{ "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, "learning_rate": 5e-06, "loss": 30.0104, "step": 5 }, { "epoch": 0.0, "learning_rate": 1e-05, "loss": 27.986, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.5e-05, "loss": 23.6395, "step": 15 }, { "epoch": 0.0, "learning_rate": 2e-05, "loss": 17.7111, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.5e-05, "loss": 13.0408, "step": 25 }, { "epoch": 0.0, "learning_rate": 3e-05, "loss": 11.0499, "step": 30 }, { "epoch": 0.0, "learning_rate": 3.5e-05, "loss": 9.9476, "step": 35 }, { "epoch": 0.0, "learning_rate": 4e-05, "loss": 9.1079, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.5e-05, "loss": 8.8318, "step": 45 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 8.8517, "step": 50 }, { "epoch": 0.0, "learning_rate": 5.500000000000001e-05, "loss": 8.8265, "step": 55 }, { "epoch": 0.0, "learning_rate": 6e-05, "loss": 8.7785, "step": 60 }, { "epoch": 0.0, "learning_rate": 6.500000000000001e-05, "loss": 8.7632, "step": 65 }, { "epoch": 0.0, "learning_rate": 7e-05, "loss": 8.707, "step": 70 }, { "epoch": 0.0, "learning_rate": 7.500000000000001e-05, "loss": 8.6049, "step": 75 }, { "epoch": 0.0, "learning_rate": 8e-05, "loss": 8.4293, "step": 80 }, { "epoch": 0.0, "learning_rate": 8.5e-05, "loss": 8.3325, "step": 85 }, { "epoch": 0.0, "learning_rate": 9e-05, "loss": 8.1294, "step": 90 }, { "epoch": 0.0, "learning_rate": 9.5e-05, "loss": 8.012, "step": 95 }, { "epoch": 0.0, "learning_rate": 0.0001, "loss": 7.8133, "step": 100 }, { "epoch": 0.0, "learning_rate": 9.999999462732701e-05, "loss": 7.5378, "step": 105 }, { "epoch": 0.0, "learning_rate": 9.999997850930916e-05, "loss": 7.4495, "step": 110 }, { "epoch": 0.0, "learning_rate": 9.999995164594991e-05, "loss": 7.3195, "step": 115 }, { "epoch": 0.0, "learning_rate": 9.999991403725507e-05, "loss": 7.1785, "step": 120 }, { "epoch": 0.0, "learning_rate": 9.99998656832327e-05, "loss": 6.9352, "step": 125 }, { "epoch": 0.0, "learning_rate": 9.999980658389317e-05, "loss": 6.8662, "step": 130 }, { "epoch": 0.0, "learning_rate": 9.999973673924921e-05, "loss": 6.7574, "step": 135 }, { "epoch": 0.0, "learning_rate": 9.999965614931583e-05, "loss": 6.5434, "step": 140 }, { "epoch": 0.0, "learning_rate": 9.999956481411035e-05, "loss": 6.6469, "step": 145 }, { "epoch": 0.0, "learning_rate": 9.999946273365238e-05, "loss": 6.4677, "step": 150 }, { "epoch": 0.0, "learning_rate": 9.999934990796387e-05, "loss": 6.4762, "step": 155 }, { "epoch": 0.0, "learning_rate": 9.999922633706906e-05, "loss": 6.3507, "step": 160 }, { "epoch": 0.0, "learning_rate": 9.999909202099453e-05, "loss": 6.1945, "step": 165 }, { "epoch": 0.01, "learning_rate": 9.999894695976912e-05, "loss": 6.2177, "step": 170 }, { "epoch": 0.01, "learning_rate": 9.9998791153424e-05, "loss": 6.0936, "step": 175 }, { "epoch": 0.01, "learning_rate": 9.999862460199267e-05, "loss": 6.1226, "step": 180 }, { "epoch": 0.01, "learning_rate": 9.999844730551092e-05, "loss": 5.9323, "step": 185 }, { "epoch": 0.01, "learning_rate": 9.999825926401686e-05, "loss": 5.9803, "step": 190 }, { "epoch": 0.01, "learning_rate": 9.999806047755088e-05, "loss": 5.8003, "step": 195 }, { "epoch": 0.01, "learning_rate": 9.999785094615573e-05, "loss": 5.935, "step": 200 }, { "epoch": 0.01, "learning_rate": 9.99976306698764e-05, "loss": 5.8743, "step": 205 }, { "epoch": 0.01, "learning_rate": 9.999739964876026e-05, "loss": 5.85, "step": 210 }, { "epoch": 0.01, "learning_rate": 9.999715788285695e-05, "loss": 5.8635, "step": 215 }, { "epoch": 0.01, "learning_rate": 9.999690537221844e-05, "loss": 5.7312, "step": 220 }, { "epoch": 0.01, "learning_rate": 9.999664211689897e-05, "loss": 5.6263, "step": 225 }, { "epoch": 0.01, "learning_rate": 9.999636811695513e-05, "loss": 5.7016, "step": 230 }, { "epoch": 0.01, "learning_rate": 9.99960833724458e-05, "loss": 5.6857, "step": 235 }, { "epoch": 0.01, "learning_rate": 9.999578788343218e-05, "loss": 5.6411, "step": 240 }, { "epoch": 0.01, "learning_rate": 9.999548164997778e-05, "loss": 5.5853, "step": 245 }, { "epoch": 0.01, "learning_rate": 9.999516467214839e-05, "loss": 5.5338, "step": 250 }, { "epoch": 0.01, "learning_rate": 9.999483695001214e-05, "loss": 5.4931, "step": 255 }, { "epoch": 0.01, "learning_rate": 9.999449848363946e-05, "loss": 5.4754, "step": 260 }, { "epoch": 0.01, "learning_rate": 9.999414927310309e-05, "loss": 5.4821, "step": 265 }, { "epoch": 0.01, "learning_rate": 9.99937893184781e-05, "loss": 5.3167, "step": 270 }, { "epoch": 0.01, "learning_rate": 9.99934186198418e-05, "loss": 5.3791, "step": 275 }, { "epoch": 0.01, "learning_rate": 9.999303717727389e-05, "loss": 5.4523, "step": 280 }, { "epoch": 0.01, "learning_rate": 9.999264499085633e-05, "loss": 5.4004, "step": 285 }, { "epoch": 0.01, "learning_rate": 9.999224206067341e-05, "loss": 5.2604, "step": 290 }, { "epoch": 0.01, "learning_rate": 9.999182838681171e-05, "loss": 5.295, "step": 295 }, { "epoch": 0.01, "learning_rate": 9.999140396936016e-05, "loss": 5.313, "step": 300 }, { "epoch": 0.01, "learning_rate": 9.999096880840994e-05, "loss": 5.2235, "step": 305 }, { "epoch": 0.01, "learning_rate": 9.999052290405459e-05, "loss": 5.27, "step": 310 }, { "epoch": 0.01, "learning_rate": 9.999006625638994e-05, "loss": 5.0792, "step": 315 }, { "epoch": 0.01, "learning_rate": 9.998959886551411e-05, "loss": 5.1927, "step": 320 }, { "epoch": 0.01, "learning_rate": 9.998912073152755e-05, "loss": 5.3568, "step": 325 }, { "epoch": 0.01, "learning_rate": 9.998863185453302e-05, "loss": 5.1588, "step": 330 }, { "epoch": 0.01, "learning_rate": 9.998813223463557e-05, "loss": 5.0924, "step": 335 }, { "epoch": 0.01, "learning_rate": 9.998762187194258e-05, "loss": 5.0127, "step": 340 }, { "epoch": 0.01, "learning_rate": 9.998710076656373e-05, "loss": 5.1553, "step": 345 }, { "epoch": 0.01, "learning_rate": 9.998656891861102e-05, "loss": 5.0836, "step": 350 }, { "epoch": 0.01, "learning_rate": 9.998602632819874e-05, "loss": 5.0929, "step": 355 }, { "epoch": 0.01, "learning_rate": 9.998547299544348e-05, "loss": 5.122, "step": 360 }, { "epoch": 0.01, "learning_rate": 9.99849089204642e-05, "loss": 4.9976, "step": 365 }, { "epoch": 0.01, "learning_rate": 9.998433410338206e-05, "loss": 5.0557, "step": 370 }, { "epoch": 0.01, "learning_rate": 9.998374854432064e-05, "loss": 5.0537, "step": 375 }, { "epoch": 0.01, "learning_rate": 9.998315224340577e-05, "loss": 5.0935, "step": 380 }, { "epoch": 0.01, "learning_rate": 9.99825452007656e-05, "loss": 4.9919, "step": 385 }, { "epoch": 0.01, "learning_rate": 9.998192741653057e-05, "loss": 5.0295, "step": 390 }, { "epoch": 0.01, "learning_rate": 9.998129889083346e-05, "loss": 4.9885, "step": 395 }, { "epoch": 0.01, "learning_rate": 9.998065962380936e-05, "loss": 5.0529, "step": 400 }, { "epoch": 0.01, "learning_rate": 9.998000961559562e-05, "loss": 5.013, "step": 405 }, { "epoch": 0.01, "learning_rate": 9.997934886633195e-05, "loss": 4.9495, "step": 410 }, { "epoch": 0.01, "learning_rate": 9.997867737616035e-05, "loss": 4.9346, "step": 415 }, { "epoch": 0.01, "learning_rate": 9.997799514522513e-05, "loss": 4.9806, "step": 420 }, { "epoch": 0.01, "learning_rate": 9.99773021736729e-05, "loss": 4.8289, "step": 425 }, { "epoch": 0.01, "learning_rate": 9.99765984616526e-05, "loss": 4.7909, "step": 430 }, { "epoch": 0.01, "learning_rate": 9.997588400931542e-05, "loss": 4.8708, "step": 435 }, { "epoch": 0.01, "learning_rate": 9.997515881681496e-05, "loss": 4.7989, "step": 440 }, { "epoch": 0.01, "learning_rate": 9.9974422884307e-05, "loss": 4.8636, "step": 445 }, { "epoch": 0.01, "learning_rate": 9.997367621194977e-05, "loss": 4.8722, "step": 450 }, { "epoch": 0.01, "learning_rate": 9.99729187999037e-05, "loss": 4.8804, "step": 455 }, { "epoch": 0.01, "learning_rate": 9.997215064833155e-05, "loss": 4.8629, "step": 460 }, { "epoch": 0.01, "learning_rate": 9.99713717573984e-05, "loss": 4.8167, "step": 465 }, { "epoch": 0.01, "learning_rate": 9.997058212727169e-05, "loss": 4.7899, "step": 470 }, { "epoch": 0.01, "learning_rate": 9.996978175812105e-05, "loss": 4.7899, "step": 475 }, { "epoch": 0.01, "learning_rate": 9.996897065011852e-05, "loss": 4.7709, "step": 480 }, { "epoch": 0.01, "learning_rate": 9.996814880343842e-05, "loss": 4.9026, "step": 485 }, { "epoch": 0.01, "learning_rate": 9.996731621825736e-05, "loss": 4.718, "step": 490 }, { "epoch": 0.01, "learning_rate": 9.996647289475426e-05, "loss": 4.8441, "step": 495 }, { "epoch": 0.01, "learning_rate": 9.996561883311036e-05, "loss": 4.8473, "step": 500 }, { "epoch": 0.01, "learning_rate": 9.996475403350921e-05, "loss": 4.742, "step": 505 }, { "epoch": 0.02, "learning_rate": 9.996387849613667e-05, "loss": 4.6555, "step": 510 }, { "epoch": 0.02, "learning_rate": 9.996299222118086e-05, "loss": 4.7345, "step": 515 }, { "epoch": 0.02, "learning_rate": 9.99620952088323e-05, "loss": 4.7299, "step": 520 }, { "epoch": 0.02, "learning_rate": 9.996118745928373e-05, "loss": 4.789, "step": 525 }, { "epoch": 0.02, "learning_rate": 9.996026897273024e-05, "loss": 4.7117, "step": 530 }, { "epoch": 0.02, "learning_rate": 9.995933974936922e-05, "loss": 4.71, "step": 535 }, { "epoch": 0.02, "learning_rate": 9.995839978940036e-05, "loss": 4.671, "step": 540 }, { "epoch": 0.02, "learning_rate": 9.995744909302567e-05, "loss": 4.6464, "step": 545 }, { "epoch": 0.02, "learning_rate": 9.995648766044947e-05, "loss": 4.5087, "step": 550 }, { "epoch": 0.02, "learning_rate": 9.995551549187836e-05, "loss": 4.6573, "step": 555 }, { "epoch": 0.02, "learning_rate": 9.995453258752127e-05, "loss": 4.6923, "step": 560 }, { "epoch": 0.02, "learning_rate": 9.995353894758945e-05, "loss": 4.6669, "step": 565 }, { "epoch": 0.02, "learning_rate": 9.995253457229643e-05, "loss": 4.6168, "step": 570 }, { "epoch": 0.02, "learning_rate": 9.995151946185804e-05, "loss": 4.7275, "step": 575 }, { "epoch": 0.02, "learning_rate": 9.995049361649247e-05, "loss": 4.6274, "step": 580 }, { "epoch": 0.02, "learning_rate": 9.994945703642013e-05, "loss": 4.5075, "step": 585 }, { "epoch": 0.02, "learning_rate": 9.994840972186386e-05, "loss": 4.6039, "step": 590 }, { "epoch": 0.02, "learning_rate": 9.994735167304866e-05, "loss": 4.5737, "step": 595 }, { "epoch": 0.02, "learning_rate": 9.994628289020195e-05, "loss": 4.609, "step": 600 }, { "epoch": 0.02, "learning_rate": 9.994520337355344e-05, "loss": 4.5973, "step": 605 }, { "epoch": 0.02, "learning_rate": 9.994411312333507e-05, "loss": 4.569, "step": 610 }, { "epoch": 0.02, "learning_rate": 9.99430121397812e-05, "loss": 4.5129, "step": 615 }, { "epoch": 0.02, "learning_rate": 9.994190042312839e-05, "loss": 4.4482, "step": 620 }, { "epoch": 0.02, "learning_rate": 9.994077797361559e-05, "loss": 4.4489, "step": 625 }, { "epoch": 0.02, "learning_rate": 9.9939644791484e-05, "loss": 4.5746, "step": 630 }, { "epoch": 0.02, "learning_rate": 9.993850087697716e-05, "loss": 4.5344, "step": 635 }, { "epoch": 0.02, "learning_rate": 9.99373462303409e-05, "loss": 4.5364, "step": 640 }, { "epoch": 0.02, "learning_rate": 9.993618085182337e-05, "loss": 4.5124, "step": 645 }, { "epoch": 0.02, "learning_rate": 9.993500474167501e-05, "loss": 4.5076, "step": 650 }, { "epoch": 0.02, "learning_rate": 9.99338179001486e-05, "loss": 4.5371, "step": 655 }, { "epoch": 0.02, "learning_rate": 9.993262032749914e-05, "loss": 4.5011, "step": 660 }, { "epoch": 0.02, "learning_rate": 9.993141202398407e-05, "loss": 4.6287, "step": 665 }, { "epoch": 0.02, "learning_rate": 9.993019298986301e-05, "loss": 4.4824, "step": 670 }, { "epoch": 0.02, "learning_rate": 9.992896322539796e-05, "loss": 4.4851, "step": 675 }, { "epoch": 0.02, "learning_rate": 9.992772273085319e-05, "loss": 4.5572, "step": 680 }, { "epoch": 0.02, "learning_rate": 9.99264715064953e-05, "loss": 4.497, "step": 685 }, { "epoch": 0.02, "learning_rate": 9.992520955259319e-05, "loss": 4.408, "step": 690 }, { "epoch": 0.02, "learning_rate": 9.992393686941808e-05, "loss": 4.5223, "step": 695 }, { "epoch": 0.02, "learning_rate": 9.992265345724344e-05, "loss": 4.4169, "step": 700 }, { "epoch": 0.02, "learning_rate": 9.99213593163451e-05, "loss": 4.5091, "step": 705 }, { "epoch": 0.02, "learning_rate": 9.992005444700121e-05, "loss": 4.4778, "step": 710 }, { "epoch": 0.02, "learning_rate": 9.991873884949214e-05, "loss": 4.399, "step": 715 }, { "epoch": 0.02, "learning_rate": 9.991741252410067e-05, "loss": 4.511, "step": 720 }, { "epoch": 0.02, "learning_rate": 9.99160754711118e-05, "loss": 4.4506, "step": 725 }, { "epoch": 0.02, "learning_rate": 9.99147276908129e-05, "loss": 4.4764, "step": 730 }, { "epoch": 0.02, "learning_rate": 9.991336918349361e-05, "loss": 4.4094, "step": 735 }, { "epoch": 0.02, "learning_rate": 9.991199994944586e-05, "loss": 4.4094, "step": 740 }, { "epoch": 0.02, "learning_rate": 9.991061998896393e-05, "loss": 4.3604, "step": 745 }, { "epoch": 0.02, "learning_rate": 9.99092293023444e-05, "loss": 4.4135, "step": 750 }, { "epoch": 0.02, "learning_rate": 9.990782788988611e-05, "loss": 4.5705, "step": 755 }, { "epoch": 0.02, "learning_rate": 9.990641575189022e-05, "loss": 4.2939, "step": 760 }, { "epoch": 0.02, "learning_rate": 9.990499288866024e-05, "loss": 4.3473, "step": 765 }, { "epoch": 0.02, "learning_rate": 9.990355930050194e-05, "loss": 4.5073, "step": 770 }, { "epoch": 