998, + "learning_rate": 9.98625226150269e-05, + "loss": 4.0601, + "step": 900 + }, + { + "epoch": 0.03, + "grad_norm": 1.7289196252822876, + "learning_rate": 9.986079957781038e-05, + "loss": 4.1762, + "step": 905 + }, + { + "epoch": 0.03, + "grad_norm": 1.3119356632232666, + "learning_rate": 9.9859065825163e-05, + "loss": 4.0798, + "step": 910 + }, + { + "epoch": 0.03, + "grad_norm": 1.3670289516448975, + "learning_rate": 9.985732135745733e-05, + "loss": 4.0833, + "step": 915 + }, + { + "epoch": 0.03, + "grad_norm": 1.3917007446289062, + "learning_rate": 9.985556617506828e-05, + "loss": 4.0507, + "step": 920 + }, + { + "epoch": 0.03, + "grad_norm": 1.314989447593689, + "learning_rate": 9.985380027837308e-05, + "loss": 4.1653, + "step": 925 + }, + { + "epoch": 0.03, + "grad_norm": 1.396897315979004, + "learning_rate": 9.98520236677512e-05, + "loss": 4.1426, + "step": 930 + }, + { + "epoch": 0.03, + "grad_norm": 1.406545639038086, + "learning_rate": 9.985023634358443e-05, + "loss": 4.0727, + "step": 935 + }, + { + "epoch": 0.03, + "grad_norm": 1.3325015306472778, + "learning_rate": 9.984843830625691e-05, + "loss": 3.9963, + "step": 940 + }, + { + "epoch": 0.03, + "grad_norm": 1.3812161684036255, + "learning_rate": 9.984662955615507e-05, + "loss": 3.9669, + "step": 945 + }, + { + "epoch": 0.03, + "grad_norm": 1.4124517440795898, + "learning_rate": 9.984481009366758e-05, + "loss": 4.137, + "step": 950 + }, + { + "epoch": 0.03, + "grad_norm": 1.550917625427246, + "learning_rate": 9.984297991918548e-05, + "loss": 3.9109, + "step": 955 + }, + { + "epoch": 0.03, + "grad_norm": 1.5635894536972046, + "learning_rate": 9.984113903310206e-05, + "loss": 4.0374, + "step": 960 + }, + { + "epoch": 0.03, + "grad_norm": 1.4015119075775146, + "learning_rate": 9.983928743581296e-05, + "loss": 3.9637, + "step": 965 + }, + { + "epoch": 0.03, + "grad_norm": 1.352587103843689, + "learning_rate": 9.983742512771611e-05, + "loss": 4.0087, + "step": 970 + }, + { + "epoch": 0.03, + "grad_norm": 1.6774641275405884, + "learning_rate": 9.983555210921171e-05, + "loss": 4.0078, + "step": 975 + }, + { + "epoch": 0.03, + "grad_norm": 1.4251840114593506, + "learning_rate": 9.983366838070231e-05, + "loss": 4.0014, + "step": 980 + }, + { + "epoch": 0.03, + "grad_norm": 1.8969238996505737, + "learning_rate": 9.983177394259271e-05, + "loss": 3.9955, + "step": 985 + }, + { + "epoch": 0.03, + "grad_norm": 1.8307636976242065, + "learning_rate": 9.982986879529006e-05, + "loss": 4.0473, + "step": 990 + }, + { + "epoch": 0.03, + "grad_norm": 1.3204973936080933, + "learning_rate": 9.982795293920377e-05, + "loss": 4.0044, + "step": 995 + }, + { + "epoch": 0.03, + "grad_norm": 1.6905070543289185, + "learning_rate": 9.982602637474558e-05, + "loss": 4.0892, + "step": 1000 + }, + { + "epoch": 0.03, + "grad_norm": 1.2660964727401733, + "learning_rate": 9.982408910232952e-05, + "loss": 4.0071, + "step": 1005 + }, + { + "epoch": 0.03, + "grad_norm": 1.4075195789337158, + "learning_rate": 9.982214112237193e-05, + "loss": 3.9502, + "step": 1010 + }, + { + "epoch": 0.03, + "grad_norm": 1.6859352588653564, + "learning_rate": 9.982018243529146e-05, + "loss": 3.9774, + "step": 1015 + }, + { + "epoch": 0.03, + "grad_norm": 1.674471378326416, + "learning_rate": 9.981821304150901e-05, + "loss": 4.0626, + "step": 1020 + }, + { + "epoch": 0.03, + "grad_norm": 1.5338214635849, + "learning_rate": 9.981623294144782e-05, + "loss": 3.9948, + "step": 1025 + }, + { + "epoch": 0.03, + "grad_norm": 1.482784390449524, + "learning_rate": 9.981424213553344e-05, + "loss": 3.9176, + "step": 1030 + }, + { + "epoch": 0.03, + "grad_norm": 1.4336692094802856, + "learning_rate": 9.981224062419372e-05, + "loss": 4.0609, + "step": 1035 + }, + { + "epoch": 0.03, + "grad_norm": 1.4444942474365234, + "learning_rate": 9.981022840785877e-05, + "loss": 4.0358, + "step": 1040 + }, + { + "epoch": 0.03, + "grad_norm": 1.3189609050750732, + "learning_rate": 9.980820548696106e-05, + "loss": 4.0662, + "step": 1045 + }, + { + "epoch": 0.03, + "grad_norm": 1.3676936626434326, + "learning_rate": 9.98061718619353e-05, + "loss": 3.963, + "step": 1050 + }, + { + "epoch": 0.03, + "grad_norm": 1.372725248336792, + "learning_rate": 9.980412753321856e-05, + "loss": 3.9721, + "step": 1055 + }, + { + "epoch": 0.03, + "grad_norm": 1.3731446266174316, + "learning_rate": 9.980207250125015e-05, + "loss": 4.0267, + "step": 1060 + }, + { + "epoch": 0.03, + "grad_norm": 1.3433283567428589, + "learning_rate": 9.980000676647174e-05, + "loss": 3.8895, + "step": 1065 + }, + { + "epoch": 0.03, + "grad_norm": 1.3674778938293457, + "learning_rate": 9.979793032932725e-05, + "loss": 3.8503, + "step": 1070 + }, + { + "epoch": 0.03, + "grad_norm": 1.63779878616333, + "learning_rate": 9.979584319026292e-05, + "loss": 3.8553, + "step": 1075 + }, + { + "epoch": 0.03, + "grad_norm": 1.4132360219955444, + "learning_rate": 9.97937453497273e-05, + "loss": 3.8523, + "step": 1080 + }, + { + "epoch": 0.03, + "grad_norm": 1.690613865852356, + "learning_rate": 9.979163680817124e-05, + "loss": 3.8742, + "step": 1085 + }, + { + "epoch": 0.03, + "grad_norm": 1.4929730892181396, + "learning_rate": 9.978951756604786e-05, + "loss": 3.9432, + "step": 1090 + }, + { + "epoch": 0.03, + "grad_norm": 1.3180598020553589, + "learning_rate": 9.978738762381262e-05, + "loss": 3.8802, + "step": 1095 + }, + { + "epoch": 0.03, + "grad_norm": 1.2295198440551758, + "learning_rate": 9.978524698192322e-05, + "loss": 4.0083, + "step": 1100 + }, + { + "epoch": 0.03, + "grad_norm": 1.2866483926773071, + "learning_rate": 9.978309564083975e-05, + "loss": 3.9818, + "step": 1105 + }, + { + "epoch": 0.03, + "grad_norm": 1.2235194444656372, + "learning_rate": 9.97809336010245e-05, + "loss": 3.9192, + "step": 1110 + }, + { + "epoch": 0.03, + "grad_norm": 3.3693974018096924, + "learning_rate": 9.977876086294216e-05, + "loss": 3.9671, + "step": 1115 + }, + { + "epoch": 0.03, + "grad_norm": 1.6152361631393433, + "learning_rate": 9.977657742705962e-05, + "loss": 4.0059, + "step": 1120 + }, + { + "epoch": 0.03, + "grad_norm": 1.7285512685775757, + "learning_rate": 9.977438329384614e-05, + "loss": 3.8421, + "step": 1125 + }, + { + "epoch": 0.03, + "grad_norm": 1.3544543981552124, + "learning_rate": 9.977217846377324e-05, + "loss": 4.0165, + "step": 1130 + }, + { + "epoch": 0.03, + "grad_norm": 1.3274308443069458, + "learning_rate": 9.976996293731476e-05, + "loss": 3.9339, + "step": 1135 + }, + { + "epoch": 0.03, + "grad_norm": 1.5880746841430664, + "learning_rate": 9.976773671494684e-05, + "loss": 4.0722, + "step": 1140 + }, + { + "epoch": 0.03, + "grad_norm": 1.3821755647659302, + "learning_rate": 9.976549979714791e-05, + "loss": 3.96, + "step": 1145 + }, + { + "epoch": 0.03, + "grad_norm": 1.4131568670272827, + "learning_rate": 9.976325218439868e-05, + "loss": 4.006, + "step": 1150 + }, + { + "epoch": 0.03, + "grad_norm": 1.360153079032898, + "learning_rate": 9.976099387718218e-05, + "loss": 3.9398, + "step": 1155 + }, + { + "epoch": 0.03, + "grad_norm": 1.2989766597747803, + "learning_rate": 9.975872487598377e-05, + "loss": 3.9837, + "step": 1160 + }, + { + "epoch": 0.03, + "grad_norm": 1.5363812446594238, + "learning_rate": 9.975644518129104e-05, + "loss": 3.9198, + "step": 1165 + }, + { + "epoch": 0.03, + "grad_norm": 1.3748068809509277, + "learning_rate": 9.975415479359394e-05, + "loss": 3.8987, + "step": 1170 + }, + { + "epoch": 0.03, + "grad_norm": 1.3851642608642578, + "learning_rate": 9.975185371338464e-05, + "loss": 3.9655, + "step": 1175 + }, + { + "epoch": 0.03, + "grad_norm": 1.2977722883224487, + "learning_rate": 9.97495419411577e-05, + "loss": 3.9315, + "step": 1180 + }, + { + "epoch": 0.03, + "grad_norm": 1.447440266609192, + "learning_rate": 9.974721947740995e-05, + "loss": 3.8491, + "step": 1185 + }, + { + "epoch": 0.04, + "grad_norm": 1.2443561553955078, + "learning_rate": 9.974488632264046e-05, + "loss": 3.844, + "step": 1190 + }, + { + "epoch": 0.04, + "grad_norm": 1.4530870914459229, + "learning_rate": 9.974254247735066e-05, + "loss": 3.8786, + "step": 1195 + }, + { + "epoch": 0.04, + "grad_norm": 1.469127893447876, + "learning_rate": 9.974018794204426e-05, + "loss": 3.9679, + "step": 1200 + }, + { + "epoch": 0.04, + "grad_norm": 1.2994585037231445, + "learning_rate": 9.973782271722727e-05, + "loss": 3.9077, + "step": 1205 + }, + { + "epoch": 0.04, + "grad_norm": 1.6947623491287231, + "learning_rate": 9.973544680340799e-05, + "loss": 3.9191, + "step": 1210 + }, + { + "epoch": 0.04, + "grad_norm": 1.6417971849441528, + "learning_rate": 9.973306020109701e-05, + "loss": 3.8768, + "step": 1215 + }, + { + "epoch": 0.04, + "grad_norm": 1.6052528619766235, + "learning_rate": 9.973066291080724e-05, + "loss": 3.9687, + "step": 1220 + }, + { + "epoch": 0.04, + "grad_norm": 1.2613556385040283, + "learning_rate": 9.972825493305387e-05, + "loss": 3.8199, + "step": 1225 + }, + { + "epoch": 0.04, + "grad_norm": 1.2953615188598633, + "learning_rate": 9.97258362683544e-05, + "loss": 3.8852, + "step": 1230 + }, + { + "epoch": 0.04, + "grad_norm": 1.3388254642486572, + "learning_rate": 9.972340691722858e-05, + "loss": 3.8348, + "step": 1235 + }, + { + "epoch": 0.04, + "grad_norm": 1.2172560691833496, + "learning_rate": 9.972096688019856e-05, + "loss": 3.8791, + "step": 1240 + }, + { + "epoch": 0.04, + "grad_norm": 1.2731328010559082, + "learning_rate": 9.971851615778868e-05, + "loss": 3.8566, + "step": 1245 + }, + { + "epoch": 0.04, + "grad_norm": 1.2738902568817139, + "learning_rate": 9.971605475052561e-05, + "loss": 4.0365, + "step": 1250 + }, + { + "epoch": 0.04, + "grad_norm": 1.3196699619293213, + "learning_rate": 9.971358265893834e-05, + "loss": 3.8147, + "step": 1255 + }, + { + "epoch": 0.04, + "grad_norm": 1.3698867559432983, + "learning_rate": 9.971109988355814e-05, + "loss": 3.8507, + "step": 1260 + }, + { + "epoch": 0.04, + "grad_norm": 1.2317814826965332, + "learning_rate": 9.970860642491854e-05, + "loss": 3.7834, + "step": 1265 + }, + { + "epoch": 0.04, + "grad_norm": 1.250658631324768, + "learning_rate": 9.970610228355548e-05, + "loss": 3.9033, + "step": 1270 + }, + { + "epoch": 0.04, + "grad_norm": 1.209499716758728, + "learning_rate": 9.970358746000704e-05, + "loss": 3.8984, + "step": 1275 + }, + { + "epoch": 0.04, + "grad_norm": 1.1895166635513306, + "learning_rate": 9.97010619548137e-05, + "loss": 3.8162, + "step": 1280 + }, + { + "epoch": 0.04, + "grad_norm": 1.210045337677002, + "learning_rate": 9.969852576851822e-05, + "loss": 3.9672, + "step": 1285 + }, + { + "epoch": 0.04, + "grad_norm": 1.298948049545288, + "learning_rate": 9.969597890166565e-05, + "loss": 3.8016, + "step": 1290 + }, + { + "epoch": 0.04, + "grad_norm": 1.429323673248291, + "learning_rate": 9.969342135480328e-05, + "loss": 3.7862, + "step": 1295 + }, + { + "epoch": 0.04, + "grad_norm": 1.4861477613449097, + "learning_rate": 9.969085312848081e-05, + "loss": 3.843, + "step": 1300 + }, + { + "epoch": 0.04, + "grad_norm": 1.55156672000885, + "learning_rate": 9.968827422325012e-05, + "loss": 3.7755, + "step": 1305 + }, + { + "epoch": 0.04, + "grad_norm": 1.2622857093811035, + "learning_rate": 9.968568463966545e-05, + "loss": 3.8329, + "step": 1310 + }, + { + "epoch": 0.04, + "grad_norm": 1.3178651332855225, + "learning_rate": 9.968308437828333e-05, + "loss": 3.8049, + "step": 1315 + }, + { + "epoch": 0.04, + "grad_norm": 1.1905425786972046, + "learning_rate": 9.968047343966256e-05, + "loss": 3.7873, + "step": 1320 + }, + { + "epoch": 0.04, + "grad_norm": 1.4271931648254395, + "learning_rate": 9.967785182436426e-05, + "loss": 3.8192, + "step": 1325 + }, + { + "epoch": 0.04, + "grad_norm": 1.2527958154678345, + "learning_rate": 9.967521953295185e-05, + "loss": 3.7842, + "step": 1330 + }, + { + "epoch": 0.04, + "grad_norm": 1.201155424118042, + "learning_rate": 9.967257656599099e-05, + "loss": 3.8645, + "step": 1335 + }, + { + "epoch": 0.04, + "grad_norm": 1.3820469379425049, + "learning_rate": 9.966992292404967e-05, + "loss": 3.7031, + "step": 1340 + }, + { + "epoch": 0.04, + "grad_norm": 1.248611569404602, + "learning_rate": 9.966725860769821e-05, + "loss": 3.7997, + "step": 1345 + }, + { + "epoch": 0.04, + "grad_norm": 1.5514973402023315, + "learning_rate": 9.966458361750916e-05, + "loss": 3.854, + "step": 1350 + }, + { + "epoch": 0.04, + "grad_norm": 1.4789297580718994, + "learning_rate": 9.966189795405743e-05, + "loss": 3.8269, + "step": 1355 + }, + { + "epoch": 0.04, + "grad_norm": 1.665956974029541, + "learning_rate": 9.965920161792016e-05, + "loss": 3.802, + "step": 1360 + }, + { + "epoch": 0.04, + "grad_norm": 1.5528392791748047, + "learning_rate": 9.96564946096768e-05, + "loss": 3.7707, + "step": 1365 + }, + { + "epoch": 0.04, + "grad_norm": 1.3726645708084106, + "learning_rate": 9.965377692990914e-05, + "loss": 3.8331, + "step": 1370 + }, + { + "epoch": 0.04, + "grad_norm": 1.3941130638122559, + "learning_rate": 9.965104857920121e-05, + "loss": 3.6449, + "step": 1375 + }, + { + "epoch": 0.04, + "grad_norm": 1.26205313205719, + "learning_rate": 9.964830955813935e-05, + "loss": 3.8732, + "step": 1380 + }, + { + "epoch": 0.04, + "grad_norm": 1.4130189418792725, + "learning_rate": 9.964555986731219e-05, + "loss": 3.8514, + "step": 1385 + }, + { + "epoch": 0.04, + "grad_norm": 1.3098586797714233, + "learning_rate": 9.964279950731066e-05, + "loss": 3.7325, + "step": 1390 + }, + { + "epoch": 0.04, + "grad_norm": 1.1799935102462769, + "learning_rate": 9.9640028478728e-05, + "loss": 3.8093, + "step": 1395 + }, + { + "epoch": 0.04, + "grad_norm": 1.1888481378555298, + "learning_rate": 9.963724678215968e-05, + "loss": 3.7819, + "step": 1400 + }, + { + "epoch": 0.04, + "grad_norm": 1.3047587871551514, + "learning_rate": 9.963445441820356e-05, + "loss": 3.819, + "step": 1405 + }, + { + "epoch": 0.04, + "grad_norm": 1.199859380722046, + "learning_rate": 9.963165138745969e-05, + "loss": 3.7111, + "step": 1410 + }, + { + "epoch": 0.04, + "grad_norm": 1.271870493888855, + "learning_rate": 9.962883769053051e-05, + "loss": 3.7756, + "step": 1415 + }, + { + "epoch": 0.04, + "grad_norm": 1.5547105073928833, + "learning_rate": 9.962601332802065e-05, + "loss": 3.9523, + "step": 1420 + }, + { + "epoch": 0.04, + "grad_norm": 1.189194917678833, + "learning_rate": 9.962317830053711e-05, + "loss": 3.8383, + "step": 1425 + }, + { + "epoch": 0.04, + "grad_norm": 1.2303493022918701, + "learning_rate": 9.962033260868916e-05, + "loss": 3.8695, + "step": 1430 + }, + { + "epoch": 0.04, + "grad_norm": 1.2686803340911865, + "learning_rate": 9.961747625308838e-05, + "loss": 3.7101, + "step": 1435 + }, + { + "epoch": 0.04, + "grad_norm": 1.5070773363113403, + "learning_rate": 9.961460923434858e-05, + "loss": 3.803, + "step": 1440 + }, + { + "epoch": 0.04, + "grad_norm": 1.3258230686187744, + "learning_rate": 9.961173155308591e-05, + "loss": 3.8025, + "step": 1445 + }, + { + "epoch": 0.04, + "grad_norm": 1.3113139867782593, + "learning_rate": 9.960884320991881e-05, + "loss": 3.8426, + "step": 1450 + }, + { + "epoch": 0.04, + "grad_norm": 1.2064306735992432, + "learning_rate": 9.960594420546802e-05, + "loss": 3.6898, + "step": 1455 + }, + { + "epoch": 0.04, + "grad_norm": 1.3960572481155396, + "learning_rate": 9.960303454035654e-05, + "loss": 3.6818, + "step": 1460 + }, + { + "epoch": 0.04, + "grad_norm": 1.2599711418151855, + "learning_rate": 9.960011421520969e-05, + "loss": 3.714, + "step": 1465 + }, + { + "epoch": 0.04, + "grad_norm": 1.1433742046356201, + "learning_rate": 9.959718323065503e-05, + "loss": 3.6758, + "step": 1470 + }, + { + "epoch": 0.04, + "grad_norm": 1.2926206588745117, + "learning_rate": 9.959424158732252e-05, + "loss": 3.7365, + "step": 1475 + }, + { + "epoch": 0.04, + "grad_norm": 1.335339903831482, + "learning_rate": 9.959128928584426e-05, + "loss": 3.6283, + "step": 1480 + }, + { + "epoch": 0.04, + "grad_norm": 1.3005627393722534, + "learning_rate": 9.958832632685478e-05, + "loss": 3.765, + "step": 1485 + }, + { + "epoch": 0.04, + "grad_norm": 1.3055315017700195, + "learning_rate": 9.95853527109908e-05, + "loss": 3.7006, + "step": 1490 + }, + { + "epoch": 0.04, + "grad_norm": 1.3264316320419312, + "learning_rate": 9.95823684388914e-05, + "loss": 3.7067, + "step": 1495 + }, + { + "epoch": 0.04, + "grad_norm": 1.2863095998764038, + "learning_rate": 9.95793735111979e-05, + "loss": 3.7036, + "step": 1500 + }, + { + "epoch": 0.04, + "grad_norm": 1.2964081764221191, + "learning_rate": 9.957636792855394e-05, + "loss": 3.7884, + "step": 1505 + }, + { + "epoch": 0.04, + "grad_norm": 1.4622471332550049, + "learning_rate": 9.957335169160545e-05, + "loss": 3.7047, + "step": 1510 + }, + { + "epoch": 0.04, + "grad_norm": 1.2015630006790161, + "learning_rate": 9.957032480100061e-05, + "loss": 3.8186, + "step": 1515 + }, + { + "epoch": 0.04, + "grad_norm": 1.2337404489517212, + "learning_rate": 9.956728725738996e-05, + "loss": 3.7033, + "step": 1520 + }, + { + "epoch": 0.04, + "grad_norm": 1.1596229076385498, + "learning_rate": 9.956423906142626e-05, + "loss": 3.7337, + "step": 1525 + }, + { + "epoch": 0.05, + "grad_norm": 1.233035922050476, + "learning_rate": 9.956118021376461e-05, + "loss": 3.7014, + "step": 1530 + }, + { + "epoch": 0.05, + "grad_norm": 1.3229448795318604, + "learning_rate": 9.955811071506236e-05, + "loss": 3.7927, + "step": 1535 + }, + { + "epoch": 0.05, + "grad_norm": 