diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,11730 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.999769248519345, + "eval_steps": 500, + "global_step": 9750, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 4.9999979235549175e-05, + "loss": 2.5578, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 4.9999916942231176e-05, + "loss": 2.4068, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 4.999978067577844e-05, + "loss": 2.0114, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 4.999957952099119e-05, + "loss": 1.5871, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 4.9999313478391527e-05, + "loss": 1.4931, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 4.9998982548669973e-05, + "loss": 1.2123, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 4.999858673268548e-05, + "loss": 1.1602, + "step": 35 + }, + { + "epoch": 0.01, + "learning_rate": 4.999812603146542e-05, + "loss": 1.25, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 4.999760044620556e-05, + "loss": 1.2903, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 4.9997009978270085e-05, + "loss": 1.0359, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 4.999635462919159e-05, + "loss": 1.1824, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 4.999563440067108e-05, + "loss": 1.0544, + "step": 60 + }, + { + "epoch": 0.02, + "learning_rate": 4.9994849294577925e-05, + "loss": 1.1258, + "step": 65 + }, + { + "epoch": 0.02, + "learning_rate": 4.9993999312949936e-05, + "loss": 1.1289, + "step": 70 + }, + { + "epoch": 0.02, + "learning_rate": 4.999308445799328e-05, + "loss": 1.0978, + "step": 75 + }, + { + "epoch": 0.02, + "learning_rate": 4.99921047320825e-05, + "loss": 1.1779, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 4.999106013776055e-05, + "loss": 1.1674, + "step": 85 + }, + { + "epoch": 0.03, + "learning_rate": 4.9989950677738726e-05, + "loss": 1.0238, + "step": 90 + }, + { + "epoch": 0.03, + "learning_rate": 4.9988776354896694e-05, + "loss": 1.1232, + "step": 95 + }, + { + "epoch": 0.03, + "learning_rate": 4.9987537172282466e-05, + "loss": 1.09, + "step": 100 + }, + { + "epoch": 0.03, + "learning_rate": 4.998623313311242e-05, + "loss": 1.0235, + "step": 105 + }, + { + "epoch": 0.03, + "learning_rate": 4.998486424077127e-05, + "loss": 1.264, + "step": 110 + }, + { + "epoch": 0.04, + "learning_rate": 4.998343049881203e-05, + "loss": 1.0221, + "step": 115 + }, + { + "epoch": 0.04, + "learning_rate": 4.9981931910956086e-05, + "loss": 0.9855, + "step": 120 + }, + { + "epoch": 0.04, + "learning_rate": 4.998036848109309e-05, + "loss": 1.1553, + "step": 125 + }, + { + "epoch": 0.04, + "learning_rate": 4.9978740213281005e-05, + "loss": 1.1147, + "step": 130 + }, + { + "epoch": 0.04, + "learning_rate": 4.997704711174611e-05, + "loss": 1.1541, + "step": 135 + }, + { + "epoch": 0.04, + "learning_rate": 4.9975289180882936e-05, + "loss": 1.2214, + "step": 140 + }, + { + "epoch": 0.04, + "learning_rate": 4.9973466425254286e-05, + "loss": 1.1273, + "step": 145 + }, + { + "epoch": 0.05, + "learning_rate": 4.9971578849591225e-05, + "loss": 0.9362, + "step": 150 + }, + { + "epoch": 0.05, + "learning_rate": 4.996962645879308e-05, + "loss": 0.9906, + "step": 155 + }, + { + "epoch": 0.05, + "learning_rate": 4.996760925792737e-05, + "loss": 1.1289, + "step": 160 + }, + { + "epoch": 0.05, + "learning_rate": 4.996552725222986e-05, + "loss": 1.1588, + "step": 165 + }, + { + "epoch": 0.05, + "learning_rate": 4.996338044710452e-05, + "loss": 1.0884, + "step": 170 + }, + { + "epoch": 0.05, + "learning_rate": 4.996116884812349e-05, + "loss": 1.046, + "step": 175 + }, + { + "epoch": 0.06, + "learning_rate": 4.995889246102711e-05, + "loss": 1.0888, + "step": 180 + }, + { + "epoch": 0.06, + "learning_rate": 4.995655129172386e-05, + "loss": 1.1652, + "step": 185 + }, + { + "epoch": 0.06, + "learning_rate": 4.995414534629039e-05, + "loss": 1.0839, + "step": 190 + }, + { + "epoch": 0.06, + "learning_rate": 4.995167463097146e-05, + "loss": 1.0881, + "step": 195 + }, + { + "epoch": 0.06, + "learning_rate": 4.9949139152179944e-05, + "loss": 1.1025, + "step": 200 + }, + { + "epoch": 0.06, + "learning_rate": 4.994653891649683e-05, + "loss": 1.0929, + "step": 205 + }, + { + "epoch": 0.06, + "learning_rate": 4.994387393067117e-05, + "loss": 1.0747, + "step": 210 + }, + { + "epoch": 0.07, + "learning_rate": 4.994114420162008e-05, + "loss": 1.0198, + "step": 215 + }, + { + "epoch": 0.07, + "learning_rate": 4.993834973642874e-05, + "loss": 1.1541, + "step": 220 + }, + { + "epoch": 0.07, + "learning_rate": 4.993549054235032e-05, + "loss": 1.0534, + "step": 225 + }, + { + "epoch": 0.07, + "learning_rate": 4.993256662680604e-05, + "loss": 1.0273, + "step": 230 + }, + { + "epoch": 0.07, + "learning_rate": 4.992957799738507e-05, + "loss": 1.0515, + "step": 235 + }, + { + "epoch": 0.07, + "learning_rate": 4.992652466184458e-05, + "loss": 1.0648, + "step": 240 + }, + { + "epoch": 0.08, + "learning_rate": 4.9923406628109666e-05, + "loss": 1.0626, + "step": 245 + }, + { + "epoch": 0.08, + "learning_rate": 4.992022390427335e-05, + "loss": 0.9381, + "step": 250 + }, + { + "epoch": 0.08, + "learning_rate": 4.99169764985966e-05, + "loss": 1.0552, + "step": 255 + }, + { + "epoch": 0.08, + "learning_rate": 4.99136644195082e-05, + "loss": 1.0584, + "step": 260 + }, + { + "epoch": 0.08, + "learning_rate": 4.991028767560487e-05, + "loss": 1.0852, + "step": 265 + }, + { + "epoch": 0.08, + "learning_rate": 4.9906846275651125e-05, + "loss": 1.1097, + "step": 270 + }, + { + "epoch": 0.08, + "learning_rate": 4.990334022857932e-05, + "loss": 1.0932, + "step": 275 + }, + { + "epoch": 0.09, + "learning_rate": 4.989976954348958e-05, + "loss": 1.0639, + "step": 280 + }, + { + "epoch": 0.09, + "learning_rate": 