0.02, "learning_rate": 9.990211498772341e-05, "loss": 4.3242, "step": 775 }, { "epoch": 0.02, "learning_rate": 9.990065995063506e-05, "loss": 4.2982, "step": 780 }, { "epoch": 0.02, "learning_rate": 9.989919418954956e-05, "loss": 4.4117, "step": 785 }, { "epoch": 0.02, "learning_rate": 9.989771770478191e-05, "loss": 4.428, "step": 790 }, { "epoch": 0.02, "learning_rate": 9.989623049664944e-05, "loss": 4.3136, "step": 795 }, { "epoch": 0.02, "learning_rate": 9.989473256547175e-05, "loss": 4.3675, "step": 800 }, { "epoch": 0.02, "learning_rate": 9.989322391157076e-05, "loss": 4.3455, "step": 805 }, { "epoch": 0.02, "learning_rate": 9.989170453527068e-05, "loss": 4.3208, "step": 810 }, { "epoch": 0.02, "learning_rate": 9.989017443689805e-05, "loss": 4.3025, "step": 815 }, { "epoch": 0.02, "learning_rate": 9.988863361678169e-05, "loss": 4.3735, "step": 820 }, { "epoch": 0.02, "learning_rate": 9.988708207525272e-05, "loss": 4.4329, "step": 825 }, { "epoch": 0.02, "learning_rate": 9.98855198126446e-05, "loss": 4.22, "step": 830 }, { "epoch": 0.02, "learning_rate": 9.988394682929305e-05, "loss": 4.313, "step": 835 }, { "epoch": 0.02, "learning_rate": 9.988236312553614e-05, "loss": 4.2936, "step": 840 }, { "epoch": 0.02, "learning_rate": 9.988076870171419e-05, "loss": 4.3449, "step": 845 }, { "epoch": 0.03, "learning_rate": 9.987916355816988e-05, "loss": 4.3452, "step": 850 }, { "epoch": 0.03, "learning_rate": 9.987754769524814e-05, "loss": 4.3523, "step": 855 }, { "epoch": 0.03, "learning_rate": 9.987592111329626e-05, "loss": 4.2805, "step": 860 }, { "epoch": 0.03, "learning_rate": 9.987428381266377e-05, "loss": 4.2829, "step": 865 }, { "epoch": 0.03, "learning_rate": 9.987263579370257e-05, "loss": 4.2336, "step": 870 }, { "epoch": 0.03, "learning_rate": 9.987097705676681e-05, "loss": 4.3863, "step": 875 }, { "epoch": 0.03, "learning_rate": 9.986930760221296e-05, "loss": 4.2955, "step": 880 }, { "epoch": 0.03, "learning_rate": 9.986762743039983e-05, "loss": 4.3271, "step": 885 }, { "epoch": 0.03, "learning_rate": 9.986593654168846e-05, "loss": 4.2802, "step": 890 }, { "epoch": 0.03, "learning_rate": 9.986423493644225e-05, "loss": 4.2536, "step": 895 }, { "epoch": 0.03, "learning_rate": 9.98625226150269e-05, "loss": 4.1219, "step": 900 }, { "epoch": 0.03, "learning_rate": 9.986079957781038e-05, "loss": 4.2771, "step": 905 }, { "epoch": 0.03, "learning_rate": 9.9859065825163e-05, "loss": 4.3244, "step": 910 }, { "epoch": 0.03, "learning_rate": 9.985732135745733e-05, "loss": 4.1981, "step": 915 }, { "epoch": 0.03, "learning_rate": 9.985556617506828e-05, "loss": 4.275, "step": 920 }, { "epoch": 0.03, "learning_rate": 9.985380027837308e-05, "loss": 4.3209, "step": 925 }, { "epoch": 0.03, "learning_rate": 9.98520236677512e-05, "loss": 4.1942, "step": 930 }, { "epoch": 0.03, "learning_rate": 9.985023634358443e-05, "loss": 4.1957, "step": 935 }, { "epoch": 0.03, "learning_rate": 9.984843830625691e-05, "loss": 4.2713, "step": 940 }, { "epoch": 0.03, "learning_rate": 9.984662955615507e-05, "loss": 4.2086, "step": 945 }, { "epoch": 0.03, "learning_rate": 9.984481009366758e-05, "loss": 4.2981, "step": 950 }, { "epoch": 0.03, "learning_rate": 9.984297991918548e-05, "loss": 4.1756, "step": 955 }, { "epoch": 0.03, "learning_rate": 9.984113903310206e-05, "loss": 4.1073, "step": 960 }, { "epoch": 0.03, "learning_rate": 9.983928743581296e-05, "loss": 4.2986, "step": 965 }, { "epoch": 0.03, "learning_rate": 9.983742512771611e-05, "loss": 4.1608, "step": 970 }, { "epoch": 0.03, "learning_rate": 9.983555210921171e-05, "loss": 4.3653, "step": 975 }, { "epoch": 0.03, "learning_rate": 9.983366838070231e-05, "loss": 4.2126, "step": 980 }, { "epoch": 0.03, "learning_rate": 9.983177394259271e-05, "loss": 4.2142, "step": 985 }, { "epoch": 0.03, "learning_rate": 9.982986879529006e-05, "loss": 4.1543, "step": 990 }, { "epoch": 0.03, "learning_rate": 9.982795293920377e-05, "loss": 4.2624, "step": 995 }, { "epoch": 0.03, "learning_rate": 9.982602637474558e-05, "loss": 4.1611, "step": 1000 }, { "epoch": 0.03, "learning_rate": 9.982408910232952e-05, "loss": 4.2513, "step": 1005 }, { "epoch": 0.03, "learning_rate": 9.982214112237193e-05, "loss": 4.2404, "step": 1010 }, { "epoch": 0.03, "learning_rate": 9.982018243529146e-05, "loss": 4.2662, "step": 1015 }, { "epoch": 0.03, "learning_rate": 9.981821304150901e-05, "loss": 4.28, "step": 1020 }, { "epoch": 0.03, "learning_rate": 9.981623294144782e-05, "loss": 4.2233, "step": 1025 }, { "epoch": 0.03, "learning_rate": 9.981424213553344e-05, "loss": 4.2186, "step": 1030 }, { "epoch": 0.03, "learning_rate": 9.981224062419372e-05, "loss": 4.1453, "step": 1035 }, { "epoch": 0.03, "learning_rate": 9.981022840785877e-05, "loss": 4.2346, "step": 1040 }, { "epoch": 0.03, "learning_rate": 9.980820548696106e-05, "loss": 4.2862, "step": 1045 }, { "epoch": 0.03, "learning_rate": 9.98061718619353e-05, "loss": 4.2036, "step": 1050 }, { "epoch": 0.03, "learning_rate": 9.980412753321856e-05, "loss": 4.2849, "step": 1055 }, { "epoch": 0.03, "learning_rate": 9.980207250125015e-05, "loss": 4.2314, "step": 1060 }, { "epoch": 0.03, "learning_rate": 9.980000676647174e-05, "loss": 4.1407, "step": 1065 }, { "epoch": 0.03, "learning_rate": 9.979793032932725e-05, "loss": 4.2014, "step": 1070 }, { "epoch": 0.03, "learning_rate": 9.979584319026292e-05, "loss": 4.0781, "step": 1075 }, { "epoch": 0.03, "learning_rate": 9.97937453497273e-05, "loss": 4.1771, "step": 1080 }, { "epoch": 0.03, "learning_rate": 9.979163680817124e-05, "loss": 4.0947, "step": 1085 }, { "epoch": 0.03, "learning_rate": 9.978951756604786e-05, "loss": 4.1947, "step": 1090 }, { "epoch": 0.03, "learning_rate": 9.978738762381262e-05, "loss": 4.2824, "step": 1095 }, { "epoch": 0.03, "learning_rate": 9.978524698192322e-05, "loss": 4.1224, "step": 1100 }, { "epoch": 0.03, "learning_rate": 9.978309564083975e-05, "loss": 4.0452, "step": 1105 }, { "epoch": 0.03, "learning_rate": 9.97809336010245e-05, "loss": 4.1537, "step": 1110 }, { "epoch": 0.03, "learning_rate": 9.977876086294216e-05, "loss": 4.1805, "step": 1115 }, { "epoch": 0.03, "learning_rate": 9.977657742705962e-05, "loss": 4.1567, "step": 1120 }, { "epoch": 0.03, "learning_rate": 9.977438329384614e-05, "loss": 4.2354, "step": 1125 }, { "epoch": 0.03, "learning_rate": 9.977217846377324e-05, "loss": 4.2062, "step": 1130 }, { "epoch": 0.03, "learning_rate": 9.976996293731476e-05, "loss": 4.1888, "step": 1135 }, { "epoch": 0.03, "learning_rate": 9.976773671494684e-05, "loss": 4.2171, "step": 1140 }, { "epoch": 0.03, "learning_rate": 9.976549979714791e-05, "loss": 4.0381, "step": 1145 }, { "epoch": 0.03, "learning_rate": 9.976325218439868e-05, "loss": 4.0751, "step": 1150 }, { "epoch": 0.03, "learning_rate": 9.976099387718218e-05, "loss": 4.0616, "step": 1155 }, { "epoch": 0.03, "learning_rate": 9.975872487598377e-05, "loss": 4.2296, "step": 1160 }, { "epoch": 0.03, "learning_rate": 9.975644518129104e-05, "loss": 4.1641, "step": 1165 }, { "epoch": 0.03, "learning_rate": 9.975415479359394e-05, "loss": 4.1444, "step": 1170 }, { "epoch": 0.03, "learning_rate": 9.975185371338464e-05, "loss": 4.104, "step": 1175 }, { "epoch": 0.03, "learning_rate": 9.97495419411577e-05, "loss": 4.191, "step": 1180 }, { "epoch": 0.03, "learning_rate": 9.974721947740995e-05, "loss": 4.1479, "step": 1185 }, { "epoch": 0.04, "learning_rate": 9.974488632264046e-05, "loss": 4.2479, "step": 1190 }, { "epoch": 0.04, "learning_rate": 9.974254247735066e-05, "loss": 3.9928, "step": 1195 }, { "epoch": 0.04, "learning_rate": 9.974018794204426e-05, "loss": 4.0395, "step": 1200 }, { "epoch": 0.04, "learning_rate": 9.973782271722727e-05, "loss": 4.0294, "step": 1205 }, { "epoch": 0.04, "learning_rate": 9.973544680340799e-05, "loss": 4.1699, "step": 1210 }, { "epoch": 0.04, "learning_rate": 9.973306020109701e-05, "loss": 4.1959, "step": 1215 }, { "epoch": 0.04, "learning_rate": 9.973066291080724e-05, "loss": 4.1337, "step": 1220 }, { "epoch": 0.04, "learning_rate": 9.972825493305387e-05, "loss": 3.9253, "step": 1225 }, { "epoch": 0.04, "learning_rate": 9.97258362683544e-05, "loss": 3.9745, "step": 1230 }, { "epoch": 0.04, "learning_rate": 9.972340691722858e-05, "loss": 4.0388, "step": 1235 }, { "epoch": 0.04, "learning_rate": 9.972096688019856e-05, "loss": 4.0299, "step": 1240 }, { "epoch": 0.04, "learning_rate": 9.971851615778868e-05, "loss": 4.1346, "step": 1245 }, { "epoch": 0.04, "learning_rate": 9.971605475052561e-05, "loss": 4.0309, "step": 1250 }, { "epoch": 0.04, "learning_rate": 9.971358265893834e-05, "loss": 4.0077, "step": 1255 }, { "epoch": 0.04, "learning_rate": 9.971109988355814e-05, "loss": 4.0894, "step": 1260 }, { "epoch": 0.04, "learning_rate": 9.970860642491854e-05, "loss": 3.9746, "step": 1265 }, { "epoch": 0.04, "learning_rate": 9.970610228355548e-05, "loss": 3.9986, "step": 1270 }, { "epoch": 0.04, "learning_rate": 9.970358746000704e-05, "loss": 4.0238, "step": 1275 }, { "epoch": 0.04, "learning_rate": 9.97010619548137e-05, "loss": 4.1338, "step": 1280 }, { "epoch": 0.04, "learning_rate": 9.969852576851822e-05, "loss": 3.9637, "step": 1285 }, { "epoch": 0.04, "learning_rate": 9.969597890166565e-05, "loss": 4.1273, "step": 1290 }, { "epoch": 0.04, "learning_rate": 9.969342135480328e-05, "loss": 3.9881, "step": 1295 }, { "epoch": 0.04, "learning_rate": 9.969085312848081e-05, "loss": 3.9957, "step": 1300 }, { "epoch": 0.04, "learning_rate": 9.968827422325012e-05, "loss": 3.9988, "step": 1305 }, { "epoch": 0.04, "learning_rate": 9.968568463966545e-05, "loss": 4.095, "step": 1310 }, { "epoch": 0.04, "learning_rate": 9.968308437828333e-05, "loss": 4.0647, "step": 1315 }, { "epoch": 0.04, "learning_rate": 9.968047343966256e-05, "loss": 3.9688, "step": 1320 }, { "epoch": 0.04, "learning_rate": 9.967785182436426e-05, "loss": 3.9631, "step": 1325 }, { "epoch": 0.04, "learning_rate": 9.967521953295185e-05, "loss": 4.0623, "step": 1330 }, { "epoch": 0.04, "learning_rate": 9.967257656599099e-05, "loss": 4.052, "step": 1335 }, { "epoch": 0.04, "learning_rate": 9.966992292404967e-05, "loss": 3.9196, "step": 1340 }, { "epoch": 0.04, "learning_rate": 9.966725860769821e-05, "loss": 3.9434, "step": 1345 }, { "epoch": 0.04, "learning_rate": 9.966458361750916e-05, "loss": 4.1073, "step": 1350 }, { "epoch": 0.04, "learning_rate": 9.966189795405743e-05, "loss": 3.9485, "step": 1355 }, { "epoch": 0.04, "learning_rate": 9.965920161792016e-05, "loss": 4.0475, "step": 1360 }, { "epoch": 0.04, "learning_rate": 9.96564946096768e-05, "loss": 4.0232, "step": 1365 }, { "epoch": 0.04, "learning_rate": 9.965377692990914e-05, "loss": 3.9488, "step": 1370 }, { "epoch": 0.04, "learning_rate": 9.965104857920121e-05, "loss": 3.949, "step": 1375 }, { "epoch": 0.04, "learning_rate": 9.964830955813935e-05, "loss": 3.8405, "step": 1380 }, { "epoch": 0.04, "learning_rate": 9.964555986731219e-05, "loss": 4.0324, "step": 1385 }, { "epoch": 0.04, "learning_rate": 9.964279950731066e-05, "loss": 4.0716, "step": 1390 }, { "epoch": 0.04, "learning_rate": 9.9640028478728e-05, "loss": 4.0121, "step": 1395 }, { "epoch": 0.04, "learning_rate": 9.963724678215968e-05, "loss": 3.8169, "step": 1400 }, { "epoch": 0.04, "learning_rate": 9.963445441820356e-05, "loss": 3.8104, "step": 1405 }, { "epoch": 0.04, "learning_rate": 9.963165138745969e-05, "loss": 3.9153, "step": 1410 }, { "epoch": 0.04, "learning_rate": 9.962883769053051e-05, "loss": 4.004, "step": 1415 }, { "epoch": 0.04, "learning_rate": 9.962601332802065e-05, "loss": 3.8096, "step": 1420 }, { "epoch": 0.04, "learning_rate": 9.962317830053711e-05, "loss": 4.0118, "step": 1425 }, { "epoch": 0.04, "learning_rate": 9.962033260868916e-05, "loss": 3.9132, "step": 1430 }, { "epoch": 0.04, "learning_rate": 9.961747625308838e-05, "loss": 3.9927, "step": 1435 }, { "epoch": 0.04, "learning_rate": 9.961460923434858e-05, "loss": 3.8739, "step": 1440 }, { "epoch": 0.04, "learning_rate": 9.961173155308591e-05, "loss": 3.994, "step": 1445 }, { "epoch": 0.04, "learning_rate": 9.960884320991881e-05, "loss": 3.9355, "step": 1450 }, { "epoch": 0.04, "learning_rate": 9.960594420546802e-05, "loss": 3.91, "step": 1455 }, { "epoch": 0.04, "learning_rate": 9.960303454035654e-05, "loss": 3.9702, "step": 1460 }, { "epoch": 0.04, "learning_rate": 9.960011421520969e-05, "loss": 4.0192, "step": 1465 }, { "epoch": 0.04, "learning_rate": 9.959718323065503e-05, "loss": 4.0134, "step": 1470 }, { "epoch": 0.04, "learning_rate": 9.959424158732252e-05, "loss": 3.837, "step": 1475 }, { "epoch": 0.04, "learning_rate": 9.959128928584426e-05, "loss": 4.0134, "step": 1480 }, { "epoch": 0.04, "learning_rate": 9.958832632685478e-05, "loss": 3.9343, "step": 1485 }, { "epoch": 0.04, "learning_rate": 9.95853527109908e-05, "loss": 3.9714, "step": 1490 }, { "epoch": 0.04, "learning_rate": 9.95823684388914e-05, "loss": 4.0173, "step": 1495 }, { "epoch": 0.04, "learning_rate": 9.95793735111979e-05, "loss": 3.8927, "step": 1500 }, { "epoch": 0.04, "learning_rate": 9.957636792855394e-05, "loss": 3.9544, "step": 1505 }, { "epoch": 0.04, "learning_rate": 9.957335169160545e-05, "loss": 4.0155, "step": 1510 }, { "epoch": 0.04, "learning_rate": 9.957032480100061e-05, "loss": 3.9991, "step": 1515 }, { "epoch": 0.04, "learning_rate": 9.956728725738996e-05, "loss": 3.8879, "step": 1520 }, { "epoch": 