1.4028730392456055, + "learning_rate": 9.955503056597917e-05, + "loss": 3.648, + "step": 1540 + }, + { + "epoch": 0.05, + "grad_norm": 1.1624157428741455, + "learning_rate": 9.955193976717698e-05, + "loss": 3.7398, + "step": 1545 + }, + { + "epoch": 0.05, + "grad_norm": 1.2442816495895386, + "learning_rate": 9.954883831932003e-05, + "loss": 3.771, + "step": 1550 + }, + { + "epoch": 0.05, + "grad_norm": 1.300835371017456, + "learning_rate": 9.954572622307485e-05, + "loss": 3.7425, + "step": 1555 + }, + { + "epoch": 0.05, + "grad_norm": 1.3888286352157593, + "learning_rate": 9.954260347911025e-05, + "loss": 3.5886, + "step": 1560 + }, + { + "epoch": 0.05, + "grad_norm": 1.3998864889144897, + "learning_rate": 9.953947008809731e-05, + "loss": 3.742, + "step": 1565 + }, + { + "epoch": 0.05, + "grad_norm": 1.30398428440094, + "learning_rate": 9.953632605070944e-05, + "loss": 3.6666, + "step": 1570 + }, + { + "epoch": 0.05, + "grad_norm": 1.2791111469268799, + "learning_rate": 9.95331713676223e-05, + "loss": 3.6994, + "step": 1575 + }, + { + "epoch": 0.05, + "grad_norm": 1.18183434009552, + "learning_rate": 9.953000603951386e-05, + "loss": 3.8104, + "step": 1580 + }, + { + "epoch": 0.05, + "grad_norm": 1.4292824268341064, + "learning_rate": 9.952683006706438e-05, + "loss": 3.8529, + "step": 1585 + }, + { + "epoch": 0.05, + "grad_norm": 1.3154456615447998, + "learning_rate": 9.952364345095638e-05, + "loss": 3.6518, + "step": 1590 + }, + { + "epoch": 0.05, + "grad_norm": 1.7707338333129883, + "learning_rate": 9.95204461918747e-05, + "loss": 3.5699, + "step": 1595 + }, + { + "epoch": 0.05, + "grad_norm": 1.4309860467910767, + "learning_rate": 9.951723829050646e-05, + "loss": 3.6895, + "step": 1600 + }, + { + "epoch": 0.05, + "grad_norm": 1.2321257591247559, + "learning_rate": 9.951401974754101e-05, + "loss": 3.7096, + "step": 1605 + }, + { + "epoch": 0.05, + "grad_norm": 1.2556930780410767, + "learning_rate": 9.951079056367013e-05, + "loss": 3.546, + "step": 1610 + }, + { + "epoch": 0.05, + "grad_norm": 1.4037466049194336, + "learning_rate": 9.95075507395877e-05, + "loss": 3.576, + "step": 1615 + }, + { + "epoch": 0.05, + "grad_norm": 1.2304840087890625, + "learning_rate": 9.950430027599003e-05, + "loss": 3.692, + "step": 1620 + }, + { + "epoch": 0.05, + "grad_norm": 1.3280993700027466, + "learning_rate": 9.950103917357566e-05, + "loss": 3.6979, + "step": 1625 + }, + { + "epoch": 0.05, + "grad_norm": 1.1542185544967651, + "learning_rate": 9.949776743304542e-05, + "loss": 3.6017, + "step": 1630 + }, + { + "epoch": 0.05, + "grad_norm": 1.1935598850250244, + "learning_rate": 9.949448505510243e-05, + "loss": 3.6209, + "step": 1635 + }, + { + "epoch": 0.05, + "grad_norm": 1.3208544254302979, + "learning_rate": 9.94911920404521e-05, + "loss": 3.7188, + "step": 1640 + }, + { + "epoch": 0.05, + "grad_norm": 1.1318814754486084, + "learning_rate": 9.94878883898021e-05, + "loss": 3.6449, + "step": 1645 + }, + { + "epoch": 0.05, + "grad_norm": 1.1852689981460571, + "learning_rate": 9.948457410386244e-05, + "loss": 3.61, + "step": 1650 + }, + { + "epoch": 0.05, + "grad_norm": 1.3725321292877197, + "learning_rate": 9.948124918334538e-05, + "loss": 3.5941, + "step": 1655 + }, + { + "epoch": 0.05, + "grad_norm": 1.1921747922897339, + "learning_rate": 9.947791362896543e-05, + "loss": 3.6686, + "step": 1660 + }, + { + "epoch": 0.05, + "grad_norm": 1.1278502941131592, + "learning_rate": 9.947456744143947e-05, + "loss": 3.5772, + "step": 1665 + }, + { + "epoch": 0.05, + "grad_norm": 1.1490073204040527, + "learning_rate": 9.947121062148658e-05, + "loss": 3.6413, + "step": 1670 + }, + { + "epoch": 0.05, + "grad_norm": 1.2591372728347778, + "learning_rate": 9.94678431698282e-05, + "loss": 3.7417, + "step": 1675 + }, + { + "epoch": 0.05, + "grad_norm": 1.2971771955490112, + "learning_rate": 9.946446508718801e-05, + "loss": 3.6352, + "step": 1680 + }, + { + "epoch": 0.05, + "grad_norm": 1.4097003936767578, + "learning_rate": 9.946107637429196e-05, + "loss": 3.7533, + "step": 1685 + }, + { + "epoch": 0.05, + "grad_norm": 1.2238167524337769, + "learning_rate": 9.945767703186832e-05, + "loss": 3.6449, + "step": 1690 + }, + { + "epoch": 0.05, + "grad_norm": 1.2335211038589478, + "learning_rate": 9.945426706064764e-05, + "loss": 3.7193, + "step": 1695 + }, + { + "epoch": 0.05, + "grad_norm": 1.1348868608474731, + "learning_rate": 9.945084646136275e-05, + "loss": 3.6696, + "step": 1700 + }, + { + "epoch": 0.05, + "grad_norm": 1.2900527715682983, + "learning_rate": 9.944741523474874e-05, + "loss": 3.6886, + "step": 1705 + }, + { + "epoch": 0.05, + "grad_norm": 1.6407312154769897, + "learning_rate": 9.944397338154301e-05, + "loss": 3.6885, + "step": 1710 + }, + { + "epoch": 0.05, + "grad_norm": 1.3822795152664185, + "learning_rate": 9.944052090248526e-05, + "loss": 3.567, + "step": 1715 + }, + { + "epoch": 0.05, + "grad_norm": 1.4265950918197632, + "learning_rate": 9.943705779831742e-05, + "loss": 3.5973, + "step": 1720 + }, + { + "epoch": 0.05, + "grad_norm": 1.1598773002624512, + "learning_rate": 9.943358406978377e-05, + "loss": 3.6546, + "step": 1725 + }, + { + "epoch": 0.05, + "grad_norm": 1.2290794849395752, + "learning_rate": 9.943009971763082e-05, + "loss": 3.6444, + "step": 1730 + }, + { + "epoch": 0.05, + "grad_norm": 1.3381224870681763, + "learning_rate": 9.942660474260737e-05, + "loss": 3.5604, + "step": 1735 + }, + { + "epoch": 0.05, + "grad_norm": 1.1058239936828613, + "learning_rate": 9.942309914546453e-05, + "loss": 3.6693, + "step": 1740 + }, + { + "epoch": 0.05, + "grad_norm": 1.134424090385437, + "learning_rate": 9.941958292695565e-05, + "loss": 3.5146, + "step": 1745 + }, + { + "epoch": 0.05, + "grad_norm": 1.1692516803741455, + "learning_rate": 9.941605608783644e-05, + "loss": 3.7908, + "step": 1750 + }, + { + "epoch": 0.05, + "grad_norm": 1.3359147310256958, + "learning_rate": 9.94125186288648e-05, + "loss": 3.6958, + "step": 1755 + }, + { + "epoch": 0.05, + "grad_norm": 1.1560852527618408, + "learning_rate": 9.940897055080098e-05, + "loss": 3.5964, + "step": 1760 + }, + { + "epoch": 0.05, + "grad_norm": 1.2410101890563965, + "learning_rate": 9.940541185440746e-05, + "loss": 3.6408, + "step": 1765 + }, + { + "epoch": 0.05, + "grad_norm": 1.272270917892456, + "learning_rate": 9.940184254044904e-05, + "loss": 3.6817, + "step": 1770 + }, + { + "epoch": 0.05, + "grad_norm": 1.123008370399475, + "learning_rate": 9.939826260969282e-05, + "loss": 3.5224, + "step": 1775 + }, + { + "epoch": 0.05, + "grad_norm": 1.2507129907608032, + "learning_rate": 9.93946720629081e-05, + "loss": 3.6544, + "step": 1780 + }, + { + "epoch": 0.05, + "grad_norm": 1.3066537380218506, + "learning_rate": 9.939107090086654e-05, + "loss": 3.6612, + "step": 1785 + }, + { + "epoch": 0.05, + "grad_norm": 1.2033692598342896, + "learning_rate": 9.938745912434208e-05, + "loss": 3.5675, + "step": 1790 + }, + { + "epoch": 0.05, + "grad_norm": 1.1097171306610107, + "learning_rate": 9.938383673411087e-05, + "loss": 3.5593, + "step": 1795 + }, + { + "epoch": 0.05, + "grad_norm": 1.24695885181427, + "learning_rate": 9.938020373095141e-05, + "loss": 3.5911, + "step": 1800 + }, + { + "epoch": 0.05, + "grad_norm": 1.331443190574646, + "learning_rate": 9.937656011564445e-05, + "loss": 3.6328, + "step": 1805 + }, + { + "epoch": 0.05, + "grad_norm": 1.1397649049758911, + "learning_rate": 9.937290588897302e-05, + "loss": 3.5421, + "step": 1810 + }, + { + "epoch": 0.05, + "grad_norm": 1.2707165479660034, + "learning_rate": 9.936924105172248e-05, + "loss": 3.6981, + "step": 1815 + }, + { + "epoch": 0.05, + "grad_norm": 1.172749638557434, + "learning_rate": 9.936556560468037e-05, + "loss": 3.5783, + "step": 1820 + }, + { + "epoch": 0.05, + "grad_norm": 1.1949453353881836, + "learning_rate": 9.93618795486366e-05, + "loss": 3.6535, + "step": 1825 + }, + { + "epoch": 0.05, + "grad_norm": 1.2154752016067505, + "learning_rate": 9.935818288438334e-05, + "loss": 3.5337, + "step": 1830 + }, + { + "epoch": 0.05, + "grad_norm": 1.4100817441940308, + "learning_rate": 9.935447561271501e-05, + "loss": 3.5728, + "step": 1835 + }, + { + "epoch": 0.05, + "grad_norm": 1.2381153106689453, + "learning_rate": 9.935075773442835e-05, + "loss": 3.5119, + "step": 1840 + }, + { + "epoch": 0.05, + "grad_norm": 1.1807883977890015, + "learning_rate": 9.934702925032232e-05, + "loss": 3.558, + "step": 1845 + }, + { + "epoch": 0.05, + "grad_norm": 1.2513213157653809, + "learning_rate": 9.934329016119824e-05, + "loss": 3.6233, + "step": 1850 + }, + { + "epoch": 0.05, + "grad_norm": 1.2229536771774292, + "learning_rate": 9.933954046785963e-05, + "loss": 3.5582, + "step": 1855 + }, + { + "epoch": 0.05, + "grad_norm": 1.1728299856185913, + "learning_rate": 9.933578017111233e-05, + "loss": 3.5775, + "step": 1860 + }, + { + "epoch": 0.05, + "grad_norm": 1.3789819478988647, + "learning_rate": 9.933200927176449e-05, + "loss": 3.5387, + "step": 1865 + }, + { + "epoch": 0.06, + "grad_norm": 1.1826677322387695, + "learning_rate": 9.932822777062646e-05, + "loss": 3.5662, + "step": 1870 + }, + { + "epoch": 0.06, + "grad_norm": 1.1156057119369507, + "learning_rate": 9.932443566851092e-05, + "loss": 3.5391, + "step": 1875 + }, + { + "epoch": 0.06, + "grad_norm": 1.1320744752883911, + "learning_rate": 9.932063296623284e-05, + "loss": 3.6027, + "step": 1880 + }, + { + "epoch": 0.06, + "grad_norm": 1.3161002397537231, + "learning_rate": 9.931681966460944e-05, + "loss": 3.5726, + "step": 1885 + }, + { + "epoch": 0.06, + "grad_norm": 1.229048728942871, + "learning_rate": 9.93129957644602e-05, + "loss": 3.5726, + "step": 1890 + }, + { + "epoch": 0.06, + "grad_norm": 1.1488388776779175, + "learning_rate": 9.930916126660691e-05, + "loss": 3.6015, + "step": 1895 + }, + { + "epoch": 0.06, + "grad_norm": 1.9094411134719849, + "learning_rate": 9.930531617187367e-05, + "loss": 3.4992, + "step": 1900 + }, + { + "epoch": 0.06, + "grad_norm": 11.372000694274902, + "learning_rate": 9.930146048108678e-05, + "loss": 3.4827, + "step": 1905 + }, + { + "epoch": 0.06, + "grad_norm": 1.1878877878189087, + "learning_rate": 9.929759419507487e-05, + "loss": 3.6137, + "step": 1910 + }, + { + "epoch": 0.06, + "grad_norm": 1.1705623865127563, + "learning_rate": 9.92937173146688e-05, + "loss": 3.5913, + "step": 1915 + }, + { + "epoch": 0.06, + "grad_norm": 1.1690722703933716, + "learning_rate": 9.928982984070179e-05, + "loss": 3.5711, + "step": 1920 + }, + { + "epoch": 0.06, + "grad_norm": 1.115605354309082, + "learning_rate": 9.928593177400925e-05, + "loss": 3.5134, + "step": 1925 + }, + { + "epoch": 0.06, + "grad_norm": 1.0696338415145874, + "learning_rate": 9.928202311542889e-05, + "loss": 3.5044, + "step": 1930 + }, + { + "epoch": 0.06, + "grad_norm": 1.3975670337677002, + "learning_rate": 9.927810386580075e-05, + "loss": 3.5602, + "step": 1935 + }, + { + "epoch": 0.06, + "grad_norm": 1.1429855823516846, + "learning_rate": 9.927417402596707e-05, + "loss": 3.5935, + "step": 1940 + }, + { + "epoch": 0.06, + "grad_norm": 1.1475850343704224, + "learning_rate": 9.927023359677241e-05, + "loss": 3.5764, + "step": 1945 + }, + { + "epoch": 0.06, + "grad_norm": 1.1566569805145264, + "learning_rate": 9.92662825790636e-05, + "loss": 3.562, + "step": 1950 + }, + { + "epoch": 0.06, + "grad_norm": 1.0919015407562256, + "learning_rate": 9.926232097368974e-05, + "loss": 3.5645, + "step": 1955 + }, + { + "epoch": 0.06, + "grad_norm": 1.128484845161438, + "learning_rate": 9.925834878150219e-05, + "loss": 3.5364, + "step": 1960 + }, + { + "epoch": 0.06, + "grad_norm": 1.1300374269485474, + "learning_rate": 9.925436600335464e-05, + "loss": 3.5287, + "step": 1965 + }, + { + "epoch": 0.06, + "grad_norm": 1.2811986207962036, + "learning_rate": 9.925037264010298e-05, + "loss": 3.5021, + "step": 1970 + }, + { + "epoch": 0.06, + "grad_norm": 1.1949442625045776, + "learning_rate": 9.924636869260542e-05, + "loss": 3.4392, + "step": 1975 + }, + { + "epoch": 0.06, + "grad_norm": 1.0614773035049438, + "learning_rate": 9.924235416172243e-05, + "loss": 3.3968, + "step": 1980 + }, + { + "epoch": 0.06, + "grad_norm": 1.0944995880126953, + "learning_rate": 9.923832904831679e-05, + "loss": 3.5237, + "step": 1985 + }, + { + "epoch": 0.06, + "grad_norm": 1.0735448598861694, + "learning_rate": 9.923429335325349e-05, + "loss": 3.5365, + "step": 1990 + }, + { + "epoch": 0.06, + "grad_norm": 1.2306127548217773, + "learning_rate": 9.923024707739985e-05, + "loss": 3.5476, + "step": 1995 + }, + { + "epoch": 0.06, + "grad_norm": 1.0836900472640991, + "learning_rate": 9.922619022162545e-05, + "loss": 3.5553, + "step": 2000 + }, + { + "epoch": 0.06, + "grad_norm": 1.171622633934021, + "learning_rate": 9.92221227868021e-05, + "loss": 3.5227, + "step": 2005 + }, + { + "epoch": 0.06, + "grad_norm": 1.1157162189483643, + "learning_rate": 9.921804477380394e-05, + "loss": 3.5679, + "step": 2010 + }, + { + "epoch": 0.06, + "grad_norm": 1.1095036268234253, + "learning_rate": 9.921395618350739e-05, + "loss": 3.4219, + "step": 2015 + }, + { + "epoch": 0.06, + "grad_norm": 1.3070727586746216, + "learning_rate": 9.920985701679106e-05, + "loss": 3.4546, + "step": 2020 + }, + { + "epoch": 0.06, + "grad_norm": 1.4068920612335205, + "learning_rate": 9.920574727453595e-05, + "loss": 3.492, + "step": 2025 + }, + { + "epoch": 0.06, + "grad_norm": 1.1070573329925537, + "learning_rate": 9.920162695762522e-05, + "loss": 3.4512, + "step": 2030 + }, + { + "epoch": 0.06, + "grad_norm": 1.0470702648162842, + "learning_rate": 9.919749606694439e-05, + "loss": 3.5969, + "step": 2035 + }, + { + "epoch": 0.06, + "grad_norm": 1.1105985641479492, + "learning_rate": 9.919335460338122e-05, + "loss": 3.5896, + "step": 2040 + }, + { + "epoch": 0.06, + "grad_norm": 1.1698545217514038, + "learning_rate": 9.918920256782572e-05, + "loss": 3.4161, + "step": 2045 + }, + { + "epoch": 0.06, + "grad_norm": 1.1736584901809692, + "learning_rate": 9.918503996117016e-05, + "loss": 3.5549, + "step": 2050 + }, + { + "epoch": 0.06, + "grad_norm": 1.239270567893982, + "learning_rate": 9.918086678430917e-05, + "loss": 3.4994, + "step": 2055 + }, + { + "epoch": 0.06, + "grad_norm": 1.1683988571166992, + "learning_rate": 9.917668303813959e-05, + "loss": 3.4777, + "step": 2060 + }, + { + "epoch": 0.06, + "grad_norm": 1.2546718120574951, + "learning_rate": 9.917248872356051e-05, + "loss": 3.501, + "step": 2065 + }, + { + "epoch": 0.06, + "grad_norm": 1.161113977432251, + "learning_rate": 9.916828384147331e-05, + "loss": 3.5274, + "step": 2070 + }, + { + "epoch": 0.06, + "grad_norm": 1.2397959232330322, + "learning_rate": 9.916406839278167e-05, + "loss": 3.4254, + "step": 2075 + }, + { + "epoch": 0.06, + "grad_norm": 1.135114073753357, + "learning_rate": 9.915984237839152e-05, + "loss": 3.5601, + "step": 2080 + }, + { + "epoch": 0.06, + "grad_norm": 1.2423151731491089, + "learning_rate": 9.915560579921106e-05, + "loss": 3.4426, + "step": 2085 + }, + { + "epoch": 0.06, + "grad_norm": 1.1082744598388672, + "learning_rate": 9.915135865615074e-05, + "loss": 3.6305, + "step": 2090 + }, + { + "epoch": 0.06, + "grad_norm": 1.1283408403396606, + "learning_rate": 9.914710095012332e-05, + "loss": 3.3979, + "step": 2095 + }, + { + "epoch": 0.06, + "grad_norm": 1.3209794759750366, + "learning_rate": 9.91428326820438e-05, + "loss": 3.4129, + "step": 2100 + }, + { + "epoch": 0.06, + "grad_norm": 1.1583530902862549, + "learning_rate": 9.913855385282947e-05, + "loss": 3.5324, + "step": 2105 + }, + { + "epoch": 0.06, + "grad_norm": 1.1608564853668213, + "learning_rate": 9.913426446339986e-05, + "loss": 3.5435, + "step": 2110 + }, + { + "epoch": 0.06, + "grad_norm": 1.0904818773269653, + "learning_rate": 9.912996451467682e-05, + "loss": 3.5937, + "step": 2115 + }, + { + "epoch": 0.06, + "grad_norm": 1.1132928133010864, + "learning_rate": 9.912565400758443e-05, + "loss": 3.4965, + "step": 2120 + }, + { + "epoch": 0.06, + "grad_norm": 1.1378835439682007, + "learning_rate": 9.912133294304901e-05, + "loss": 3.4342, + "step": 2125 + }, + { + "epoch": 0.06, + "grad_norm": 1.0805127620697021, + "learning_rate": 9.911700132199924e-05, + "loss": 3.4143, + "step": 2130 + }, + { + "epoch": 0.06, + "grad_norm": 1.2845242023468018, + "learning_rate": 9.911265914536598e-05, + "loss": 3.5065, + "step": 2135 + }, + { + "epoch": 0.06, + "grad_norm": 1.2314151525497437, + "learning_rate": 9.910830641408242e-05, + "loss": 3.4214, + "step": 2140 + }, + { + "epoch": 0.06, + "grad_norm": 1.2160024642944336, + "learning_rate": 9.910394312908397e-05, + "loss": 3.5221, + "step": 2145 + }, + { + "epoch": 0.06, + "grad_norm": 1.084956169128418, + "learning_rate": 9.909956929130834e-05, + "loss": 3.4286, + "step": 2150 + }, + { + "epoch": 0.06, + "grad_norm": 1.1384857892990112, + "learning_rate": 9.90951849016955e-05, + "loss": 3.3824, + "step": 2155 + }, + { + "epoch": 0.06, + "grad_norm": 1.0912495851516724, + "learning_rate": 9.909078996118768e-05, + "loss": 3.5683, + "step": 2160 + }, + { + "epoch": 0.06, + "grad_norm": 1.04654061794281, + "learning_rate": 9.908638447072939e-05, + "loss": 3.4216, + "step": 2165 + }, + { + "epoch": 0.06, + "grad_norm": 1.1329139471054077, + "learning_rate": 9.90819684312674e-05, + "loss": 3.5459, + "step": 2170 + }, + { + "epoch": 0.06, + "grad_norm": 1.1518476009368896, + "learning_rate": 9.907754184375074e-05, + "loss": 