4.989613422964983e-05, + "loss": 1.1047, + "step": 285 + }, + { + "epoch": 0.09, + "learning_rate": 4.989243429649573e-05, + "loss": 1.0451, + "step": 290 + }, + { + "epoch": 0.09, + "learning_rate": 4.988866975363067e-05, + "loss": 0.9397, + "step": 295 + }, + { + "epoch": 0.09, + "learning_rate": 4.988484061082572e-05, + "loss": 1.1546, + "step": 300 + }, + { + "epoch": 0.09, + "learning_rate": 4.988094687801964e-05, + "loss": 1.115, + "step": 305 + }, + { + "epoch": 0.1, + "learning_rate": 4.987698856531884e-05, + "loss": 1.021, + "step": 310 + }, + { + "epoch": 0.1, + "learning_rate": 4.987296568299733e-05, + "loss": 0.9338, + "step": 315 + }, + { + "epoch": 0.1, + "learning_rate": 4.986887824149674e-05, + "loss": 0.9197, + "step": 320 + }, + { + "epoch": 0.1, + "learning_rate": 4.9864726251426255e-05, + "loss": 1.0489, + "step": 325 + }, + { + "epoch": 0.1, + "learning_rate": 4.9860509723562573e-05, + "loss": 0.9879, + "step": 330 + }, + { + "epoch": 0.1, + "learning_rate": 4.985622866884996e-05, + "loss": 1.1831, + "step": 335 + }, + { + "epoch": 0.1, + "learning_rate": 4.985188309840012e-05, + "loss": 1.0671, + "step": 340 + }, + { + "epoch": 0.11, + "learning_rate": 4.984747302349222e-05, + "loss": 1.057, + "step": 345 + }, + { + "epoch": 0.11, + "learning_rate": 4.984299845557287e-05, + "loss": 1.1265, + "step": 350 + }, + { + "epoch": 0.11, + "learning_rate": 4.9838459406256045e-05, + "loss": 1.0464, + "step": 355 + }, + { + "epoch": 0.11, + "learning_rate": 4.983385588732312e-05, + "loss": 1.0961, + "step": 360 + }, + { + "epoch": 0.11, + "learning_rate": 4.982918791072278e-05, + "loss": 1.0167, + "step": 365 + }, + { + "epoch": 0.11, + "learning_rate": 4.982445548857102e-05, + "loss": 0.98, + "step": 370 + }, + { + "epoch": 0.12, + "learning_rate": 4.981965863315111e-05, + "loss": 1.0808, + "step": 375 + }, + { + "epoch": 0.12, + "learning_rate": 4.981479735691354e-05, + "loss": 1.1325, + "step": 380 + }, + { + "epoch": 0.12, + "learning_rate": 4.980987167247606e-05, + "loss": 1.0781, + "step": 385 + }, + { + "epoch": 0.12, + "learning_rate": 4.980488159262353e-05, + "loss": 1.0557, + "step": 390 + }, + { + "epoch": 0.12, + "learning_rate": 4.9799827130307993e-05, + "loss": 1.1231, + "step": 395 + }, + { + "epoch": 0.12, + "learning_rate": 4.97947082986486e-05, + "loss": 0.956, + "step": 400 + }, + { + "epoch": 0.12, + "learning_rate": 4.9789525110931545e-05, + "loss": 1.1428, + "step": 405 + }, + { + "epoch": 0.13, + "learning_rate": 4.978427758061009e-05, + "loss": 1.0477, + "step": 410 + }, + { + "epoch": 0.13, + "learning_rate": 4.9778965721304516e-05, + "loss": 1.0781, + "step": 415 + }, + { + "epoch": 0.13, + "learning_rate": 4.977358954680202e-05, + "loss": 1.1137, + "step": 420 + }, + { + "epoch": 0.13, + "learning_rate": 4.976814907105679e-05, + "loss": 1.0836, + "step": 425 + }, + { + "epoch": 0.13, + "learning_rate": 4.9762644308189865e-05, + "loss": 1.1098, + "step": 430 + }, + { + "epoch": 0.13, + "learning_rate": 4.975707527248918e-05, + "loss": 1.1296, + "step": 435 + }, + { + "epoch": 0.14, + "learning_rate": 4.9751441978409484e-05, + "loss": 1.095, + "step": 440 + }, + { + "epoch": 0.14, + "learning_rate": 4.9745744440572296e-05, + "loss": 1.1201, + "step": 445 + }, + { + "epoch": 0.14, + "learning_rate": 4.97399826737659e-05, + "loss": 1.0732, + "step": 450 + }, + { + "epoch": 0.14, + "learning_rate": 4.973415669294529e-05, + "loss": 1.0732, + "step": 455 + }, + { + "epoch": 0.14, + "learning_rate": 4.972826651323211e-05, + "loss": 1.0432, + "step": 460 + }, + { + "epoch": 0.14, + "learning_rate": 4.972231214991468e-05, + "loss": 1.1121, + "step": 465 + }, + { + "epoch": 0.14, + "learning_rate": 4.971629361844785e-05, + "loss": 1.0443, + "step": 470 + }, + { + "epoch": 0.15, + "learning_rate": 4.971021093445308e-05, + "loss": 1.0765, + "step": 475 + }, + { + "epoch": 0.15, + "learning_rate": 4.9704064113718326e-05, + "loss": 1.045, + "step": 480 + }, + { + "epoch": 0.15, + "learning_rate": 4.9697853172197985e-05, + "loss": 1.0619, + "step": 485 + }, + { + "epoch": 0.15, + "learning_rate": 4.9691578126012905e-05, + "loss": 1.0598, + "step": 490 + }, + { + "epoch": 0.15, + "learning_rate": 4.968523899145033e-05, + "loss": 1.1654, + "step": 495 + }, + { + "epoch": 0.15, + "learning_rate": 4.967883578496385e-05, + "loss": 1.1356, + "step": 500 + }, + { + "epoch": 0.16, + "learning_rate": 4.967236852317333e-05, + "loss": 1.0416, + "step": 505 + }, + { + "epoch": 0.16, + "learning_rate": 4.966583722286491e-05, + "loss": 1.0538, + "step": 510 + }, + { + "epoch": 0.16, + "learning_rate": 4.965924190099096e-05, + "loss": 0.9972, + "step": 515 + }, + { + "epoch": 0.16, + "learning_rate": 4.9652582574669995e-05, + "loss": 1.179, + "step": 520 + }, + { + "epoch": 0.16, + "learning_rate": 4.9645859261186675e-05, + "loss": 0.9667, + "step": 525 + }, + { + "epoch": 0.16, + "learning_rate": 4.963907197799174e-05, + "loss": 1.08, + "step": 530 + }, + { + "epoch": 0.16, + "learning_rate": 4.9632220742701965e-05, + "loss": 0.966, + "step": 535 + }, + { + "epoch": 0.17, + "learning_rate": 4.9625305573100115e-05, + "loss": 1.0907, + "step": 540 + }, + { + "epoch": 0.17, + "learning_rate": 4.9618326487134906e-05, + "loss": 1.0568, + "step": 545 + }, + { + "epoch": 0.17, + "learning_rate": 4.961128350292094e-05, + "loss": 1.042, + "step": 550 + }, + { + "epoch": 0.17, + "learning_rate": 4.960417663873867e-05, + "loss": 1.0162, + "step": 555 + }, + { + "epoch": 0.17, + "learning_rate": 4.959700591303438e-05, + "loss": 1.0601, + "step": 560 + }, + { + "epoch": 0.17, + "learning_rate": 4.9589771344420074e-05, + "loss": 1.0976, + "step": 565 + }, + { + "epoch": 0.18, + "learning_rate": 4.95824729516735e-05, + "loss": 1.0848, + "step": 570 + }, + { + "epoch": 