0.04, "learning_rate": 9.956423906142626e-05, "loss": 3.8843, "step": 1525 }, { "epoch": 0.05, "learning_rate": 9.956118021376461e-05, "loss": 3.8918, "step": 1530 }, { "epoch": 0.05, "learning_rate": 9.955811071506236e-05, "loss": 3.8666, "step": 1535 }, { "epoch": 0.05, "learning_rate": 9.955503056597917e-05, "loss": 3.8574, "step": 1540 }, { "epoch": 0.05, "learning_rate": 9.955193976717698e-05, "loss": 3.9926, "step": 1545 }, { "epoch": 0.05, "learning_rate": 9.954883831932003e-05, "loss": 3.9262, "step": 1550 }, { "epoch": 0.05, "learning_rate": 9.954572622307485e-05, "loss": 4.0488, "step": 1555 }, { "epoch": 0.05, "learning_rate": 9.954260347911025e-05, "loss": 3.9376, "step": 1560 }, { "epoch": 0.05, "learning_rate": 9.953947008809731e-05, "loss": 3.9499, "step": 1565 }, { "epoch": 0.05, "learning_rate": 9.953632605070944e-05, "loss": 4.0606, "step": 1570 }, { "epoch": 0.05, "learning_rate": 9.95331713676223e-05, "loss": 3.8886, "step": 1575 }, { "epoch": 0.05, "learning_rate": 9.953000603951386e-05, "loss": 3.8749, "step": 1580 }, { "epoch": 0.05, "learning_rate": 9.952683006706438e-05, "loss": 3.9932, "step": 1585 }, { "epoch": 0.05, "learning_rate": 9.952364345095638e-05, "loss": 3.8857, "step": 1590 }, { "epoch": 0.05, "learning_rate": 9.95204461918747e-05, "loss": 3.8816, "step": 1595 }, { "epoch": 0.05, "learning_rate": 9.951723829050646e-05, "loss": 3.8959, "step": 1600 }, { "epoch": 0.05, "learning_rate": 9.951401974754101e-05, "loss": 3.9856, "step": 1605 }, { "epoch": 0.05, "learning_rate": 9.951079056367013e-05, "loss": 3.7517, "step": 1610 }, { "epoch": 0.05, "learning_rate": 9.95075507395877e-05, "loss": 3.934, "step": 1615 }, { "epoch": 0.05, "learning_rate": 9.950430027599003e-05, "loss": 3.8781, "step": 1620 }, { "epoch": 0.05, "learning_rate": 9.950103917357566e-05, "loss": 3.8482, "step": 1625 }, { "epoch": 0.05, "learning_rate": 9.949776743304542e-05, "loss": 3.9253, "step": 1630 }, { "epoch": 0.05, "learning_rate": 9.949448505510243e-05, "loss": 3.9341, "step": 1635 }, { "epoch": 0.05, "learning_rate": 9.94911920404521e-05, "loss": 3.9171, "step": 1640 }, { "epoch": 0.05, "learning_rate": 9.94878883898021e-05, "loss": 3.8751, "step": 1645 }, { "epoch": 0.05, "learning_rate": 9.948457410386244e-05, "loss": 3.8468, "step": 1650 }, { "epoch": 0.05, "learning_rate": 9.948124918334538e-05, "loss": 4.0173, "step": 1655 }, { "epoch": 0.05, "learning_rate": 9.947791362896543e-05, "loss": 3.9076, "step": 1660 }, { "epoch": 0.05, "learning_rate": 9.947456744143947e-05, "loss": 3.8304, "step": 1665 }, { "epoch": 0.05, "learning_rate": 9.947121062148658e-05, "loss": 3.8634, "step": 1670 }, { "epoch": 0.05, "learning_rate": 9.94678431698282e-05, "loss": 3.8487, "step": 1675 }, { "epoch": 0.05, "learning_rate": 9.946446508718801e-05, "loss": 3.792, "step": 1680 }, { "epoch": 0.05, "learning_rate": 9.946107637429196e-05, "loss": 3.9171, "step": 1685 }, { "epoch": 0.05, "learning_rate": 9.945767703186832e-05, "loss": 3.8678, "step": 1690 }, { "epoch": 0.05, "learning_rate": 9.945426706064764e-05, "loss": 3.9718, "step": 1695 }, { "epoch": 0.05, "learning_rate": 9.945084646136275e-05, "loss": 3.9278, "step": 1700 }, { "epoch": 0.05, "learning_rate": 9.944741523474874e-05, "loss": 3.8965, "step": 1705 }, { "epoch": 0.05, "learning_rate": 9.944397338154301e-05, "loss": 3.8804, "step": 1710 }, { "epoch": 0.05, "learning_rate": 9.944052090248526e-05, "loss": 3.9099, "step": 1715 }, { "epoch": 0.05, "learning_rate": 9.943705779831742e-05, "loss": 3.957, "step": 1720 }, { "epoch": 0.05, "learning_rate": 9.943358406978377e-05, "loss": 3.876, "step": 1725 }, { "epoch": 0.05, "learning_rate": 9.943009971763082e-05, "loss": 3.8351, "step": 1730 }, { "epoch": 0.05, "learning_rate": 9.942660474260737e-05, "loss": 3.8931, "step": 1735 }, { "epoch": 0.05, "learning_rate": 9.942309914546453e-05, "loss": 3.8407, "step": 1740 }, { "epoch": 0.05, "learning_rate": 9.941958292695565e-05, "loss": 3.9183, "step": 1745 }, { "epoch": 0.05, "learning_rate": 9.941605608783644e-05, "loss": 3.8446, "step": 1750 }, { "epoch": 0.05, "learning_rate": 9.94125186288648e-05, "loss": 3.7401, "step": 1755 }, { "epoch": 0.05, "learning_rate": 9.940897055080098e-05, "loss": 3.9059, "step": 1760 }, { "epoch": 0.05, "learning_rate": 9.940541185440746e-05, "loss": 3.8122, "step": 1765 }, { "epoch": 0.05, "learning_rate": 9.940184254044904e-05, "loss": 3.924, "step": 1770 }, { "epoch": 0.05, "learning_rate": 9.939826260969282e-05, "loss": 3.7909, "step": 1775 }, { "epoch": 0.05, "learning_rate": 9.93946720629081e-05, "loss": 3.89, "step": 1780 }, { "epoch": 0.05, "learning_rate": 9.939107090086654e-05, "loss": 3.8394, "step": 1785 }, { "epoch": 0.05, "learning_rate": 9.938745912434208e-05, "loss": 3.7444, "step": 1790 }, { "epoch": 0.05, "learning_rate": 9.938383673411087e-05, "loss": 3.9146, "step": 1795 }, { "epoch": 0.05, "learning_rate": 9.938020373095141e-05, "loss": 3.8807, "step": 1800 }, { "epoch": 0.05, "learning_rate": 9.937656011564445e-05, "loss": 3.833, "step": 1805 }, { "epoch": 0.05, "learning_rate": 9.937290588897302e-05, "loss": 3.7605, "step": 1810 }, { "epoch": 0.05, "learning_rate": 9.936924105172248e-05, "loss": 3.8064, "step": 1815 }, { "epoch": 0.05, "learning_rate": 9.936556560468037e-05, "loss": 3.9446, "step": 1820 }, { "epoch": 0.05, "learning_rate": 9.93618795486366e-05, "loss": 3.8674, "step": 1825 }, { "epoch": 0.05, "learning_rate": 9.935818288438334e-05, "loss": 3.8104, "step": 1830 }, { "epoch": 0.05, "learning_rate": 9.935447561271501e-05, "loss": 3.7519, "step": 1835 }, { "epoch": 0.05, "learning_rate": 9.935075773442835e-05, "loss": 3.8585, "step": 1840 }, { "epoch": 0.05, "learning_rate": 9.934702925032232e-05, "loss": 3.8825, "step": 1845 }, { "epoch": 0.05, "learning_rate": 9.934329016119824e-05, "loss": 3.8894, "step": 1850 }, { "epoch": 0.05, "learning_rate": 9.933954046785963e-05, "loss": 3.7938, "step": 1855 }, { "epoch": 0.05, "learning_rate": 9.933578017111233e-05, "loss": 3.655, "step": 1860 }, { "epoch": 0.05, "learning_rate": 9.933200927176449e-05, "loss": 3.8744, "step": 1865 }, { "epoch": 0.06, "learning_rate": 9.932822777062646e-05, "loss": 3.7506, "step": 1870 }, { "epoch": 0.06, "learning_rate": 9.932443566851092e-05, "loss": 3.8658, "step": 1875 }, { "epoch": 0.06, "learning_rate": 9.932063296623284e-05, "loss": 3.8262, "step": 1880 }, { "epoch": 0.06, "learning_rate": 9.931681966460944e-05, "loss": 3.9689, "step": 1885 }, { "epoch": 0.06, "learning_rate": 9.93129957644602e-05, "loss": 3.6744, "step": 1890 }, { "epoch": 0.06, "learning_rate": 9.930916126660691e-05, "loss": 3.8299, "step": 1895 }, { "epoch": 0.06, "learning_rate": 9.930531617187367e-05, "loss": 3.8617, "step": 1900 }, { "epoch": 0.06, "learning_rate": 9.930146048108678e-05, "loss": 3.8257, "step": 1905 }, { "epoch": 0.06, "learning_rate": 9.929759419507487e-05, "loss": 3.8561, "step": 1910 }, { "epoch": 0.06, "learning_rate": 9.92937173146688e-05, "loss": 3.7819, "step": 1915 }, { "epoch": 0.06, "learning_rate": 9.928982984070179e-05, "loss": 3.7196, "step": 1920 }, { "epoch": 0.06, "learning_rate": 9.928593177400925e-05, "loss": 3.7656, "step": 1925 }, { "epoch": 0.06, "learning_rate": 9.928202311542889e-05, "loss": 3.8188, "step": 1930 }, { "epoch": 0.06, "learning_rate": 9.927810386580075e-05, "loss": 3.813, "step": 1935 }, { "epoch": 0.06, "learning_rate": 9.927417402596707e-05, "loss": 3.8146, "step": 1940 }, { "epoch": 0.06, "learning_rate": 9.927023359677241e-05, "loss": 3.8007, "step": 1945 }, { "epoch": 0.06, "learning_rate": 9.92662825790636e-05, "loss": 3.7574, "step": 1950 }, { "epoch": 0.06, "learning_rate": 9.926232097368974e-05, "loss": 3.8817, "step": 1955 }, { "epoch": 0.06, "learning_rate": 9.925834878150219e-05, "loss": 3.8091, "step": 1960 }, { "epoch": 0.06, "learning_rate": 9.925436600335464e-05, "loss": 3.8492, "step": 1965 }, { "epoch": 0.06, "learning_rate": 9.925037264010298e-05, "loss": 3.7997, "step": 1970 }, { "epoch": 0.06, "learning_rate": 9.924636869260542e-05, "loss": 3.7635, "step": 1975 }, { "epoch": 0.06, "learning_rate": 9.924235416172243e-05, "loss": 3.8286, "step": 1980 }, { "epoch": 0.06, "learning_rate": 9.923832904831679e-05, "loss": 3.767, "step": 1985 }, { "epoch": 0.06, "learning_rate": 9.923429335325349e-05, "loss": 3.728, "step": 1990 }, { "epoch": 0.06, "learning_rate": 9.923024707739985e-05, "loss": 3.787, "step": 1995 }, { "epoch": 0.06, "learning_rate": 9.922619022162545e-05, "loss": 3.7583, "step": 2000 }, { "epoch": 0.06, "learning_rate": 9.92221227868021e-05, "loss": 3.7524, "step": 2005 }, { "epoch": 0.06, "learning_rate": 9.921804477380394e-05, "loss": 3.7763, "step": 2010 }, { "epoch": 0.06, "learning_rate": 9.921395618350739e-05, "loss": 3.7679, "step": 2015 }, { "epoch": 0.06, "learning_rate": 9.920985701679106e-05, "loss": 3.6609, "step": 2020 }, { "epoch": 0.06, "learning_rate": 9.920574727453595e-05, "loss": 3.6948, "step": 2025 }, { "epoch": 0.06, "learning_rate": 9.920162695762522e-05, "loss": 3.6551, "step": 2030 }, { "epoch": 0.06, "learning_rate": 9.919749606694439e-05, "loss": 3.8844, "step": 2035 }, { "epoch": 0.06, "learning_rate": 9.919335460338122e-05, "loss": 3.7572, "step": 2040 }, { "epoch": 0.06, "learning_rate": 9.918920256782572e-05, "loss": 3.7896, "step": 2045 }, { "epoch": 0.06, "learning_rate": 9.918503996117016e-05, "loss": 3.8745, "step": 2050 }, { "epoch": 0.06, "learning_rate": 9.918086678430917e-05, "loss": 3.7582, "step": 2055 }, { "epoch": 0.06, "learning_rate": 9.917668303813959e-05, "loss": 3.7299, "step": 2060 }, { "epoch": 0.06, "learning_rate": 9.917248872356051e-05, "loss": 3.7707, "step": 2065 }, { "epoch": 0.06, "learning_rate": 9.916828384147331e-05, "loss": 3.7567, "step": 2070 }, { "epoch": 0.06, "learning_rate": 9.916406839278167e-05, "loss": 3.8481, "step": 2075 }, { "epoch": 0.06, "learning_rate": 9.915984237839152e-05, "loss": 3.7821, "step": 2080 }, { "epoch": 0.06, "learning_rate": 9.915560579921106e-05, "loss": 3.7699, "step": 2085 }, { "epoch": 0.06, "learning_rate": 9.915135865615074e-05, "loss": 3.8605, "step": 2090 }, { "epoch": 0.06, "learning_rate": 9.914710095012332e-05, "loss": 3.8208, "step": 2095 }, { "epoch": 0.06, "learning_rate": 9.91428326820438e-05, "loss": 3.6304, "step": 2100 }, { "epoch": 0.06, "learning_rate": 9.913855385282947e-05, "loss": 3.7525, "step": 2105 }, { "epoch": 0.06, "learning_rate": 9.913426446339986e-05, "loss": 3.7882, "step": 2110 }, { "epoch": 0.06, "learning_rate": 9.912996451467682e-05, "loss": 3.8085, "step": 2115 }, { "epoch": 0.06, "learning_rate": 9.912565400758443e-05, "loss": 3.7663, "step": 2120 }, { "epoch": 0.06, "learning_rate": 9.912133294304901e-05, "loss": 3.7437, "step": 2125 }, { "epoch": 0.06, "learning_rate": 9.911700132199924e-05, "loss": 3.6844, "step": 2130 }, { "epoch": 0.06, "learning_rate": 9.911265914536598e-05, "loss": 3.7061, "step": 2135 }, { "epoch": 0.06, "learning_rate": 9.910830641408242e-05, "loss": 3.8178, "step": 2140 }, { "epoch": 0.06, "learning_rate": 9.910394312908397e-05, "loss": 3.6907, "step": 2145 }, { "epoch": 0.06, "learning_rate": 9.909956929130834e-05, "loss": 3.6389, "step": 2150 }, { "epoch": 0.06, "learning_rate": 9.90951849016955e-05, "loss": 3.7524, "step": 2155 }, { "epoch": 0.06, "learning_rate": 9.909078996118768e-05, "loss": 3.8128, "step": 2160 }, { "epoch": 0.06, "learning_rate": 9.908638447072939e-05, "loss": 3.7007, "step": 2165 }, { "epoch": 0.06, "learning_rate": 9.90819684312674e-05, "loss": 3.6257, "step": 2170 }, { "epoch": 0.06, "learning_rate": 9.907754184375074e-05, "loss": 3.785, "step": 2175 }, { "epoch": 0.06, "learning_rate": 9.90731047091307e-05, "loss": 3.7489, "step": 2180 }, { "epoch": 0.06, "learning_rate": 9.906865702836088e-05, "loss": 3.7875, "step": 2185 }, { "epoch": 0.06, "learning_rate": 9.90641988023971e-05, "loss": 3.7717, "step": 2190 }, { "epoch": 0.06, "learning_rate": 9.905973003219748e-05, "loss": 3.6764, "step": 2195 }, { "epoch": 0.06, "learning_rate": 9.905525071872237e-05, "loss": 3.7371, "step": 2200 }, { "epoch": 0.06, "learning_rate": 9.905076086293443e-05, "loss": 3.7509, "step": 2205 }, { "epoch": 0.07, "learning_rate": 9.904626046579852e-05, "loss": 3.782, "step": 2210 }, { "epoch": 0.07, "learning_rate": 9.904174952828186e-05, "loss": 3.7391, "step": 2215 }, { "epoch": 0.07, "learning_rate": 9.903722805135385e-05, "loss": 3.7291, "step": 2220 }, { "epoch": 0.07, "learning_rate": 9.903269603598617e-05, "loss": 3.6544, "step": 2225 }, { "epoch": 0.07, "learning_rate": 9.902815348315282e-05, "loss": 3.7834, "step": 2230 }, { "epoch": 0.07, "learning_rate": 9.902360039383e-05, "loss": 3.7681, "step": 2235 }, { "epoch": 0.07, "learning_rate": 9.901903676899621e-05, "loss": 3.7269, "step": 2240 }, { "epoch": 0.07, "learning_rate": 9.901446260963221e-05, "loss": 3.8613, "step": 2245 }, { "epoch": 0.07, "learning_rate": 9.9009877916721e-05, "loss": 3.5739, "step": 2250 }, { "epoch": 0.07, "learning_rate": 9.900528269124787e-05, "loss": 3.7783, "step": 2255 }, { "epoch": 0.07, "learning_rate": 9.900067693420039e-05, "loss": 3.8964, "step": 2260 }, { "epoch": 0.07, "learning_rate": 9.899606064656832e-05, "loss": 3.6615, "step": 2265 }, { "epoch": 0.07, "learning_rate": 9.899143382934379e-05, "loss": 