3.4809, + "step": 2175 + }, + { + "epoch": 0.06, + "grad_norm": 1.1370935440063477, + "learning_rate": 9.90731047091307e-05, + "loss": 3.5388, + "step": 2180 + }, + { + "epoch": 0.06, + "grad_norm": 1.2988322973251343, + "learning_rate": 9.906865702836088e-05, + "loss": 3.5395, + "step": 2185 + }, + { + "epoch": 0.06, + "grad_norm": 1.2206143140792847, + "learning_rate": 9.90641988023971e-05, + "loss": 3.483, + "step": 2190 + }, + { + "epoch": 0.06, + "grad_norm": 1.0459623336791992, + "learning_rate": 9.905973003219748e-05, + "loss": 3.534, + "step": 2195 + }, + { + "epoch": 0.06, + "grad_norm": 1.1394981145858765, + "learning_rate": 9.905525071872237e-05, + "loss": 3.5048, + "step": 2200 + }, + { + "epoch": 0.06, + "grad_norm": 1.0579923391342163, + "learning_rate": 9.905076086293443e-05, + "loss": 3.5077, + "step": 2205 + }, + { + "epoch": 0.07, + "grad_norm": 1.176694631576538, + "learning_rate": 9.904626046579852e-05, + "loss": 3.428, + "step": 2210 + }, + { + "epoch": 0.07, + "grad_norm": 1.1430487632751465, + "learning_rate": 9.904174952828186e-05, + "loss": 3.4802, + "step": 2215 + }, + { + "epoch": 0.07, + "grad_norm": 1.1191006898880005, + "learning_rate": 9.903722805135385e-05, + "loss": 3.524, + "step": 2220 + }, + { + "epoch": 0.07, + "grad_norm": 1.2326856851577759, + "learning_rate": 9.903269603598617e-05, + "loss": 3.533, + "step": 2225 + }, + { + "epoch": 0.07, + "grad_norm": 1.3681950569152832, + "learning_rate": 9.902815348315282e-05, + "loss": 3.4235, + "step": 2230 + }, + { + "epoch": 0.07, + "grad_norm": 1.1448445320129395, + "learning_rate": 9.902360039383e-05, + "loss": 3.5322, + "step": 2235 + }, + { + "epoch": 0.07, + "grad_norm": 1.1583805084228516, + "learning_rate": 9.901903676899621e-05, + "loss": 3.5336, + "step": 2240 + }, + { + "epoch": 0.07, + "grad_norm": 1.1489771604537964, + "learning_rate": 9.901446260963221e-05, + "loss": 3.507, + "step": 2245 + }, + { + "epoch": 0.07, + "grad_norm": 1.1143720149993896, + "learning_rate": 9.9009877916721e-05, + "loss": 3.4261, + "step": 2250 + }, + { + "epoch": 0.07, + "grad_norm": 1.121553659439087, + "learning_rate": 9.900528269124787e-05, + "loss": 3.6087, + "step": 2255 + }, + { + "epoch": 0.07, + "grad_norm": 1.0763742923736572, + "learning_rate": 9.900067693420039e-05, + "loss": 3.4677, + "step": 2260 + }, + { + "epoch": 0.07, + "grad_norm": 1.0556530952453613, + "learning_rate": 9.899606064656832e-05, + "loss": 3.5344, + "step": 2265 + }, + { + "epoch": 0.07, + "grad_norm": 1.182672142982483, + "learning_rate": 9.899143382934379e-05, + "loss": 3.4531, + "step": 2270 + }, + { + "epoch": 0.07, + "grad_norm": 1.1016777753829956, + "learning_rate": 9.898679648352108e-05, + "loss": 3.3865, + "step": 2275 + }, + { + "epoch": 0.07, + "grad_norm": 1.1757758855819702, + "learning_rate": 9.898214861009683e-05, + "loss": 3.3833, + "step": 2280 + }, + { + "epoch": 0.07, + "grad_norm": 1.198561668395996, + "learning_rate": 9.897749021006988e-05, + "loss": 3.5921, + "step": 2285 + }, + { + "epoch": 0.07, + "grad_norm": 1.259002447128296, + "learning_rate": 9.897282128444135e-05, + "loss": 3.4573, + "step": 2290 + }, + { + "epoch": 0.07, + "grad_norm": 1.158482313156128, + "learning_rate": 9.896814183421464e-05, + "loss": 3.53, + "step": 2295 + }, + { + "epoch": 0.07, + "grad_norm": 1.121313452720642, + "learning_rate": 9.896345186039538e-05, + "loss": 3.3255, + "step": 2300 + }, + { + "epoch": 0.07, + "grad_norm": 1.0576354265213013, + "learning_rate": 9.895875136399149e-05, + "loss": 3.5605, + "step": 2305 + }, + { + "epoch": 0.07, + "grad_norm": 1.0766839981079102, + "learning_rate": 9.895404034601313e-05, + "loss": 3.4699, + "step": 2310 + }, + { + "epoch": 0.07, + "grad_norm": 1.122793436050415, + "learning_rate": 9.894931880747273e-05, + "loss": 3.3475, + "step": 2315 + }, + { + "epoch": 0.07, + "grad_norm": 1.152417778968811, + "learning_rate": 9.894458674938499e-05, + "loss": 3.4169, + "step": 2320 + }, + { + "epoch": 0.07, + "grad_norm": 1.2037073373794556, + "learning_rate": 9.893984417276686e-05, + "loss": 3.3987, + "step": 2325 + }, + { + "epoch": 0.07, + "grad_norm": 1.0625969171524048, + "learning_rate": 9.893509107863754e-05, + "loss": 3.4897, + "step": 2330 + }, + { + "epoch": 0.07, + "grad_norm": 1.2452435493469238, + "learning_rate": 9.893032746801854e-05, + "loss": 3.4455, + "step": 2335 + }, + { + "epoch": 0.07, + "grad_norm": 1.1815768480300903, + "learning_rate": 9.892555334193353e-05, + "loss": 3.4669, + "step": 2340 + }, + { + "epoch": 0.07, + "grad_norm": 1.291759729385376, + "learning_rate": 9.892076870140857e-05, + "loss": 3.4508, + "step": 2345 + }, + { + "epoch": 0.07, + "grad_norm": 1.2819740772247314, + "learning_rate": 9.891597354747186e-05, + "loss": 3.4187, + "step": 2350 + }, + { + "epoch": 0.07, + "grad_norm": 1.1181094646453857, + "learning_rate": 9.891116788115393e-05, + "loss": 3.4426, + "step": 2355 + }, + { + "epoch": 0.07, + "grad_norm": 1.0060194730758667, + "learning_rate": 9.890635170348756e-05, + "loss": 3.4742, + "step": 2360 + }, + { + "epoch": 0.07, + "grad_norm": 1.0563273429870605, + "learning_rate": 9.890152501550777e-05, + "loss": 3.5502, + "step": 2365 + }, + { + "epoch": 0.07, + "grad_norm": 1.1242426633834839, + "learning_rate": 9.889668781825185e-05, + "loss": 3.4226, + "step": 2370 + }, + { + "epoch": 0.07, + "grad_norm": 1.3223096132278442, + "learning_rate": 9.889184011275936e-05, + "loss": 3.3523, + "step": 2375 + }, + { + "epoch": 0.07, + "grad_norm": 1.2018234729766846, + "learning_rate": 9.888698190007209e-05, + "loss": 3.4708, + "step": 2380 + }, + { + "epoch": 0.07, + "grad_norm": 1.116727590560913, + "learning_rate": 9.88821131812341e-05, + "loss": 3.4198, + "step": 2385 + }, + { + "epoch": 0.07, + "grad_norm": 1.2352378368377686, + "learning_rate": 9.887723395729174e-05, + "loss": 3.4779, + "step": 2390 + }, + { + "epoch": 0.07, + "grad_norm": 1.1054075956344604, + "learning_rate": 9.887234422929355e-05, + "loss": 3.4659, + "step": 2395 + }, + { + "epoch": 0.07, + "grad_norm": 1.2472244501113892, + "learning_rate": 9.886744399829039e-05, + "loss": 3.489, + "step": 2400 + }, + { + "epoch": 0.07, + "grad_norm": 1.175764799118042, + "learning_rate": 9.886253326533535e-05, + "loss": 3.4551, + "step": 2405 + }, + { + "epoch": 0.07, + "grad_norm": 1.130882978439331, + "learning_rate": 9.885761203148378e-05, + "loss": 3.2901, + "step": 2410 + }, + { + "epoch": 0.07, + "grad_norm": 1.165326714515686, + "learning_rate": 9.885268029779327e-05, + "loss": 3.5147, + "step": 2415 + }, + { + "epoch": 0.07, + "grad_norm": 1.065177083015442, + "learning_rate": 9.884773806532373e-05, + "loss": 3.4107, + "step": 2420 + }, + { + "epoch": 0.07, + "grad_norm": 1.1022299528121948, + "learning_rate": 9.884278533513723e-05, + "loss": 3.3516, + "step": 2425 + }, + { + "epoch": 0.07, + "grad_norm": 1.233691930770874, + "learning_rate": 9.883782210829816e-05, + "loss": 3.447, + "step": 2430 + }, + { + "epoch": 0.07, + "grad_norm": 1.0388953685760498, + "learning_rate": 9.883284838587317e-05, + "loss": 3.4108, + "step": 2435 + }, + { + "epoch": 0.07, + "grad_norm": 1.018480896949768, + "learning_rate": 9.882786416893112e-05, + "loss": 3.325, + "step": 2440 + }, + { + "epoch": 0.07, + "grad_norm": 1.0984917879104614, + "learning_rate": 9.882286945854319e-05, + "loss": 3.4115, + "step": 2445 + }, + { + "epoch": 0.07, + "grad_norm": 1.1359695196151733, + "learning_rate": 9.881786425578274e-05, + "loss": 3.3971, + "step": 2450 + }, + { + "epoch": 0.07, + "grad_norm": 1.0665357112884521, + "learning_rate": 9.881284856172545e-05, + "loss": 3.4957, + "step": 2455 + }, + { + "epoch": 0.07, + "grad_norm": 1.0459110736846924, + "learning_rate": 9.880782237744921e-05, + "loss": 3.5812, + "step": 2460 + }, + { + "epoch": 0.07, + "grad_norm": 1.14045250415802, + "learning_rate": 9.880278570403419e-05, + "loss": 3.4124, + "step": 2465 + }, + { + "epoch": 0.07, + "grad_norm": 1.033178448677063, + "learning_rate": 9.87977385425628e-05, + "loss": 3.455, + "step": 2470 + }, + { + "epoch": 0.07, + "grad_norm": 1.0426372289657593, + "learning_rate": 9.87926808941197e-05, + "loss": 3.3799, + "step": 2475 + }, + { + "epoch": 0.07, + "grad_norm": 1.1203879117965698, + "learning_rate": 9.878761275979184e-05, + "loss": 3.3785, + "step": 2480 + }, + { + "epoch": 0.07, + "grad_norm": 1.0577795505523682, + "learning_rate": 9.878253414066839e-05, + "loss": 3.5137, + "step": 2485 + }, + { + "epoch": 0.07, + "grad_norm": 1.056058406829834, + "learning_rate": 9.877744503784076e-05, + "loss": 3.3817, + "step": 2490 + }, + { + "epoch": 0.07, + "grad_norm": 1.0504395961761475, + "learning_rate": 9.877234545240265e-05, + "loss": 3.4085, + "step": 2495 + }, + { + "epoch": 0.07, + "grad_norm": 1.0361164808273315, + "learning_rate": 9.876723538545001e-05, + "loss": 3.2976, + "step": 2500 + }, + { + "epoch": 0.07, + "grad_norm": 1.1712521314620972, + "learning_rate": 9.8762114838081e-05, + "loss": 3.4922, + "step": 2505 + }, + { + "epoch": 0.07, + "grad_norm": 1.1452791690826416, + "learning_rate": 9.875698381139607e-05, + "loss": 3.3463, + "step": 2510 + }, + { + "epoch": 0.07, + "grad_norm": 1.0299887657165527, + "learning_rate": 9.875184230649792e-05, + "loss": 3.4343, + "step": 2515 + }, + { + "epoch": 0.07, + "grad_norm": 1.237331509590149, + "learning_rate": 9.874669032449149e-05, + "loss": 3.4067, + "step": 2520 + }, + { + "epoch": 0.07, + "grad_norm": 1.1920459270477295, + "learning_rate": 9.874152786648399e-05, + "loss": 3.4217, + "step": 2525 + }, + { + "epoch": 0.07, + "grad_norm": 1.1390721797943115, + "learning_rate": 9.873635493358484e-05, + "loss": 3.4699, + "step": 2530 + }, + { + "epoch": 0.07, + "grad_norm": 1.1805967092514038, + "learning_rate": 9.873117152690576e-05, + "loss": 3.4463, + "step": 2535 + }, + { + "epoch": 0.07, + "grad_norm": 1.1282398700714111, + "learning_rate": 9.872597764756071e-05, + "loss": 3.3637, + "step": 2540 + }, + { + "epoch": 0.07, + "grad_norm": 1.1326520442962646, + "learning_rate": 9.872077329666587e-05, + "loss": 3.4084, + "step": 2545 + }, + { + "epoch": 0.08, + "grad_norm": 1.19896399974823, + "learning_rate": 9.871555847533968e-05, + "loss": 3.4736, + "step": 2550 + }, + { + "epoch": 0.08, + "grad_norm": 1.0882346630096436, + "learning_rate": 9.871033318470289e-05, + "loss": 3.364, + "step": 2555 + }, + { + "epoch": 0.08, + "grad_norm": 1.0608967542648315, + "learning_rate": 9.870509742587838e-05, + "loss": 3.4367, + "step": 2560 + }, + { + "epoch": 0.08, + "grad_norm": 1.1200900077819824, + "learning_rate": 9.869985119999143e-05, + "loss": 3.4134, + "step": 2565 + }, + { + "epoch": 0.08, + "grad_norm": 1.0936342477798462, + "learning_rate": 9.869459450816942e-05, + "loss": 3.3918, + "step": 2570 + }, + { + "epoch": 0.08, + "grad_norm": 1.1664392948150635, + "learning_rate": 9.86893273515421e-05, + "loss": 3.3535, + "step": 2575 + }, + { + "epoch": 0.08, + "grad_norm": 1.0741019248962402, + "learning_rate": 9.868404973124139e-05, + "loss": 3.3092, + "step": 2580 + }, + { + "epoch": 0.08, + "grad_norm": 1.0989964008331299, + "learning_rate": 9.86787616484015e-05, + "loss": 3.4099, + "step": 2585 + }, + { + "epoch": 0.08, + "grad_norm": 1.0745898485183716, + "learning_rate": 9.867346310415886e-05, + "loss": 3.4144, + "step": 2590 + }, + { + "epoch": 0.08, + "grad_norm": 1.1228419542312622, + "learning_rate": 9.866815409965219e-05, + "loss": 3.4599, + "step": 2595 + }, + { + "epoch": 0.08, + "grad_norm": 1.0642093420028687, + "learning_rate": 9.86628346360224e-05, + "loss": 3.3887, + "step": 2600 + }, + { + "epoch": 0.08, + "grad_norm": 1.1604464054107666, + "learning_rate": 9.865750471441271e-05, + "loss": 3.4637, + "step": 2605 + }, + { + "epoch": 0.08, + "grad_norm": 1.2347286939620972, + "learning_rate": 9.865216433596853e-05, + "loss": 3.3825, + "step": 2610 + }, + { + "epoch": 0.08, + "grad_norm": 1.2223398685455322, + "learning_rate": 9.864681350183758e-05, + "loss": 3.3713, + "step": 2615 + }, + { + "epoch": 0.08, + "grad_norm": 1.0664856433868408, + "learning_rate": 9.864145221316975e-05, + "loss": 3.3459, + "step": 2620 + }, + { + "epoch": 0.08, + "grad_norm": 1.057586431503296, + "learning_rate": 9.863608047111725e-05, + "loss": 3.4113, + "step": 2625 + }, + { + "epoch": 0.08, + "grad_norm": 1.0008800029754639, + "learning_rate": 9.863069827683446e-05, + "loss": 3.3723, + "step": 2630 + }, + { + "epoch": 0.08, + "grad_norm": 1.14852774143219, + "learning_rate": 9.862530563147812e-05, + "loss": 3.3748, + "step": 2635 + }, + { + "epoch": 0.08, + "grad_norm": 1.1226527690887451, + "learning_rate": 9.861990253620709e-05, + "loss": 3.4066, + "step": 2640 + }, + { + "epoch": 0.08, + "grad_norm": 1.0443450212478638, + "learning_rate": 9.861448899218255e-05, + "loss": 3.4365, + "step": 2645 + }, + { + "epoch": 0.08, + "grad_norm": 1.1397427320480347, + "learning_rate": 9.860906500056792e-05, + "loss": 3.4138, + "step": 2650 + }, + { + "epoch": 0.08, + "grad_norm": 1.0453484058380127, + "learning_rate": 9.860363056252883e-05, + "loss": 3.3828, + "step": 2655 + }, + { + "epoch": 0.08, + "grad_norm": 1.1816262006759644, + "learning_rate": 9.859818567923321e-05, + "loss": 3.3993, + "step": 2660 + }, + { + "epoch": 0.08, + "grad_norm": 1.0515496730804443, + "learning_rate": 9.859273035185118e-05, + "loss": 3.4077, + "step": 2665 + }, + { + "epoch": 0.08, + "grad_norm": 1.0940170288085938, + "learning_rate": 9.858726458155512e-05, + "loss": 3.3206, + "step": 2670 + }, + { + "epoch": 0.08, + "grad_norm": 1.021521806716919, + "learning_rate": 9.858178836951967e-05, + "loss": 3.3998, + "step": 2675 + }, + { + "epoch": 0.08, + "grad_norm": 1.1276389360427856, + "learning_rate": 9.857630171692174e-05, + "loss": 3.4172, + "step": 2680 + }, + { + "epoch": 0.08, + "grad_norm": 1.042746901512146, + "learning_rate": 9.857080462494039e-05, + "loss": 3.3914, + "step": 2685 + }, + { + "epoch": 0.08, + "grad_norm": 1.0213004350662231, + "learning_rate": 9.856529709475703e-05, + "loss": 3.3805, + "step": 2690 + }, + { + "epoch": 0.08, + "grad_norm": 1.0166774988174438, + "learning_rate": 9.855977912755524e-05, + "loss": 3.3577, + "step": 2695 + }, + { + "epoch": 0.08, + "grad_norm": 1.1209036111831665, + "learning_rate": 9.855425072452086e-05, + "loss": 3.4344, + "step": 2700 + }, + { + "epoch": 0.08, + "grad_norm": 1.0883934497833252, + "learning_rate": 9.854871188684204e-05, + "loss": 3.4697, + "step": 2705 + }, + { + "epoch": 0.08, + "grad_norm": 1.0857993364334106, + "learning_rate": 9.854316261570905e-05, + "loss": 3.3999, + "step": 2710 + }, + { + "epoch": 0.08, + "grad_norm": 1.426241397857666, + "learning_rate": 9.85376029123145e-05, + "loss": 3.3162, + "step": 2715 + }, + { + "epoch": 0.08, + "grad_norm": 1.1627262830734253, + "learning_rate": 9.853203277785318e-05, + "loss": 3.4148, + "step": 2720 + }, + { + "epoch": 0.08, + "grad_norm": 1.0585832595825195, + "learning_rate": 9.85264522135222e-05, + "loss": 3.3133, + "step": 2725 + }, + { + "epoch": 0.08, + "grad_norm": 1.1181896924972534, + "learning_rate": 9.852086122052083e-05, + "loss": 3.4377, + "step": 2730 + }, + { + "epoch": 0.08, + "grad_norm": 1.1979148387908936, + "learning_rate": 9.85152598000506e-05, + "loss": 3.1993, + "step": 2735 + }, + { + "epoch": 0.08, + "grad_norm": 1.1172866821289062, + "learning_rate": 9.850964795331531e-05, + "loss": 3.4593, + "step": 2740 + }, + { + "epoch": 0.08, + "grad_norm": 1.021009087562561, + "learning_rate": 9.8504025681521e-05, + "loss": 3.2912, + "step": 2745 + }, + { + "epoch": 0.08, + "grad_norm": 1.0908108949661255, + "learning_rate": 9.84983929858759e-05, + "loss": 3.3506, + "step": 2750 + }, + { + "epoch": 0.08, + "grad_norm": 1.1938234567642212, + "learning_rate": 9.849274986759054e-05, + "loss": 3.3088, + "step": 2755 + }, + { + "epoch": 0.08, + "grad_norm": 0.9920837879180908, + "learning_rate": 9.848709632787768e-05, + "loss": 3.3696, + "step": 2760 + }, + { + "epoch": 0.08, + "grad_norm": 1.0236974954605103, + "learning_rate": 9.848143236795225e-05, + "loss": 3.2657, + "step": 2765 + }, + { + "epoch": 0.08, + "grad_norm": 1.02934992313385, + "learning_rate": 9.847575798903152e-05, + "loss": 3.3358, + "step": 2770 + }, + { + "epoch": 0.08, + "grad_norm": 1.135703682899475, + "learning_rate": 9.847007319233496e-05, + "loss": 3.3709, + "step": 2775 + }, + { + "epoch": 0.08, + "grad_norm": 1.1426535844802856, + "learning_rate": 9.846437797908424e-05, + "loss": 3.2658, + "step": 2780 + }, + { + "epoch": 0.08, + "grad_norm": 1.3116892576217651, + "learning_rate": 9.845867235050333e-05, + "loss": 3.3952, + "step": 2785 + }, + { + "epoch": 0.08, + "grad_norm": 1.0452803373336792, + "learning_rate": 9.845295630781837e-05, + "loss": 3.3946, + "step": 2790 + }, + { + "epoch": 0.08, + "grad_norm": 1.055330514907837, + "learning_rate": 9.844722985225783e-05, + "loss": 3.3978, + "step": 2795 + }, + { + "epoch": 0.08, + "grad_norm": 1.1368728876113892, + "learning_rate": 9.844149298505232e-05, + "loss": 3.3929, + "step": 2800 + }, + { + "epoch": 0.08, + "grad_norm": 1.0991214513778687, + "learning_rate": 9.843574570743474e-05, + "loss": 3.2969, + "step": 2805 + }, + { + "epoch": 0.08, + "grad_norm": 1.2326358556747437, + "learning_rate": 9.842998802064025e-05, + "loss": 3.302, + "step": 2810 + }, + { + "epoch": 0.08, + "grad_norm": 1.1568881273269653, + "learning_rate": 