0.18, + "learning_rate": 4.957511075373803e-05, + "loss": 1.0207, + "step": 575 + }, + { + "epoch": 0.18, + "learning_rate": 4.9567684769722664e-05, + "loss": 1.0877, + "step": 580 + }, + { + "epoch": 0.18, + "learning_rate": 4.956019501890198e-05, + "loss": 1.0506, + "step": 585 + }, + { + "epoch": 0.18, + "learning_rate": 4.9552641520716014e-05, + "loss": 1.0129, + "step": 590 + }, + { + "epoch": 0.18, + "learning_rate": 4.954502429477032e-05, + "loss": 0.9247, + "step": 595 + }, + { + "epoch": 0.18, + "learning_rate": 4.953734336083583e-05, + "loss": 1.0508, + "step": 600 + }, + { + "epoch": 0.19, + "learning_rate": 4.9529598738848824e-05, + "loss": 1.0406, + "step": 605 + }, + { + "epoch": 0.19, + "learning_rate": 4.9521790448910925e-05, + "loss": 1.1159, + "step": 610 + }, + { + "epoch": 0.19, + "learning_rate": 4.951391851128898e-05, + "loss": 1.2021, + "step": 615 + }, + { + "epoch": 0.19, + "learning_rate": 4.950598294641504e-05, + "loss": 1.1157, + "step": 620 + }, + { + "epoch": 0.19, + "learning_rate": 4.949798377488631e-05, + "loss": 1.0757, + "step": 625 + }, + { + "epoch": 0.19, + "learning_rate": 4.94899210174651e-05, + "loss": 1.1126, + "step": 630 + }, + { + "epoch": 0.2, + "learning_rate": 4.948179469507874e-05, + "loss": 1.0995, + "step": 635 + }, + { + "epoch": 0.2, + "learning_rate": 4.947360482881956e-05, + "loss": 1.2224, + "step": 640 + }, + { + "epoch": 0.2, + "learning_rate": 4.9465351439944815e-05, + "loss": 1.1432, + "step": 645 + }, + { + "epoch": 0.2, + "learning_rate": 4.9457034549876636e-05, + "loss": 1.0593, + "step": 650 + }, + { + "epoch": 0.2, + "learning_rate": 4.9448654180201984e-05, + "loss": 1.0801, + "step": 655 + }, + { + "epoch": 0.2, + "learning_rate": 4.944021035267257e-05, + "loss": 1.0837, + "step": 660 + }, + { + "epoch": 0.2, + "learning_rate": 4.943170308920484e-05, + "loss": 1.0951, + "step": 665 + }, + { + "epoch": 0.21, + "learning_rate": 4.942313241187986e-05, + "loss": 1.0628, + "step": 670 + }, + { + "epoch": 0.21, + "learning_rate": 4.9414498342943305e-05, + "loss": 1.1334, + "step": 675 + }, + { + "epoch": 0.21, + "learning_rate": 4.9405800904805386e-05, + "loss": 1.0021, + "step": 680 + }, + { + "epoch": 0.21, + "learning_rate": 4.93970401200408e-05, + "loss": 1.0818, + "step": 685 + }, + { + "epoch": 0.21, + "learning_rate": 4.9388216011388656e-05, + "loss": 1.1248, + "step": 690 + }, + { + "epoch": 0.21, + "learning_rate": 4.937932860175241e-05, + "loss": 1.0823, + "step": 695 + }, + { + "epoch": 0.22, + "learning_rate": 4.937037791419984e-05, + "loss": 1.1675, + "step": 700 + }, + { + "epoch": 0.22, + "learning_rate": 4.9361363971962965e-05, + "loss": 1.1806, + "step": 705 + }, + { + "epoch": 0.22, + "learning_rate": 4.935228679843796e-05, + "loss": 1.0891, + "step": 710 + }, + { + "epoch": 0.22, + "learning_rate": 4.9343146417185164e-05, + "loss": 0.9939, + "step": 715 + }, + { + "epoch": 0.22, + "learning_rate": 4.9333942851928934e-05, + "loss": 1.0225, + "step": 720 + }, + { + "epoch": 0.22, + "learning_rate": 4.932467612655764e-05, + "loss": 1.1713, + "step": 725 + }, + { + "epoch": 0.22, + "learning_rate": 4.9315346265123594e-05, + "loss": 1.0893, + "step": 730 + }, + { + "epoch": 0.23, + "learning_rate": 4.930595329184296e-05, + "loss": 1.0905, + "step": 735 + }, + { + "epoch": 0.23, + "learning_rate": 4.929649723109575e-05, + "loss": 1.0916, + "step": 740 + }, + { + "epoch": 0.23, + "learning_rate": 4.9286978107425696e-05, + "loss": 1.0566, + "step": 745 + }, + { + "epoch": 0.23, + "learning_rate": 4.9277395945540206e-05, + "loss": 1.0974, + "step": 750 + }, + { + "epoch": 0.23, + "learning_rate": 4.926775077031032e-05, + "loss": 0.9807, + "step": 755 + }, + { + "epoch": 0.23, + "learning_rate": 4.9258042606770646e-05, + "loss": 1.0656, + "step": 760 + }, + { + "epoch": 0.24, + "learning_rate": 4.9248271480119255e-05, + "loss": 1.0792, + "step": 765 + }, + { + "epoch": 0.24, + "learning_rate": 4.923843741571766e-05, + "loss": 1.1384, + "step": 770 + }, + { + "epoch": 0.24, + "learning_rate": 4.922854043909072e-05, + "loss": 1.0754, + "step": 775 + }, + { + "epoch": 0.24, + "learning_rate": 4.921858057592661e-05, + "loss": 1.0395, + "step": 780 + }, + { + "epoch": 0.24, + "learning_rate": 4.92085578520767e-05, + "loss": 1.1031, + "step": 785 + }, + { + "epoch": 0.24, + "learning_rate": 4.919847229355553e-05, + "loss": 1.0629, + "step": 790 + }, + { + "epoch": 0.24, + "learning_rate": 4.9188323926540746e-05, + "loss": 1.055, + "step": 795 + }, + { + "epoch": 0.25, + "learning_rate": 4.9178112777373e-05, + "loss": 1.0913, + "step": 800 + }, + { + "epoch": 0.25, + "learning_rate": 4.9167838872555904e-05, + "loss": 1.0009, + "step": 805 + }, + { + "epoch": 0.25, + "learning_rate": 4.9157502238755945e-05, + "loss": 0.9669, + "step": 810 + }, + { + "epoch": 0.25, + "learning_rate": 4.914710290280245e-05, + "loss": 1.1028, + "step": 815 + }, + { + "epoch": 0.25, + "learning_rate": 4.913664089168747e-05, + "loss": 1.0596, + "step": 820 + }, + { + "epoch": 0.25, + "learning_rate": 4.9126116232565744e-05, + "loss": 1.1413, + "step": 825 + }, + { + "epoch": 0.26, + "learning_rate": 4.911552895275462e-05, + "loss": 0.9488, + "step": 830 + }, + { + "epoch": 0.26, + "learning_rate": 4.910487907973397e-05, + "loss": 1.1634, + "step": 835 + }, + { + "epoch": 0.26, + "learning_rate": 4.909416664114614e-05, + "loss": 1.0243, + "step": 840 + }, + { + "epoch": 0.26, + "learning_rate": 4.908339166479588e-05, + "loss": 1.0199, + "step": 845 + }, + { + "epoch": 0.26, + "learning_rate": 4.907255417865023e-05, + "loss": 1.1234, + "step": 850 + }, + { + "epoch": 0.26, + "learning_rate": 4.9061654210838494e-05, + "loss": 0.988, + "step": 855 + }, + { + "epoch": 0.26, + "learning_rate": 4.905069178965215e-05, + "loss": 