3.7902, "step": 2270 }, { "epoch": 0.07, "learning_rate": 9.898679648352108e-05, "loss": 3.7225, "step": 2275 }, { "epoch": 0.07, "learning_rate": 9.898214861009683e-05, "loss": 3.7407, "step": 2280 }, { "epoch": 0.07, "learning_rate": 9.897749021006988e-05, "loss": 3.805, "step": 2285 }, { "epoch": 0.07, "learning_rate": 9.897282128444135e-05, "loss": 3.7967, "step": 2290 }, { "epoch": 0.07, "learning_rate": 9.896814183421464e-05, "loss": 3.6607, "step": 2295 }, { "epoch": 0.07, "learning_rate": 9.896345186039538e-05, "loss": 3.7044, "step": 2300 }, { "epoch": 0.07, "learning_rate": 9.895875136399149e-05, "loss": 3.6989, "step": 2305 }, { "epoch": 0.07, "learning_rate": 9.895404034601313e-05, "loss": 3.6222, "step": 2310 }, { "epoch": 0.07, "learning_rate": 9.894931880747273e-05, "loss": 3.6806, "step": 2315 }, { "epoch": 0.07, "learning_rate": 9.894458674938499e-05, "loss": 3.6711, "step": 2320 }, { "epoch": 0.07, "learning_rate": 9.893984417276686e-05, "loss": 3.5707, "step": 2325 }, { "epoch": 0.07, "learning_rate": 9.893509107863754e-05, "loss": 3.6347, "step": 2330 }, { "epoch": 0.07, "learning_rate": 9.893032746801854e-05, "loss": 3.7042, "step": 2335 }, { "epoch": 0.07, "learning_rate": 9.892555334193353e-05, "loss": 3.6306, "step": 2340 }, { "epoch": 0.07, "learning_rate": 9.892076870140857e-05, "loss": 3.7553, "step": 2345 }, { "epoch": 0.07, "learning_rate": 9.891597354747186e-05, "loss": 3.6957, "step": 2350 }, { "epoch": 0.07, "learning_rate": 9.891116788115393e-05, "loss": 3.7179, "step": 2355 }, { "epoch": 0.07, "learning_rate": 9.890635170348756e-05, "loss": 3.6699, "step": 2360 }, { "epoch": 0.07, "learning_rate": 9.890152501550777e-05, "loss": 3.6383, "step": 2365 }, { "epoch": 0.07, "learning_rate": 9.889668781825185e-05, "loss": 3.7099, "step": 2370 }, { "epoch": 0.07, "learning_rate": 9.889184011275936e-05, "loss": 3.8588, "step": 2375 }, { "epoch": 0.07, "learning_rate": 9.888698190007209e-05, "loss": 3.6491, "step": 2380 }, { "epoch": 0.07, "learning_rate": 9.88821131812341e-05, "loss": 3.6457, "step": 2385 }, { "epoch": 0.07, "learning_rate": 9.887723395729174e-05, "loss": 3.7847, "step": 2390 }, { "epoch": 0.07, "learning_rate": 9.887234422929355e-05, "loss": 3.6919, "step": 2395 }, { "epoch": 0.07, "learning_rate": 9.886744399829039e-05, "loss": 3.6915, "step": 2400 }, { "epoch": 0.07, "learning_rate": 9.886253326533535e-05, "loss": 3.7435, "step": 2405 }, { "epoch": 0.07, "learning_rate": 9.885761203148378e-05, "loss": 3.766, "step": 2410 }, { "epoch": 0.07, "learning_rate": 9.885268029779327e-05, "loss": 3.6636, "step": 2415 }, { "epoch": 0.07, "learning_rate": 9.884773806532373e-05, "loss": 3.7015, "step": 2420 }, { "epoch": 0.07, "learning_rate": 9.884278533513723e-05, "loss": 3.7592, "step": 2425 }, { "epoch": 0.07, "learning_rate": 9.883782210829816e-05, "loss": 3.6633, "step": 2430 }, { "epoch": 0.07, "learning_rate": 9.883284838587317e-05, "loss": 3.6871, "step": 2435 }, { "epoch": 0.07, "learning_rate": 9.882786416893112e-05, "loss": 3.6322, "step": 2440 }, { "epoch": 0.07, "learning_rate": 9.882286945854319e-05, "loss": 3.6646, "step": 2445 }, { "epoch": 0.07, "learning_rate": 9.881786425578274e-05, "loss": 3.5715, "step": 2450 }, { "epoch": 0.07, "learning_rate": 9.881284856172545e-05, "loss": 3.6029, "step": 2455 }, { "epoch": 0.07, "learning_rate": 9.880782237744921e-05, "loss": 3.7067, "step": 2460 }, { "epoch": 0.07, "learning_rate": 9.880278570403419e-05, "loss": 3.6391, "step": 2465 }, { "epoch": 0.07, "learning_rate": 9.87977385425628e-05, "loss": 3.615, "step": 2470 }, { "epoch": 0.07, "learning_rate": 9.87926808941197e-05, "loss": 3.7216, "step": 2475 }, { "epoch": 0.07, "learning_rate": 9.878761275979184e-05, "loss": 3.5785, "step": 2480 }, { "epoch": 0.07, "learning_rate": 9.878253414066839e-05, "loss": 3.6584, "step": 2485 }, { "epoch": 0.07, "learning_rate": 9.877744503784076e-05, "loss": 3.6272, "step": 2490 }, { "epoch": 0.07, "learning_rate": 9.877234545240265e-05, "loss": 3.5458, "step": 2495 }, { "epoch": 0.07, "learning_rate": 9.876723538545001e-05, "loss": 3.7147, "step": 2500 }, { "epoch": 0.07, "learning_rate": 9.8762114838081e-05, "loss": 3.6937, "step": 2505 }, { "epoch": 0.07, "learning_rate": 9.875698381139607e-05, "loss": 3.6202, "step": 2510 }, { "epoch": 0.07, "learning_rate": 9.875184230649792e-05, "loss": 3.5451, "step": 2515 }, { "epoch": 0.07, "learning_rate": 9.874669032449149e-05, "loss": 3.6884, "step": 2520 }, { "epoch": 0.07, "learning_rate": 9.874152786648399e-05, "loss": 3.6449, "step": 2525 }, { "epoch": 0.07, "learning_rate": 9.873635493358484e-05, "loss": 3.5987, "step": 2530 }, { "epoch": 0.07, "learning_rate": 9.873117152690576e-05, "loss": 3.6785, "step": 2535 }, { "epoch": 0.07, "learning_rate": 9.872597764756071e-05, "loss": 3.5647, "step": 2540 }, { "epoch": 0.07, "learning_rate": 9.872077329666587e-05, "loss": 3.5217, "step": 2545 }, { "epoch": 0.08, "learning_rate": 9.871555847533968e-05, "loss": 3.6951, "step": 2550 }, { "epoch": 0.08, "learning_rate": 9.871033318470289e-05, "loss": 3.6929, "step": 2555 }, { "epoch": 0.08, "learning_rate": 9.870509742587838e-05, "loss": 3.7907, "step": 2560 }, { "epoch": 0.08, "learning_rate": 9.869985119999143e-05, "loss": 3.6542, "step": 2565 }, { "epoch": 0.08, "learning_rate": 9.869459450816942e-05, "loss": 3.6009, "step": 2570 }, { "epoch": 0.08, "learning_rate": 9.86893273515421e-05, "loss": 3.7162, "step": 2575 }, { "epoch": 0.08, "learning_rate": 9.868404973124139e-05, "loss": 3.6235, "step": 2580 }, { "epoch": 0.08, "learning_rate": 9.86787616484015e-05, "loss": 3.8202, "step": 2585 }, { "epoch": 0.08, "learning_rate": 9.867346310415886e-05, "loss": 3.6709, "step": 2590 }, { "epoch": 0.08, "learning_rate": 9.866815409965219e-05, "loss": 3.5926, "step": 2595 }, { "epoch": 0.08, "learning_rate": 9.86628346360224e-05, "loss": 3.6341, "step": 2600 }, { "epoch": 0.08, "learning_rate": 9.865750471441271e-05, "loss": 3.6455, "step": 2605 }, { "epoch": 0.08, "learning_rate": 9.865216433596853e-05, "loss": 3.6004, "step": 2610 }, { "epoch": 0.08, "learning_rate": 9.864681350183758e-05, "loss": 3.5626, "step": 2615 }, { "epoch": 0.08, "learning_rate": 9.864145221316975e-05, "loss": 3.6064, "step": 2620 }, { "epoch": 0.08, "learning_rate": 9.863608047111725e-05, "loss": 3.6997, "step": 2625 }, { "epoch": 0.08, "learning_rate": 9.863069827683446e-05, "loss": 3.6119, "step": 2630 }, { "epoch": 0.08, "learning_rate": 9.862530563147812e-05, "loss": 3.6654, "step": 2635 }, { "epoch": 0.08, "learning_rate": 9.861990253620709e-05, "loss": 3.7743, "step": 2640 }, { "epoch": 0.08, "learning_rate": 9.861448899218255e-05, "loss": 3.6977, "step": 2645 }, { "epoch": 0.08, "learning_rate": 9.860906500056792e-05, "loss": 3.6086, "step": 2650 }, { "epoch": 0.08, "learning_rate": 9.860363056252883e-05, "loss": 3.7039, "step": 2655 }, { "epoch": 0.08, "learning_rate": 9.859818567923321e-05, "loss": 3.5752, "step": 2660 }, { "epoch": 0.08, "learning_rate": 9.859273035185118e-05, "loss": 3.7488, "step": 2665 }, { "epoch": 0.08, "learning_rate": 9.858726458155512e-05, "loss": 3.6323, "step": 2670 }, { "epoch": 0.08, "learning_rate": 9.858178836951967e-05, "loss": 3.6568, "step": 2675 }, { "epoch": 0.08, "learning_rate": 9.857630171692174e-05, "loss": 3.5942, "step": 2680 }, { "epoch": 0.08, "learning_rate": 9.857080462494039e-05, "loss": 3.6181, "step": 2685 }, { "epoch": 0.08, "learning_rate": 9.856529709475703e-05, "loss": 3.5677, "step": 2690 }, { "epoch": 0.08, "learning_rate": 9.855977912755524e-05, "loss": 3.6621, "step": 2695 }, { "epoch": 0.08, "learning_rate": 9.855425072452086e-05, "loss": 3.5985, "step": 2700 }, { "epoch": 0.08, "learning_rate": 9.854871188684204e-05, "loss": 3.6207, "step": 2705 }, { "epoch": 0.08, "learning_rate": 9.854316261570905e-05, "loss": 3.659, "step": 2710 }, { "epoch": 0.08, "learning_rate": 9.85376029123145e-05, "loss": 3.6571, "step": 2715 }, { "epoch": 0.08, "learning_rate": 9.853203277785318e-05, "loss": 3.596, "step": 2720 }, { "epoch": 0.08, "learning_rate": 9.85264522135222e-05, "loss": 3.4375, "step": 2725 }, { "epoch": 0.08, "learning_rate": 9.852086122052083e-05, "loss": 3.514, "step": 2730 }, { "epoch": 0.08, "learning_rate": 9.85152598000506e-05, "loss": 3.6249, "step": 2735 }, { "epoch": 0.08, "learning_rate": 9.850964795331531e-05, "loss": 3.6201, "step": 2740 }, { "epoch": 0.08, "learning_rate": 9.8504025681521e-05, "loss": 3.658, "step": 2745 }, { "epoch": 0.08, "learning_rate": 9.84983929858759e-05, "loss": 3.6492, "step": 2750 }, { "epoch": 0.08, "learning_rate": 9.849274986759054e-05, "loss": 3.5838, "step": 2755 }, { "epoch": 0.08, "learning_rate": 9.848709632787768e-05, "loss": 3.5243, "step": 2760 }, { "epoch": 0.08, "learning_rate": 9.848143236795225e-05, "loss": 3.5901, "step": 2765 }, { "epoch": 0.08, "learning_rate": 9.847575798903152e-05, "loss": 3.6736, "step": 2770 }, { "epoch": 0.08, "learning_rate": 9.847007319233496e-05, "loss": 3.6411, "step": 2775 }, { "epoch": 0.08, "learning_rate": 9.846437797908424e-05, "loss": 3.5555, "step": 2780 }, { "epoch": 0.08, "learning_rate": 9.845867235050333e-05, "loss": 3.6226, "step": 2785 }, { "epoch": 0.08, "learning_rate": 9.845295630781837e-05, "loss": 3.5828, "step": 2790 }, { "epoch": 0.08, "learning_rate": 9.844722985225783e-05, "loss": 3.6022, "step": 2795 }, { "epoch": 0.08, "learning_rate": 9.844149298505232e-05, "loss": 3.7086, "step": 2800 }, { "epoch": 0.08, "learning_rate": 9.843574570743474e-05, "loss": 3.5993, "step": 2805 }, { "epoch": 0.08, "learning_rate": 9.842998802064025e-05, "loss": 3.7182, "step": 2810 }, { "epoch": 0.08, "learning_rate": 9.842421992590619e-05, "loss": 3.6104, "step": 2815 }, { "epoch": 0.08, "learning_rate": 9.841844142447215e-05, "loss": 3.4157, "step": 2820 }, { "epoch": 0.08, "learning_rate": 9.841265251758e-05, "loss": 3.564, "step": 2825 }, { "epoch": 0.08, "learning_rate": 9.840685320647381e-05, "loss": 3.6557, "step": 2830 }, { "epoch": 0.08, "learning_rate": 9.840104349239988e-05, "loss": 3.7682, "step": 2835 }, { "epoch": 0.08, "learning_rate": 9.839522337660677e-05, "loss": 3.711, "step": 2840 }, { "epoch": 0.08, "learning_rate": 9.838939286034526e-05, "loss": 3.6355, "step": 2845 }, { "epoch": 0.08, "learning_rate": 9.838355194486836e-05, "loss": 3.5413, "step": 2850 }, { "epoch": 0.08, "learning_rate": 9.837770063143133e-05, "loss": 3.6182, "step": 2855 }, { "epoch": 0.08, "learning_rate": 9.837183892129164e-05, "loss": 3.6445, "step": 2860 }, { "epoch": 0.08, "learning_rate": 9.836596681570906e-05, "loss": 3.6774, "step": 2865 }, { "epoch": 0.08, "learning_rate": 9.83600843159455e-05, "loss": 3.5605, "step": 2870 }, { "epoch": 0.08, "learning_rate": 9.835419142326517e-05, "loss": 3.5925, "step": 2875 }, { "epoch": 0.08, "learning_rate": 9.834828813893448e-05, "loss": 3.7148, "step": 2880 }, { "epoch": 0.08, "learning_rate": 9.83423744642221e-05, "loss": 3.5206, "step": 2885 }, { "epoch": 0.09, "learning_rate": 9.833645040039892e-05, "loss": 3.5177, "step": 2890 }, { "epoch": 0.09, "learning_rate": 9.833051594873805e-05, "loss": 3.6005, "step": 2895 }, { "epoch": 0.09, "learning_rate": 9.832457111051485e-05, "loss": 3.5484, "step": 2900 }, { "epoch": 0.09, "learning_rate": 9.831861588700692e-05, "loss": 3.5738, "step": 2905 }, { "epoch": 0.09, "learning_rate": 9.831265027949406e-05, "loss": 3.6355, "step": 2910 }, { "epoch": 0.09, "learning_rate": 9.830667428925833e-05, "loss": 3.5998, "step": 2915 }, { "epoch": 0.09, "learning_rate": 9.8300687917584e-05, "loss": 3.5742, "step": 2920 }, { "epoch": 0.09, "learning_rate": 9.82946911657576e-05, "loss": 3.6207, "step": 2925 }, { "epoch": 0.09, "learning_rate": 9.828868403506788e-05, "loss": 3.5179, "step": 2930 }, { "epoch": 0.09, "learning_rate": 9.828266652680577e-05, "loss": 3.5834, "step": 2935 }, { "epoch": 0.09, "learning_rate": 9.827663864226453e-05, "loss": 3.6086, "step": 2940 }, { "epoch": 0.09, "learning_rate": 9.827060038273956e-05, "loss": 3.63, "step": 2945 }, { "epoch": 0.09, "learning_rate": 9.826455174952852e-05, "loss": 3.5566, "step": 2950 }, { "epoch": 0.09, "learning_rate": 9.825849274393133e-05, "loss": 3.596, "step": 2955 }, { "epoch": 0.09, "learning_rate": 9.825242336725009e-05, "loss": 3.6521, "step": 2960 }, { "epoch": 0.09, "learning_rate": 9.824634362078915e-05, "loss": 3.5593, "step": 2965 }, { "epoch": 0.09, "learning_rate": 9.824025350585511e-05, "loss": 3.5156, "step": 2970 }, { "epoch": 0.09, "learning_rate": 9.823415302375676e-05, "loss": 3.4972, "step": 2975 }, { "epoch": 0.09, "learning_rate": 9.822804217580515e-05, "loss": 3.5705, "step": 2980 }, { "epoch": 0.09, "learning_rate": 9.822192096331351e-05, "loss": 3.5638, "step": 2985 }, { "epoch": 0.09, "learning_rate": 9.82157893875974e-05, "loss": 3.5142, "step": 2990 }, { "epoch": 0.09, "learning_rate": 9.820964744997445e-05, "loss": 3.5986, "step": 2995 }, { "epoch": 0.09, "learning_rate": 9.820349515176466e-05, "loss": 3.4652, "step": 3000 }, { "epoch": 0.09, "learning_rate": 9.819733249429019e-05, "loss": 3.5524, "step": 3005 }, { "epoch": 0.09, "learning_rate": 9.819115947887545e-05, "loss": 3.4913, "step": 3010 }, { "epoch": 0.09, "learning_rate": 9.818497610684705e-05, "loss": 3.5506, "step": 3015 }, { "epoch": 0.09, "learning_rate": 9.817878237953383e-05, "loss": 3.587, "step": 3020 }, { "epoch": 0.09, "learning_rate": 9.817257829826689e-05, "loss": 3.6845, "step": 3025 }, { "epoch": 0.09, "learning_rate": 9.81663638643795e-05, "loss": 3.565, "step": 3030 }, { "epoch": 0.09, "learning_rate": 9.81601390792072e-05, "loss": 3.5565, "step": 3035 }, { "epoch": 0.09, "learning_rate": 9.815390394408776e-05, "loss": 3.5882, "step": 3040 }, { "epoch": 0.09, "learning_rate": 9.814765846036111e-05, "loss": 3.504, "step": 3045 }, { "epoch": 0.09, "learning_rate": 9.81414026293695e-05, "loss": 3.5471, "step": 3050 }, { "epoch": 0.09, "learning_rate": 9.813513645245729e-05, "loss": 3.475, "step": 3055 }, { "epoch": 0.09, "learning_rate": 9.812885993097118e-05, "loss": 3.5046, "step": 3060 }, { "epoch": 0.09, "learning_rate": 9.812257306626002e-05, "loss": 3.4921, "step": 3065 }, { "epoch": 0.09, "learning_rate": 9.811627585967487e-05, "loss": 3.6266, "step": 3070 }, { "epoch": 0.09, "learning_rate": 9.810996831256909e-05, "loss": 3.6082, "step": 3075 }, { "epoch": 0.09, "learning_rate": 9.81036504262982e-05, "loss": 3.61, "step": 3080 }, { "epoch": 0.09, "learning_rate": 9.809732220221995e-05, "loss": 3.5418, "step": 3085 }, { "epoch": 0.09, "learning_rate": 9.80909836416943e-05, "loss": 3.5818, "step": 3090 }, { "epoch": 0.09, "learning_rate": 9.80846347460835e-05, "loss": 3.5288, "step": 3095 }, { "epoch": 0.09, "learning_rate": 9.807827551675192e-05, "loss": 3.6082, "step": 3100 }, { "epoch": 0.09, "learning_rate": 9.807190595506623e-05, "loss": 3.489, "step": 3105 }, { "epoch": 0.09, "learning_rate": 9.80655260623953e-05, "loss": 3.5294, "step": 3110 }, { "epoch": 0.09, "learning_rate": 9.805913584011019e-05, "loss": 3.4579, "step": 3115 }, { "epoch": 0.09, "learning_rate": 9.805273528958423e-05, "loss": 3.5594, "step": 3120 }, { "epoch": 0.09, "learning_rate": 9.804632441219293e-05, "loss": 3.4222, "step": 3125 }, { "epoch": 0.09, "learning_rate": 9.803990320931402e-05, "loss": 3.5012, "step": 3130 }, { "epoch": 0.09, "learning_rate": 9.803347168232746e-05, "loss": 3.5813, "step": 3135 }, { "epoch": 0.09, "learning_rate": 9.802702983261545e-05, "loss": 3.4701, "step": 3140 }, { "epoch": 0.09, "learning_rate": 9.802057766156238e-05, "loss": 3.5323, "step": 3145 }, { "epoch": 0.09, "learning_rate": 9.801411517055488e-05, "loss": 3.5833, "step": 3150 }, { "epoch": 0.09, "learning_rate": 9.800764236098175e-05, "loss": 3.5418, "step": 3155 }, { "epoch": 0.09, "learning_rate": 9.800115923423405e-05, "loss": 3.5789, "step": 3160 }, { "epoch": 0.09, "learning_rate": 9.799466579170508e-05, "loss": 3.4224, "step": 3165 }, { "epoch": 0.09, "learning_rate": 9.798816203479029e-05, "loss": 3.5892, "step": 3170 }, { "epoch": 0.09, "learning_rate": 9.79816479648874e-05, "loss": 3.6522, "step": 3175 }, { "epoch": 0.09, "learning_rate": 9.797512358339633e-05, "loss": 3.5452, "step": 3180 }, { "epoch": 0.09, "learning_rate": 9.796858889171921e-05, "loss": 3.5649, "step": 3185 }, { "epoch": 0.09, "learning_rate": 9.79620438912604e-05, "loss": 3.5465, "step": 3190 }, { "epoch": 0.09, "learning_rate": 9.795548858342645e-05, "loss": 3.4591, "step": 3195 }, { "epoch": 0.09, "learning_rate": 9.794892296962615e-05, "loss": 3.5206, "step": 3200 }, { "epoch": 0.09, "learning_rate": 9.794234705127047e-05, "loss": 3.5793, "step": 3205 }, { "epoch": 0.09, "learning_rate": 9.793576082977267e-05, "loss": 3.5698, "step": 3210 }, { "epoch": 0.09, "learning_rate": 9.792916430654815e-05, "loss": 3.4982, "step": 3215 }, { "epoch": 0.09, "learning_rate": 9.792255748301453e-05, "loss": 3.5577, "step": 3220 }, { "epoch": 0.09, "learning_rate": 9.791594036059168e-05, "loss": 3.5415, "step": 3225 }, { "epoch": 0.1, "learning_rate": 9.790931294070169e-05, "loss": 3.6462, "step": 3230 }, { "epoch": 0.1, "learning_rate": 9.79026752247688e-05, "loss": 3.4927, "step": 3235 }, { "epoch": 0.1, "learning_rate": 9.78960272142195e-05, "loss": 3.4249, "step": 3240 }, { "epoch": 0.1, "learning_rate": 9.788936891048251e-05, "loss": 3.5307, "step": 3245 }, { "epoch": 0.1, "learning_rate": 9.788270031498876e-05, "loss": 3.6074, "step": 3250 }, { "epoch": 0.1, "learning_rate": 9.787602142917135e-05, "loss": 3.4884, "step": 3255 }, { "epoch": 0.1, "learning_rate": 9.786933225446563e-05, "loss": 3.4562, "step": 3260 }, { "epoch": 0.1, "learning_rate": 9.786263279230915e-05, "loss": 3.4785, "step": 3265 }, { "epoch": 0.1, "learning_rate": 9.785592304414167e-05, "loss": 3.4605, "step": 3270 }, { "epoch": 0.1, "learning_rate": 9.784920301140516e-05, "loss": 3.5619, "step": 3275 }, { "epoch": 0.1, "learning_rate": 9.78424726955438e-05, "loss": 3.4984, "step": 3280 }, { "epoch": 0.1, "learning_rate": 9.783573209800398e-05, "loss": 3.5464, "step": 3285 }, { "epoch": 0.1, "learning_rate": 9.782898122023432e-05, "loss": 3.5217, "step": 3290 }, { "epoch": 0.1, "learning_rate": 9.78222200636856e-05, "loss": 3.4728, "step": 3295 }, { "epoch": 0.1, "learning_rate": 9.781544862981087e-05, "loss": 3.4702, "step": 3300 }, { "epoch": 0.1, "learning_rate": 9.780866692006534e-05, "loss": 3.5615, "step": 3305 }, { "epoch": 0.1, "learning_rate": 9.780187493590645e-05, "loss": 3.434, "step": 3310 }, { "epoch": 0.1, "learning_rate": 9.779507267879385e-05, "loss": 3.4237, "step": 3315 }, { "epoch": 0.1, "learning_rate": 9.778826015018936e-05, "loss": 3.4534, "step": 3320 }, { "epoch": 0.1, "learning_rate": 9.778143735155709e-05, "loss": 3.5577, "step": 3325 }, { "epoch": 0.1, "learning_rate": 9.777460428436327e-05, "loss": 3.6206, "step": 3330 }, { "epoch": 0.1, "learning_rate": 9.776776095007638e-05, "loss": 3.5863, "step": 3335 }, { "epoch": 0.1, "learning_rate": 9.776090735016711e-05, "loss": 3.49, "step": 3340 }, { "epoch": 0.1, "learning_rate": 9.775404348610835e-05, "loss": 3.4714, "step": 3345 }, { "epoch": 0.1, "learning_rate": 9.774716935937518e-05, "loss": 3.4186, "step": 3350 }, { "epoch": 0.1, "learning_rate": 9.77402849714449e-05, "loss": 3.4826, "step": 3355 }, { "epoch": 0.1, "learning_rate": 9.773339032379701e-05, "loss": 3.4647, "step": 3360 }, { "epoch": 0.1, "learning_rate": 9.772648541791323e-05, "loss": 3.5205, "step": 3365 }, { "epoch": 0.1, "learning_rate": 9.771957025527745e-05, "loss": 3.5438, "step": 3370 }, { "epoch": 0.1, "learning_rate": 9.771264483737581e-05, "loss": 3.5663, "step": 3375 }, { "epoch": 0.1, "learning_rate": 9.770570916569661e-05, "loss": 3.6438, "step": 3380 }, { "epoch": 0.1, "learning_rate": 9.76987632417304e-05, "loss": 3.5451, "step": 3385 }, { "epoch": 0.1, "learning_rate": 9.769180706696988e-05, "loss": 3.4981, "step": 3390 }, { "epoch": 0.1, "learning_rate": 9.768484064290997e-05, "loss": 3.43, "step": 3395 }, { "epoch": 0.1, "learning_rate": 9.767786397104786e-05, "loss": 3.3975, "step": 3400 }, { "epoch": 0.1, "learning_rate": 9.767087705288283e-05, "loss": 3.5235, "step": 3405 }, { "epoch": 0.1, "learning_rate": 9.766387988991642e-05, "loss": 3.484, "step": 3410 }, { "epoch": 0.1, "learning_rate": 9.765687248365241e-05, "loss": 3.6093, "step": 3415 }, { "epoch": 0.1, "learning_rate": 9.76498548355967e-05, "loss": 3.5762, "step": 3420 }, { "epoch": 0.1, "learning_rate": 9.764282694725744e-05, "loss": 3.4666, "step": 3425 }, { "epoch": 0.1, "learning_rate": 9.7635788820145e-05, "loss": 3.4313, "step": 3430 }, { "epoch": 0.1, "learning_rate": 9.762874045577189e-05, "loss": 3.4088, "step": 3435 }, { "epoch": 0.1, "learning_rate": 9.762168185565284e-05, "loss": 3.4477, "step": 3440 }, { "epoch": 0.1, "learning_rate": 9.761461302130484e-05, "loss": 3.526, "step": 3445 }, { "epoch": 0.1, "learning_rate": 9.760753395424699e-05, "loss": 3.5055, "step": 3450 }, { "epoch": 0.1, "learning_rate": 9.760044465600065e-05, "loss": 3.4135, "step": 3455 }, { "epoch": 0.1, "learning_rate": 9.759334512808935e-05, "loss": 3.4834, "step": 3460 }, { "epoch": 0.1, "learning_rate": 9.758623537203883e-05, "loss": 3.5054, "step": 3465 }, { "epoch": 0.1, "learning_rate": 9.757911538937706e-05, "loss": 3.4846, "step": 3470 }, { "epoch": 0.1, "learning_rate": 9.757198518163411e-05, "loss": 3.3804, "step": 3475 }, { "epoch": 0.1, "learning_rate": 9.756484475034235e-05, "loss": 3.4798, "step": 3480 }, { "epoch": 0.1, "learning_rate": 9.75576940970363e-05, "loss": 3.4752, "step": 3485 }, { "epoch": 0.1, "learning_rate": 9.755053322325269e-05, "loss": 3.5148, "step": 3490 }, { "epoch": 0.1, "learning_rate": 9.754336213053043e-05, "loss": 3.503, "step": 3495 }, { "epoch": 0.1, "learning_rate": 9.753618082041066e-05, "loss": 3.4043, "step": 3500 }, { "epoch": 0.1, "learning_rate": 9.752898929443666e-05, "loss": 3.6043, "step": 3505 }, { "epoch": 0.1, "learning_rate": 9.752178755415396e-05, "loss": 3.463, "step": 3510 }, { "epoch": 0.1, "learning_rate": 9.751457560111026e-05, "loss": 3.548, "step": 3515 }, { "epoch": 0.1, "learning_rate": 9.750735343685547e-05, "loss": 3.5364, "step": 3520 }, { "epoch": 0.1, "learning_rate": 9.750012106294167e-05, "loss": 3.4854, "step": 3525 }, { "epoch": 0.1, "learning_rate": 9.749287848092315e-05, "loss": 3.4073, "step": 3530 }, { "epoch": 0.1, "learning_rate": 9.748562569235637e-05, "loss": 3.3942, "step": 3535 }, { "epoch": 0.1, "learning_rate": 9.747836269880003e-05, "loss": 3.421, "step": 3540 }, { "epoch": 0.1, "learning_rate": 9.7471089501815e-05, "loss": 3.4606, "step": 3545 }, { "epoch": 0.1, "learning_rate": 9.746380610296435e-05, "loss": 3.5276, "step": 3550 }, { "epoch": 0.1, "learning_rate": 9.74565125038133e-05, "loss": 3.5111, "step": 3555 }, { "epoch": 0.1, "learning_rate": 9.744920870592931e-05, "loss": 3.4115, "step": 3560 }, { "epoch": 0.1, "learning_rate": 9.7441894710882e-05, "loss": 3.4755, "step": 3565 }, { "epoch": 0.11, "learning_rate": 9.743457052024323e-05, "loss": 3.5168, "step": 3570 }, { "epoch": 0.11, "learning_rate": 9.742723613558702e-05, "loss": 3.4832, "step": 3575 }, { "epoch": 0.11, "learning_rate": 9.741989155848953e-05, "loss": 3.5103, "step": 3580 }, { "epoch": 0.11, "learning_rate": 9.741253679052923e-05, "loss": 3.5214, "step": 3585 }, { "epoch": 0.11, "learning_rate": 9.740517183328666e-05, "loss": 3.4799, "step": 3590 }, { "epoch": 0.11, "learning_rate": 9.739779668834461e-05, "loss": 3.4453, "step": 3595 }, { "epoch": 0.11, "learning_rate": 9.739041135728806e-05, "loss": 3.4108, "step": 3600 }, { "epoch": 0.11, "learning_rate": 9.738301584170417e-05, "loss": 3.3736, "step": 3605 }, { "epoch": 0.11, "learning_rate": 9.737561014318228e-05, "loss": 3.5427, "step": 3610 }, { "epoch": 0.11, "learning_rate": 9.736819426331392e-05, "loss": 3.3774, "step": 3615 }, { "epoch": 0.11, "learning_rate": 9.736076820369283e-05, "loss": 3.5135, "step": 3620 }, { "epoch": 0.11, "learning_rate": 9.735333196591492e-05, "loss": 3.3976, "step": 3625 }, { "epoch": 0.11, "learning_rate": 9.734588555157826e-05, "loss": 3.4378, "step": 3630 }, { "epoch": 0.11, "learning_rate": 9.733842896228318e-05, "loss": 3.4348, "step": 3635 }, { "epoch": 0.11, "learning_rate": 9.733096219963211e-05, "loss": 3.4797, "step": 3640 }, { "epoch": 0.11, "learning_rate": 9.732348526522974e-05, "loss": 3.4273, "step": 3645 }, { "epoch": 0.11, "learning_rate": 9.73159981606829e-05, "loss": 3.4319, "step": 3650 }, { "epoch": 0.11, "learning_rate": 9.730850088760064e-05, "loss": 3.4461, "step": 3655 }, { "epoch": 0.11, "learning_rate": 9.730099344759413e-05, "loss": 3.4054, "step": 3660 }, { "epoch": 0.11, "learning_rate": 9.729347584227684e-05, "loss": 3.4341, "step": 3665 }, { "epoch": 0.11, "learning_rate": 9.72859480732643e-05, "loss": 3.3793, "step": 3670 }, { "epoch": 0.11, "learning_rate": 9.727841014217429e-05, "loss": 3.5353, "step": 3675 }, { "epoch": 0.11, "learning_rate": 9.727086205062676e-05, "loss": 3.4559, "step": 3680 }, { "epoch": 0.11, "learning_rate": 9.72633038002439e-05, "loss": 3.4973, "step": 3685 }, { "epoch": 0.11, "learning_rate": 9.725573539264994e-05, "loss": 3.4418, "step": 3690 }, { "epoch": 0.11, "learning_rate": 9.724815682947146e-05, "loss": 3.4889, "step": 3695 }, { "epoch": 0.11, "learning_rate": 9.72405681123371e-05, "loss": 3.524, "step": 3700 }, { "epoch": 0.11, "learning_rate": 9.723296924287775e-05, "loss": 3.4541, "step": 3705 }, { "epoch": 0.11, "learning_rate": 9.722536022272646e-05, "loss": 3.3191, "step": 3710 }, { "epoch": 0.11, "learning_rate": 9.721774105351845e-05, "loss": 3.4628, "step": 3715 }, { "epoch": 0.11, "learning_rate": 9.721011173689114e-05, "loss": 3.4685, "step": 3720 }, { "epoch": 0.11, "learning_rate": 9.720247227448411e-05, "loss": 3.454, "step": 3725 }, { "epoch": 0.11, "learning_rate": 9.719482266793916e-05, "loss": 3.4463, "step": 3730 }, { "epoch": 0.11, "learning_rate": 9.718716291890022e-05, "loss": 3.4836, "step": 3735 }, { "epoch": 0.11, "learning_rate": 9.717949302901343e-05, "loss": 3.435, "step": 3740 }, { "epoch": 0.11, "learning_rate": 9.71718129999271e-05, "loss": 3.4135, "step": 3745 }, { "epoch": 0.11, "learning_rate": 9.716412283329172e-05, "loss": 3.3847, "step": 3750 }, { "epoch": 0.11, "learning_rate": 9.715642253075998e-05, "loss": 3.5253, "step": 3755 }, { "epoch": 0.11, "learning_rate": 9.714871209398671e-05, "loss": 3.4378, "step": 3760 }, { "epoch": 0.11, "learning_rate": 9.714099152462892e-05, "loss": 3.5142, "step": 3765 }, { "epoch": 0.11, "learning_rate": 