9.842421992590619e-05, + "loss": 3.2712, + "step": 2815 + }, + { + "epoch": 0.08, + "grad_norm": 1.05539071559906, + "learning_rate": 9.841844142447215e-05, + "loss": 3.2998, + "step": 2820 + }, + { + "epoch": 0.08, + "grad_norm": 1.2128485441207886, + "learning_rate": 9.841265251758e-05, + "loss": 3.3318, + "step": 2825 + }, + { + "epoch": 0.08, + "grad_norm": 1.0557286739349365, + "learning_rate": 9.840685320647381e-05, + "loss": 3.3783, + "step": 2830 + }, + { + "epoch": 0.08, + "grad_norm": 1.2186791896820068, + "learning_rate": 9.840104349239988e-05, + "loss": 3.3781, + "step": 2835 + }, + { + "epoch": 0.08, + "grad_norm": 1.1459729671478271, + "learning_rate": 9.839522337660677e-05, + "loss": 3.3517, + "step": 2840 + }, + { + "epoch": 0.08, + "grad_norm": 1.1700817346572876, + "learning_rate": 9.838939286034526e-05, + "loss": 3.3014, + "step": 2845 + }, + { + "epoch": 0.08, + "grad_norm": 1.0370063781738281, + "learning_rate": 9.838355194486836e-05, + "loss": 3.2981, + "step": 2850 + }, + { + "epoch": 0.08, + "grad_norm": 1.6378519535064697, + "learning_rate": 9.837770063143133e-05, + "loss": 3.3349, + "step": 2855 + }, + { + "epoch": 0.08, + "grad_norm": 1.0988675355911255, + "learning_rate": 9.837183892129164e-05, + "loss": 3.4475, + "step": 2860 + }, + { + "epoch": 0.08, + "grad_norm": 1.0614280700683594, + "learning_rate": 9.836596681570906e-05, + "loss": 3.1829, + "step": 2865 + }, + { + "epoch": 0.08, + "grad_norm": 1.0539418458938599, + "learning_rate": 9.83600843159455e-05, + "loss": 3.3959, + "step": 2870 + }, + { + "epoch": 0.08, + "grad_norm": 1.061159372329712, + "learning_rate": 9.835419142326517e-05, + "loss": 3.3587, + "step": 2875 + }, + { + "epoch": 0.08, + "grad_norm": 1.0594666004180908, + "learning_rate": 9.834828813893448e-05, + "loss": 3.3501, + "step": 2880 + }, + { + "epoch": 0.08, + "grad_norm": 1.0276389122009277, + "learning_rate": 9.83423744642221e-05, + "loss": 3.3536, + "step": 2885 + }, + { + "epoch": 0.09, + "grad_norm": 1.0819017887115479, + "learning_rate": 9.833645040039892e-05, + "loss": 3.3122, + "step": 2890 + }, + { + "epoch": 0.09, + "grad_norm": 1.0561038255691528, + "learning_rate": 9.833051594873805e-05, + "loss": 3.3057, + "step": 2895 + }, + { + "epoch": 0.09, + "grad_norm": 1.115460753440857, + "learning_rate": 9.832457111051485e-05, + "loss": 3.3948, + "step": 2900 + }, + { + "epoch": 0.09, + "grad_norm": 0.98100745677948, + "learning_rate": 9.831861588700692e-05, + "loss": 3.3954, + "step": 2905 + }, + { + "epoch": 0.09, + "grad_norm": 1.112380027770996, + "learning_rate": 9.831265027949406e-05, + "loss": 3.3128, + "step": 2910 + }, + { + "epoch": 0.09, + "grad_norm": 1.1612874269485474, + "learning_rate": 9.830667428925833e-05, + "loss": 3.442, + "step": 2915 + }, + { + "epoch": 0.09, + "grad_norm": 1.0861941576004028, + "learning_rate": 9.8300687917584e-05, + "loss": 3.3483, + "step": 2920 + }, + { + "epoch": 0.09, + "grad_norm": 1.0207030773162842, + "learning_rate": 9.82946911657576e-05, + "loss": 3.223, + "step": 2925 + }, + { + "epoch": 0.09, + "grad_norm": 1.019600510597229, + "learning_rate": 9.828868403506788e-05, + "loss": 3.3136, + "step": 2930 + }, + { + "epoch": 0.09, + "grad_norm": 1.24674654006958, + "learning_rate": 9.828266652680577e-05, + "loss": 3.3306, + "step": 2935 + }, + { + "epoch": 0.09, + "grad_norm": 1.1440917253494263, + "learning_rate": 9.827663864226453e-05, + "loss": 3.3834, + "step": 2940 + }, + { + "epoch": 0.09, + "grad_norm": 1.07683265209198, + "learning_rate": 9.827060038273956e-05, + "loss": 3.3468, + "step": 2945 + }, + { + "epoch": 0.09, + "grad_norm": 1.3300790786743164, + "learning_rate": 9.826455174952852e-05, + "loss": 3.365, + "step": 2950 + }, + { + "epoch": 0.09, + "grad_norm": 0.960716962814331, + "learning_rate": 9.825849274393133e-05, + "loss": 3.2648, + "step": 2955 + }, + { + "epoch": 0.09, + "grad_norm": 1.1142815351486206, + "learning_rate": 9.825242336725009e-05, + "loss": 3.272, + "step": 2960 + }, + { + "epoch": 0.09, + "grad_norm": 1.200162649154663, + "learning_rate": 9.824634362078915e-05, + "loss": 3.3159, + "step": 2965 + }, + { + "epoch": 0.09, + "grad_norm": 1.0301605463027954, + "learning_rate": 9.824025350585511e-05, + "loss": 3.287, + "step": 2970 + }, + { + "epoch": 0.09, + "grad_norm": 1.014706015586853, + "learning_rate": 9.823415302375676e-05, + "loss": 3.3637, + "step": 2975 + }, + { + "epoch": 0.09, + "grad_norm": 1.067028284072876, + "learning_rate": 9.822804217580515e-05, + "loss": 3.3526, + "step": 2980 + }, + { + "epoch": 0.09, + "grad_norm": 1.0247745513916016, + "learning_rate": 9.822192096331351e-05, + "loss": 3.329, + "step": 2985 + }, + { + "epoch": 0.09, + "grad_norm": 1.1649980545043945, + "learning_rate": 9.82157893875974e-05, + "loss": 3.3351, + "step": 2990 + }, + { + "epoch": 0.09, + "grad_norm": 1.2622787952423096, + "learning_rate": 9.820964744997445e-05, + "loss": 3.2787, + "step": 2995 + }, + { + "epoch": 0.09, + "grad_norm": 1.0350223779678345, + "learning_rate": 9.820349515176466e-05, + "loss": 3.273, + "step": 3000 + }, + { + "epoch": 0.09, + "grad_norm": 1.0727938413619995, + "learning_rate": 9.819733249429019e-05, + "loss": 3.3342, + "step": 3005 + }, + { + "epoch": 0.09, + "grad_norm": 1.063476324081421, + "learning_rate": 9.819115947887545e-05, + "loss": 3.3356, + "step": 3010 + }, + { + "epoch": 0.09, + "grad_norm": 1.016035556793213, + "learning_rate": 9.818497610684705e-05, + "loss": 3.4115, + "step": 3015 + }, + { + "epoch": 0.09, + "grad_norm": 1.1418989896774292, + "learning_rate": 9.817878237953383e-05, + "loss": 3.2155, + "step": 3020 + }, + { + "epoch": 0.09, + "grad_norm": 1.0644562244415283, + "learning_rate": 9.817257829826689e-05, + "loss": 3.3691, + "step": 3025 + }, + { + "epoch": 0.09, + "grad_norm": 1.1053295135498047, + "learning_rate": 9.81663638643795e-05, + "loss": 3.3381, + "step": 3030 + }, + { + "epoch": 0.09, + "grad_norm": 0.976543664932251, + "learning_rate": 9.81601390792072e-05, + "loss": 3.2442, + "step": 3035 + }, + { + "epoch": 0.09, + "grad_norm": 1.0978988409042358, + "learning_rate": 9.815390394408776e-05, + "loss": 3.2416, + "step": 3040 + }, + { + "epoch": 0.09, + "grad_norm": 1.0420993566513062, + "learning_rate": 9.814765846036111e-05, + "loss": 3.3476, + "step": 3045 + }, + { + "epoch": 0.09, + "grad_norm": 1.1441594362258911, + "learning_rate": 9.81414026293695e-05, + "loss": 3.2752, + "step": 3050 + }, + { + "epoch": 0.09, + "grad_norm": 1.0094490051269531, + "learning_rate": 9.813513645245729e-05, + "loss": 3.2776, + "step": 3055 + }, + { + "epoch": 0.09, + "grad_norm": 1.1460001468658447, + "learning_rate": 9.812885993097118e-05, + "loss": 3.297, + "step": 3060 + }, + { + "epoch": 0.09, + "grad_norm": 1.114621877670288, + "learning_rate": 9.812257306626002e-05, + "loss": 3.2507, + "step": 3065 + }, + { + "epoch": 0.09, + "grad_norm": 1.1293511390686035, + "learning_rate": 9.811627585967487e-05, + "loss": 3.2922, + "step": 3070 + }, + { + "epoch": 0.09, + "grad_norm": 1.0641099214553833, + "learning_rate": 9.810996831256909e-05, + "loss": 3.3379, + "step": 3075 + }, + { + "epoch": 0.09, + "grad_norm": 0.9949053525924683, + "learning_rate": 9.81036504262982e-05, + "loss": 3.2959, + "step": 3080 + }, + { + "epoch": 0.09, + "grad_norm": 1.1378787755966187, + "learning_rate": 9.809732220221995e-05, + "loss": 3.2674, + "step": 3085 + }, + { + "epoch": 0.09, + "grad_norm": 1.2493293285369873, + "learning_rate": 9.80909836416943e-05, + "loss": 3.3685, + "step": 3090 + }, + { + "epoch": 0.09, + "grad_norm": 1.0391952991485596, + "learning_rate": 9.80846347460835e-05, + "loss": 3.2901, + "step": 3095 + }, + { + "epoch": 0.09, + "grad_norm": 0.99662184715271, + "learning_rate": 9.807827551675192e-05, + "loss": 3.2846, + "step": 3100 + }, + { + "epoch": 0.09, + "grad_norm": 1.0067330598831177, + "learning_rate": 9.807190595506623e-05, + "loss": 3.2748, + "step": 3105 + }, + { + "epoch": 0.09, + "grad_norm": 1.188556432723999, + "learning_rate": 9.80655260623953e-05, + "loss": 3.2996, + "step": 3110 + }, + { + "epoch": 0.09, + "grad_norm": 1.1176966428756714, + "learning_rate": 9.805913584011019e-05, + "loss": 3.3144, + "step": 3115 + }, + { + "epoch": 0.09, + "grad_norm": 1.0369738340377808, + "learning_rate": 9.805273528958423e-05, + "loss": 3.3915, + "step": 3120 + }, + { + "epoch": 0.09, + "grad_norm": 1.2094932794570923, + "learning_rate": 9.804632441219293e-05, + "loss": 3.3566, + "step": 3125 + }, + { + "epoch": 0.09, + "grad_norm": 1.0953224897384644, + "learning_rate": 9.803990320931402e-05, + "loss": 3.3231, + "step": 3130 + }, + { + "epoch": 0.09, + "grad_norm": 1.0981478691101074, + "learning_rate": 9.803347168232746e-05, + "loss": 3.2904, + "step": 3135 + }, + { + "epoch": 0.09, + "grad_norm": 1.0416529178619385, + "learning_rate": 9.802702983261545e-05, + "loss": 3.1855, + "step": 3140 + }, + { + "epoch": 0.09, + "grad_norm": 1.049074649810791, + "learning_rate": 9.802057766156238e-05, + "loss": 3.2878, + "step": 3145 + }, + { + "epoch": 0.09, + "grad_norm": 1.062117338180542, + "learning_rate": 9.801411517055488e-05, + "loss": 3.3188, + "step": 3150 + }, + { + "epoch": 0.09, + "grad_norm": 1.029934287071228, + "learning_rate": 9.800764236098175e-05, + "loss": 3.2383, + "step": 3155 + }, + { + "epoch": 0.09, + "grad_norm": 1.0264325141906738, + "learning_rate": 9.800115923423405e-05, + "loss": 3.3148, + "step": 3160 + }, + { + "epoch": 0.09, + "grad_norm": 1.1122504472732544, + "learning_rate": 9.799466579170508e-05, + "loss": 3.3511, + "step": 3165 + }, + { + "epoch": 0.09, + "grad_norm": 1.0294831991195679, + "learning_rate": 9.798816203479029e-05, + "loss": 3.2759, + "step": 3170 + }, + { + "epoch": 0.09, + "grad_norm": 0.992550253868103, + "learning_rate": 9.79816479648874e-05, + "loss": 3.258, + "step": 3175 + }, + { + "epoch": 0.09, + "grad_norm": 0.9870030283927917, + "learning_rate": 9.797512358339633e-05, + "loss": 3.2301, + "step": 3180 + }, + { + "epoch": 0.09, + "grad_norm": 0.9789267778396606, + "learning_rate": 9.796858889171921e-05, + "loss": 3.2391, + "step": 3185 + }, + { + "epoch": 0.09, + "grad_norm": 1.3775835037231445, + "learning_rate": 9.79620438912604e-05, + "loss": 3.296, + "step": 3190 + }, + { + "epoch": 0.09, + "grad_norm": 1.1433501243591309, + "learning_rate": 9.795548858342645e-05, + "loss": 3.3368, + "step": 3195 + }, + { + "epoch": 0.09, + "grad_norm": 1.1027940511703491, + "learning_rate": 9.794892296962615e-05, + "loss": 3.3382, + "step": 3200 + }, + { + "epoch": 0.09, + "grad_norm": 1.0624799728393555, + "learning_rate": 9.794234705127047e-05, + "loss": 3.2651, + "step": 3205 + }, + { + "epoch": 0.09, + "grad_norm": 0.986177384853363, + "learning_rate": 9.793576082977267e-05, + "loss": 3.2777, + "step": 3210 + }, + { + "epoch": 0.09, + "grad_norm": 1.0513557195663452, + "learning_rate": 9.792916430654815e-05, + "loss": 3.2493, + "step": 3215 + }, + { + "epoch": 0.09, + "grad_norm": 1.0633766651153564, + "learning_rate": 9.792255748301453e-05, + "loss": 3.2656, + "step": 3220 + }, + { + "epoch": 0.09, + "grad_norm": 1.0628114938735962, + "learning_rate": 9.791594036059168e-05, + "loss": 3.2167, + "step": 3225 + }, + { + "epoch": 0.1, + "grad_norm": 1.023634910583496, + "learning_rate": 9.790931294070169e-05, + "loss": 3.2915, + "step": 3230 + }, + { + "epoch": 0.1, + "grad_norm": 1.0147098302841187, + "learning_rate": 9.79026752247688e-05, + "loss": 3.3614, + "step": 3235 + }, + { + "epoch": 0.1, + "grad_norm": 1.102491855621338, + "learning_rate": 9.78960272142195e-05, + "loss": 3.2644, + "step": 3240 + }, + { + "epoch": 0.1, + "grad_norm": 0.9646819233894348, + "learning_rate": 9.788936891048251e-05, + "loss": 3.3257, + "step": 3245 + }, + { + "epoch": 0.1, + "grad_norm": 1.031937599182129, + "learning_rate": 9.788270031498876e-05, + "loss": 3.4071, + "step": 3250 + }, + { + "epoch": 0.1, + "grad_norm": 1.0518944263458252, + "learning_rate": 9.787602142917135e-05, + "loss": 3.2514, + "step": 3255 + }, + { + "epoch": 0.1, + "grad_norm": 1.0549023151397705, + "learning_rate": 9.786933225446563e-05, + "loss": 3.3785, + "step": 3260 + }, + { + "epoch": 0.1, + "grad_norm": 1.0625032186508179, + "learning_rate": 9.786263279230915e-05, + "loss": 3.3509, + "step": 3265 + }, + { + "epoch": 0.1, + "grad_norm": 1.0469131469726562, + "learning_rate": 9.785592304414167e-05, + "loss": 3.3547, + "step": 3270 + }, + { + "epoch": 0.1, + "grad_norm": 1.0542151927947998, + "learning_rate": 9.784920301140516e-05, + "loss": 3.3403, + "step": 3275 + }, + { + "epoch": 0.1, + "grad_norm": 1.0412989854812622, + "learning_rate": 9.78424726955438e-05, + "loss": 3.3187, + "step": 3280 + }, + { + "epoch": 0.1, + "grad_norm": 1.048068881034851, + "learning_rate": 9.783573209800398e-05, + "loss": 3.2692, + "step": 3285 + }, + { + "epoch": 0.1, + "grad_norm": 1.0363472700119019, + "learning_rate": 9.782898122023432e-05, + "loss": 3.3502, + "step": 3290 + }, + { + "epoch": 0.1, + "grad_norm": 0.9634552597999573, + "learning_rate": 9.78222200636856e-05, + "loss": 3.3025, + "step": 3295 + }, + { + "epoch": 0.1, + "grad_norm": 1.0056029558181763, + "learning_rate": 9.781544862981087e-05, + "loss": 3.2824, + "step": 3300 + }, + { + "epoch": 0.1, + "grad_norm": 1.0057324171066284, + "learning_rate": 9.780866692006534e-05, + "loss": 3.3254, + "step": 3305 + }, + { + "epoch": 0.1, + "grad_norm": 1.0660400390625, + "learning_rate": 9.780187493590645e-05, + "loss": 3.176, + "step": 3310 + }, + { + "epoch": 0.1, + "grad_norm": 1.1009554862976074, + "learning_rate": 9.779507267879385e-05, + "loss": 3.3518, + "step": 3315 + }, + { + "epoch": 0.1, + "grad_norm": 1.1846927404403687, + "learning_rate": 9.778826015018936e-05, + "loss": 3.232, + "step": 3320 + }, + { + "epoch": 0.1, + "grad_norm": 1.1132549047470093, + "learning_rate": 9.778143735155709e-05, + "loss": 3.3043, + "step": 3325 + }, + { + "epoch": 0.1, + "grad_norm": 0.9825388193130493, + "learning_rate": 9.777460428436327e-05, + "loss": 3.3504, + "step": 3330 + }, + { + "epoch": 0.1, + "grad_norm": 1.173962116241455, + "learning_rate": 9.776776095007638e-05, + "loss": 3.2796, + "step": 3335 + }, + { + "epoch": 0.1, + "grad_norm": 1.0070972442626953, + "learning_rate": 9.776090735016711e-05, + "loss": 3.2256, + "step": 3340 + }, + { + "epoch": 0.1, + "grad_norm": 1.0383707284927368, + "learning_rate": 9.775404348610835e-05, + "loss": 3.2062, + "step": 3345 + }, + { + "epoch": 0.1, + "grad_norm": 1.0273122787475586, + "learning_rate": 9.774716935937518e-05, + "loss": 3.231, + "step": 3350 + }, + { + "epoch": 0.1, + "grad_norm": 1.0111980438232422, + "learning_rate": 9.77402849714449e-05, + "loss": 3.1955, + "step": 3355 + }, + { + "epoch": 0.1, + "grad_norm": 1.0524190664291382, + "learning_rate": 9.773339032379701e-05, + "loss": 3.1444, + "step": 3360 + }, + { + "epoch": 0.1, + "grad_norm": 0.9937483072280884, + "learning_rate": 9.772648541791323e-05, + "loss": 3.2675, + "step": 3365 + }, + { + "epoch": 0.1, + "grad_norm": 1.0116652250289917, + "learning_rate": 9.771957025527745e-05, + "loss": 3.2636, + "step": 3370 + }, + { + "epoch": 0.1, + "grad_norm": 1.0126763582229614, + "learning_rate": 9.771264483737581e-05, + "loss": 3.2921, + "step": 3375 + }, + { + "epoch": 0.1, + "grad_norm": 1.054195523262024, + "learning_rate": 9.770570916569661e-05, + "loss": 3.1773, + "step": 3380 + }, + { + "epoch": 0.1, + "grad_norm": 1.1624895334243774, + "learning_rate": 9.76987632417304e-05, + "loss": 3.2985, + "step": 3385 + }, + { + "epoch": 0.1, + "grad_norm": 1.0580755472183228, + "learning_rate": 9.769180706696988e-05, + "loss": 3.3159, + "step": 3390 + }, + { + "epoch": 0.1, + "grad_norm": 1.1071776151657104, + "learning_rate": 9.768484064290997e-05, + "loss": 3.1794, + "step": 3395 + }, + { + "epoch": 0.1, + "grad_norm": 1.031856894493103, + "learning_rate": 9.767786397104786e-05, + "loss": 3.3067, + "step": 3400 + }, + { + "epoch": 0.1, + "grad_norm": 1.090532898902893, + "learning_rate": 9.767087705288283e-05, + "loss": 3.2101, + "step": 3405 + }, + { + "epoch": 0.1, + "grad_norm": 1.0135164260864258, + "learning_rate": 9.766387988991642e-05, + "loss": 3.3106, + "step": 3410 + }, + { + "epoch": 0.1, + "grad_norm": 1.0624022483825684, + "learning_rate": 9.765687248365241e-05, + "loss": 3.2475, + "step": 3415 + }, + { + "epoch": 0.1, + "grad_norm": 0.9920370578765869, + "learning_rate": 9.76498548355967e-05, + "loss": 3.2311, + "step": 3420 + }, + { + "epoch": 0.1, + "grad_norm": 1.239082932472229, + "learning_rate": 9.764282694725744e-05, + "loss": 3.3208, + "step": 3425 + }, + { + "epoch": 0.1, + "grad_norm": 1.0729447603225708, + "learning_rate": 9.7635788820145e-05, + "loss": 3.2946, + "step": 3430 + }, + { + "epoch": 0.1, + "grad_norm": 1.0100903511047363, + "learning_rate": 9.762874045577189e-05, + "loss": 3.1716, + "step": 3435 + }, + { + "epoch": 0.1, + "grad_norm": 1.050044059753418, + "learning_rate": 9.762168185565284e-05, + "loss": 3.2466, + "step": 3440 + }, + { + "epoch": 0.1, + "grad_norm": 0.9647949934005737, + "learning_rate": 9.761461302130484e-05, + "loss": 3.1862, + "step": 3445 + }, + { + "epoch": 0.1, + "grad_norm": 1.1227718591690063, + "learning_rate": 9.760753395424699e-05, + "loss": 3.2379, + "step": 3450 + }, + { + "epoch": 0.1, + "grad_norm": 1.1006200313568115, + "learning_rate": 9.760044465600065e-05, + "loss": 