1.0122, + "step": 860 + }, + { + "epoch": 0.27, + "learning_rate": 4.903966694354478e-05, + "loss": 1.0742, + "step": 865 + }, + { + "epoch": 0.27, + "learning_rate": 4.9028579701132006e-05, + "loss": 0.9571, + "step": 870 + }, + { + "epoch": 0.27, + "learning_rate": 4.901743009119137e-05, + "loss": 0.9499, + "step": 875 + }, + { + "epoch": 0.27, + "learning_rate": 4.900621814266231e-05, + "loss": 1.0478, + "step": 880 + }, + { + "epoch": 0.27, + "learning_rate": 4.899494388464608e-05, + "loss": 1.167, + "step": 885 + }, + { + "epoch": 0.27, + "learning_rate": 4.8983607346405646e-05, + "loss": 1.0833, + "step": 890 + }, + { + "epoch": 0.28, + "learning_rate": 4.8972208557365616e-05, + "loss": 1.1151, + "step": 895 + }, + { + "epoch": 0.28, + "learning_rate": 4.896074754711221e-05, + "loss": 1.0109, + "step": 900 + }, + { + "epoch": 0.28, + "learning_rate": 4.894922434539311e-05, + "loss": 1.1618, + "step": 905 + }, + { + "epoch": 0.28, + "learning_rate": 4.893763898211743e-05, + "loss": 1.094, + "step": 910 + }, + { + "epoch": 0.28, + "learning_rate": 4.892599148735564e-05, + "loss": 0.9429, + "step": 915 + }, + { + "epoch": 0.28, + "learning_rate": 4.891428189133947e-05, + "loss": 1.0533, + "step": 920 + }, + { + "epoch": 0.28, + "learning_rate": 4.890251022446181e-05, + "loss": 1.0832, + "step": 925 + }, + { + "epoch": 0.29, + "learning_rate": 4.889067651727671e-05, + "loss": 1.1952, + "step": 930 + }, + { + "epoch": 0.29, + "learning_rate": 4.887878080049921e-05, + "loss": 1.0382, + "step": 935 + }, + { + "epoch": 0.29, + "learning_rate": 4.8866823105005306e-05, + "loss": 0.9861, + "step": 940 + }, + { + "epoch": 0.29, + "learning_rate": 4.885480346183188e-05, + "loss": 1.1674, + "step": 945 + }, + { + "epoch": 0.29, + "learning_rate": 4.884272190217658e-05, + "loss": 1.0596, + "step": 950 + }, + { + "epoch": 0.29, + "learning_rate": 4.883057845739778e-05, + "loss": 1.0223, + "step": 955 + }, + { + "epoch": 0.3, + "learning_rate": 4.8818373159014464e-05, + "loss": 1.1016, + "step": 960 + }, + { + "epoch": 0.3, + "learning_rate": 4.880610603870618e-05, + "loss": 1.1162, + "step": 965 + }, + { + "epoch": 0.3, + "learning_rate": 4.879377712831292e-05, + "loss": 1.0396, + "step": 970 + }, + { + "epoch": 0.3, + "learning_rate": 4.8781386459835066e-05, + "loss": 1.1468, + "step": 975 + }, + { + "epoch": 0.3, + "learning_rate": 4.8768934065433295e-05, + "loss": 1.0262, + "step": 980 + }, + { + "epoch": 0.3, + "learning_rate": 4.875641997742849e-05, + "loss": 1.1216, + "step": 985 + }, + { + "epoch": 0.3, + "learning_rate": 4.874384422830167e-05, + "loss": 0.9971, + "step": 990 + }, + { + "epoch": 0.31, + "learning_rate": 4.8731206850693913e-05, + "loss": 1.0378, + "step": 995 + }, + { + "epoch": 0.31, + "learning_rate": 4.8718507877406224e-05, + "loss": 1.0317, + "step": 1000 + }, + { + "epoch": 0.31, + "learning_rate": 4.870574734139951e-05, + "loss": 1.0811, + "step": 1005 + }, + { + "epoch": 0.31, + "learning_rate": 4.8692925275794454e-05, + "loss": 1.0898, + "step": 1010 + }, + { + "epoch": 0.31, + "learning_rate": 4.868004171387145e-05, + "loss": 0.9244, + "step": 1015 + }, + { + "epoch": 0.31, + "learning_rate": 4.8667096689070514e-05, + "loss": 1.064, + "step": 1020 + }, + { + "epoch": 0.32, + "learning_rate": 4.865409023499117e-05, + "loss": 1.1773, + "step": 1025 + }, + { + "epoch": 0.32, + "learning_rate": 4.8641022385392406e-05, + "loss": 1.0189, + "step": 1030 + }, + { + "epoch": 0.32, + "learning_rate": 4.862789317419256e-05, + "loss": 0.9959, + "step": 1035 + }, + { + "epoch": 0.32, + "learning_rate": 4.8614702635469236e-05, + "loss": 1.1068, + "step": 1040 + }, + { + "epoch": 0.32, + "learning_rate": 4.860145080345922e-05, + "loss": 1.0368, + "step": 1045 + }, + { + "epoch": 0.32, + "learning_rate": 4.858813771255839e-05, + "loss": 1.0131, + "step": 1050 + }, + { + "epoch": 0.32, + "learning_rate": 4.8574763397321614e-05, + "loss": 1.0841, + "step": 1055 + }, + { + "epoch": 0.33, + "learning_rate": 4.8561327892462684e-05, + "loss": 1.0428, + "step": 1060 + }, + { + "epoch": 0.33, + "learning_rate": 4.8547831232854197e-05, + "loss": 1.0757, + "step": 1065 + }, + { + "epoch": 0.33, + "learning_rate": 4.85342734535275e-05, + "loss": 1.1231, + "step": 1070 + }, + { + "epoch": 0.33, + "learning_rate": 4.852065458967256e-05, + "loss": 1.0533, + "step": 1075 + }, + { + "epoch": 0.33, + "learning_rate": 4.850697467663793e-05, + "loss": 0.9773, + "step": 1080 + }, + { + "epoch": 0.33, + "learning_rate": 4.8493233749930556e-05, + "loss": 0.9959, + "step": 1085 + }, + { + "epoch": 0.34, + "learning_rate": 4.8479431845215804e-05, + "loss": 1.0637, + "step": 1090 + }, + { + "epoch": 0.34, + "learning_rate": 4.84655689983173e-05, + "loss": 1.0259, + "step": 1095 + }, + { + "epoch": 0.34, + "learning_rate": 4.845164524521682e-05, + "loss": 1.0969, + "step": 1100 + }, + { + "epoch": 0.34, + "learning_rate": 4.843766062205428e-05, + "loss": 1.1443, + "step": 1105 + }, + { + "epoch": 0.34, + "learning_rate": 4.842361516512752e-05, + "loss": 1.0425, + "step": 1110 + }, + { + "epoch": 0.34, + "learning_rate": 4.8409508910892326e-05, + "loss": 1.0704, + "step": 1115 + }, + { + "epoch": 0.34, + "learning_rate": 4.839534189596228e-05, + "loss": 1.0601, + "step": 1120 + }, + { + "epoch": 0.35, + "learning_rate": 4.838111415710864e-05, + "loss": 1.0754, + "step": 1125 + }, + { + "epoch": 0.35, + "learning_rate": 4.8366825731260336e-05, + "loss": 0.9877, + "step": 1130 + }, + { + "epoch": 0.35, + "learning_rate": 4.835247665550376e-05, + "loss": 1.006, + "step": 1135 + }, + { + "epoch": 0.35, + "learning_rate": 4.833806696708275e-05, + "loss": 1.1268, + "step": 1140 + }, + { + "epoch": 0.35, + "learning_rate": 4.832359670339845e-05, + "loss": 1.0366, + "step": 