9.713326082434584e-05, "loss": 3.4947, "step": 3770 }, { "epoch": 0.11, "learning_rate": 9.712551999479884e-05, "loss": 3.4238, "step": 3775 }, { "epoch": 0.11, "learning_rate": 9.71177690376515e-05, "loss": 3.5428, "step": 3780 }, { "epoch": 0.11, "learning_rate": 9.711000795456951e-05, "loss": 3.4005, "step": 3785 }, { "epoch": 0.11, "learning_rate": 9.710223674722083e-05, "loss": 3.4442, "step": 3790 }, { "epoch": 0.11, "learning_rate": 9.709445541727551e-05, "loss": 3.443, "step": 3795 }, { "epoch": 0.11, "learning_rate": 9.708666396640582e-05, "loss": 3.4811, "step": 3800 }, { "epoch": 0.11, "learning_rate": 9.707886239628621e-05, "loss": 3.3853, "step": 3805 }, { "epoch": 0.11, "learning_rate": 9.707105070859326e-05, "loss": 3.5006, "step": 3810 }, { "epoch": 0.11, "learning_rate": 9.706322890500579e-05, "loss": 3.3997, "step": 3815 }, { "epoch": 0.11, "learning_rate": 9.705539698720476e-05, "loss": 3.4439, "step": 3820 }, { "epoch": 0.11, "learning_rate": 9.704755495687328e-05, "loss": 3.4211, "step": 3825 }, { "epoch": 0.11, "learning_rate": 9.703970281569668e-05, "loss": 3.4226, "step": 3830 }, { "epoch": 0.11, "learning_rate": 9.70318405653624e-05, "loss": 3.3889, "step": 3835 }, { "epoch": 0.11, "learning_rate": 9.702396820756014e-05, "loss": 3.4591, "step": 3840 }, { "epoch": 0.11, "learning_rate": 9.701608574398169e-05, "loss": 3.3168, "step": 3845 }, { "epoch": 0.11, "learning_rate": 9.700819317632107e-05, "loss": 3.4179, "step": 3850 }, { "epoch": 0.11, "learning_rate": 9.700029050627443e-05, "loss": 3.4312, "step": 3855 }, { "epoch": 0.11, "learning_rate": 9.699237773554012e-05, "loss": 3.4045, "step": 3860 }, { "epoch": 0.11, "learning_rate": 9.698445486581864e-05, "loss": 3.4217, "step": 3865 }, { "epoch": 0.11, "learning_rate": 9.697652189881266e-05, "loss": 3.4608, "step": 3870 }, { "epoch": 0.11, "learning_rate": 9.696857883622706e-05, "loss": 3.4561, "step": 3875 }, { "epoch": 0.11, "learning_rate": 9.696062567976885e-05, "loss": 3.3915, "step": 3880 }, { "epoch": 0.11, "learning_rate": 9.695266243114718e-05, "loss": 3.3788, "step": 3885 }, { "epoch": 0.11, "learning_rate": 9.694468909207345e-05, "loss": 3.3768, "step": 3890 }, { "epoch": 0.11, "learning_rate": 9.693670566426119e-05, "loss": 3.3526, "step": 3895 }, { "epoch": 0.11, "learning_rate": 9.692871214942608e-05, "loss": 3.2118, "step": 3900 }, { "epoch": 0.11, "learning_rate": 9.692070854928596e-05, "loss": 3.3762, "step": 3905 }, { "epoch": 0.12, "learning_rate": 9.691269486556088e-05, "loss": 3.3282, "step": 3910 }, { "epoch": 0.12, "learning_rate": 9.690467109997303e-05, "loss": 3.4298, "step": 3915 }, { "epoch": 0.12, "learning_rate": 9.68966372542468e-05, "loss": 3.5684, "step": 3920 }, { "epoch": 0.12, "learning_rate": 9.688859333010866e-05, "loss": 3.4513, "step": 3925 }, { "epoch": 0.12, "learning_rate": 9.688053932928737e-05, "loss": 3.4303, "step": 3930 }, { "epoch": 0.12, "learning_rate": 9.687247525351375e-05, "loss": 3.4452, "step": 3935 }, { "epoch": 0.12, "learning_rate": 9.686440110452084e-05, "loss": 3.4473, "step": 3940 }, { "epoch": 0.12, "learning_rate": 9.685631688404384e-05, "loss": 3.35, "step": 3945 }, { "epoch": 0.12, "learning_rate": 9.684822259382008e-05, "loss": 3.5458, "step": 3950 }, { "epoch": 0.12, "learning_rate": 9.684011823558909e-05, "loss": 3.3833, "step": 3955 }, { "epoch": 0.12, "learning_rate": 9.683200381109254e-05, "loss": 3.454, "step": 3960 }, { "epoch": 0.12, "learning_rate": 9.68238793220743e-05, "loss": 3.4288, "step": 3965 }, { "epoch": 0.12, "learning_rate": 9.681574477028039e-05, "loss": 3.5821, "step": 3970 }, { "epoch": 0.12, "learning_rate": 9.680760015745894e-05, "loss": 3.3348, "step": 3975 }, { "epoch": 0.12, "learning_rate": 9.67994454853603e-05, "loss": 3.5115, "step": 3980 }, { "epoch": 0.12, "learning_rate": 9.679128075573698e-05, "loss": 3.4398, "step": 3985 }, { "epoch": 0.12, "learning_rate": 9.678310597034364e-05, "loss": 3.4904, "step": 3990 }, { "epoch": 0.12, "learning_rate": 9.677492113093706e-05, "loss": 3.3131, "step": 3995 }, { "epoch": 0.12, "learning_rate": 9.676672623927625e-05, "loss": 3.4131, "step": 4000 }, { "epoch": 0.12, "learning_rate": 9.675852129712234e-05, "loss": 3.3852, "step": 4005 }, { "epoch": 0.12, "learning_rate": 9.675030630623864e-05, "loss": 3.3877, "step": 4010 }, { "epoch": 0.12, "learning_rate": 9.67420812683906e-05, "loss": 3.4349, "step": 4015 }, { "epoch": 0.12, "learning_rate": 9.673384618534582e-05, "loss": 3.3508, "step": 4020 }, { "epoch": 0.12, "learning_rate": 9.672560105887409e-05, "loss": 3.4768, "step": 4025 }, { "epoch": 0.12, "learning_rate": 9.671734589074736e-05, "loss": 3.321, "step": 4030 }, { "epoch": 0.12, "learning_rate": 9.670908068273972e-05, "loss": 3.4776, "step": 4035 }, { "epoch": 0.12, "learning_rate": 9.67008054366274e-05, "loss": 3.4067, "step": 4040 }, { "epoch": 0.12, "learning_rate": 9.669252015418882e-05, "loss": 3.5532, "step": 4045 }, { "epoch": 0.12, "learning_rate": 9.668422483720456e-05, "loss": 3.4502, "step": 4050 }, { "epoch": 0.12, "learning_rate": 9.66759194874573e-05, "loss": 3.383, "step": 4055 }, { "epoch": 0.12, "learning_rate": 9.666760410673195e-05, "loss": 3.3702, "step": 4060 }, { "epoch": 0.12, "learning_rate": 9.665927869681553e-05, "loss": 3.4244, "step": 4065 }, { "epoch": 0.12, "learning_rate": 9.665094325949723e-05, "loss": 3.3792, "step": 4070 }, { "epoch": 0.12, "learning_rate": 9.664259779656842e-05, "loss": 3.4093, "step": 4075 }, { "epoch": 0.12, "learning_rate": 9.663424230982254e-05, "loss": 3.3988, "step": 4080 }, { "epoch": 0.12, "learning_rate": 9.66258768010553e-05, "loss": 3.2906, "step": 4085 }, { "epoch": 0.12, "learning_rate": 9.661750127206445e-05, "loss": 3.3418, "step": 4090 }, { "epoch": 0.12, "learning_rate": 9.660911572465e-05, "loss": 3.3516, "step": 4095 }, { "epoch": 0.12, "learning_rate": 9.660072016061406e-05, "loss": 3.4349, "step": 4100 }, { "epoch": 0.12, "learning_rate": 9.659231458176085e-05, "loss": 3.4432, "step": 4105 }, { "epoch": 0.12, "learning_rate": 9.658389898989684e-05, "loss": 3.3825, "step": 4110 }, { "epoch": 0.12, "learning_rate": 9.657547338683055e-05, "loss": 3.2531, "step": 4115 }, { "epoch": 0.12, "learning_rate": 9.656703777437273e-05, "loss": 3.3713, "step": 4120 }, { "epoch": 0.12, "learning_rate": 9.655859215433625e-05, "loss": 3.4513, "step": 4125 }, { "epoch": 0.12, "learning_rate": 9.655013652853612e-05, "loss": 3.405, "step": 4130 }, { "epoch": 0.12, "learning_rate": 9.654167089878952e-05, "loss": 3.313, "step": 4135 }, { "epoch": 0.12, "learning_rate": 9.653319526691577e-05, "loss": 3.3539, "step": 4140 }, { "epoch": 0.12, "learning_rate": 9.652470963473635e-05, "loss": 3.3997, "step": 4145 }, { "epoch": 0.12, "learning_rate": 9.651621400407487e-05, "loss": 3.5007, "step": 4150 }, { "epoch": 0.12, "learning_rate": 9.650770837675712e-05, "loss": 3.3426, "step": 4155 }, { "epoch": 0.12, "learning_rate": 9.649919275461099e-05, "loss": 3.396, "step": 4160 }, { "epoch": 0.12, "learning_rate": 9.649066713946655e-05, "loss": 3.3567, "step": 4165 }, { "epoch": 0.12, "learning_rate": 9.648213153315603e-05, "loss": 3.4276, "step": 4170 }, { "epoch": 0.12, "learning_rate": 9.647358593751378e-05, "loss": 3.3737, "step": 4175 }, { "epoch": 0.12, "learning_rate": 9.646503035437633e-05, "loss": 3.4441, "step": 4180 }, { "epoch": 0.12, "learning_rate": 9.64564647855823e-05, "loss": 3.3679, "step": 4185 }, { "epoch": 0.12, "learning_rate": 9.644788923297253e-05, "loss": 3.3334, "step": 4190 }, { "epoch": 0.12, "learning_rate": 9.64393036983899e-05, "loss": 3.3404, "step": 4195 }, { "epoch": 0.12, "learning_rate": 9.643070818367957e-05, "loss": 3.4324, "step": 4200 }, { "epoch": 0.12, "learning_rate": 9.642210269068874e-05, "loss": 3.2615, "step": 4205 }, { "epoch": 0.12, "learning_rate": 9.641348722126679e-05, "loss": 3.2892, "step": 4210 }, { "epoch": 0.12, "learning_rate": 9.640486177726527e-05, "loss": 3.3699, "step": 4215 }, { "epoch": 0.12, "learning_rate": 9.639622636053781e-05, "loss": 3.4975, "step": 4220 }, { "epoch": 0.12, "learning_rate": 9.638758097294024e-05, "loss": 3.3043, "step": 4225 }, { "epoch": 0.12, "learning_rate": 9.637892561633054e-05, "loss": 3.3647, "step": 4230 }, { "epoch": 0.12, "learning_rate": 9.637026029256874e-05, "loss": 3.3362, "step": 4235 }, { "epoch": 0.12, "learning_rate": 9.636158500351713e-05, "loss": 3.4874, "step": 4240 }, { "epoch": 0.12, "learning_rate": 9.635289975104009e-05, "loss": 3.2991, "step": 4245 }, { "epoch": 0.13, "learning_rate": 9.634420453700413e-05, "loss": 3.4028, "step": 4250 }, { "epoch": 0.13, "learning_rate": 9.633549936327789e-05, "loss": 3.352, "step": 4255 }, { "epoch": 0.13, "learning_rate": 9.63267842317322e-05, "loss": 3.3447, "step": 4260 }, { "epoch": 0.13, "learning_rate": 9.631805914424e-05, "loss": 3.388, "step": 4265 }, { "epoch": 0.13, "learning_rate": 9.630932410267637e-05, "loss": 3.3391, "step": 4270 }, { "epoch": 0.13, "learning_rate": 9.630057910891852e-05, "loss": 3.4093, "step": 4275 }, { "epoch": 0.13, "learning_rate": 9.629182416484582e-05, "loss": 3.3773, "step": 4280 }, { "epoch": 0.13, "learning_rate": 9.628305927233975e-05, "loss": 3.4344, "step": 4285 }, { "epoch": 0.13, "learning_rate": 9.627428443328397e-05, "loss": 3.4174, "step": 4290 }, { "epoch": 0.13, "learning_rate": 9.626549964956424e-05, "loss": 3.3622, "step": 4295 }, { "epoch": 0.13, "learning_rate": 9.625670492306848e-05, "loss": 3.3917, "step": 4300 }, { "epoch": 0.13, "learning_rate": 9.624790025568673e-05, "loss": 3.3958, "step": 4305 }, { "epoch": 0.13, "learning_rate": 9.623908564931117e-05, "loss": 3.3683, "step": 4310 }, { "epoch": 0.13, "learning_rate": 9.623026110583612e-05, "loss": 3.4463, "step": 4315 }, { "epoch": 0.13, "learning_rate": 9.622142662715808e-05, "loss": 3.4756, "step": 4320 }, { "epoch": 0.13, "learning_rate": 9.621258221517556e-05, "loss": 3.4132, "step": 4325 }, { "epoch": 0.13, "learning_rate": 9.620372787178934e-05, "loss": 3.3038, "step": 4330 }, { "epoch": 0.13, "learning_rate": 9.619486359890228e-05, "loss": 3.4097, "step": 4335 }, { "epoch": 0.13, "learning_rate": 9.618598939841934e-05, "loss": 3.4928, "step": 4340 }, { "epoch": 0.13, "learning_rate": 9.617710527224768e-05, "loss": 3.3449, "step": 4345 }, { "epoch": 0.13, "learning_rate": 9.616821122229656e-05, "loss": 3.4373, "step": 4350 }, { "epoch": 0.13, "learning_rate": 9.615930725047732e-05, "loss": 3.3819, "step": 4355 }, { "epoch": 0.13, "learning_rate": 9.615039335870357e-05, "loss": 3.444, "step": 4360 }, { "epoch": 0.13, "learning_rate": 9.614146954889091e-05, "loss": 3.4625, "step": 4365 }, { "epoch": 0.13, "learning_rate": 9.613253582295714e-05, "loss": 3.2717, "step": 4370 }, { "epoch": 0.13, "learning_rate": 9.612359218282216e-05, "loss": 3.3392, "step": 4375 }, { "epoch": 0.13, "learning_rate": 9.611463863040807e-05, "loss": 3.2773, "step": 4380 }, { "epoch": 0.13, "learning_rate": 9.610567516763902e-05, "loss": 3.3818, "step": 4385 }, { "epoch": 0.13, "learning_rate": 9.60967017964413e-05, "loss": 3.4804, "step": 4390 }, { "epoch": 0.13, "learning_rate": 9.608771851874338e-05, "loss": 3.3522, "step": 4395 }, { "epoch": 0.13, "learning_rate": 9.607872533647584e-05, "loss": 3.3653, "step": 4400 }, { "epoch": 0.13, "learning_rate": 9.606972225157134e-05, "loss": 3.4196, "step": 4405 }, { "epoch": 0.13, "learning_rate": 9.606070926596471e-05, "loss": 3.2794, "step": 4410 }, { "epoch": 0.13, "learning_rate": 9.605168638159294e-05, "loss": 3.3001, "step": 4415 }, { "epoch": 0.13, "learning_rate": 9.604265360039507e-05, "loss": 3.3381, "step": 4420 }, { "epoch": 0.13, "learning_rate": 9.603361092431234e-05, "loss": 3.2356, "step": 4425 }, { "epoch": 0.13, "learning_rate": 9.602455835528804e-05, "loss": 3.2953, "step": 4430 }, { "epoch": 0.13, "learning_rate": 9.601549589526767e-05, "loss": 3.3068, "step": 4435 }, { "epoch": 0.13, "learning_rate": 9.60064235461988e-05, "loss": 3.3654, "step": 4440 }, { "epoch": 0.13, "learning_rate": 9.599734131003114e-05, "loss": 3.3661, "step": 4445 }, { "epoch": 0.13, "learning_rate": 9.598824918871654e-05, "loss": 3.4177, "step": 4450 }, { "epoch": 0.13, "learning_rate": 9.597914718420892e-05, "loss": 3.329, "step": 4455 }, { "epoch": 0.13, "learning_rate": 9.597003529846442e-05, "loss": 3.4091, "step": 4460 }, { "epoch": 0.13, "learning_rate": 9.59609135334412e-05, "loss": 3.4259, "step": 4465 }, { "epoch": 0.13, "learning_rate": 9.595178189109963e-05, "loss": 3.4893, "step": 4470 }, { "epoch": 0.13, "learning_rate": 9.594264037340212e-05, "loss": 3.3317, "step": 4475 }, { "epoch": 0.13, "learning_rate": 9.593348898231326e-05, "loss": 3.3579, "step": 4480 }, { "epoch": 0.13, "learning_rate": 9.592432771979978e-05, "loss": 3.3556, "step": 4485 }, { "epoch": 0.13, "learning_rate": 9.591515658783046e-05, "loss": 3.4588, "step": 4490 }, { "epoch": 0.13, "learning_rate": 9.590597558837626e-05, "loss": 3.3479, "step": 4495 }, { "epoch": 0.13, "learning_rate": 9.589678472341023e-05, "loss": 3.3917, "step": 4500 }, { "epoch": 0.13, "learning_rate": 9.588758399490755e-05, "loss": 3.3577, "step": 4505 }, { "epoch": 0.13, "learning_rate": 9.587837340484553e-05, "loss": 3.3464, "step": 4510 }, { "epoch": 0.13, "learning_rate": 9.586915295520358e-05, "loss": 3.4005, "step": 