3.3244, + "step": 3455 + }, + { + "epoch": 0.1, + "grad_norm": 1.0457630157470703, + "learning_rate": 9.759334512808935e-05, + "loss": 3.3315, + "step": 3460 + }, + { + "epoch": 0.1, + "grad_norm": 1.0084728002548218, + "learning_rate": 9.758623537203883e-05, + "loss": 3.1374, + "step": 3465 + }, + { + "epoch": 0.1, + "grad_norm": 0.9979555606842041, + "learning_rate": 9.757911538937706e-05, + "loss": 3.2328, + "step": 3470 + }, + { + "epoch": 0.1, + "grad_norm": 0.9979318976402283, + "learning_rate": 9.757198518163411e-05, + "loss": 3.147, + "step": 3475 + }, + { + "epoch": 0.1, + "grad_norm": 0.9489660263061523, + "learning_rate": 9.756484475034235e-05, + "loss": 3.1912, + "step": 3480 + }, + { + "epoch": 0.1, + "grad_norm": 1.041934609413147, + "learning_rate": 9.75576940970363e-05, + "loss": 3.2969, + "step": 3485 + }, + { + "epoch": 0.1, + "grad_norm": 0.9783023595809937, + "learning_rate": 9.755053322325269e-05, + "loss": 3.3139, + "step": 3490 + }, + { + "epoch": 0.1, + "grad_norm": 0.9634132385253906, + "learning_rate": 9.754336213053043e-05, + "loss": 3.246, + "step": 3495 + }, + { + "epoch": 0.1, + "grad_norm": 0.978161096572876, + "learning_rate": 9.753618082041066e-05, + "loss": 3.1686, + "step": 3500 + }, + { + "epoch": 0.1, + "grad_norm": 1.3386309146881104, + "learning_rate": 9.752898929443666e-05, + "loss": 3.3441, + "step": 3505 + }, + { + "epoch": 0.1, + "grad_norm": 1.09601628780365, + "learning_rate": 9.752178755415396e-05, + "loss": 3.2182, + "step": 3510 + }, + { + "epoch": 0.1, + "grad_norm": 1.059962511062622, + "learning_rate": 9.751457560111026e-05, + "loss": 3.3105, + "step": 3515 + }, + { + "epoch": 0.1, + "grad_norm": 0.9673791527748108, + "learning_rate": 9.750735343685547e-05, + "loss": 3.2631, + "step": 3520 + }, + { + "epoch": 0.1, + "grad_norm": 0.9784255027770996, + "learning_rate": 9.750012106294167e-05, + "loss": 3.2662, + "step": 3525 + }, + { + "epoch": 0.1, + "grad_norm": 4.847470283508301, + "learning_rate": 9.749287848092315e-05, + "loss": 3.1393, + "step": 3530 + }, + { + "epoch": 0.1, + "grad_norm": 1.013223648071289, + "learning_rate": 9.748562569235637e-05, + "loss": 3.2794, + "step": 3535 + }, + { + "epoch": 0.1, + "grad_norm": 1.08577561378479, + "learning_rate": 9.747836269880003e-05, + "loss": 3.2719, + "step": 3540 + }, + { + "epoch": 0.1, + "grad_norm": 1.0656304359436035, + "learning_rate": 9.7471089501815e-05, + "loss": 3.2232, + "step": 3545 + }, + { + "epoch": 0.1, + "grad_norm": 1.1528815031051636, + "learning_rate": 9.746380610296435e-05, + "loss": 3.2001, + "step": 3550 + }, + { + "epoch": 0.1, + "grad_norm": 1.0031167268753052, + "learning_rate": 9.74565125038133e-05, + "loss": 3.303, + "step": 3555 + }, + { + "epoch": 0.1, + "grad_norm": 1.2080628871917725, + "learning_rate": 9.744920870592931e-05, + "loss": 3.2576, + "step": 3560 + }, + { + "epoch": 0.1, + "grad_norm": 1.0465301275253296, + "learning_rate": 9.7441894710882e-05, + "loss": 3.2699, + "step": 3565 + }, + { + "epoch": 0.11, + "grad_norm": 0.9583694934844971, + "learning_rate": 9.743457052024323e-05, + "loss": 3.3456, + "step": 3570 + }, + { + "epoch": 0.11, + "grad_norm": 0.9784826040267944, + "learning_rate": 9.742723613558702e-05, + "loss": 3.1787, + "step": 3575 + }, + { + "epoch": 0.11, + "grad_norm": 1.0103017091751099, + "learning_rate": 9.741989155848953e-05, + "loss": 3.2687, + "step": 3580 + }, + { + "epoch": 0.11, + "grad_norm": 1.1582177877426147, + "learning_rate": 9.741253679052923e-05, + "loss": 3.1584, + "step": 3585 + }, + { + "epoch": 0.11, + "grad_norm": 0.9797343611717224, + "learning_rate": 9.740517183328666e-05, + "loss": 3.1365, + "step": 3590 + }, + { + "epoch": 0.11, + "grad_norm": 1.0125981569290161, + "learning_rate": 9.739779668834461e-05, + "loss": 3.2496, + "step": 3595 + }, + { + "epoch": 0.11, + "grad_norm": 1.1603243350982666, + "learning_rate": 9.739041135728806e-05, + "loss": 3.2848, + "step": 3600 + }, + { + "epoch": 0.11, + "grad_norm": 1.0229432582855225, + "learning_rate": 9.738301584170417e-05, + "loss": 3.2777, + "step": 3605 + }, + { + "epoch": 0.11, + "grad_norm": 1.0532867908477783, + "learning_rate": 9.737561014318228e-05, + "loss": 3.228, + "step": 3610 + }, + { + "epoch": 0.11, + "grad_norm": 1.020656943321228, + "learning_rate": 9.736819426331392e-05, + "loss": 3.2704, + "step": 3615 + }, + { + "epoch": 0.11, + "grad_norm": 1.0407886505126953, + "learning_rate": 9.736076820369283e-05, + "loss": 3.2099, + "step": 3620 + }, + { + "epoch": 0.11, + "grad_norm": 1.075788974761963, + "learning_rate": 9.735333196591492e-05, + "loss": 3.3471, + "step": 3625 + }, + { + "epoch": 0.11, + "grad_norm": 1.04179048538208, + "learning_rate": 9.734588555157826e-05, + "loss": 3.2312, + "step": 3630 + }, + { + "epoch": 0.11, + "grad_norm": 1.074134349822998, + "learning_rate": 9.733842896228318e-05, + "loss": 3.2014, + "step": 3635 + }, + { + "epoch": 0.11, + "grad_norm": 1.0459777116775513, + "learning_rate": 9.733096219963211e-05, + "loss": 3.2439, + "step": 3640 + }, + { + "epoch": 0.11, + "grad_norm": 1.116286039352417, + "learning_rate": 9.732348526522974e-05, + "loss": 3.201, + "step": 3645 + }, + { + "epoch": 0.11, + "grad_norm": 1.0456223487854004, + "learning_rate": 9.73159981606829e-05, + "loss": 3.161, + "step": 3650 + }, + { + "epoch": 0.11, + "grad_norm": 0.9911007285118103, + "learning_rate": 9.730850088760064e-05, + "loss": 3.1654, + "step": 3655 + }, + { + "epoch": 0.11, + "grad_norm": 0.9365742802619934, + "learning_rate": 9.730099344759413e-05, + "loss": 3.2499, + "step": 3660 + }, + { + "epoch": 0.11, + "grad_norm": 0.9555690884590149, + "learning_rate": 9.729347584227684e-05, + "loss": 3.2099, + "step": 3665 + }, + { + "epoch": 0.11, + "grad_norm": 1.012406349182129, + "learning_rate": 9.72859480732643e-05, + "loss": 3.2235, + "step": 3670 + }, + { + "epoch": 0.11, + "grad_norm": 1.0403445959091187, + "learning_rate": 9.727841014217429e-05, + "loss": 3.3225, + "step": 3675 + }, + { + "epoch": 0.11, + "grad_norm": 1.0465917587280273, + "learning_rate": 9.727086205062676e-05, + "loss": 3.199, + "step": 3680 + }, + { + "epoch": 0.11, + "grad_norm": 1.005976676940918, + "learning_rate": 9.72633038002439e-05, + "loss": 3.2454, + "step": 3685 + }, + { + "epoch": 0.11, + "grad_norm": 1.056708574295044, + "learning_rate": 9.725573539264994e-05, + "loss": 3.2731, + "step": 3690 + }, + { + "epoch": 0.11, + "grad_norm": 1.0521161556243896, + "learning_rate": 9.724815682947146e-05, + "loss": 3.274, + "step": 3695 + }, + { + "epoch": 0.11, + "grad_norm": 1.0160726308822632, + "learning_rate": 9.72405681123371e-05, + "loss": 3.1829, + "step": 3700 + }, + { + "epoch": 0.11, + "grad_norm": 1.0059547424316406, + "learning_rate": 9.723296924287775e-05, + "loss": 3.2402, + "step": 3705 + }, + { + "epoch": 0.11, + "grad_norm": 0.9979924559593201, + "learning_rate": 9.722536022272646e-05, + "loss": 3.2353, + "step": 3710 + }, + { + "epoch": 0.11, + "grad_norm": 1.0275793075561523, + "learning_rate": 9.721774105351845e-05, + "loss": 3.2155, + "step": 3715 + }, + { + "epoch": 0.11, + "grad_norm": 1.0295485258102417, + "learning_rate": 9.721011173689114e-05, + "loss": 3.1871, + "step": 3720 + }, + { + "epoch": 0.11, + "grad_norm": 1.0929020643234253, + "learning_rate": 9.720247227448411e-05, + "loss": 3.2225, + "step": 3725 + }, + { + "epoch": 0.11, + "grad_norm": 1.0383379459381104, + "learning_rate": 9.719482266793916e-05, + "loss": 3.2908, + "step": 3730 + }, + { + "epoch": 0.11, + "grad_norm": 5.556280136108398, + "learning_rate": 9.718716291890022e-05, + "loss": 3.3103, + "step": 3735 + }, + { + "epoch": 0.11, + "grad_norm": 1.1070916652679443, + "learning_rate": 9.717949302901343e-05, + "loss": 3.2119, + "step": 3740 + }, + { + "epoch": 0.11, + "grad_norm": 1.0047955513000488, + "learning_rate": 9.71718129999271e-05, + "loss": 3.3143, + "step": 3745 + }, + { + "epoch": 0.11, + "grad_norm": 0.967634916305542, + "learning_rate": 9.716412283329172e-05, + "loss": 3.1705, + "step": 3750 + }, + { + "epoch": 0.11, + "grad_norm": 0.9875706434249878, + "learning_rate": 9.715642253075998e-05, + "loss": 3.1666, + "step": 3755 + }, + { + "epoch": 0.11, + "grad_norm": 1.0046371221542358, + "learning_rate": 9.714871209398671e-05, + "loss": 3.2213, + "step": 3760 + }, + { + "epoch": 0.11, + "grad_norm": 1.0251070261001587, + "learning_rate": 9.714099152462892e-05, + "loss": 3.2301, + "step": 3765 + }, + { + "epoch": 0.11, + "grad_norm": 1.0146162509918213, + "learning_rate": 9.713326082434584e-05, + "loss": 3.2381, + "step": 3770 + }, + { + "epoch": 0.11, + "grad_norm": 1.1744074821472168, + "learning_rate": 9.712551999479884e-05, + "loss": 3.2566, + "step": 3775 + }, + { + "epoch": 0.11, + "grad_norm": 0.9853100776672363, + "learning_rate": 9.71177690376515e-05, + "loss": 3.1612, + "step": 3780 + }, + { + "epoch": 0.11, + "grad_norm": 0.9493988752365112, + "learning_rate": 9.711000795456951e-05, + "loss": 3.2541, + "step": 3785 + }, + { + "epoch": 0.11, + "grad_norm": 1.0125104188919067, + "learning_rate": 9.710223674722083e-05, + "loss": 3.3245, + "step": 3790 + }, + { + "epoch": 0.11, + "grad_norm": 0.9618760943412781, + "learning_rate": 9.709445541727551e-05, + "loss": 3.2638, + "step": 3795 + }, + { + "epoch": 0.11, + "grad_norm": 0.9417055249214172, + "learning_rate": 9.708666396640582e-05, + "loss": 3.2045, + "step": 3800 + }, + { + "epoch": 0.11, + "grad_norm": 1.0970779657363892, + "learning_rate": 9.707886239628621e-05, + "loss": 3.281, + "step": 3805 + }, + { + "epoch": 0.11, + "grad_norm": 1.138805866241455, + "learning_rate": 9.707105070859326e-05, + "loss": 3.1976, + "step": 3810 + }, + { + "epoch": 0.11, + "grad_norm": 1.1180046796798706, + "learning_rate": 9.706322890500579e-05, + "loss": 3.1572, + "step": 3815 + }, + { + "epoch": 0.11, + "grad_norm": 1.028878092765808, + "learning_rate": 9.705539698720476e-05, + "loss": 3.2817, + "step": 3820 + }, + { + "epoch": 0.11, + "grad_norm": 1.1804286241531372, + "learning_rate": 9.704755495687328e-05, + "loss": 3.1559, + "step": 3825 + }, + { + "epoch": 0.11, + "grad_norm": 1.0877653360366821, + "learning_rate": 9.703970281569668e-05, + "loss": 3.2691, + "step": 3830 + }, + { + "epoch": 0.11, + "grad_norm": 0.9572838544845581, + "learning_rate": 9.70318405653624e-05, + "loss": 3.162, + "step": 3835 + }, + { + "epoch": 0.11, + "grad_norm": 0.9814620614051819, + "learning_rate": 9.702396820756014e-05, + "loss": 3.2957, + "step": 3840 + }, + { + "epoch": 0.11, + "grad_norm": 1.1293951272964478, + "learning_rate": 9.701608574398169e-05, + "loss": 3.1476, + "step": 3845 + }, + { + "epoch": 0.11, + "grad_norm": 0.953865647315979, + "learning_rate": 9.700819317632107e-05, + "loss": 3.1507, + "step": 3850 + }, + { + "epoch": 0.11, + "grad_norm": 1.1356327533721924, + "learning_rate": 9.700029050627443e-05, + "loss": 3.2085, + "step": 3855 + }, + { + "epoch": 0.11, + "grad_norm": 0.9980968236923218, + "learning_rate": 9.699237773554012e-05, + "loss": 3.1539, + "step": 3860 + }, + { + "epoch": 0.11, + "grad_norm": 1.0363867282867432, + "learning_rate": 9.698445486581864e-05, + "loss": 3.1899, + "step": 3865 + }, + { + "epoch": 0.11, + "grad_norm": 1.1382116079330444, + "learning_rate": 9.697652189881266e-05, + "loss": 3.2687, + "step": 3870 + }, + { + "epoch": 0.11, + "grad_norm": 0.9376232028007507, + "learning_rate": 9.696857883622706e-05, + "loss": 3.2268, + "step": 3875 + }, + { + "epoch": 0.11, + "grad_norm": 1.0130692720413208, + "learning_rate": 9.696062567976885e-05, + "loss": 3.12, + "step": 3880 + }, + { + "epoch": 0.11, + "grad_norm": 0.9634711146354675, + "learning_rate": 9.695266243114718e-05, + "loss": 3.165, + "step": 3885 + }, + { + "epoch": 0.11, + "grad_norm": 0.9566382765769958, + "learning_rate": 9.694468909207345e-05, + "loss": 3.2132, + "step": 3890 + }, + { + "epoch": 0.11, + "grad_norm": 1.1243633031845093, + "learning_rate": 9.693670566426119e-05, + "loss": 3.2486, + "step": 3895 + }, + { + "epoch": 0.11, + "grad_norm": 0.9307610988616943, + "learning_rate": 9.692871214942608e-05, + "loss": 3.2388, + "step": 3900 + }, + { + "epoch": 0.11, + "grad_norm": 0.9809203147888184, + "learning_rate": 9.692070854928596e-05, + "loss": 3.1955, + "step": 3905 + }, + { + "epoch": 0.12, + "grad_norm": 1.2277921438217163, + "learning_rate": 9.691269486556088e-05, + "loss": 3.2626, + "step": 3910 + }, + { + "epoch": 0.12, + "grad_norm": 1.124157190322876, + "learning_rate": 9.690467109997303e-05, + "loss": 3.1605, + "step": 3915 + }, + { + "epoch": 0.12, + "grad_norm": 0.9832596182823181, + "learning_rate": 9.68966372542468e-05, + "loss": 3.1653, + "step": 3920 + }, + { + "epoch": 0.12, + "grad_norm": 0.9972193837165833, + "learning_rate": 9.688859333010866e-05, + "loss": 3.1743, + "step": 3925 + }, + { + "epoch": 0.12, + "grad_norm": 1.0058636665344238, + "learning_rate": 9.688053932928737e-05, + "loss": 3.1502, + "step": 3930 + }, + { + "epoch": 0.12, + "grad_norm": 0.9747347235679626, + "learning_rate": 9.687247525351375e-05, + "loss": 3.1896, + "step": 3935 + }, + { + "epoch": 0.12, + "grad_norm": 0.9446713924407959, + "learning_rate": 9.686440110452084e-05, + "loss": 3.2145, + "step": 3940 + }, + { + "epoch": 0.12, + "grad_norm": 1.192351222038269, + "learning_rate": 9.685631688404384e-05, + "loss": 3.1464, + "step": 3945 + }, + { + "epoch": 0.12, + "grad_norm": 1.1033145189285278, + "learning_rate": 9.684822259382008e-05, + "loss": 3.1901, + "step": 3950 + }, + { + "epoch": 0.12, + "grad_norm": 1.0484864711761475, + "learning_rate": 9.684011823558909e-05, + "loss": 3.1339, + "step": 3955 + }, + { + "epoch": 0.12, + "grad_norm": 1.0042369365692139, + "learning_rate": 9.683200381109254e-05, + "loss": 3.2072, + "step": 3960 + }, + { + "epoch": 0.12, + "grad_norm": 1.0181210041046143, + "learning_rate": 9.68238793220743e-05, + "loss": 3.1887, + "step": 3965 + }, + { + "epoch": 0.12, + "grad_norm": 0.9293723106384277, + "learning_rate": 9.681574477028039e-05, + "loss": 3.2442, + "step": 3970 + }, + { + "epoch": 0.12, + "grad_norm": 0.9407383799552917, + "learning_rate": 9.680760015745894e-05, + "loss": 3.0638, + "step": 3975 + }, + { + "epoch": 0.12, + "grad_norm": 1.1896600723266602, + "learning_rate": 9.67994454853603e-05, + "loss": 3.12, + "step": 3980 + }, + { + "epoch": 0.12, + "grad_norm": 1.0163270235061646, + "learning_rate": 9.679128075573698e-05, + "loss": 3.2767, + "step": 3985 + }, + { + "epoch": 0.12, + "grad_norm": 0.9911450147628784, + "learning_rate": 9.678310597034364e-05, + "loss": 3.2639, + "step": 3990 + }, + { + "epoch": 0.12, + "grad_norm": 0.9702705144882202, + "learning_rate": 9.677492113093706e-05, + "loss": 3.2586, + "step": 3995 + }, + { + "epoch": 0.12, + "grad_norm": 0.9476492404937744, + "learning_rate": 9.676672623927625e-05, + "loss": 3.2005, + "step": 4000 + }, + { + "epoch": 0.12, + "grad_norm": 1.1955156326293945, + "learning_rate": 9.675852129712234e-05, + "loss": 3.184, + "step": 4005 + }, + { + "epoch": 0.12, + "grad_norm": 0.9895833730697632, + "learning_rate": 9.675030630623864e-05, + "loss": 2.9834, + "step": 4010 + }, + { + "epoch": 0.12, + "grad_norm": 0.9855818152427673, + "learning_rate": 9.67420812683906e-05, + "loss": 3.1643, + "step": 4015 + }, + { + "epoch": 0.12, + "grad_norm": 1.1091126203536987, + "learning_rate": 9.673384618534582e-05, + "loss": 3.1418, + "step": 4020 + }, + { + "epoch": 0.12, + "grad_norm": 0.9791935086250305, + "learning_rate": 9.672560105887409e-05, + "loss": 3.1436, + "step": 4025 + }, + { + "epoch": 0.12, + "grad_norm": 0.9897438883781433, + "learning_rate": 9.671734589074736e-05, + "loss": 3.1678, + "step": 4030 + }, + { + "epoch": 0.12, + "grad_norm": 1.007053256034851, + "learning_rate": 9.670908068273972e-05, + "loss": 3.1932, + "step": 4035 + }, + { + "epoch": 0.12, + "grad_norm": 0.994600772857666, + "learning_rate": 9.67008054366274e-05, + "loss": 3.2138, + "step": 4040 + }, + { + "epoch": 0.12, + "grad_norm": 0.917945146560669, + "learning_rate": 9.669252015418882e-05, + "loss": 3.228, + "step": 4045 + }, + { + "epoch": 0.12, + "grad_norm": 0.9876474142074585, + "learning_rate": 9.668422483720456e-05, + "loss": 3.2799, + "step": 4050 + }, + { + "epoch": 0.12, + "grad_norm": 0.9300573468208313, + "learning_rate": 9.66759194874573e-05, + "loss": 3.178, + "step": 4055 + }, + { + "epoch": 0.12, + "grad_norm": 0.9726993441581726, + "learning_rate": 9.666760410673195e-05, + "loss": 3.2248, + "step": 4060 + }, + { + "epoch": 0.12, + "grad_norm": 1.0185961723327637, + "learning_rate": 9.665927869681553e-05, + "loss": 3.2422, + "step": 4065 + }, + { + "epoch": 0.12, + "grad_norm": 0.9282801747322083, + "learning_rate": 9.665094325949723e-05, + "loss": 3.1877, + "step": 4070 + }, + { + "epoch": 0.12, + "grad_norm": 0.974884033203125, + "learning_rate": 9.664259779656842e-05, + "loss": 3.171, + "step": 4075 + }, + { + "epoch": 0.12, + "grad_norm": 1.0273559093475342, + "learning_rate": 9.663424230982254e-05, + "loss": 3.217, + "step": 4080 + }, + { + "epoch": 0.12, + "grad_norm": 0.9975438714027405, + "learning_rate": 9.66258768010553e-05, + "loss": 3.1924, + "step": 4085 + }, + { + "epoch": 0.12, + "grad_norm": 0.9452067613601685, + "learning_rate": 9.661750127206445e-05, + "loss": 3.1949, + "step": 4090 + }, + { + "epoch": 0.12, + "grad_norm": 1.1542057991027832, + "learning_rate": 9.660911572465e-05, + "loss": 