1145 + }, + { + "epoch": 0.35, + "learning_rate": 4.8309065902009254e-05, + "loss": 0.9914, + "step": 1150 + }, + { + "epoch": 0.36, + "learning_rate": 4.829447460063067e-05, + "loss": 1.0464, + "step": 1155 + }, + { + "epoch": 0.36, + "learning_rate": 4.827982283713523e-05, + "loss": 0.9479, + "step": 1160 + }, + { + "epoch": 0.36, + "learning_rate": 4.826511064955243e-05, + "loss": 1.0767, + "step": 1165 + }, + { + "epoch": 0.36, + "learning_rate": 4.825033807606856e-05, + "loss": 1.0902, + "step": 1170 + }, + { + "epoch": 0.36, + "learning_rate": 4.823550515502667e-05, + "loss": 1.0831, + "step": 1175 + }, + { + "epoch": 0.36, + "learning_rate": 4.822061192492644e-05, + "loss": 0.9966, + "step": 1180 + }, + { + "epoch": 0.36, + "learning_rate": 4.820565842442408e-05, + "loss": 1.0378, + "step": 1185 + }, + { + "epoch": 0.37, + "learning_rate": 4.819064469233225e-05, + "loss": 0.9169, + "step": 1190 + }, + { + "epoch": 0.37, + "learning_rate": 4.817557076761994e-05, + "loss": 1.0553, + "step": 1195 + }, + { + "epoch": 0.37, + "learning_rate": 4.8160436689412346e-05, + "loss": 1.0782, + "step": 1200 + }, + { + "epoch": 0.37, + "learning_rate": 4.814524249699084e-05, + "loss": 1.0319, + "step": 1205 + }, + { + "epoch": 0.37, + "learning_rate": 4.812998822979278e-05, + "loss": 1.1362, + "step": 1210 + }, + { + "epoch": 0.37, + "learning_rate": 4.811467392741149e-05, + "loss": 1.0532, + "step": 1215 + }, + { + "epoch": 0.38, + "learning_rate": 4.809929962959609e-05, + "loss": 1.0872, + "step": 1220 + }, + { + "epoch": 0.38, + "learning_rate": 4.808386537625145e-05, + "loss": 1.1151, + "step": 1225 + }, + { + "epoch": 0.38, + "learning_rate": 4.8068371207438026e-05, + "loss": 1.0061, + "step": 1230 + }, + { + "epoch": 0.38, + "learning_rate": 4.805281716337181e-05, + "loss": 1.0723, + "step": 1235 + }, + { + "epoch": 0.38, + "learning_rate": 4.8037203284424194e-05, + "loss": 1.1954, + "step": 1240 + }, + { + "epoch": 0.38, + "learning_rate": 4.802152961112188e-05, + "loss": 1.0454, + "step": 1245 + }, + { + "epoch": 0.38, + "learning_rate": 4.800579618414676e-05, + "loss": 1.103, + "step": 1250 + }, + { + "epoch": 0.39, + "learning_rate": 4.7990003044335833e-05, + "loss": 0.9541, + "step": 1255 + }, + { + "epoch": 0.39, + "learning_rate": 4.7974150232681084e-05, + "loss": 1.0924, + "step": 1260 + }, + { + "epoch": 0.39, + "learning_rate": 4.795823779032937e-05, + "loss": 1.0226, + "step": 1265 + }, + { + "epoch": 0.39, + "learning_rate": 4.7942265758582326e-05, + "loss": 1.0829, + "step": 1270 + }, + { + "epoch": 0.39, + "learning_rate": 4.792623417889627e-05, + "loss": 1.0497, + "step": 1275 + }, + { + "epoch": 0.39, + "learning_rate": 4.791014309288205e-05, + "loss": 1.0474, + "step": 1280 + }, + { + "epoch": 0.4, + "learning_rate": 4.7893992542305e-05, + "loss": 1.0973, + "step": 1285 + }, + { + "epoch": 0.4, + "learning_rate": 4.7877782569084766e-05, + "loss": 1.0942, + "step": 1290 + }, + { + "epoch": 0.4, + "learning_rate": 4.7861513215295255e-05, + "loss": 1.0356, + "step": 1295 + }, + { + "epoch": 0.4, + "learning_rate": 4.784518452316448e-05, + "loss": 1.0284, + "step": 1300 + }, + { + "epoch": 0.4, + "learning_rate": 4.782879653507449e-05, + "loss": 1.1028, + "step": 1305 + }, + { + "epoch": 0.4, + "learning_rate": 4.7812349293561206e-05, + "loss": 1.0029, + "step": 1310 + }, + { + "epoch": 0.4, + "learning_rate": 4.77958428413144e-05, + "loss": 1.0146, + "step": 1315 + }, + { + "epoch": 0.41, + "learning_rate": 4.777927722117746e-05, + "loss": 1.0442, + "step": 1320 + }, + { + "epoch": 0.41, + "learning_rate": 4.776265247614741e-05, + "loss": 1.0519, + "step": 1325 + }, + { + "epoch": 0.41, + "learning_rate": 4.77459686493747e-05, + "loss": 1.03, + "step": 1330 + }, + { + "epoch": 0.41, + "learning_rate": 4.772922578416313e-05, + "loss": 1.0757, + "step": 1335 + }, + { + "epoch": 0.41, + "learning_rate": 4.771242392396977e-05, + "loss": 1.0494, + "step": 1340 + }, + { + "epoch": 0.41, + "learning_rate": 4.7695563112404765e-05, + "loss": 1.1324, + "step": 1345 + }, + { + "epoch": 0.42, + "learning_rate": 4.767864339323132e-05, + "loss": 1.0177, + "step": 1350 + }, + { + "epoch": 0.42, + "learning_rate": 4.76616648103655e-05, + "loss": 1.0216, + "step": 1355 + }, + { + "epoch": 0.42, + "learning_rate": 4.7644627407876175e-05, + "loss": 0.9145, + "step": 1360 + }, + { + "epoch": 0.42, + "learning_rate": 4.7627531229984884e-05, + "loss": 1.0646, + "step": 1365 + }, + { + "epoch": 0.42, + "learning_rate": 4.761037632106572e-05, + "loss": 1.0356, + "step": 1370 + }, + { + "epoch": 0.42, + "learning_rate": 4.75931627256452e-05, + "loss": 1.1168, + "step": 1375 + }, + { + "epoch": 0.42, + "learning_rate": 4.7575890488402185e-05, + "loss": 1.0297, + "step": 1380 + }, + { + "epoch": 0.43, + "learning_rate": 4.7558559654167736e-05, + "loss": 1.0279, + "step": 1385 + }, + { + "epoch": 0.43, + "learning_rate": 4.754117026792501e-05, + "loss": 1.0293, + "step": 1390 + }, + { + "epoch": 0.43, + "learning_rate": 4.752372237480912e-05, + "loss": 1.0697, + "step": 1395 + }, + { + "epoch": 0.43, + "learning_rate": 4.750621602010708e-05, + "loss": 1.0334, + "step": 1400 + }, + { + "epoch": 0.43, + "learning_rate": 4.748865124925759e-05, + "loss": 1.0819, + "step": 1405 + }, + { + "epoch": 0.43, + "learning_rate": 4.747102810785101e-05, + "loss": 0.9419, + "step": 1410 + }, + { + "epoch": 0.44, + "learning_rate": 4.745334664162919e-05, + "loss": 1.0451, + "step": 1415 + }, + { + "epoch": 0.44, + "learning_rate": 4.7435606896485374e-05, + "loss": 1.0981, + "step": 1420 + }, + { + "epoch": 0.44, + "learning_rate": 4.741780891846407e-05, + "loss": 1.0632, + "step": 1425 + }, + { + "epoch": 0.44, + "learning_rate": 4.739995275376092e-05, + "loss": 1.0788, + "step": 1430 + }, + { + "epoch": 0.44, + "learning_rate": 4.73820384487226e-05, + "loss": 1.1371, + "step": 1435 + }, + { + "epoch": 0.44, + "learning_rate": 4.73640660498467e-05, + "loss": 1.0411, + "step": 1440 + }, + { + "epoch": 0.44, + "learning_rate": 4.73460356037816e-05, + "loss": 0.9905, + "step": 1445 + }, + { + "epoch": 0.45, + "learning_rate": 4.7327947157326314e-05, + "loss": 1.1467, + "step": 1450 + }, + { + "epoch": 0.45, + "learning_rate": 4.730980075743044e-05, + "loss": 1.1207, + "step": 1455 + }, + { + "epoch": 0.45, + "learning_rate": 4.729159645119396e-05, + "loss": 1.1177, + "step": 1460 + }, + { + "epoch": 0.45, + "learning_rate": 4.7273334285867196e-05, + "loss": 1.1426, + "step": 1465 + }, + { + "epoch": 0.45, + "learning_rate": 4.72550143088506e-05, + "loss": 1.0281, + "step": 1470 + }, + { + "epoch": 0.45, + "learning_rate": 4.723663656769472e-05, + "loss": 1.0426, + "step": 1475 + }, + { + "epoch": 0.46, + "learning_rate": 4.72182011101e-05, + "loss": 1.1024, + "step": 1480 + }, + { + "epoch": 0.46, + "learning_rate": 4.719970798391672e-05, + "loss": 1.1707, + "step": 1485 + }, + { + "epoch": 0.46, + "learning_rate": 4.7181157237144814e-05, + "loss": 1.0526, + "step": 1490 + }, + { + "epoch": 0.46, + "learning_rate": 4.716254891793379e-05, + "loss": 1.0504, + "step": 1495 + }, + { + "epoch": 0.46, + "learning_rate": 4.714388307458259e-05, + "loss": 0.9978, + "step": 1500 + }, + { + "epoch": 0.46, + "learning_rate": 4.712515975553945e-05, + "loss": 1.0464, + "step": 1505 + }, + { + "epoch": 0.46, + "learning_rate": 4.710637900940181e-05, + "loss": 1.0067, + "step": 1510 + }, + { + "epoch": 0.47, + "learning_rate": 4.7087540884916146e-05, + "loss": 1.002, + "step": 1515 + }, + { + "epoch": 0.47, + "learning_rate": 4.706864543097785e-05, + "loss": 1.0072, + "step": 1520 + }, + { + "epoch": 0.47, + "learning_rate": 4.7049692696631164e-05, + "loss": 1.0186, + "step": 1525 + }, + { + "epoch": 0.47, + "learning_rate": 4.7030682731068954e-05, + "loss": 1.0628, + "step": 1530 + }, + { + "epoch": 0.47, + "learning_rate": 4.701161558363266e-05, + "loss": 1.0407, + "step": 1535 + }, + { + "epoch": 0.47, + "learning_rate": 4.699249130381213e-05, + "loss": 1.102, + "step": 1540 + }, + { + "epoch": 0.48, + "learning_rate": 4.6973309941245506e-05, + "loss": 1.0515, + "step": 1545 + }, + { + "epoch": 0.48, + "learning_rate": 4.69540715457191e-05, + "loss": 1.0299, + "step": 1550 + }, + { + "epoch": 0.48, + "learning_rate": 4.6934776167167235e-05, + "loss": 0.962, + "step": 1555 + }, + { + "epoch": 0.48, + "learning_rate": 4.6915423855672166e-05, + "loss": 1.0396, + "step": 1560 + }, + { + "epoch": 0.48, + "learning_rate": 4.689601466146388e-05, + "loss": 0.9753, + "step": 1565 + }, + { + "epoch": 0.48, + "learning_rate": 4.6876548634920057e-05, + "loss": 1.052, + "step": 1570 + }, + { + "epoch": 0.48, + "learning_rate": 4.685702582656584e-05, + "loss": 1.0641, + "step": 1575 + }, + { + "epoch": 0.49, + "learning_rate": 4.68374462870738e-05, + "loss": 1.112, + "step": 1580 + }, + { + "epoch": 0.49, + "learning_rate": 4.68178100672637e-05, + "loss": 0.9603, + "step": 1585 + }, + { + "epoch": 0.49, + "learning_rate": 4.679811721810247e-05, + "loss": 1.0824, + "step": 1590 + }, + { + "epoch": 0.49, + "learning_rate": 4.6778367790704e-05, + "loss": 0.993, + "step": 1595 + }, + { + "epoch": 0.49, + "learning_rate": 4.675856183632904e-05, + "loss": 1.1299, + "step": 1600 + }, + { + "epoch": 0.49, + "learning_rate": 4.6738699406385054e-05, + "loss": 1.101, + "step": 1605 + }, + { + "epoch": 0.5, + "learning_rate": 4.67187805524261e-05, + "loss": 1.0934, + "step": 1610 + }, + { + "epoch": 0.5, + "learning_rate": 4.669880532615268e-05, + "loss": 0.9486, + "step": 1615 + }, + { + "epoch": 0.5, + "learning_rate": 4.6678773779411614e-05, + "loss": 0.8742, + "step": 1620 + }, + { + "epoch": 0.5, + "learning_rate": 4.6658685964195907e-05, + "loss": 1.0317, + "step": 1625 + }, + { + "epoch": 0.5, + "learning_rate": 4.663854193264461e-05, + "loss": 1.1559, + "step": 1630 + }, + { + "epoch": 0.5, + "learning_rate": 4.661834173704269e-05, + "loss": 1.0961, + "step": 1635 + }, + { + "epoch": 0.5, + "learning_rate": 4.659808542982088e-05, + "loss": 1.026, + "step": 1640 + }, + { + "epoch": 0.51, + "learning_rate": 4.6577773063555585e-05, + "loss": 1.0197, + "step": 1645 + }, + { + "epoch": 0.51, + "learning_rate": 4.6557404690968665e-05, + "loss": 1.1052, + "step": 1650 + }, + { + "epoch": 0.51, + "learning_rate": 4.65369803649274e-05, + "loss": 0.9945, + "step": 1655 + }, + { + "epoch": 0.51, + "learning_rate": 4.651650013844425e-05, + "loss": 1.0487, + "step": 1660 + }, + { + "epoch": 0.51, + "learning_rate": 4.649596406467681e-05, + "loss": 0.9734, + "step": 1665 + }, + { + "epoch": 0.51, + "learning_rate": 4.647537219692761e-05, + "loss": 1.0409, + "step": 1670 + }, + { + "epoch": 0.52, + "learning_rate": 4.6454724588643995e-05, + "loss": 1.025, + "step": 1675 + }, + { + "epoch": 0.52, + "learning_rate": 4.6434021293417993e-05, + "loss": 1.0995, + "step": 1680 + }, + { + "epoch": 0.52, + "learning_rate": 4.641326236498617e-05, + "loss": 1.1503, + "step": 1685 + }, + { + "epoch": 0.52, + "learning_rate": 4.639244785722949e-05, + "loss": 1.0707, + "step": 1690 + }, + { + "epoch": 0.52, + "learning_rate": 4.637157782417319e-05, + "loss": 0.9672, + "step": 1695 + }, + { + "epoch": 0.52, + "learning_rate": 4.635065231998658e-05, + "loss": 1.0637, + "step": 1700 + }, + { + "epoch": 0.52, + "learning_rate": 4.632967139898301e-05, + "loss": 1.0157, + "step": 1705 + }, + { + "epoch": 0.53, + "learning_rate": 4.630863511561963e-05, + "loss": 1.2091, + "step": 1710 + }, + { + "epoch": 0.53, + "learning_rate": 4.628754352449729e-05, + "loss": 1.0324, + "step": 1715 + }, + { + "epoch": 