4515 }, { "epoch": 0.13, "learning_rate": 9.585992264796324e-05, "loss": 3.2996, "step": 4520 }, { "epoch": 0.13, "learning_rate": 9.585068248510817e-05, "loss": 3.2989, "step": 4525 }, { "epoch": 0.13, "learning_rate": 9.584143246862414e-05, "loss": 3.3607, "step": 4530 }, { "epoch": 0.13, "learning_rate": 9.583217260049907e-05, "loss": 3.4189, "step": 4535 }, { "epoch": 0.13, "learning_rate": 9.582290288272291e-05, "loss": 3.3947, "step": 4540 }, { "epoch": 0.13, "learning_rate": 9.581362331728784e-05, "loss": 3.3322, "step": 4545 }, { "epoch": 0.13, "learning_rate": 9.580433390618808e-05, "loss": 3.1919, "step": 4550 }, { "epoch": 0.13, "learning_rate": 9.579503465142e-05, "loss": 3.3667, "step": 4555 }, { "epoch": 0.13, "learning_rate": 9.578572555498207e-05, "loss": 3.3281, "step": 4560 }, { "epoch": 0.13, "learning_rate": 9.577640661887488e-05, "loss": 3.4226, "step": 4565 }, { "epoch": 0.13, "learning_rate": 9.576707784510113e-05, "loss": 3.271, "step": 4570 }, { "epoch": 0.13, "learning_rate": 9.575773923566563e-05, "loss": 3.2916, "step": 4575 }, { "epoch": 0.13, "learning_rate": 9.574839079257532e-05, "loss": 3.4393, "step": 4580 }, { "epoch": 0.13, "learning_rate": 9.573903251783925e-05, "loss": 3.4136, "step": 4585 }, { "epoch": 0.14, "learning_rate": 9.572966441346858e-05, "loss": 3.2594, "step": 4590 }, { "epoch": 0.14, "learning_rate": 9.572028648147657e-05, "loss": 3.4518, "step": 4595 }, { "epoch": 0.14, "learning_rate": 9.571089872387862e-05, "loss": 3.2994, "step": 4600 }, { "epoch": 0.14, "learning_rate": 9.57015011426922e-05, "loss": 3.2572, "step": 4605 }, { "epoch": 0.14, "learning_rate": 9.569209373993693e-05, "loss": 3.3915, "step": 4610 }, { "epoch": 0.14, "learning_rate": 9.568267651763451e-05, "loss": 3.4086, "step": 4615 }, { "epoch": 0.14, "learning_rate": 9.56732494778088e-05, "loss": 3.4126, "step": 4620 }, { "epoch": 0.14, "learning_rate": 9.56638126224857e-05, "loss": 3.4058, "step": 4625 }, { "epoch": 0.14, "learning_rate": 9.565436595369328e-05, "loss": 3.2934, "step": 4630 }, { "epoch": 0.14, "learning_rate": 9.564490947346168e-05, "loss": 3.4265, "step": 4635 }, { "epoch": 0.14, "learning_rate": 9.563544318382316e-05, "loss": 3.2207, "step": 4640 }, { "epoch": 0.14, "learning_rate": 9.56259670868121e-05, "loss": 3.3307, "step": 4645 }, { "epoch": 0.14, "learning_rate": 9.5616481184465e-05, "loss": 3.3516, "step": 4650 }, { "epoch": 0.14, "learning_rate": 9.560698547882041e-05, "loss": 3.3334, "step": 4655 }, { "epoch": 0.14, "learning_rate": 9.559747997191903e-05, "loss": 3.2952, "step": 4660 }, { "epoch": 0.14, "learning_rate": 9.558796466580366e-05, "loss": 3.3193, "step": 4665 }, { "epoch": 0.14, "learning_rate": 9.557843956251924e-05, "loss": 3.3539, "step": 4670 }, { "epoch": 0.14, "learning_rate": 9.556890466411272e-05, "loss": 3.3381, "step": 4675 }, { "epoch": 0.14, "learning_rate": 9.555935997263328e-05, "loss": 3.2442, "step": 4680 }, { "epoch": 0.14, "learning_rate": 9.554980549013209e-05, "loss": 3.3313, "step": 4685 }, { "epoch": 0.14, "learning_rate": 9.554024121866249e-05, "loss": 3.2815, "step": 4690 }, { "epoch": 0.14, "learning_rate": 9.55306671602799e-05, "loss": 3.3177, "step": 4695 }, { "epoch": 0.14, "learning_rate": 9.552108331704189e-05, "loss": 3.3499, "step": 4700 }, { "epoch": 0.14, "learning_rate": 9.551148969100804e-05, "loss": 3.3807, "step": 4705 }, { "epoch": 0.14, "learning_rate": 9.550188628424012e-05, "loss": 3.3183, "step": 4710 }, { "epoch": 0.14, "learning_rate": 9.549227309880197e-05, "loss": 3.3598, "step": 4715 }, { "epoch": 0.14, "learning_rate": 9.548265013675951e-05, "loss": 3.3721, "step": 4720 }, { "epoch": 0.14, "learning_rate": 9.54730174001808e-05, "loss": 3.3086, "step": 4725 }, { "epoch": 0.14, "learning_rate": 9.546337489113597e-05, "loss": 3.26, "step": 4730 }, { "epoch": 0.14, "learning_rate": 9.545372261169725e-05, "loss": 3.3393, "step": 4735 }, { "epoch": 0.14, "learning_rate": 9.544406056393903e-05, "loss": 3.3281, "step": 4740 }, { "epoch": 0.14, "learning_rate": 9.54343887499377e-05, "loss": 3.3703, "step": 4745 }, { "epoch": 0.14, "learning_rate": 9.542470717177182e-05, "loss": 3.371, "step": 4750 }, { "epoch": 0.14, "learning_rate": 9.541501583152204e-05, "loss": 3.275, "step": 4755 }, { "epoch": 0.14, "learning_rate": 9.540531473127106e-05, "loss": 3.3468, "step": 4760 }, { "epoch": 0.14, "learning_rate": 9.539560387310375e-05, "loss": 3.4358, "step": 4765 }, { "epoch": 0.14, "learning_rate": 9.538588325910702e-05, "loss": 3.338, "step": 4770 }, { "epoch": 0.14, "learning_rate": 9.53761528913699e-05, "loss": 3.3311, "step": 4775 }, { "epoch": 0.14, "learning_rate": 9.536641277198353e-05, "loss": 3.3156, "step": 4780 }, { "epoch": 0.14, "learning_rate": 9.535666290304111e-05, "loss": 3.3695, "step": 4785 }, { "epoch": 0.14, "learning_rate": 9.534690328663796e-05, "loss": 3.2454, "step": 4790 }, { "epoch": 0.14, "learning_rate": 9.533713392487149e-05, "loss": 3.3274, "step": 4795 }, { "epoch": 0.14, "learning_rate": 9.532735481984121e-05, "loss": 3.2908, "step": 4800 }, { "epoch": 0.14, "learning_rate": 9.53175659736487e-05, "loss": 3.3121, "step": 4805 }, { "epoch": 0.14, "learning_rate": 9.530776738839766e-05, "loss": 3.4286, "step": 4810 }, { "epoch": 0.14, "learning_rate": 9.529795906619387e-05, "loss": 3.303, "step": 4815 }, { "epoch": 0.14, "learning_rate": 9.528814100914522e-05, "loss": 3.2957, "step": 4820 }, { "epoch": 0.14, "learning_rate": 9.52783132193617e-05, "loss": 3.3665, "step": 4825 }, { "epoch": 0.14, "learning_rate": 9.52684756989553e-05, "loss": 3.3626, "step": 4830 }, { "epoch": 0.14, "learning_rate": 9.525862845004022e-05, "loss": 3.4983, "step": 4835 }, { "epoch": 0.14, "learning_rate": 9.52487714747327e-05, "loss": 3.3202, "step": 4840 }, { "epoch": 0.14, "learning_rate": 9.523890477515105e-05, "loss": 3.3656, "step": 4845 }, { "epoch": 0.14, "learning_rate": 9.522902835341573e-05, "loss": 3.342, "step": 4850 }, { "epoch": 0.14, "learning_rate": 9.521914221164922e-05, "loss": 3.317, "step": 4855 }, { "epoch": 0.14, "learning_rate": 9.520924635197612e-05, "loss": 3.3173, "step": 4860 }, { "epoch": 0.14, "learning_rate": 9.519934077652312e-05, "loss": 3.3294, "step": 4865 }, { "epoch": 0.14, "learning_rate": 9.518942548741902e-05, "loss": 3.4037, "step": 4870 }, { "epoch": 0.14, "learning_rate": 9.517950048679467e-05, "loss": 3.3612, "step": 4875 }, { "epoch": 0.14, "learning_rate": 9.516956577678302e-05, "loss": 3.3419, "step": 4880 }, { "epoch": 0.14, "learning_rate": 9.51596213595191e-05, "loss": 3.311, "step": 4885 }, { "epoch": 0.14, "learning_rate": 9.514966723714004e-05, "loss": 3.312, "step": 4890 }, { "epoch": 0.14, "learning_rate": 9.513970341178506e-05, "loss": 3.4311, "step": 4895 }, { "epoch": 0.14, "learning_rate": 9.512972988559545e-05, "loss": 3.3428, "step": 4900 }, { "epoch": 0.14, "learning_rate": 9.511974666071458e-05, "loss": 3.2884, "step": 4905 }, { "epoch": 0.14, "learning_rate": 9.510975373928792e-05, "loss": 3.3346, "step": 4910 }, { "epoch": 0.14, "learning_rate": 9.509975112346303e-05, "loss": 3.3547, "step": 4915 }, { "epoch": 0.14, "learning_rate": 9.508973881538952e-05, "loss": 3.2405, "step": 4920 }, { "epoch": 0.14, "learning_rate": 9.507971681721911e-05, "loss": 3.2311, "step": 4925 }, { "epoch": 0.15, "learning_rate": 9.50696851311056e-05, "loss": 3.308, "step": 4930 }, { "epoch": 0.15, "learning_rate": 9.505964375920489e-05, "loss": 3.3177, "step": 4935 }, { "epoch": 0.15, "learning_rate": 9.50495927036749e-05, "loss": 3.2557, "step": 4940 }, { "epoch": 0.15, "learning_rate": 9.503953196667569e-05, "loss": 3.3237, "step": 4945 }, { "epoch": 0.15, "learning_rate": 9.502946155036938e-05, "loss": 3.247, "step": 4950 }, { "epoch": 0.15, "learning_rate": 9.50193814569202e-05, "loss": 3.2742, "step": 4955 }, { "epoch": 0.15, "learning_rate": 9.500929168849439e-05, "loss": 3.2732, "step": 4960 }, { "epoch": 0.15, "learning_rate": 9.499919224726032e-05, "loss": 3.3952, "step": 4965 }, { "epoch": 0.15, "learning_rate": 9.498908313538846e-05, "loss": 3.1348, "step": 4970 }, { "epoch": 0.15, "learning_rate": 9.49789643550513e-05, "loss": 3.3573, "step": 4975 }, { "epoch": 0.15, "learning_rate": 9.496883590842343e-05, "loss": 3.3269, "step": 4980 }, { "epoch": 0.15, "learning_rate": 9.495869779768154e-05, "loss": 3.3636, "step": 4985 }, { "epoch": 0.15, "learning_rate": 9.494855002500438e-05, "loss": 3.3299, "step": 4990 }, { "epoch": 0.15, "learning_rate": 9.493839259257278e-05, "loss": 3.3819, "step": 4995 }, { "epoch": 0.15, "learning_rate": 9.492822550256962e-05, "loss": 3.3019, "step": 5000 }, { "epoch": 0.15, "learning_rate": 9.49180487571799e-05, "loss": 3.2621, "step": 5005 }, { "epoch": 0.15, "learning_rate": 9.490786235859066e-05, "loss": 3.29, "step": 5010 }, { "epoch": 0.15, "learning_rate": 9.489766630899104e-05, "loss": 3.3693, "step": 5015 }, { "epoch": 0.15, "learning_rate": 9.488746061057223e-05, "loss": 3.2379, "step": 5020 }, { "epoch": 0.15, "learning_rate": 9.487724526552753e-05, "loss": 3.3407, "step": 5025 }, { "epoch": 0.15, "learning_rate": 9.486702027605224e-05, "loss": 3.2699, "step": 5030 }, { "epoch": 0.15, "learning_rate": 9.485678564434381e-05, "loss": 3.3146, "step": 5035 }, { "epoch": 0.15, "learning_rate": 9.484654137260176e-05, "loss": 3.2477, "step": 5040 }, { "epoch": 0.15, "learning_rate": 9.483628746302761e-05, "loss": 3.3624, "step": 5045 }, { "epoch": 0.15, "learning_rate": 9.482602391782501e-05, "loss": 3.3015, "step": 5050 }, { "epoch": 0.15, "learning_rate": 9.481575073919968e-05, "loss": 3.3278, "step": 5055 }, { "epoch": 0.15, "learning_rate": 9.480546792935939e-05, "loss": 3.3241, "step": 5060 }, { "epoch": 0.15, "learning_rate": 9.479517549051398e-05, "loss": 3.3348, "step": 5065 }, { "epoch": 0.15, "learning_rate": 9.478487342487536e-05, "loss": 3.3635, "step": 5070 }, { "epoch": 0.15, "learning_rate": 9.477456173465754e-05, "loss": 3.353, "step": 5075 }, { "epoch": 0.15, "learning_rate": 9.476424042207655e-05, "loss": 3.3332, "step": 5080 }, { "epoch": 0.15, "learning_rate": 9.475390948935054e-05, "loss": 3.2202, "step": 5085 }, { "epoch": 0.15, "learning_rate": 9.474356893869967e-05, "loss": 3.3003, "step": 5090 }, { "epoch": 0.15, "learning_rate": 9.47332187723462e-05, "loss": 3.3221, "step": 5095 }, { "epoch": 0.15, "learning_rate": 9.472285899251445e-05, "loss": 3.2182, "step": 5100 }, { "epoch": 0.15, "learning_rate": 9.471248960143085e-05, "loss": 3.2567, "step": 5105 }, { "epoch": 0.15, "learning_rate": 9.47021106013238e-05, "loss": 3.2818, "step": 5110 }, { "epoch": 0.15, "learning_rate": 9.469172199442383e-05, "loss": 3.1988, "step": 5115 }, { "epoch": 0.15, "learning_rate": 9.468132378296355e-05, "loss": 3.3245, "step": 5120 }, { "epoch": 0.15, "learning_rate": 9.467091596917758e-05, "loss": 3.2606, "step": 5125 }, { "epoch": 0.15, "learning_rate": 9.466049855530265e-05, "loss": 3.2542, "step": 5130 }, { "epoch": 0.15, "learning_rate": 9.465007154357752e-05, "loss": 3.2094, "step": 5135 }, { "epoch": 0.15, "learning_rate": 9.463963493624304e-05, "loss": 3.195, "step": 5140 }, { "epoch": 0.15, "learning_rate": 9.462918873554209e-05, "loss": 3.1936, "step": 5145 }, { "epoch": 0.15, "learning_rate": 9.461873294371965e-05, "loss": 3.3237, "step": 5150 }, { "epoch": 0.15, "learning_rate": 9.460826756302275e-05, "loss": 3.3569, "step": 5155 }, { "epoch": 0.15, "learning_rate": 9.459779259570045e-05, "loss": 3.2792, "step": 5160 }, { "epoch": 0.15, "learning_rate": 9.458730804400389e-05, "loss": 3.2756, "step": 5165 }, { "epoch": 0.15, "learning_rate": 9.457681391018629e-05, "loss": 3.341, "step": 5170 }, { "epoch": 0.15, "learning_rate": 9.45663101965029e-05, "loss": 3.3978, "step": 5175 }, { "epoch": 0.15, "learning_rate": 9.455579690521106e-05, "loss": 3.2991, "step": 5180 }, { "epoch": 0.15, "learning_rate": 9.454527403857013e-05, "loss": 3.3523, "step": 5185 }, { "epoch": 0.15, "learning_rate": 9.453474159884155e-05, "loss": 3.2806, "step": 5190 }, { "epoch": 0.15, "learning_rate": 9.452419958828883e-05, "loss": 3.2031, "step": 5195 }, { "epoch": 0.15, "learning_rate": 9.45136480091775e-05, "loss": 3.2939, "step": 5200 }, { "epoch": 0.15, "learning_rate": 9.450308686377516e-05, "loss": 3.397, "step": 5205 }, { "epoch": 0.15, "learning_rate": 9.449251615435151e-05, "loss": 3.1577, "step": 5210 }, { "epoch": 0.15, "learning_rate": 9.448193588317822e-05, "loss": 3.3231, "step": 5215 }, { "epoch": 0.15, "learning_rate": 9.447134605252912e-05, "loss": 3.2686, "step": 5220 }, { "epoch": 0.15, "learning_rate": 9.446074666468e-05, "loss": 3.2036, "step": 5225 }, { "epoch": 0.15, "learning_rate": 9.445013772190874e-05, "loss": 3.3337, "step": 5230 }, { "epoch": 0.15, "learning_rate": 9.443951922649529e-05, "loss": 3.2076, "step": 5235 }, { "epoch": 0.15, "learning_rate": 9.442889118072162e-05, "loss": 3.3085, "step": 5240 }, { "epoch": 0.15, "learning_rate": 9.441825358687181e-05, "loss": 3.2284, "step": 5245 }, { "epoch": 0.15, "learning_rate": 9.440760644723192e-05, "loss": 3.207, "step": 5250 }, { "epoch": 0.15, "learning_rate": 9.439694976409008e-05, "loss": 3.3082, "step": 5255 }, { "epoch": 0.15, "learning_rate": 9.438628353973653e-05, "loss": 3.3179, "step": 5260 }, { "epoch": 0.15, "learning_rate": 