3.1446, + "step": 4095 + }, + { + "epoch": 0.12, + "grad_norm": 1.1220757961273193, + "learning_rate": 9.660072016061406e-05, + "loss": 3.1061, + "step": 4100 + }, + { + "epoch": 0.12, + "grad_norm": 0.984806478023529, + "learning_rate": 9.659231458176085e-05, + "loss": 3.1159, + "step": 4105 + }, + { + "epoch": 0.12, + "grad_norm": 0.9188714027404785, + "learning_rate": 9.658389898989684e-05, + "loss": 3.1992, + "step": 4110 + }, + { + "epoch": 0.12, + "grad_norm": 0.998063325881958, + "learning_rate": 9.657547338683055e-05, + "loss": 3.2972, + "step": 4115 + }, + { + "epoch": 0.12, + "grad_norm": 1.1119955778121948, + "learning_rate": 9.656703777437273e-05, + "loss": 3.1035, + "step": 4120 + }, + { + "epoch": 0.12, + "grad_norm": 0.9921952486038208, + "learning_rate": 9.655859215433625e-05, + "loss": 3.227, + "step": 4125 + }, + { + "epoch": 0.12, + "grad_norm": 0.9861551523208618, + "learning_rate": 9.655013652853612e-05, + "loss": 3.1481, + "step": 4130 + }, + { + "epoch": 0.12, + "grad_norm": 0.9659957885742188, + "learning_rate": 9.654167089878952e-05, + "loss": 3.118, + "step": 4135 + }, + { + "epoch": 0.12, + "grad_norm": 0.9291513562202454, + "learning_rate": 9.653319526691577e-05, + "loss": 3.2441, + "step": 4140 + }, + { + "epoch": 0.12, + "grad_norm": 1.414057970046997, + "learning_rate": 9.652470963473635e-05, + "loss": 3.1662, + "step": 4145 + }, + { + "epoch": 0.12, + "grad_norm": 0.9417849183082581, + "learning_rate": 9.651621400407487e-05, + "loss": 3.2096, + "step": 4150 + }, + { + "epoch": 0.12, + "grad_norm": 0.9371815919876099, + "learning_rate": 9.650770837675712e-05, + "loss": 3.1353, + "step": 4155 + }, + { + "epoch": 0.12, + "grad_norm": 0.9965218305587769, + "learning_rate": 9.649919275461099e-05, + "loss": 3.2309, + "step": 4160 + }, + { + "epoch": 0.12, + "grad_norm": 1.0609004497528076, + "learning_rate": 9.649066713946655e-05, + "loss": 3.186, + "step": 4165 + }, + { + "epoch": 0.12, + "grad_norm": 0.9502406120300293, + "learning_rate": 9.648213153315603e-05, + "loss": 3.1905, + "step": 4170 + }, + { + "epoch": 0.12, + "grad_norm": 0.9332844614982605, + "learning_rate": 9.647358593751378e-05, + "loss": 3.185, + "step": 4175 + }, + { + "epoch": 0.12, + "grad_norm": 1.0272194147109985, + "learning_rate": 9.646503035437633e-05, + "loss": 3.1724, + "step": 4180 + }, + { + "epoch": 0.12, + "grad_norm": 1.2085559368133545, + "learning_rate": 9.64564647855823e-05, + "loss": 3.1716, + "step": 4185 + }, + { + "epoch": 0.12, + "grad_norm": 1.00102961063385, + "learning_rate": 9.644788923297253e-05, + "loss": 3.0681, + "step": 4190 + }, + { + "epoch": 0.12, + "grad_norm": 1.077770471572876, + "learning_rate": 9.64393036983899e-05, + "loss": 3.1457, + "step": 4195 + }, + { + "epoch": 0.12, + "grad_norm": 0.956590473651886, + "learning_rate": 9.643070818367957e-05, + "loss": 3.154, + "step": 4200 + }, + { + "epoch": 0.12, + "grad_norm": 1.0999020338058472, + "learning_rate": 9.642210269068874e-05, + "loss": 3.1355, + "step": 4205 + }, + { + "epoch": 0.12, + "grad_norm": 0.9278420209884644, + "learning_rate": 9.641348722126679e-05, + "loss": 3.1517, + "step": 4210 + }, + { + "epoch": 0.12, + "grad_norm": 0.9793952107429504, + "learning_rate": 9.640486177726527e-05, + "loss": 3.2223, + "step": 4215 + }, + { + "epoch": 0.12, + "grad_norm": 1.0725845098495483, + "learning_rate": 9.639622636053781e-05, + "loss": 3.2416, + "step": 4220 + }, + { + "epoch": 0.12, + "grad_norm": 1.0750727653503418, + "learning_rate": 9.638758097294024e-05, + "loss": 3.2245, + "step": 4225 + }, + { + "epoch": 0.12, + "grad_norm": 22.925918579101562, + "learning_rate": 9.637892561633054e-05, + "loss": 3.0923, + "step": 4230 + }, + { + "epoch": 0.12, + "grad_norm": 1.0242000818252563, + "learning_rate": 9.637026029256874e-05, + "loss": 3.1602, + "step": 4235 + }, + { + "epoch": 0.12, + "grad_norm": 0.9416380524635315, + "learning_rate": 9.636158500351713e-05, + "loss": 3.1801, + "step": 4240 + }, + { + "epoch": 0.12, + "grad_norm": 0.9604207277297974, + "learning_rate": 9.635289975104009e-05, + "loss": 3.14, + "step": 4245 + }, + { + "epoch": 0.13, + "grad_norm": 0.9354197978973389, + "learning_rate": 9.634420453700413e-05, + "loss": 3.1713, + "step": 4250 + }, + { + "epoch": 0.13, + "grad_norm": 0.9732345342636108, + "learning_rate": 9.633549936327789e-05, + "loss": 3.0962, + "step": 4255 + }, + { + "epoch": 0.13, + "grad_norm": 0.9559193849563599, + "learning_rate": 9.63267842317322e-05, + "loss": 3.2126, + "step": 4260 + }, + { + "epoch": 0.13, + "grad_norm": 0.9427840709686279, + "learning_rate": 9.631805914424e-05, + "loss": 3.2705, + "step": 4265 + }, + { + "epoch": 0.13, + "grad_norm": 0.9436772465705872, + "learning_rate": 9.630932410267637e-05, + "loss": 3.1428, + "step": 4270 + }, + { + "epoch": 0.13, + "grad_norm": 0.9520498514175415, + "learning_rate": 9.630057910891852e-05, + "loss": 3.2115, + "step": 4275 + }, + { + "epoch": 0.13, + "grad_norm": 0.9842592477798462, + "learning_rate": 9.629182416484582e-05, + "loss": 3.2112, + "step": 4280 + }, + { + "epoch": 0.13, + "grad_norm": 0.9719690680503845, + "learning_rate": 9.628305927233975e-05, + "loss": 3.196, + "step": 4285 + }, + { + "epoch": 0.13, + "grad_norm": 0.9838169813156128, + "learning_rate": 9.627428443328397e-05, + "loss": 3.1423, + "step": 4290 + }, + { + "epoch": 0.13, + "grad_norm": 1.1199114322662354, + "learning_rate": 9.626549964956424e-05, + "loss": 3.2253, + "step": 4295 + }, + { + "epoch": 0.13, + "grad_norm": 0.992247462272644, + "learning_rate": 9.625670492306848e-05, + "loss": 3.1605, + "step": 4300 + }, + { + "epoch": 0.13, + "grad_norm": 0.9515162110328674, + "learning_rate": 9.624790025568673e-05, + "loss": 3.0991, + "step": 4305 + }, + { + "epoch": 0.13, + "grad_norm": 0.9793718457221985, + "learning_rate": 9.623908564931117e-05, + "loss": 3.1371, + "step": 4310 + }, + { + "epoch": 0.13, + "grad_norm": 0.9227912425994873, + "learning_rate": 9.623026110583612e-05, + "loss": 3.1254, + "step": 4315 + }, + { + "epoch": 0.13, + "grad_norm": 0.9420730471611023, + "learning_rate": 9.622142662715808e-05, + "loss": 3.1866, + "step": 4320 + }, + { + "epoch": 0.13, + "grad_norm": 0.9549916386604309, + "learning_rate": 9.621258221517556e-05, + "loss": 3.0825, + "step": 4325 + }, + { + "epoch": 0.13, + "grad_norm": 0.9246135950088501, + "learning_rate": 9.620372787178934e-05, + "loss": 3.0439, + "step": 4330 + }, + { + "epoch": 0.13, + "grad_norm": 0.9154413342475891, + "learning_rate": 9.619486359890228e-05, + "loss": 3.1001, + "step": 4335 + }, + { + "epoch": 0.13, + "grad_norm": 0.9776270389556885, + "learning_rate": 9.618598939841934e-05, + "loss": 3.1465, + "step": 4340 + }, + { + "epoch": 0.13, + "grad_norm": 0.9864166975021362, + "learning_rate": 9.617710527224768e-05, + "loss": 3.0429, + "step": 4345 + }, + { + "epoch": 0.13, + "grad_norm": 1.1089850664138794, + "learning_rate": 9.616821122229656e-05, + "loss": 3.2106, + "step": 4350 + }, + { + "epoch": 0.13, + "grad_norm": 0.9036588072776794, + "learning_rate": 9.615930725047732e-05, + "loss": 3.0918, + "step": 4355 + }, + { + "epoch": 0.13, + "grad_norm": 0.938593864440918, + "learning_rate": 9.615039335870357e-05, + "loss": 3.1548, + "step": 4360 + }, + { + "epoch": 0.13, + "grad_norm": 0.9223225712776184, + "learning_rate": 9.614146954889091e-05, + "loss": 3.1185, + "step": 4365 + }, + { + "epoch": 0.13, + "grad_norm": 1.0628299713134766, + "learning_rate": 9.613253582295714e-05, + "loss": 3.2035, + "step": 4370 + }, + { + "epoch": 0.13, + "grad_norm": 0.963643491268158, + "learning_rate": 9.612359218282216e-05, + "loss": 3.0519, + "step": 4375 + }, + { + "epoch": 0.13, + "grad_norm": 0.9788914322853088, + "learning_rate": 9.611463863040807e-05, + "loss": 3.1374, + "step": 4380 + }, + { + "epoch": 0.13, + "grad_norm": 1.0176095962524414, + "learning_rate": 9.610567516763902e-05, + "loss": 3.1764, + "step": 4385 + }, + { + "epoch": 0.13, + "grad_norm": 0.9673143029212952, + "learning_rate": 9.60967017964413e-05, + "loss": 3.2011, + "step": 4390 + }, + { + "epoch": 0.13, + "grad_norm": 0.9602774977684021, + "learning_rate": 9.608771851874338e-05, + "loss": 3.1547, + "step": 4395 + }, + { + "epoch": 0.13, + "grad_norm": 0.9239591956138611, + "learning_rate": 9.607872533647584e-05, + "loss": 3.2295, + "step": 4400 + }, + { + "epoch": 0.13, + "grad_norm": 0.8919190764427185, + "learning_rate": 9.606972225157134e-05, + "loss": 3.1286, + "step": 4405 + }, + { + "epoch": 0.13, + "grad_norm": 0.959225058555603, + "learning_rate": 9.606070926596471e-05, + "loss": 3.1202, + "step": 4410 + }, + { + "epoch": 0.13, + "grad_norm": 1.0456562042236328, + "learning_rate": 9.605168638159294e-05, + "loss": 3.2411, + "step": 4415 + }, + { + "epoch": 0.13, + "grad_norm": 1.075289249420166, + "learning_rate": 9.604265360039507e-05, + "loss": 3.2417, + "step": 4420 + }, + { + "epoch": 0.13, + "grad_norm": 0.9902351498603821, + "learning_rate": 9.603361092431234e-05, + "loss": 3.1832, + "step": 4425 + }, + { + "epoch": 0.13, + "grad_norm": 0.9459065794944763, + "learning_rate": 9.602455835528804e-05, + "loss": 3.2088, + "step": 4430 + }, + { + "epoch": 0.13, + "grad_norm": 0.9626405239105225, + "learning_rate": 9.601549589526767e-05, + "loss": 3.2646, + "step": 4435 + }, + { + "epoch": 0.13, + "grad_norm": 0.9072127342224121, + "learning_rate": 9.60064235461988e-05, + "loss": 3.1363, + "step": 4440 + }, + { + "epoch": 0.13, + "grad_norm": 0.9629626870155334, + "learning_rate": 9.599734131003114e-05, + "loss": 3.1225, + "step": 4445 + }, + { + "epoch": 0.13, + "grad_norm": 1.0885331630706787, + "learning_rate": 9.598824918871654e-05, + "loss": 3.0919, + "step": 4450 + }, + { + "epoch": 0.13, + "grad_norm": 0.8944528698921204, + "learning_rate": 9.597914718420892e-05, + "loss": 3.1677, + "step": 4455 + }, + { + "epoch": 0.13, + "grad_norm": 0.952925443649292, + "learning_rate": 9.597003529846442e-05, + "loss": 3.2258, + "step": 4460 + }, + { + "epoch": 0.13, + "grad_norm": 0.8903512954711914, + "learning_rate": 9.59609135334412e-05, + "loss": 3.1708, + "step": 4465 + }, + { + "epoch": 0.13, + "grad_norm": 0.9634586572647095, + "learning_rate": 9.595178189109963e-05, + "loss": 3.1921, + "step": 4470 + }, + { + "epoch": 0.13, + "grad_norm": 0.9447696208953857, + "learning_rate": 9.594264037340212e-05, + "loss": 3.1636, + "step": 4475 + }, + { + "epoch": 0.13, + "grad_norm": 1.0390684604644775, + "learning_rate": 9.593348898231326e-05, + "loss": 3.2445, + "step": 4480 + }, + { + "epoch": 0.13, + "grad_norm": 0.9162504076957703, + "learning_rate": 9.592432771979978e-05, + "loss": 3.1568, + "step": 4485 + }, + { + "epoch": 0.13, + "grad_norm": 1.1015783548355103, + "learning_rate": 9.591515658783046e-05, + "loss": 3.19, + "step": 4490 + }, + { + "epoch": 0.13, + "grad_norm": 0.9759737253189087, + "learning_rate": 9.590597558837626e-05, + "loss": 3.2005, + "step": 4495 + }, + { + "epoch": 0.13, + "grad_norm": 0.9446675777435303, + "learning_rate": 9.589678472341023e-05, + "loss": 3.1061, + "step": 4500 + }, + { + "epoch": 0.13, + "grad_norm": 0.9221960306167603, + "learning_rate": 9.588758399490755e-05, + "loss": 3.0276, + "step": 4505 + }, + { + "epoch": 0.13, + "grad_norm": 0.9297600984573364, + "learning_rate": 9.587837340484553e-05, + "loss": 3.1172, + "step": 4510 + }, + { + "epoch": 0.13, + "grad_norm": 1.0142234563827515, + "learning_rate": 9.586915295520358e-05, + "loss": 3.1665, + "step": 4515 + }, + { + "epoch": 0.13, + "grad_norm": 1.0056792497634888, + "learning_rate": 9.585992264796324e-05, + "loss": 3.1312, + "step": 4520 + }, + { + "epoch": 0.13, + "grad_norm": 0.9645575284957886, + "learning_rate": 9.585068248510817e-05, + "loss": 3.13, + "step": 4525 + }, + { + "epoch": 0.13, + "grad_norm": 0.9708833694458008, + "learning_rate": 9.584143246862414e-05, + "loss": 3.1203, + "step": 4530 + }, + { + "epoch": 0.13, + "grad_norm": 1.0611423254013062, + "learning_rate": 9.583217260049907e-05, + "loss": 3.222, + "step": 4535 + }, + { + "epoch": 0.13, + "grad_norm": 0.9717942476272583, + "learning_rate": 9.582290288272291e-05, + "loss": 3.2117, + "step": 4540 + }, + { + "epoch": 0.13, + "grad_norm": 0.9020853042602539, + "learning_rate": 9.581362331728784e-05, + "loss": 3.0173, + "step": 4545 + }, + { + "epoch": 0.13, + "grad_norm": 1.0137736797332764, + "learning_rate": 9.580433390618808e-05, + "loss": 3.1873, + "step": 4550 + }, + { + "epoch": 0.13, + "grad_norm": 0.9122316241264343, + "learning_rate": 9.579503465142e-05, + "loss": 3.153, + "step": 4555 + }, + { + "epoch": 0.13, + "grad_norm": 0.9816657304763794, + "learning_rate": 9.578572555498207e-05, + "loss": 3.1384, + "step": 4560 + }, + { + "epoch": 0.13, + "grad_norm": 0.8921579122543335, + "learning_rate": 9.577640661887488e-05, + "loss": 3.2227, + "step": 4565 + }, + { + "epoch": 0.13, + "grad_norm": 0.9225032329559326, + "learning_rate": 9.576707784510113e-05, + "loss": 3.0866, + "step": 4570 + }, + { + "epoch": 0.13, + "grad_norm": 0.8976895213127136, + "learning_rate": 9.575773923566563e-05, + "loss": 3.1282, + "step": 4575 + }, + { + "epoch": 0.13, + "grad_norm": 0.8989979028701782, + "learning_rate": 9.574839079257532e-05, + "loss": 3.1176, + "step": 4580 + }, + { + "epoch": 0.13, + "grad_norm": 0.9586094617843628, + "learning_rate": 9.573903251783925e-05, + "loss": 3.1647, + "step": 4585 + }, + { + "epoch": 0.14, + "grad_norm": 1.0168122053146362, + "learning_rate": 9.572966441346858e-05, + "loss": 3.1484, + "step": 4590 + }, + { + "epoch": 0.14, + "grad_norm": 1.0488462448120117, + "learning_rate": 9.572028648147657e-05, + "loss": 3.143, + "step": 4595 + }, + { + "epoch": 0.14, + "grad_norm": 0.9871429204940796, + "learning_rate": 9.571089872387862e-05, + "loss": 3.0555, + "step": 4600 + }, + { + "epoch": 0.14, + "grad_norm": 0.9694020748138428, + "learning_rate": 9.57015011426922e-05, + "loss": 3.106, + "step": 4605 + }, + { + "epoch": 0.14, + "grad_norm": 0.9993796348571777, + "learning_rate": 9.569209373993693e-05, + "loss": 3.051, + "step": 4610 + }, + { + "epoch": 0.14, + "grad_norm": 0.9226096272468567, + "learning_rate": 9.568267651763451e-05, + "loss": 3.088, + "step": 4615 + }, + { + "epoch": 0.14, + "grad_norm": 0.9481409788131714, + "learning_rate": 9.56732494778088e-05, + "loss": 3.1625, + "step": 4620 + }, + { + "epoch": 0.14, + "grad_norm": 0.986202597618103, + "learning_rate": 9.56638126224857e-05, + "loss": 3.1433, + "step": 4625 + }, + { + "epoch": 0.14, + "grad_norm": 0.9785463809967041, + "learning_rate": 9.565436595369328e-05, + "loss": 3.0595, + "step": 4630 + }, + { + "epoch": 0.14, + "grad_norm": 1.0816633701324463, + "learning_rate": 9.564490947346168e-05, + "loss": 3.1522, + "step": 4635 + }, + { + "epoch": 0.14, + "grad_norm": 0.9942913055419922, + "learning_rate": 9.563544318382316e-05, + "loss": 3.1329, + "step": 4640 + }, + { + "epoch": 0.14, + "grad_norm": 0.9319725632667542, + "learning_rate": 9.56259670868121e-05, + "loss": 3.1554, + "step": 4645 + }, + { + "epoch": 0.14, + "grad_norm": 0.9760807752609253, + "learning_rate": 9.5616481184465e-05, + "loss": 3.1543, + "step": 4650 + }, + { + "epoch": 0.14, + "grad_norm": 0.9634469747543335, + "learning_rate": 9.560698547882041e-05, + "loss": 3.1396, + "step": 4655 + }, + { + "epoch": 0.14, + "grad_norm": 0.8733280301094055, + "learning_rate": 9.559747997191903e-05, + "loss": 3.0607, + "step": 4660 + }, + { + "epoch": 0.14, + "grad_norm": 0.9256601929664612, + "learning_rate": 9.558796466580366e-05, + "loss": 3.0816, + "step": 4665 + }, + { + "epoch": 0.14, + "grad_norm": 0.9982563257217407, + "learning_rate": 9.557843956251924e-05, + "loss": 3.1793, + "step": 4670 + }, + { + "epoch": 0.14, + "grad_norm": 0.9599419236183167, + "learning_rate": 9.556890466411272e-05, + "loss": 3.0765, + "step": 4675 + }, + { + "epoch": 0.14, + "grad_norm": 1.1619309186935425, + "learning_rate": 9.555935997263328e-05, + "loss": 3.1075, + "step": 4680 + }, + { + "epoch": 0.14, + "grad_norm": 0.972918689250946, + "learning_rate": 9.554980549013209e-05, + "loss": 3.0646, + "step": 4685 + }, + { + "epoch": 0.14, + "grad_norm": 1.02189040184021, + "learning_rate": 9.554024121866249e-05, + "loss": 3.0762, + "step": 4690 + }, + { + "epoch": 0.14, + "grad_norm": 0.9249009490013123, + "learning_rate": 9.55306671602799e-05, + "loss": 3.0952, + "step": 4695 + }, + { + "epoch": 0.14, + "grad_norm": 0.930340588092804, + "learning_rate": 9.552108331704189e-05, + "loss": 3.1389, + "step": 4700 + }, + { + "epoch": 0.14, + "grad_norm": 1.0409553050994873, + "learning_rate": 9.551148969100804e-05, + "loss": 3.1314, + "step": 4705 + }, + { + "epoch": 0.14, + "grad_norm": 0.9474987983703613, + "learning_rate": 9.550188628424012e-05, + "loss": 3.0758, + "step": 4710 + }, + { + "epoch": 0.14, + "grad_norm": 1.1047687530517578, + "learning_rate": 9.549227309880197e-05, + "loss": 3.0588, + "step": 4715 + }, + { + "epoch": 0.14, + "grad_norm": 0.9622979760169983, + "learning_rate": 9.548265013675951e-05, + "loss": 3.0415, + "step": 4720 + }, + { + "epoch": 0.14, + "grad_norm": 1.026153564453125, + "learning_rate": 9.54730174001808e-05, + "loss": 3.127, + "step": 4725 + }, + { + "epoch": 0.14, + "grad_norm": 0.999793529510498, + "learning_rate": 9.546337489113597e-05, + "loss": 3.0971, + "step": 4730 + }, + { + "epoch": 0.14, + "grad_norm": 