0.53, + "learning_rate": 4.6266396680360415e-05, + "loss": 1.0442, + "step": 1720 + }, + { + "epoch": 0.53, + "learning_rate": 4.62451946380968e-05, + "loss": 0.9929, + "step": 1725 + }, + { + "epoch": 0.53, + "learning_rate": 4.6223937452737556e-05, + "loss": 0.9119, + "step": 1730 + }, + { + "epoch": 0.53, + "learning_rate": 4.62026251794569e-05, + "loss": 1.1521, + "step": 1735 + }, + { + "epoch": 0.54, + "learning_rate": 4.618125787357203e-05, + "loss": 1.1598, + "step": 1740 + }, + { + "epoch": 0.54, + "learning_rate": 4.615983559054298e-05, + "loss": 1.0319, + "step": 1745 + }, + { + "epoch": 0.54, + "learning_rate": 4.61383583859725e-05, + "loss": 1.0767, + "step": 1750 + }, + { + "epoch": 0.54, + "learning_rate": 4.611682631560589e-05, + "loss": 1.0748, + "step": 1755 + }, + { + "epoch": 0.54, + "learning_rate": 4.609523943533082e-05, + "loss": 1.0645, + "step": 1760 + }, + { + "epoch": 0.54, + "learning_rate": 4.607359780117727e-05, + "loss": 1.0733, + "step": 1765 + }, + { + "epoch": 0.54, + "learning_rate": 4.605190146931731e-05, + "loss": 1.1735, + "step": 1770 + }, + { + "epoch": 0.55, + "learning_rate": 4.6030150496064996e-05, + "loss": 1.066, + "step": 1775 + }, + { + "epoch": 0.55, + "learning_rate": 4.600834493787618e-05, + "loss": 1.164, + "step": 1780 + }, + { + "epoch": 0.55, + "learning_rate": 4.598648485134844e-05, + "loss": 1.0979, + "step": 1785 + }, + { + "epoch": 0.55, + "learning_rate": 4.596457029322083e-05, + "loss": 1.0689, + "step": 1790 + }, + { + "epoch": 0.55, + "learning_rate": 4.5942601320373847e-05, + "loss": 0.9265, + "step": 1795 + }, + { + "epoch": 0.55, + "learning_rate": 4.592057798982917e-05, + "loss": 1.1218, + "step": 1800 + }, + { + "epoch": 0.56, + "learning_rate": 4.58985003587496e-05, + "loss": 1.161, + "step": 1805 + }, + { + "epoch": 0.56, + "learning_rate": 4.587636848443885e-05, + "loss": 1.0374, + "step": 1810 + }, + { + "epoch": 0.56, + "learning_rate": 4.585418242434147e-05, + "loss": 1.1273, + "step": 1815 + }, + { + "epoch": 0.56, + "learning_rate": 4.5831942236042616e-05, + "loss": 1.0319, + "step": 1820 + }, + { + "epoch": 0.56, + "learning_rate": 4.580964797726794e-05, + "loss": 1.0766, + "step": 1825 + }, + { + "epoch": 0.56, + "learning_rate": 4.578729970588344e-05, + "loss": 1.0973, + "step": 1830 + }, + { + "epoch": 0.56, + "learning_rate": 4.5764897479895317e-05, + "loss": 1.1318, + "step": 1835 + }, + { + "epoch": 0.57, + "learning_rate": 4.5742441357449804e-05, + "loss": 1.1277, + "step": 1840 + }, + { + "epoch": 0.57, + "learning_rate": 4.5719931396833025e-05, + "loss": 1.0464, + "step": 1845 + }, + { + "epoch": 0.57, + "learning_rate": 4.569736765647085e-05, + "loss": 1.1164, + "step": 1850 + }, + { + "epoch": 0.57, + "learning_rate": 4.567475019492874e-05, + "loss": 1.029, + "step": 1855 + }, + { + "epoch": 0.57, + "learning_rate": 4.565207907091157e-05, + "loss": 1.0611, + "step": 1860 + }, + { + "epoch": 0.57, + "learning_rate": 4.562935434326353e-05, + "loss": 1.1184, + "step": 1865 + }, + { + "epoch": 0.58, + "learning_rate": 4.5606576070967916e-05, + "loss": 1.0881, + "step": 1870 + }, + { + "epoch": 0.58, + "learning_rate": 4.5583744313147026e-05, + "loss": 1.0486, + "step": 1875 + }, + { + "epoch": 0.58, + "learning_rate": 4.556085912906195e-05, + "loss": 0.9676, + "step": 1880 + }, + { + "epoch": 0.58, + "learning_rate": 4.553792057811249e-05, + "loss": 1.0419, + "step": 1885 + }, + { + "epoch": 0.58, + "learning_rate": 4.551492871983693e-05, + "loss": 0.9601, + "step": 1890 + }, + { + "epoch": 0.58, + "learning_rate": 4.5491883613911934e-05, + "loss": 1.123, + "step": 1895 + }, + { + "epoch": 0.58, + "learning_rate": 4.5468785320152365e-05, + "loss": 0.9876, + "step": 1900 + }, + { + "epoch": 0.59, + "learning_rate": 4.544563389851114e-05, + "loss": 0.9866, + "step": 1905 + }, + { + "epoch": 0.59, + "learning_rate": 4.542242940907909e-05, + "loss": 1.0857, + "step": 1910 + }, + { + "epoch": 0.59, + "learning_rate": 4.5399171912084743e-05, + "loss": 1.0982, + "step": 1915 + }, + { + "epoch": 0.59, + "learning_rate": 4.537586146789426e-05, + "loss": 1.0641, + "step": 1920 + }, + { + "epoch": 0.59, + "learning_rate": 4.5352498137011195e-05, + "loss": 1.1106, + "step": 1925 + }, + { + "epoch": 0.59, + "learning_rate": 4.532908198007639e-05, + "loss": 1.048, + "step": 1930 + }, + { + "epoch": 0.6, + "learning_rate": 4.5305613057867805e-05, + "loss": 1.1287, + "step": 1935 + }, + { + "epoch": 0.6, + "learning_rate": 4.5282091431300334e-05, + "loss": 0.9996, + "step": 1940 + }, + { + "epoch": 0.6, + "learning_rate": 4.5258517161425676e-05, + "loss": 1.1478, + "step": 1945 + }, + { + "epoch": 0.6, + "learning_rate": 4.523489030943218e-05, + "loss": 1.0896, + "step": 1950 + }, + { + "epoch": 0.6, + "learning_rate": 4.521121093664467e-05, + "loss": 1.056, + "step": 1955 + }, + { + "epoch": 0.6, + "learning_rate": 4.5187479104524275e-05, + "loss": 1.0293, + "step": 1960 + }, + { + "epoch": 0.6, + "learning_rate": 4.516369487466832e-05, + "loss": 0.9746, + "step": 1965 + }, + { + "epoch": 0.61, + "learning_rate": 4.513985830881008e-05, + "loss": 0.9957, + "step": 1970 + }, + { + "epoch": 0.61, + "learning_rate": 4.511596946881872e-05, + "loss": 0.9588, + "step": 1975 + }, + { + "epoch": 0.61, + "learning_rate": 4.509202841669906e-05, + "loss": 1.0514, + "step": 1980 + }, + { + "epoch": 0.61, + "learning_rate": 4.506803521459144e-05, + "loss": 1.0063, + "step": 1985 + }, + { + "epoch": 0.61, + "learning_rate": 4.504398992477158e-05, + "loss": 1.0231, + "step": 1990 + }, + { + "epoch": 0.61, + "learning_rate": 4.501989260965036e-05, + "loss": 1.1287, + "step": 1995 + }, + { + "epoch": 0.62, + "learning_rate": 4.499574333177373e-05, + "loss": 0.9807, + "step": 2000 + }, + { + "epoch": 