9.437560777646348e-05, "loss": 3.2069, "step": 5265 }, { "epoch": 0.16, "learning_rate": 9.436492247656523e-05, "loss": 3.2634, "step": 5270 }, { "epoch": 0.16, "learning_rate": 9.435422764233814e-05, "loss": 3.319, "step": 5275 }, { "epoch": 0.16, "learning_rate": 9.43435232760806e-05, "loss": 3.3166, "step": 5280 }, { "epoch": 0.16, "learning_rate": 9.433280938009304e-05, "loss": 3.1611, "step": 5285 }, { "epoch": 0.16, "learning_rate": 9.432208595667797e-05, "loss": 3.3107, "step": 5290 }, { "epoch": 0.16, "learning_rate": 9.431135300813989e-05, "loss": 3.2306, "step": 5295 }, { "epoch": 0.16, "learning_rate": 9.430061053678542e-05, "loss": 3.2194, "step": 5300 }, { "epoch": 0.16, "learning_rate": 9.42898585449232e-05, "loss": 3.2067, "step": 5305 }, { "epoch": 0.16, "learning_rate": 9.427909703486388e-05, "loss": 3.2945, "step": 5310 }, { "epoch": 0.16, "learning_rate": 9.426832600892019e-05, "loss": 3.2228, "step": 5315 }, { "epoch": 0.16, "learning_rate": 9.425754546940691e-05, "loss": 3.3152, "step": 5320 }, { "epoch": 0.16, "learning_rate": 9.424675541864081e-05, "loss": 3.2153, "step": 5325 }, { "epoch": 0.16, "learning_rate": 9.423595585894082e-05, "loss": 3.2751, "step": 5330 }, { "epoch": 0.16, "learning_rate": 9.422514679262778e-05, "loss": 3.1989, "step": 5335 }, { "epoch": 0.16, "learning_rate": 9.421432822202465e-05, "loss": 3.2175, "step": 5340 }, { "epoch": 0.16, "learning_rate": 9.420350014945643e-05, "loss": 3.2294, "step": 5345 }, { "epoch": 0.16, "learning_rate": 9.419266257725013e-05, "loss": 3.2789, "step": 5350 }, { "epoch": 0.16, "learning_rate": 9.418181550773483e-05, "loss": 3.3754, "step": 5355 }, { "epoch": 0.16, "learning_rate": 9.417095894324163e-05, "loss": 3.1684, "step": 5360 }, { "epoch": 0.16, "learning_rate": 9.41600928861037e-05, "loss": 3.1588, "step": 5365 }, { "epoch": 0.16, "learning_rate": 9.41492173386562e-05, "loss": 3.2185, "step": 5370 }, { "epoch": 0.16, "learning_rate": 9.413833230323639e-05, "loss": 3.1885, "step": 5375 }, { "epoch": 0.16, "learning_rate": 9.412743778218351e-05, "loss": 3.2473, "step": 5380 }, { "epoch": 0.16, "learning_rate": 9.41165337778389e-05, "loss": 3.3871, "step": 5385 }, { "epoch": 0.16, "learning_rate": 9.410562029254589e-05, "loss": 3.234, "step": 5390 }, { "epoch": 0.16, "learning_rate": 9.409469732864986e-05, "loss": 3.2022, "step": 5395 }, { "epoch": 0.16, "learning_rate": 9.408376488849824e-05, "loss": 3.2827, "step": 5400 }, { "epoch": 0.16, "learning_rate": 9.407282297444048e-05, "loss": 3.3055, "step": 5405 }, { "epoch": 0.16, "learning_rate": 9.406187158882808e-05, "loss": 3.2995, "step": 5410 }, { "epoch": 0.16, "learning_rate": 9.405091073401458e-05, "loss": 3.3357, "step": 5415 }, { "epoch": 0.16, "learning_rate": 9.40399404123555e-05, "loss": 3.2916, "step": 5420 }, { "epoch": 0.16, "learning_rate": 9.402896062620848e-05, "loss": 3.2231, "step": 5425 }, { "epoch": 0.16, "learning_rate": 9.401797137793313e-05, "loss": 3.2433, "step": 5430 }, { "epoch": 0.16, "learning_rate": 9.400697266989112e-05, "loss": 3.2179, "step": 5435 }, { "epoch": 0.16, "learning_rate": 9.399596450444617e-05, "loss": 3.2686, "step": 5440 }, { "epoch": 0.16, "learning_rate": 9.398494688396398e-05, "loss": 3.3104, "step": 5445 }, { "epoch": 0.16, "learning_rate": 9.397391981081232e-05, "loss": 3.282, "step": 5450 }, { "epoch": 0.16, "learning_rate": 9.3962883287361e-05, "loss": 3.2298, "step": 5455 }, { "epoch": 0.16, "learning_rate": 9.395183731598183e-05, "loss": 3.3857, "step": 5460 }, { "epoch": 0.16, "learning_rate": 9.394078189904867e-05, "loss": 3.2588, "step": 5465 }, { "epoch": 0.16, "learning_rate": 9.392971703893741e-05, "loss": 3.138, "step": 5470 }, { "epoch": 0.16, "learning_rate": 9.391864273802595e-05, "loss": 3.1927, "step": 5475 }, { "epoch": 0.16, "learning_rate": 9.390755899869425e-05, "loss": 3.2027, "step": 5480 }, { "epoch": 0.16, "learning_rate": 9.38964658233243e-05, "loss": 3.1946, "step": 5485 }, { "epoch": 0.16, "learning_rate": 9.388536321430005e-05, "loss": 3.3785, "step": 5490 }, { "epoch": 0.16, "learning_rate": 9.387425117400757e-05, "loss": 3.2013, "step": 5495 }, { "epoch": 0.16, "learning_rate": 9.386312970483489e-05, "loss": 3.3167, "step": 5500 }, { "epoch": 0.16, "learning_rate": 9.385199880917211e-05, "loss": 3.2372, "step": 5505 }, { "epoch": 0.16, "learning_rate": 9.384085848941132e-05, "loss": 3.1363, "step": 5510 }, { "epoch": 0.16, "learning_rate": 9.382970874794666e-05, "loss": 3.3119, "step": 5515 }, { "epoch": 0.16, "learning_rate": 9.381854958717429e-05, "loss": 3.2506, "step": 5520 }, { "epoch": 0.16, "learning_rate": 9.380738100949238e-05, "loss": 3.2233, "step": 5525 }, { "epoch": 0.16, "learning_rate": 9.379620301730114e-05, "loss": 3.2886, "step": 5530 }, { "epoch": 0.16, "learning_rate": 9.378501561300282e-05, "loss": 3.1359, "step": 5535 }, { "epoch": 0.16, "learning_rate": 9.377381879900162e-05, "loss": 3.2307, "step": 5540 }, { "epoch": 0.16, "learning_rate": 9.376261257770384e-05, "loss": 3.2334, "step": 5545 }, { "epoch": 0.16, "learning_rate": 9.37513969515178e-05, "loss": 3.199, "step": 5550 }, { "epoch": 0.16, "learning_rate": 9.374017192285378e-05, "loss": 3.1897, "step": 5555 }, { "epoch": 0.16, "learning_rate": 9.372893749412413e-05, "loss": 3.1353, "step": 5560 }, { "epoch": 0.16, "learning_rate": 9.371769366774321e-05, "loss": 3.2906, "step": 5565 }, { "epoch": 0.16, "learning_rate": 9.370644044612738e-05, "loss": 3.1799, "step": 5570 }, { "epoch": 0.16, "learning_rate": 9.369517783169505e-05, "loss": 3.3707, "step": 5575 }, { "epoch": 0.16, "learning_rate": 9.368390582686663e-05, "loss": 3.292, "step": 5580 }, { "epoch": 0.16, "learning_rate": 9.367262443406456e-05, "loss": 3.2882, "step": 5585 }, { "epoch": 0.16, "learning_rate": 9.366133365571328e-05, "loss": 3.2441, "step": 5590 }, { "epoch": 0.16, "learning_rate": 9.365003349423924e-05, "loss": 3.207, "step": 5595 }, { "epoch": 0.16, "learning_rate": 9.363872395207097e-05, "loss": 3.3215, "step": 5600 }, { "epoch": 0.16, "learning_rate": 9.362740503163891e-05, "loss": 3.2892, "step": 5605 }, { "epoch": 0.17, "learning_rate": 9.361607673537563e-05, "loss": 3.2035, "step": 5610 }, { "epoch": 0.17, "learning_rate": 9.36047390657156e-05, "loss": 3.3535, "step": 5615 }, { "epoch": 0.17, "learning_rate": 9.359339202509543e-05, "loss": 3.2756, "step": 5620 }, { "epoch": 0.17, "learning_rate": 9.358203561595363e-05, "loss": 3.2523, "step": 5625 }, { "epoch": 0.17, "learning_rate": 9.357066984073079e-05, "loss": 3.2303, "step": 5630 }, { "epoch": 0.17, "learning_rate": 9.355929470186948e-05, "loss": 3.1659, "step": 5635 }, { "epoch": 0.17, "learning_rate": 9.354791020181431e-05, "loss": 3.3589, "step": 5640 }, { "epoch": 0.17, "learning_rate": 9.353651634301188e-05, "loss": 3.261, "step": 5645 }, { "epoch": 0.17, "learning_rate": 9.352511312791083e-05, "loss": 3.0563, "step": 5650 }, { "epoch": 0.17, "learning_rate": 9.351370055896175e-05, "loss": 3.1946, "step": 5655 }, { "epoch": 0.17, "learning_rate": 9.350227863861732e-05, "loss": 3.2893, "step": 5660 }, { "epoch": 0.17, "learning_rate": 9.349084736933215e-05, "loss": 3.2163, "step": 5665 }, { "epoch": 0.17, "learning_rate": 9.347940675356292e-05, "loss": 3.268, "step": 5670 }, { "epoch": 0.17, "learning_rate": 9.346795679376832e-05, "loss": 3.2412, "step": 5675 }, { "epoch": 0.17, "learning_rate": 9.345649749240898e-05, "loss": 3.296, "step": 5680 }, { "epoch": 0.17, "learning_rate": 9.344502885194763e-05, "loss": 3.2986, "step": 5685 }, { "epoch": 0.17, "learning_rate": 9.343355087484894e-05, "loss": 3.2975, "step": 5690 }, { "epoch": 0.17, "learning_rate": 9.342206356357956e-05, "loss": 3.2961, "step": 5695 }, { "epoch": 0.17, "learning_rate": 9.341056692060828e-05, "loss": 3.1958, "step": 5700 }, { "epoch": 0.17, "learning_rate": 9.339906094840575e-05, "loss": 3.1428, "step": 5705 }, { "epoch": 0.17, "learning_rate": 9.33875456494447e-05, "loss": 3.1454, "step": 5710 }, { "epoch": 0.17, "learning_rate": 9.337602102619985e-05, "loss": 3.2243, "step": 5715 }, { "epoch": 0.17, "learning_rate": 9.336448708114789e-05, "loss": 3.2566, "step": 5720 }, { "epoch": 0.17, "learning_rate": 9.33529438167676e-05, "loss": 3.146, "step": 5725 }, { "epoch": 0.17, "learning_rate": 9.334139123553967e-05, "loss": 3.2375, "step": 5730 }, { "epoch": 0.17, "learning_rate": 9.332982933994683e-05, "loss": 3.1254, "step": 5735 }, { "epoch": 0.17, "learning_rate": 9.331825813247381e-05, "loss": 3.3195, "step": 5740 }, { "epoch": 0.17, "learning_rate": 9.330667761560739e-05, "loss": 3.2382, "step": 5745 }, { "epoch": 0.17, "learning_rate": 9.329508779183623e-05, "loss": 3.2424, "step": 5750 }, { "epoch": 0.17, "learning_rate": 9.328348866365111e-05, "loss": 3.2218, "step": 5755 }, { "epoch": 0.17, "learning_rate": 9.327188023354474e-05, "loss": 3.2264, "step": 5760 }, { "epoch": 0.17, "learning_rate": 9.326026250401188e-05, "loss": 3.2483, "step": 5765 }, { "epoch": 0.17, "learning_rate": 9.324863547754924e-05, "loss": 3.296, "step": 5770 }, { "epoch": 0.17, "learning_rate": 9.323699915665555e-05, "loss": 3.2337, "step": 5775 }, { "epoch": 0.17, "learning_rate": 9.322535354383153e-05, "loss": 3.1583, "step": 5780 }, { "epoch": 0.17, "learning_rate": 9.321369864157992e-05, "loss": 3.2086, "step": 5785 }, { "epoch": 0.17, "learning_rate": 9.320203445240543e-05, "loss": 3.2841, "step": 5790 }, { "epoch": 0.17, "learning_rate": 9.319036097881477e-05, "loss": 3.1322, "step": 5795 }, { "epoch": 0.17, "learning_rate": 9.317867822331667e-05, "loss": 3.1885, "step": 5800 }, { "epoch": 0.17, "learning_rate": 9.31669861884218e-05, "loss": 3.3078, "step": 5805 }, { "epoch": 0.17, "learning_rate": 9.315528487664289e-05, "loss": 3.1713, "step": 5810 }, { "epoch": 0.17, "learning_rate": 9.314357429049462e-05, "loss": 3.2282, "step": 5815 }, { "epoch": 0.17, "learning_rate": 9.313185443249367e-05, "loss": 3.2884, "step": 5820 }, { "epoch": 0.17, "learning_rate": 9.312012530515875e-05, "loss": 3.2441, "step": 5825 }, { "epoch": 0.17, "learning_rate": 9.310838691101048e-05, "loss": 3.3607, "step": 5830 }, { "epoch": 0.17, "learning_rate": 9.309663925257156e-05, "loss": 3.2086, "step": 5835 }, { "epoch": 0.17, "learning_rate": 9.308488233236663e-05, "loss": 3.2626, "step": 5840 }, { "epoch": 0.17, "learning_rate": 9.307311615292234e-05, "loss": 3.1298, "step": 5845 }, { "epoch": 0.17, "learning_rate": 9.306134071676734e-05, "loss": 3.0569, "step": 5850 }, { "epoch": 0.17, "learning_rate": 9.30495560264322e-05, "loss": 3.2851, "step": 5855 }, { "epoch": 0.17, "learning_rate": 9.303776208444958e-05, "loss": 3.2771, "step": 5860 }, { "epoch": 0.17, "learning_rate": 9.302595889335405e-05, "loss": 3.2027, "step": 5865 }, { "epoch": 0.17, "learning_rate": 9.301414645568223e-05, "loss": 3.2394, "step": 5870 }, { "epoch": 0.17, "learning_rate": 9.300232477397266e-05, "loss": 3.2169, "step": 5875 }, { "epoch": 0.17, "learning_rate": 9.299049385076592e-05, "loss": 3.1772, "step": 5880 }, { "epoch": 0.17, "learning_rate": 9.297865368860456e-05, "loss": 3.0846, "step": 5885 }, { "epoch": 0.17, "learning_rate": 9.296680429003311e-05, "loss": 3.1612, "step": 5890 }, { "epoch": 0.17, "learning_rate": 9.295494565759808e-05, "loss": 3.1998, "step": 5895 }, { "epoch": 0.17, "learning_rate": 9.294307779384796e-05, "loss": 3.2693, "step": 5900 }, { "epoch": 0.17, "learning_rate": 9.293120070133327e-05, "loss": 3.2011, "step": 5905 }, { "epoch": 0.17, "learning_rate": 9.291931438260646e-05, "loss": 3.2596, "step": 5910 }, { "epoch": 0.17, "learning_rate": 9.290741884022198e-05, "loss": 3.2519, "step": 5915 }, { "epoch": 0.17, "learning_rate": 9.28955140767363e-05, "loss": 3.1639, "step": 5920 }, { "epoch": 0.17, "learning_rate": 9.288360009470776e-05, "loss": 3.2562, "step": 5925 }, { "epoch": 0.17, "learning_rate": 9.287167689669683e-05, "loss": 3.2143, "step": 5930 }, { "epoch": 0.17, "learning_rate": 9.285974448526585e-05, "loss": 3.304, "step": 5935 }, { "epoch": 0.17, "learning_rate": 9.284780286297918e-05, "loss": 3.3079, "step": 5940 }, { "epoch": 0.17, "learning_rate": 9.283585203240317e-05, "loss": 3.1674, "step": 5945 }, { "epoch": 0.18, "learning_rate": 9.282389199610613e-05, "loss": 3.1334, "step": 5950 }, { "epoch": 0.18, "learning_rate": 9.281192275665836e-05, "loss": 3.2737, "step": 5955 }, { "epoch": 0.18, "learning_rate": 9.279994431663212e-05, "loss": 3.1755, "step": 5960 }, { "epoch": 0.18, "learning_rate": 9.278795667860167e-05, "loss": 3.2864, "step": 5965 }, { "epoch": 0.18, "learning_rate": 9.277595984514325e-05, "loss": 3.1439, "step": 5970 }, { "epoch": 0.18, "learning_rate": 9.276395381883502e-05, "loss": 3.2498, "step": 5975 }, { "epoch": 0.18, "learning_rate": 9.275193860225722e-05, "loss": 3.2156, "step": 5980 }, { "epoch": 0.18, "learning_rate": 9.273991419799193e-05, "loss": 3.2667, "step": 5985 }, { "epoch": 0.18, "learning_rate": 9.272788060862334e-05, "loss": 3.2057, "step": 5990 }, { "epoch": 0.18, "learning_rate": 9.271583783673751e-05, "loss": 3.2252, "step": 5995 }, { "epoch": 0.18, "learning_rate": 9.270378588492255e-05, "loss": 3.2112, "step": 6000 } ], "logging_steps": 5, "max_steps": 33984, "num_train_epochs": 1, "save_steps": 1000, "total_flos": 1.7753342881038336e+19, "trial_name": null, "trial_params": null }