0.9588595628738403, + "learning_rate": 9.545372261169725e-05, + "loss": 3.1047, + "step": 4735 + }, + { + "epoch": 0.14, + "grad_norm": 0.9668876528739929, + "learning_rate": 9.544406056393903e-05, + "loss": 3.1188, + "step": 4740 + }, + { + "epoch": 0.14, + "grad_norm": 0.9264355897903442, + "learning_rate": 9.54343887499377e-05, + "loss": 3.175, + "step": 4745 + }, + { + "epoch": 0.14, + "grad_norm": 0.9462378621101379, + "learning_rate": 9.542470717177182e-05, + "loss": 3.1262, + "step": 4750 + }, + { + "epoch": 0.14, + "grad_norm": 1.1726490259170532, + "learning_rate": 9.541501583152204e-05, + "loss": 3.1027, + "step": 4755 + }, + { + "epoch": 0.14, + "grad_norm": 1.0557887554168701, + "learning_rate": 9.540531473127106e-05, + "loss": 3.2259, + "step": 4760 + }, + { + "epoch": 0.14, + "grad_norm": 0.9835221171379089, + "learning_rate": 9.539560387310375e-05, + "loss": 3.1416, + "step": 4765 + }, + { + "epoch": 0.14, + "grad_norm": 1.0208748579025269, + "learning_rate": 9.538588325910702e-05, + "loss": 3.1589, + "step": 4770 + }, + { + "epoch": 0.14, + "grad_norm": 0.9093871116638184, + "learning_rate": 9.53761528913699e-05, + "loss": 3.2232, + "step": 4775 + }, + { + "epoch": 0.14, + "grad_norm": 1.0152350664138794, + "learning_rate": 9.536641277198353e-05, + "loss": 3.0421, + "step": 4780 + }, + { + "epoch": 0.14, + "grad_norm": 0.8923962116241455, + "learning_rate": 9.535666290304111e-05, + "loss": 2.9991, + "step": 4785 + }, + { + "epoch": 0.14, + "grad_norm": 1.0020900964736938, + "learning_rate": 9.534690328663796e-05, + "loss": 3.101, + "step": 4790 + }, + { + "epoch": 0.14, + "grad_norm": 0.9471282958984375, + "learning_rate": 9.533713392487149e-05, + "loss": 3.1177, + "step": 4795 + }, + { + "epoch": 0.14, + "grad_norm": 0.9707486033439636, + "learning_rate": 9.532735481984121e-05, + "loss": 3.2052, + "step": 4800 + }, + { + "epoch": 0.14, + "grad_norm": 0.9802972674369812, + "learning_rate": 9.53175659736487e-05, + "loss": 3.1157, + "step": 4805 + }, + { + "epoch": 0.14, + "grad_norm": 1.022676706314087, + "learning_rate": 9.530776738839766e-05, + "loss": 3.3467, + "step": 4810 + }, + { + "epoch": 0.14, + "grad_norm": 0.9314159154891968, + "learning_rate": 9.529795906619387e-05, + "loss": 3.1474, + "step": 4815 + }, + { + "epoch": 0.14, + "grad_norm": 1.0583088397979736, + "learning_rate": 9.528814100914522e-05, + "loss": 3.0841, + "step": 4820 + }, + { + "epoch": 0.14, + "grad_norm": 0.9484631419181824, + "learning_rate": 9.52783132193617e-05, + "loss": 3.1767, + "step": 4825 + }, + { + "epoch": 0.14, + "grad_norm": 0.9330920577049255, + "learning_rate": 9.52684756989553e-05, + "loss": 3.1109, + "step": 4830 + }, + { + "epoch": 0.14, + "grad_norm": 0.9174767732620239, + "learning_rate": 9.525862845004022e-05, + "loss": 3.1579, + "step": 4835 + }, + { + "epoch": 0.14, + "grad_norm": 0.9538732767105103, + "learning_rate": 9.52487714747327e-05, + "loss": 3.1477, + "step": 4840 + }, + { + "epoch": 0.14, + "grad_norm": 1.0374078750610352, + "learning_rate": 9.523890477515105e-05, + "loss": 3.0845, + "step": 4845 + }, + { + "epoch": 0.14, + "grad_norm": 0.892899215221405, + "learning_rate": 9.522902835341573e-05, + "loss": 3.1093, + "step": 4850 + }, + { + "epoch": 0.14, + "grad_norm": 0.9015700221061707, + "learning_rate": 9.521914221164922e-05, + "loss": 3.1629, + "step": 4855 + }, + { + "epoch": 0.14, + "grad_norm": 0.9192642569541931, + "learning_rate": 9.520924635197612e-05, + "loss": 3.1058, + "step": 4860 + }, + { + "epoch": 0.14, + "grad_norm": 0.9244606494903564, + "learning_rate": 9.519934077652312e-05, + "loss": 3.1187, + "step": 4865 + }, + { + "epoch": 0.14, + "grad_norm": 0.9331244826316833, + "learning_rate": 9.518942548741902e-05, + "loss": 3.1199, + "step": 4870 + }, + { + "epoch": 0.14, + "grad_norm": 0.9354492425918579, + "learning_rate": 9.517950048679467e-05, + "loss": 3.0673, + "step": 4875 + }, + { + "epoch": 0.14, + "grad_norm": 0.9620521664619446, + "learning_rate": 9.516956577678302e-05, + "loss": 3.0989, + "step": 4880 + }, + { + "epoch": 0.14, + "grad_norm": 0.8998059630393982, + "learning_rate": 9.51596213595191e-05, + "loss": 3.1812, + "step": 4885 + }, + { + "epoch": 0.14, + "grad_norm": 1.0036343336105347, + "learning_rate": 9.514966723714004e-05, + "loss": 3.1168, + "step": 4890 + }, + { + "epoch": 0.14, + "grad_norm": 0.9013308882713318, + "learning_rate": 9.513970341178506e-05, + "loss": 3.1105, + "step": 4895 + }, + { + "epoch": 0.14, + "grad_norm": 0.9903246760368347, + "learning_rate": 9.512972988559545e-05, + "loss": 3.0285, + "step": 4900 + }, + { + "epoch": 0.14, + "grad_norm": 0.9553041458129883, + "learning_rate": 9.511974666071458e-05, + "loss": 3.1084, + "step": 4905 + }, + { + "epoch": 0.14, + "grad_norm": 0.9445865750312805, + "learning_rate": 9.510975373928792e-05, + "loss": 3.0825, + "step": 4910 + }, + { + "epoch": 0.14, + "grad_norm": 0.9836540818214417, + "learning_rate": 9.509975112346303e-05, + "loss": 3.0393, + "step": 4915 + }, + { + "epoch": 0.14, + "grad_norm": 0.9509047865867615, + "learning_rate": 9.508973881538952e-05, + "loss": 3.1963, + "step": 4920 + }, + { + "epoch": 0.14, + "grad_norm": 1.0266945362091064, + "learning_rate": 9.507971681721911e-05, + "loss": 3.1994, + "step": 4925 + }, + { + "epoch": 0.15, + "grad_norm": 0.9312596917152405, + "learning_rate": 9.50696851311056e-05, + "loss": 3.0886, + "step": 4930 + }, + { + "epoch": 0.15, + "grad_norm": 1.0032775402069092, + "learning_rate": 9.505964375920489e-05, + "loss": 3.1894, + "step": 4935 + }, + { + "epoch": 0.15, + "grad_norm": 1.0451502799987793, + "learning_rate": 9.50495927036749e-05, + "loss": 3.0687, + "step": 4940 + }, + { + "epoch": 0.15, + "grad_norm": 1.2992931604385376, + "learning_rate": 9.503953196667569e-05, + "loss": 3.0106, + "step": 4945 + }, + { + "epoch": 0.15, + "grad_norm": 0.9820343852043152, + "learning_rate": 9.502946155036938e-05, + "loss": 3.1293, + "step": 4950 + }, + { + "epoch": 0.15, + "grad_norm": 0.9634777903556824, + "learning_rate": 9.50193814569202e-05, + "loss": 3.1373, + "step": 4955 + }, + { + "epoch": 0.15, + "grad_norm": 0.9421091675758362, + "learning_rate": 9.500929168849439e-05, + "loss": 3.1681, + "step": 4960 + }, + { + "epoch": 0.15, + "grad_norm": 0.9735555648803711, + "learning_rate": 9.499919224726032e-05, + "loss": 3.1241, + "step": 4965 + }, + { + "epoch": 0.15, + "grad_norm": 0.9242784380912781, + "learning_rate": 9.498908313538846e-05, + "loss": 3.1531, + "step": 4970 + }, + { + "epoch": 0.15, + "grad_norm": 0.9356673359870911, + "learning_rate": 9.49789643550513e-05, + "loss": 3.0116, + "step": 4975 + }, + { + "epoch": 0.15, + "grad_norm": 0.8965728282928467, + "learning_rate": 9.496883590842343e-05, + "loss": 3.0862, + "step": 4980 + }, + { + "epoch": 0.15, + "grad_norm": 1.014461636543274, + "learning_rate": 9.495869779768154e-05, + "loss": 3.083, + "step": 4985 + }, + { + "epoch": 0.15, + "grad_norm": 0.9236679077148438, + "learning_rate": 9.494855002500438e-05, + "loss": 3.1612, + "step": 4990 + }, + { + "epoch": 0.15, + "grad_norm": 1.08620023727417, + "learning_rate": 9.493839259257278e-05, + "loss": 3.1232, + "step": 4995 + }, + { + "epoch": 0.15, + "grad_norm": 1.0171806812286377, + "learning_rate": 9.492822550256962e-05, + "loss": 3.0397, + "step": 5000 + }, + { + "epoch": 0.15, + "grad_norm": 1.0970803499221802, + "learning_rate": 9.49180487571799e-05, + "loss": 3.2259, + "step": 5005 + }, + { + "epoch": 0.15, + "grad_norm": 0.9284546971321106, + "learning_rate": 9.490786235859066e-05, + "loss": 3.062, + "step": 5010 + }, + { + "epoch": 0.15, + "grad_norm": 0.9204232096672058, + "learning_rate": 9.489766630899104e-05, + "loss": 3.0677, + "step": 5015 + }, + { + "epoch": 0.15, + "grad_norm": 0.9892458915710449, + "learning_rate": 9.488746061057223e-05, + "loss": 3.1438, + "step": 5020 + }, + { + "epoch": 0.15, + "grad_norm": 0.9285604953765869, + "learning_rate": 9.487724526552753e-05, + "loss": 3.0606, + "step": 5025 + }, + { + "epoch": 0.15, + "grad_norm": 0.8983063697814941, + "learning_rate": 9.486702027605224e-05, + "loss": 3.0695, + "step": 5030 + }, + { + "epoch": 0.15, + "grad_norm": 0.9375035166740417, + "learning_rate": 9.485678564434381e-05, + "loss": 3.0337, + "step": 5035 + }, + { + "epoch": 0.15, + "grad_norm": 0.9231756925582886, + "learning_rate": 9.484654137260176e-05, + "loss": 3.1084, + "step": 5040 + }, + { + "epoch": 0.15, + "grad_norm": 0.8873231410980225, + "learning_rate": 9.483628746302761e-05, + "loss": 3.1179, + "step": 5045 + }, + { + "epoch": 0.15, + "grad_norm": 0.9875752329826355, + "learning_rate": 9.482602391782501e-05, + "loss": 3.1626, + "step": 5050 + }, + { + "epoch": 0.15, + "grad_norm": 0.8863257765769958, + "learning_rate": 9.481575073919968e-05, + "loss": 3.1358, + "step": 5055 + }, + { + "epoch": 0.15, + "grad_norm": 0.9098997712135315, + "learning_rate": 9.480546792935939e-05, + "loss": 3.0734, + "step": 5060 + }, + { + "epoch": 0.15, + "grad_norm": 0.9136868119239807, + "learning_rate": 9.479517549051398e-05, + "loss": 3.0315, + "step": 5065 + }, + { + "epoch": 0.15, + "grad_norm": 0.9826388955116272, + "learning_rate": 9.478487342487536e-05, + "loss": 3.0784, + "step": 5070 + }, + { + "epoch": 0.15, + "grad_norm": 0.885038435459137, + "learning_rate": 9.477456173465754e-05, + "loss": 3.0042, + "step": 5075 + }, + { + "epoch": 0.15, + "grad_norm": 0.9448022246360779, + "learning_rate": 9.476424042207655e-05, + "loss": 3.1119, + "step": 5080 + }, + { + "epoch": 0.15, + "grad_norm": 0.9787291288375854, + "learning_rate": 9.475390948935054e-05, + "loss": 3.1502, + "step": 5085 + }, + { + "epoch": 0.15, + "grad_norm": 0.9000442028045654, + "learning_rate": 9.474356893869967e-05, + "loss": 3.0938, + "step": 5090 + }, + { + "epoch": 0.15, + "grad_norm": 0.9612510800361633, + "learning_rate": 9.47332187723462e-05, + "loss": 3.0464, + "step": 5095 + }, + { + "epoch": 0.15, + "grad_norm": 0.9308996796607971, + "learning_rate": 9.472285899251445e-05, + "loss": 3.1386, + "step": 5100 + }, + { + "epoch": 0.15, + "grad_norm": 1.0861619710922241, + "learning_rate": 9.471248960143085e-05, + "loss": 3.0304, + "step": 5105 + }, + { + "epoch": 0.15, + "grad_norm": 0.9617395401000977, + "learning_rate": 9.47021106013238e-05, + "loss": 3.0773, + "step": 5110 + }, + { + "epoch": 0.15, + "grad_norm": 1.0649669170379639, + "learning_rate": 9.469172199442383e-05, + "loss": 3.0922, + "step": 5115 + }, + { + "epoch": 0.15, + "grad_norm": 0.9687249064445496, + "learning_rate": 9.468132378296355e-05, + "loss": 3.1337, + "step": 5120 + }, + { + "epoch": 0.15, + "grad_norm": 0.9213399887084961, + "learning_rate": 9.467091596917758e-05, + "loss": 3.0697, + "step": 5125 + }, + { + "epoch": 0.15, + "grad_norm": 0.964924156665802, + "learning_rate": 9.466049855530265e-05, + "loss": 3.0832, + "step": 5130 + }, + { + "epoch": 0.15, + "grad_norm": 0.9299331903457642, + "learning_rate": 9.465007154357752e-05, + "loss": 3.081, + "step": 5135 + }, + { + "epoch": 0.15, + "grad_norm": 1.089355707168579, + "learning_rate": 9.463963493624304e-05, + "loss": 3.0087, + "step": 5140 + }, + { + "epoch": 0.15, + "grad_norm": 0.9571266174316406, + "learning_rate": 9.462918873554209e-05, + "loss": 2.943, + "step": 5145 + }, + { + "epoch": 0.15, + "grad_norm": 0.9856504201889038, + "learning_rate": 9.461873294371965e-05, + "loss": 3.0669, + "step": 5150 + }, + { + "epoch": 0.15, + "grad_norm": 0.9385844469070435, + "learning_rate": 9.460826756302275e-05, + "loss": 3.1456, + "step": 5155 + }, + { + "epoch": 0.15, + "grad_norm": 1.135668396949768, + "learning_rate": 9.459779259570045e-05, + "loss": 2.9911, + "step": 5160 + }, + { + "epoch": 0.15, + "grad_norm": 1.0288223028182983, + "learning_rate": 9.458730804400389e-05, + "loss": 3.0772, + "step": 5165 + }, + { + "epoch": 0.15, + "grad_norm": 0.9754334092140198, + "learning_rate": 9.457681391018629e-05, + "loss": 3.1177, + "step": 5170 + }, + { + "epoch": 0.15, + "grad_norm": 0.9688030481338501, + "learning_rate": 9.45663101965029e-05, + "loss": 3.0614, + "step": 5175 + }, + { + "epoch": 0.15, + "grad_norm": 0.9702439308166504, + "learning_rate": 9.455579690521106e-05, + "loss": 3.1666, + "step": 5180 + }, + { + "epoch": 0.15, + "grad_norm": 0.9346526265144348, + "learning_rate": 9.454527403857013e-05, + "loss": 3.0653, + "step": 5185 + }, + { + "epoch": 0.15, + "grad_norm": 0.9160826802253723, + "learning_rate": 9.453474159884155e-05, + "loss": 3.0325, + "step": 5190 + }, + { + "epoch": 0.15, + "grad_norm": 1.067042350769043, + "learning_rate": 9.452419958828883e-05, + "loss": 3.081, + "step": 5195 + }, + { + "epoch": 0.15, + "grad_norm": 0.8856215476989746, + "learning_rate": 9.45136480091775e-05, + "loss": 3.0636, + "step": 5200 + }, + { + "epoch": 0.15, + "grad_norm": 1.0516242980957031, + "learning_rate": 9.450308686377516e-05, + "loss": 3.1103, + "step": 5205 + }, + { + "epoch": 0.15, + "grad_norm": 0.9815742373466492, + "learning_rate": 9.449251615435151e-05, + "loss": 3.018, + "step": 5210 + }, + { + "epoch": 0.15, + "grad_norm": 1.0035467147827148, + "learning_rate": 9.448193588317822e-05, + "loss": 3.0403, + "step": 5215 + }, + { + "epoch": 0.15, + "grad_norm": 0.9196162223815918, + "learning_rate": 9.447134605252912e-05, + "loss": 3.0895, + "step": 5220 + }, + { + "epoch": 0.15, + "grad_norm": 0.897792398929596, + "learning_rate": 9.446074666468e-05, + "loss": 3.0325, + "step": 5225 + }, + { + "epoch": 0.15, + "grad_norm": 0.9135429263114929, + "learning_rate": 9.445013772190874e-05, + "loss": 3.0916, + "step": 5230 + }, + { + "epoch": 0.15, + "grad_norm": 0.9409388303756714, + "learning_rate": 9.443951922649529e-05, + "loss": 3.0085, + "step": 5235 + }, + { + "epoch": 0.15, + "grad_norm": 0.9783315062522888, + "learning_rate": 9.442889118072162e-05, + "loss": 2.9461, + "step": 5240 + }, + { + "epoch": 0.15, + "grad_norm": 0.9518296122550964, + "learning_rate": 9.441825358687181e-05, + "loss": 3.0404, + "step": 5245 + }, + { + "epoch": 0.15, + "grad_norm": 0.9281454086303711, + "learning_rate": 9.440760644723192e-05, + "loss": 3.0663, + "step": 5250 + }, + { + "epoch": 0.15, + "grad_norm": 0.9811400771141052, + "learning_rate": 9.439694976409008e-05, + "loss": 3.1092, + "step": 5255 + }, + { + "epoch": 0.15, + "grad_norm": 0.9171397686004639, + "learning_rate": 9.438628353973653e-05, + "loss": 3.0938, + "step": 5260 + }, + { + "epoch": 0.15, + "grad_norm": 0.9300424456596375, + "learning_rate": 9.437560777646348e-05, + "loss": 3.0173, + "step": 5265 + }, + { + "epoch": 0.16, + "grad_norm": 0.8961432576179504, + "learning_rate": 9.436492247656523e-05, + "loss": 3.1348, + "step": 5270 + }, + { + "epoch": 0.16, + "grad_norm": 0.9847704172134399, + "learning_rate": 9.435422764233814e-05, + "loss": 3.0856, + "step": 5275 + }, + { + "epoch": 0.16, + "grad_norm": 0.9468925595283508, + "learning_rate": 9.43435232760806e-05, + "loss": 3.055, + "step": 5280 + }, + { + "epoch": 0.16, + "grad_norm": 0.9075853824615479, + "learning_rate": 9.433280938009304e-05, + "loss": 3.0759, + "step": 5285 + }, + { + "epoch": 0.16, + "grad_norm": 1.0659477710723877, + "learning_rate": 9.432208595667797e-05, + "loss": 3.064, + "step": 5290 + }, + { + "epoch": 0.16, + "grad_norm": 0.9474266171455383, + "learning_rate": 9.431135300813989e-05, + "loss": 3.1393, + "step": 5295 + }, + { + "epoch": 0.16, + "grad_norm": 0.8995897769927979, + "learning_rate": 9.430061053678542e-05, + "loss": 3.0231, + "step": 5300 + }, + { + "epoch": 0.16, + "grad_norm": 0.8841486573219299, + "learning_rate": 9.42898585449232e-05, + "loss": 3.1303, + "step": 5305 + }, + { + "epoch": 0.16, + "grad_norm": 0.9356794953346252, + "learning_rate": 9.427909703486388e-05, + "loss": 3.0927, + "step": 5310 + }, + { + "epoch": 0.16, + "grad_norm": 1.0089812278747559, + "learning_rate": 9.426832600892019e-05, + "loss": 3.15, + "step": 5315 + }, + { + "epoch": 0.16, + "grad_norm": 1.0069613456726074, + "learning_rate": 9.425754546940691e-05, + "loss": 2.9146, + "step": 5320 + }, + { + "epoch": 0.16, + "grad_norm": 1.038227915763855, + "learning_rate": 9.424675541864081e-05, + "loss": 3.1017, + "step": 5325 + }, + { + "epoch": 0.16, + "grad_norm": 0.9591827392578125, + "learning_rate": 9.423595585894082e-05, + "loss": 3.0895, + "step": 5330 + }, + { + "epoch": 0.16, + "grad_norm": 0.9559281468391418, + "learning_rate": 9.422514679262778e-05, + "loss": 3.0926, + "step": 5335 + }, + { + "epoch": 0.16, + "grad_norm": 0.9427942037582397, + "learning_rate": 9.421432822202465e-05, + "loss": 3.0139, + "step": 5340 + }, + { + "epoch": 0.16, + "grad_norm": 1.0378471612930298, + "learning_rate": 9.420350014945643e-05, + "loss": 3.1508, + "step": 5345 + }, + { + "epoch": 0.16, + "grad_norm": 1.048354983329773, + "learning_rate": 9.419266257725013e-05, + "loss": 3.0393, + "step": 5350 + }, + { + "epoch": 0.16, + "grad_norm": 1.051143765449524, + "learning_rate": 9.418181550773483e-05, + "loss": 3.0653, + "step": 5355 + }, + { + "epoch": 0.16, + "grad_norm": 1.0456222295761108, + "learning_rate": 9.417095894324163e-05, + "loss": 3.0485, + "step": 5360 + }, + { + "epoch": 0.16, + "grad_norm": 0.8933046460151672, + "learning_rate": 9.41600928861037e-05, + "loss": 3.0675, + "step": 5365 + }, + { + "epoch": 0.16, + "grad_norm": 0.8699972629547119, + "learning_rate": 9.41492173386562e-05, + "loss": 3.0926, + "step": 5370 + }, + { + "epoch": 0.16, + "grad_norm": 1.0275518894195557, + "learning_rate": 9.413833230323639e-05, + "loss": 3.0697, + "step": 5375 + }, + { + "epoch": 0.16, + "grad_norm": 0.9169639945030212, + "learning_rate": 9.412743778218351e-05, + "loss": 3.0625, + "step": 5380 + }, + { + "epoch": 0.16, + "grad_norm": 1.0008935928344727, + "learning_rate": 9.41165337778389e-05, + "loss": 3.1422, + "step": 5385 + }, + { + "epoch": 0.16, + "grad_norm": 0.8949229121208191, + "learning_rate": 9.410562029254589e-05, + "loss": 3.151, + "step": 5390 + }, + { + "epoch": 0.16, + "grad_norm": 1.7710978984832764, + "learning_rate": 9.409469732864986e-05, + "loss": 3.135, + "step": 5395 + }, + { + "epoch": 0.16, + "grad_norm": 0.9376665949821472, + "learning_rate": 9.408376488849824e-05, + "loss": 3.0602, + "step": 5400 + }, + { + "epoch": 0.16, + "grad_norm": 0.9570960998535156, + "learning_rate": 9.407282297444048e-05, + "loss": 3.1216, + "step": 5405 + }, + { + "epoch": 0.16, + "grad_norm": 0.8948537111282349, + "learning_rate": 9.406187158882808e-05, + "loss": 3.1084, + "step": 5410 + }, + { + "epoch": 0.16, + "grad_norm": 0.9144769906997681, + "learning_rate": 9.405091073401458e-05, + "loss": 3.0838, + "step": 5415 + }, + { + "epoch": 0.16, + "grad_norm": 0.9140651822090149, + "learning_rate": 9.40399404123555e-05, + "loss": 3.1402, + "step": 5420 + }, + { + "epoch": 0.16, + "grad_norm": 0.9569693207740784, + "learning_rate": 9.402896062620848e-05, + "loss": 3.0683, + "step": 5425 + }, + { + "epoch": 0.16, + "grad_norm": 1.0005345344543457, + "learning_rate": 9.401797137793313e-05, + "loss": 3.0874, + "step": 5430 + }, + { + "epoch": 0.16, + "grad_norm": 1.074997067451477, + "learning_rate": 9.400697266989112e-05, + "loss": 3.0866, + "step": 5435 + }, + { + "epoch": 0.16, + "grad_norm": 0.9990468621253967, + "learning_rate": 9.399596450444617e-05, + "loss": 3.084, + "step": 5440 + }, + { + "epoch": 0.16, + "grad_norm": 0.9001169800758362, + "learning_rate": 9.398494688396398e-05, + "loss": 3.1563, + "step": 5445 + }, + { + "epoch": 0.16, + "grad_norm": 0.9728519320487976, + "learning_rate": 9.397391981081232e-05, + "loss": 3.1318, + "step": 5450 + }, + { + "epoch": 0.16, + "grad_norm": 0.9565594792366028, + "learning_rate": 9.3962883287361e-05, + "loss": 3.1149, + "step": 5455 + }, + { + "epoch": 0.16, + "grad_norm": 0.9113457202911377, + "learning_rate": 9.395183731598183e-05, + "loss": 3.0861, + "step": 5460 + }, + { + "epoch": 0.16, + "grad_norm": 0.8806849718093872, + "learning_rate": 9.394078189904867e-05, + "loss": 3.096, + "step": 5465 + }, + { + "epoch": 0.16, + "grad_norm": 0.915174663066864, + "learning_rate": 9.392971703893741e-05, + "loss": 3.0986, + "step": 5470 + }, + { + "epoch": 0.16, + "grad_norm": 0.9114915728569031, + "learning_rate": 9.391864273802595e-05, + "loss": 3.1144, + "step": 5475 + }, + { + "epoch": 0.16, + "grad_norm": 0.9522095918655396, + "learning_rate": 9.390755899869425e-05, + "loss": 3.2203, + "step": 5480 + }, + { + "epoch": 0.16, + "grad_norm": 0.9693437814712524, + "learning_rate": 9.38964658233243e-05, + "loss": 3.1767, + "step": 5485 + }, + { + "epoch": 0.16, + "grad_norm": 1.0195435285568237, + "learning_rate": 9.388536321430005e-05, + "loss": 3.1332, + "step": 5490 + }, + { + "epoch": 0.16, + "grad_norm": 0.8879653811454773, + "learning_rate": 9.387425117400757e-05, + "loss": 3.0779, + "step": 5495 + }, + { + "epoch": 0.16, + "grad_norm": 0.9125641584396362, + "learning_rate": 9.386312970483489e-05, + "loss": 3.1022, + "step": 5500 + }, + { + "epoch": 0.16, + "grad_norm": 0.9602115154266357, + "learning_rate": 9.385199880917211e-05, + "loss": 3.0031, + "step": 5505 + }, + { + "epoch": 0.16, + "grad_norm": 1.0341030359268188, + "learning_rate": 9.384085848941132e-05, + "loss": 3.0895, + "step": 5510 + }, + { + "epoch": 0.16, + "grad_norm": 0.9830625057220459, + "learning_rate": 9.382970874794666e-05, + "loss": 3.0688, + "step": 5515 + }, + { + "epoch": 0.16, + "grad_norm": 1.0639983415603638, + "learning_rate": 9.381854958717429e-05, + "loss": 3.0003, + "step": 5520 + }, + { + "epoch": 0.16, + "grad_norm": 0.9073677659034729, + "learning_rate": 9.380738100949238e-05, + "loss": 3.1641, + "step": 5525 + }, + { + "epoch": 0.16, + "grad_norm": 3.0645134449005127, + "learning_rate": 9.379620301730114e-05, + "loss": 3.0915, + "step": 5530 + }, + { + "epoch": 0.16, + "grad_norm": 0.9002934098243713, + "learning_rate": 9.378501561300282e-05, + "loss": 3.0467, + "step": 5535 + }, + { + "epoch": 0.16, + "grad_norm": 0.9292882084846497, + "learning_rate": 9.377381879900162e-05, + "loss": 3.0605, + "step": 5540 + }, + { + "epoch": 0.16, + "grad_norm": 1.0092555284500122, + "learning_rate": 9.376261257770384e-05, + "loss": 3.0534, + "step": 5545 + }, + { + "epoch": 0.16, + "grad_norm": 0.9227961897850037, + "learning_rate": 9.37513969515178e-05, + "loss": 3.0049, + "step": 5550 + }, + { + "epoch": 0.16, + "grad_norm": 0.9298681616783142, + "learning_rate": 9.374017192285378e-05, + "loss": 3.0657, + "step": 5555 + }, + { + "epoch": 0.16, + "grad_norm": 1.0186702013015747, + "learning_rate": 9.372893749412413e-05, + "loss": 2.9714, + "step": 5560 + }, + { + "epoch": 0.16, + "grad_norm": 0.9279263615608215, + "learning_rate": 9.371769366774321e-05, + "loss": 3.0813, + "step": 5565 + }, + { + "epoch": 0.16, + "grad_norm": 0.9058989882469177, + "learning_rate": 9.370644044612738e-05, + "loss": 3.0831, + "step": 5570 + }, + { + "epoch": 0.16, + "grad_norm": 1.07100510597229, + "learning_rate": 9.369517783169505e-05, + "loss": 3.0151, + "step": 5575 + }, + { + "epoch": 0.16, + "grad_norm": 0.8887584805488586, + "learning_rate": 9.368390582686663e-05, + "loss": 3.1176, + "step": 5580 + }, + { + "epoch": 0.16, + "grad_norm": 0.9808238744735718, + "learning_rate": 9.367262443406456e-05, + "loss": 3.0446, + "step": 5585 + }, + { + "epoch": 0.16, + "grad_norm": 0.9783839583396912, + "learning_rate": 9.366133365571328e-05, + "loss": 3.1415, + "step": 5590 + }, + { + "epoch": 0.16, + "grad_norm": 1.1077193021774292, + "learning_rate": 9.365003349423924e-05, + "loss": 3.0162, + "step": 5595 + }, + { + "epoch": 0.16, + "grad_norm": 0.923333466053009, + "learning_rate": 9.363872395207097e-05, + "loss": 3.0176, + "step": 5600 + }, + { + "epoch": 0.16, + "grad_norm": 0.954433798789978, + "learning_rate": 9.362740503163891e-05, + "loss": 3.0693, + "step": 5605 + }, + { + "epoch": 0.17, + "grad_norm": 0.9824564456939697, + "learning_rate": 9.361607673537563e-05, + "loss": 3.0451, + "step": 5610 + }, + { + "epoch": 0.17, + "grad_norm": 0.9252722859382629, + "learning_rate": 9.36047390657156e-05, + "loss": 2.9954, + "step": 5615 + }, + { + "epoch": 0.17, + "grad_norm": 0.9171250462532043, + "learning_rate": 9.359339202509543e-05, + "loss": 3.0212, + "step": 5620 + }, + { + "epoch": 0.17, + "grad_norm": 0.9020513892173767, + "learning_rate": 9.358203561595363e-05, + "loss": 2.9942, + "step": 5625 + }, + { + "epoch": 0.17, + "grad_norm": 0.863895058631897, + "learning_rate": 9.357066984073079e-05, + "loss": 3.1678, + "step": 5630 + }, + { + "epoch": 0.17, + "grad_norm": 0.909782886505127, + "learning_rate": 9.355929470186948e-05, + "loss": 3.0788, + "step": 5635 + }, + { + "epoch": 0.17, + "grad_norm": 0.9356575608253479, + "learning_rate": 9.354791020181431e-05, + "loss": 3.0337, + "step": 5640 + }, + { + "epoch": 0.17, + "grad_norm": 0.9049946665763855, + "learning_rate": 9.353651634301188e-05, + "loss": 3.1495, + "step": 5645 + }, + { + "epoch": 0.17, + "grad_norm": 1.0070096254348755, + "learning_rate": 9.352511312791083e-05, + "loss": 3.0585, + "step": 5650 + }, + { + "epoch": 0.17, + "grad_norm": 0.982075035572052, + "learning_rate": 9.351370055896175e-05, + "loss": 3.0193, + "step": 5655 + }, + { + "epoch": 0.17, + "grad_norm": 0.9161244034767151, + "learning_rate": 9.350227863861732e-05, + "loss": 3.1001, + "step": 5660 + }, + { + "epoch": 0.17, + "grad_norm": 0.9532464146614075, + "learning_rate": 9.349084736933215e-05, + "loss": 3.0325, + "step": 5665 + }, + { + "epoch": 0.17, + "grad_norm": 0.9546010494232178, + "learning_rate": 9.347940675356292e-05, + "loss": 3.0102, + "step": 5670 + }, + { + "epoch": 0.17, + "grad_norm": 0.8775145411491394, + "learning_rate": 9.346795679376832e-05, + "loss": 3.0969, + "step": 5675 + }, + { + "epoch": 0.17, + "grad_norm": 0.9970228672027588, + "learning_rate": 9.345649749240898e-05, + "loss": 3.026, + "step": 5680 + }, + { + "epoch": 0.17, + "grad_norm": 1.0569418668746948, + "learning_rate": 9.344502885194763e-05, + "loss": 3.1053, + "step": 5685 + }, + { + "epoch": 0.17, + "grad_norm": 0.9413895010948181, + "learning_rate": 9.343355087484894e-05, + "loss": 3.212, + "step": 5690 + }, + { + "epoch": 0.17, + "grad_norm": 0.8960546255111694, + "learning_rate": 9.342206356357956e-05, + "loss": 3.0312, + "step": 5695 + }, + { + "epoch": 0.17, + "grad_norm": 0.9607859253883362, + "learning_rate": 9.341056692060828e-05, + "loss": 3.1139, + "step": 5700 + }, + { + "epoch": 0.17, + "grad_norm": 1.0707714557647705, + "learning_rate": 9.339906094840575e-05, + "loss": 3.031, + "step": 5705 + }, + { + "epoch": 0.17, + "grad_norm": 0.8942881226539612, + "learning_rate": 9.33875456494447e-05, + "loss": 3.0716, + "step": 5710 + }, + { + "epoch": 0.17, + "grad_norm": 0.8838489651679993, + "learning_rate": 9.337602102619985e-05, + "loss": 3.0491, + "step": 5715 + }, + { + "epoch": 0.17, + "grad_norm": 0.9379225373268127, + "learning_rate": 9.336448708114789e-05, + "loss": 3.0897, + "step": 5720 + }, + { + "epoch": 0.17, + "grad_norm": 0.9009203314781189, + "learning_rate": 9.33529438167676e-05, + "loss": 3.0508, + "step": 5725 + }, + { + "epoch": 0.17, + "grad_norm": 0.9227913618087769, + "learning_rate": 9.334139123553967e-05, + "loss": 3.0185, + "step": 5730 + }, + { + "epoch": 0.17, + "grad_norm": 0.903289794921875, + "learning_rate": 9.332982933994683e-05, + "loss": 3.1071, + "step": 5735 + }, + { + "epoch": 0.17, + "grad_norm": 0.8733929991722107, + "learning_rate": 9.331825813247381e-05, + "loss": 3.1138, + "step": 5740 + }, + { + "epoch": 0.17, + "grad_norm": 0.9137933850288391, + "learning_rate": 9.330667761560739e-05, + "loss": 3.1246, + "step": 5745 + }, + { + "epoch": 0.17, + "grad_norm": 0.9088469743728638, + "learning_rate": 9.329508779183623e-05, + "loss": 3.1022, + "step": 5750 + }, + { + "epoch": 0.17, + "grad_norm": 0.9645144939422607, + "learning_rate": 9.328348866365111e-05, + "loss": 3.0103, + "step": 5755 + }, + { + "epoch": 0.17, + "grad_norm": 0.8542119264602661, + "learning_rate": 9.327188023354474e-05, + "loss": 3.0196, + "step": 5760 + }, + { + "epoch": 0.17, + "grad_norm": 0.9103429913520813, + "learning_rate": 9.326026250401188e-05, + "loss": 2.9316, + "step": 5765 + }, + { + "epoch": 0.17, + "grad_norm": 0.884625256061554, + "learning_rate": 9.324863547754924e-05, + "loss": 2.9978, + "step": 5770 + }, + { + "epoch": 0.17, + "grad_norm": 0.9044618606567383, + "learning_rate": 9.323699915665555e-05, + "loss": 3.1545, + "step": 5775 + }, + { + "epoch": 0.17, + "grad_norm": 0.9126080274581909, + "learning_rate": 9.322535354383153e-05, + "loss": 3.0609, + "step": 5780 + }, + { + "epoch": 0.17, + "grad_norm": 0.8592166900634766, + "learning_rate": 9.321369864157992e-05, + "loss": 2.9966, + "step": 5785 + }, + { + "epoch": 0.17, + "grad_norm": 0.9289705753326416, + "learning_rate": 9.320203445240543e-05, + "loss": 3.0706, + "step": 5790 + }, + { + "epoch": 0.17, + "grad_norm": 0.9263682961463928, + "learning_rate": 9.319036097881477e-05, + "loss": 3.0623, + "step": 5795 + }, + { + "epoch": 0.17, + "grad_norm": 0.9361786246299744, + "learning_rate": 9.317867822331667e-05, + "loss": 3.0399, + "step": 5800 + }, + { + "epoch": 0.17, + "grad_norm": 1.0559189319610596, + "learning_rate": 9.31669861884218e-05, + "loss": 3.0121, + "step": 5805 + }, + { + "epoch": 0.17, + "grad_norm": 1.1284879446029663, + "learning_rate": 9.315528487664289e-05, + "loss": 3.1219, + "step": 5810 + }, + { + "epoch": 0.17, + "grad_norm": 0.8799806237220764, + "learning_rate": 9.314357429049462e-05, + "loss": 2.988, + "step": 5815 + }, + { + "epoch": 0.17, + "grad_norm": 0.9400696754455566, + "learning_rate": 9.313185443249367e-05, + "loss": 3.0069, + "step": 5820 + }, + { + "epoch": 0.17, + "grad_norm": 0.8967433571815491, + "learning_rate": 9.312012530515875e-05, + "loss": 2.9988, + "step": 5825 + }, + { + "epoch": 0.17, + "grad_norm": 0.8952714800834656, + "learning_rate": 9.310838691101048e-05, + "loss": 3.0182, + "step": 5830 + }, + { + "epoch": 0.17, + "grad_norm": 0.8868004083633423, + "learning_rate": 9.309663925257156e-05, + "loss": 3.1065, + "step": 5835 + }, + { + "epoch": 0.17, + "grad_norm": 0.9409423470497131, + "learning_rate": 9.308488233236663e-05, + "loss": 3.0056, + "step": 5840 + }, + { + "epoch": 0.17, + "grad_norm": 0.9090366959571838, + "learning_rate": 9.307311615292234e-05, + "loss": 3.0804, + "step": 5845 + }, + { + "epoch": 0.17, + "grad_norm": 0.8912823796272278, + "learning_rate": 9.306134071676734e-05, + "loss": 3.1234, + "step": 5850 + }, + { + "epoch": 0.17, + "grad_norm": 0.9315927624702454, + "learning_rate": 9.30495560264322e-05, + "loss": 2.9337, + "step": 5855 + }, + { + "epoch": 0.17, + "grad_norm": 0.8894001841545105, + "learning_rate": 9.303776208444958e-05, + "loss": 3.1096, + "step": 5860 + }, + { + "epoch": 0.17, + "grad_norm": 0.8975959420204163, + "learning_rate": 9.302595889335405e-05, + "loss": 3.0436, + "step": 5865 + }, + { + "epoch": 0.17, + "grad_norm": 1.018897533416748, + "learning_rate": 9.301414645568223e-05, + "loss": 3.0479, + "step": 5870 + }, + { + "epoch": 0.17, + "grad_norm": 0.9984728097915649, + "learning_rate": 9.300232477397266e-05, + "loss": 3.1387, + "step": 5875 + }, + { + "epoch": 0.17, + "grad_norm": 0.9613728523254395, + "learning_rate": 9.299049385076592e-05, + "loss": 3.1193, + "step": 5880 + }, + { + "epoch": 0.17, + "grad_norm": 0.9925873279571533, + "learning_rate": 9.297865368860456e-05, + "loss": 2.9969, + "step": 5885 + }, + { + "epoch": 0.17, + "grad_norm": 0.8893855214118958, + "learning_rate": 9.296680429003311e-05, + "loss": 3.0197, + "step": 5890 + }, + { + "epoch": 0.17, + "grad_norm": 0.9727075695991516, + "learning_rate": 9.295494565759808e-05, + "loss": 3.0039, + "step": 5895 + }, + { + "epoch": 0.17, + "grad_norm": 0.9391651153564453, + "learning_rate": 9.294307779384796e-05, + "loss": 3.0504, + "step": 5900 + }, + { + "epoch": 0.17, + "grad_norm": 0.8870016932487488, + "learning_rate": 9.293120070133327e-05, + "loss": 3.026, + "step": 5905 + }, + { + "epoch": 0.17, + "grad_norm": 0.9716064929962158, + "learning_rate": 9.291931438260646e-05, + "loss": 3.1189, + "step": 5910 + }, + { + "epoch": 0.17, + "grad_norm": 0.9531162977218628, + "learning_rate": 9.290741884022198e-05, + "loss": 3.0919, + "step": 5915 + }, + { + "epoch": 0.17, + "grad_norm": 0.8801805377006531, + "learning_rate": 9.28955140767363e-05, + "loss": 3.0755, + "step": 5920 + }, + { + "epoch": 0.17, + "grad_norm": 0.8843389749526978, + "learning_rate": 9.288360009470776e-05, + "loss": 3.0561, + "step": 5925 + }, + { + "epoch": 0.17, + "grad_norm": 0.9179670214653015, + "learning_rate": 9.287167689669683e-05, + "loss": 3.0457, + "step": 5930 + }, + { + "epoch": 0.17, + "grad_norm": 0.9082323312759399, + "learning_rate": 9.285974448526585e-05, + "loss": 3.0574, + "step": 5935 + }, + { + "epoch": 0.17, + "grad_norm": 0.9412997961044312, + "learning_rate": 9.284780286297918e-05, + "loss": 3.0019, + "step": 5940 + }, + { + "epoch": 0.17, + "grad_norm": 0.957048237323761, + "learning_rate": 9.283585203240317e-05, + "loss": 3.0066, + "step": 5945 + }, + { + "epoch": 0.18, + "grad_norm": 0.9767757654190063, + "learning_rate": 9.282389199610613e-05, + "loss": 2.9845, + "step": 5950 + }, + { + "epoch": 0.18, + "grad_norm": 0.948574960231781, + "learning_rate": 9.281192275665836e-05, + "loss": 3.0659, + "step": 5955 + }, + { + "epoch": 0.18, + "grad_norm": 0.9179603457450867, + "learning_rate": 9.279994431663212e-05, + "loss": 3.0879, + "step": 5960 + }, + { + "epoch": 0.18, + "grad_norm": 1.0086420774459839, + "learning_rate": 9.278795667860167e-05, + "loss": 3.0935, + "step": 5965 + }, + { + "epoch": 0.18, + "grad_norm": 0.919671893119812, + "learning_rate": 9.277595984514325e-05, + "loss": 3.0571, + "step": 5970 + }, + { + "epoch": 0.18, + "grad_norm": 0.9324042797088623, + "learning_rate": 9.276395381883502e-05, + "loss": 3.1011, + "step": 5975 + }, + { + "epoch": 0.18, + "grad_norm": 1.0073186159133911, + "learning_rate": 9.275193860225722e-05, + "loss": 3.101, + "step": 5980 + }, + { + "epoch": 0.18, + "grad_norm": 0.8804324269294739, + "learning_rate": 9.273991419799193e-05, + "loss": 2.9656, + "step": 5985 + }, + { + "epoch": 0.18, + "grad_norm": 0.9571539759635925, + "learning_rate": 9.272788060862334e-05, + "loss": 3.0818, + "step": 5990 + }, + { + "epoch": 0.18, + "grad_norm": 0.9541386365890503, + "learning_rate": 9.271583783673751e-05, + "loss": 3.0246, + "step": 5995 + }, + { + "epoch": 0.18, + "grad_norm": 0.8818163275718689, + "learning_rate": 9.270378588492255e-05, + "loss": 3.0014, + "step": 6000 + } + ], + "logging_steps": 5, + "max_steps": 33984, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 1000, + "total_flos": 1.7753342881038336e+19, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}