0.62, + "learning_rate": 4.497154215382249e-05, + "loss": 0.9944, + "step": 2005 + }, + { + "epoch": 0.62, + "learning_rate": 4.494728913861215e-05, + "loss": 1.0149, + "step": 2010 + }, + { + "epoch": 0.62, + "learning_rate": 4.492298434909278e-05, + "loss": 0.8783, + "step": 2015 + }, + { + "epoch": 0.62, + "learning_rate": 4.489862784834883e-05, + "loss": 1.0137, + "step": 2020 + }, + { + "epoch": 0.62, + "learning_rate": 4.487421969959896e-05, + "loss": 1.0178, + "step": 2025 + }, + { + "epoch": 0.62, + "learning_rate": 4.484975996619589e-05, + "loss": 1.0005, + "step": 2030 + }, + { + "epoch": 0.63, + "learning_rate": 4.482524871162623e-05, + "loss": 1.0349, + "step": 2035 + }, + { + "epoch": 0.63, + "learning_rate": 4.4800685999510315e-05, + "loss": 1.1509, + "step": 2040 + }, + { + "epoch": 0.63, + "learning_rate": 4.477607189360206e-05, + "loss": 1.0146, + "step": 2045 + }, + { + "epoch": 0.63, + "learning_rate": 4.4751406457788744e-05, + "loss": 0.9787, + "step": 2050 + }, + { + "epoch": 0.63, + "learning_rate": 4.472668975609089e-05, + "loss": 1.119, + "step": 2055 + }, + { + "epoch": 0.63, + "learning_rate": 4.47019218526621e-05, + "loss": 1.113, + "step": 2060 + }, + { + "epoch": 0.64, + "learning_rate": 4.467710281178884e-05, + "loss": 1.0364, + "step": 2065 + }, + { + "epoch": 0.64, + "learning_rate": 4.4652232697890334e-05, + "loss": 1.1773, + "step": 2070 + }, + { + "epoch": 0.64, + "learning_rate": 4.462731157551835e-05, + "loss": 1.0428, + "step": 2075 + }, + { + "epoch": 0.64, + "learning_rate": 4.4602339509357074e-05, + "loss": 1.13, + "step": 2080 + }, + { + "epoch": 0.64, + "learning_rate": 4.4577316564222905e-05, + "loss": 1.134, + "step": 2085 + }, + { + "epoch": 0.64, + "learning_rate": 4.4552242805064286e-05, + "loss": 1.1042, + "step": 2090 + }, + { + "epoch": 0.64, + "learning_rate": 4.452711829696158e-05, + "loss": 1.0635, + "step": 2095 + }, + { + "epoch": 0.65, + "learning_rate": 4.4501943105126864e-05, + "loss": 1.0237, + "step": 2100 + }, + { + "epoch": 0.65, + "learning_rate": 4.447671729490375e-05, + "loss": 0.987, + "step": 2105 + }, + { + "epoch": 0.65, + "learning_rate": 4.445144093176724e-05, + "loss": 1.1039, + "step": 2110 + }, + { + "epoch": 0.65, + "learning_rate": 4.4426114081323575e-05, + "loss": 1.1445, + "step": 2115 + }, + { + "epoch": 0.65, + "learning_rate": 4.440073680931e-05, + "loss": 0.9769, + "step": 2120 + }, + { + "epoch": 0.65, + "learning_rate": 4.437530918159465e-05, + "loss": 1.0396, + "step": 2125 + }, + { + "epoch": 0.66, + "learning_rate": 4.434983126417637e-05, + "loss": 0.9372, + "step": 2130 + }, + { + "epoch": 0.66, + "learning_rate": 4.4324303123184517e-05, + "loss": 1.0454, + "step": 2135 + }, + { + "epoch": 0.66, + "learning_rate": 4.429872482487882e-05, + "loss": 0.9945, + "step": 2140 + }, + { + "epoch": 0.66, + "learning_rate": 4.4273096435649194e-05, + "loss": 1.011, + "step": 2145 + }, + { + "epoch": 0.66, + "learning_rate": 4.424741802201554e-05, + "loss": 1.07, + "step": 2150 + }, + { + "epoch": 0.66, + "learning_rate": 4.422168965062765e-05, + "loss": 1.1419, + "step": 2155 + }, + { + "epoch": 0.66, + "learning_rate": 4.4195911388264946e-05, + "loss": 1.0535, + "step": 2160 + }, + { + "epoch": 0.67, + "learning_rate": 4.417008330183634e-05, + "loss": 1.1407, + "step": 2165 + }, + { + "epoch": 0.67, + "learning_rate": 4.41442054583801e-05, + "loss": 1.0387, + "step": 2170 + }, + { + "epoch": 0.67, + "learning_rate": 4.411827792506361e-05, + "loss": 1.1442, + "step": 2175 + }, + { + "epoch": 0.67, + "learning_rate": 4.409230076918326e-05, + "loss": 1.0498, + "step": 2180 + }, + { + "epoch": 0.67, + "learning_rate": 4.4066274058164185e-05, + "loss": 0.9523, + "step": 2185 + }, + { + "epoch": 0.67, + "learning_rate": 4.4040197859560196e-05, + "loss": 0.9737, + "step": 2190 + }, + { + "epoch": 0.68, + "learning_rate": 4.401407224105353e-05, + "loss": 1.034, + "step": 2195 + }, + { + "epoch": 0.68, + "learning_rate": 4.398789727045469e-05, + "loss": 0.9203, + "step": 2200 + }, + { + "epoch": 0.68, + "learning_rate": 4.396167301570228e-05, + "loss": 0.9979, + "step": 2205 + }, + { + "epoch": 0.68, + "learning_rate": 4.393539954486283e-05, + "loss": 1.0786, + "step": 2210 + }, + { + "epoch": 0.68, + "learning_rate": 4.390907692613062e-05, + "loss": 1.0129, + "step": 2215 + }, + { + "epoch": 0.68, + "learning_rate": 4.388270522782746e-05, + "loss": 0.9616, + "step": 2220 + }, + { + "epoch": 0.68, + "learning_rate": 4.3856284518402594e-05, + "loss": 1.0415, + "step": 2225 + }, + { + "epoch": 0.69, + "learning_rate": 4.382981486643246e-05, + "loss": 0.9654, + "step": 2230 + }, + { + "epoch": 0.69, + "learning_rate": 4.38032963406205e-05, + "loss": 1.0333, + "step": 2235 + }, + { + "epoch": 0.69, + "learning_rate": 4.3776729009797064e-05, + "loss": 1.0319, + "step": 2240 + }, + { + "epoch": 0.69, + "learning_rate": 4.3750112942919136e-05, + "loss": 0.9943, + "step": 2245 + }, + { + "epoch": 0.69, + "learning_rate": 4.3723448209070215e-05, + "loss": 1.1758, + "step": 2250 + }, + { + "epoch": 0.69, + "learning_rate": 4.369673487746012e-05, + "loss": 1.0614, + "step": 2255 + }, + { + "epoch": 0.7, + "learning_rate": 4.36699730174248e-05, + "loss": 1.0226, + "step": 2260 + }, + { + "epoch": 0.7, + "learning_rate": 4.3643162698426165e-05, + "loss": 1.1838, + "step": 2265 + }, + { + "epoch": 0.7, + "learning_rate": 4.36163039900519e-05, + "loss": 1.0863, + "step": 2270 + }, + { + "epoch": 0.7, + "learning_rate": 4.3589396962015294e-05, + "loss": 0.9998, + "step": 2275 + }, + { + "epoch": 0.7, + "learning_rate": 4.356244168415505e-05, + "loss": 1.0702, + "step": 2280 + }, + { + "epoch": 0.7, + "learning_rate": 4.353543822643511e-05, + "loss": 1.0564, + "step": 2285 + }, + { + "epoch": 0.7, + "learning_rate": 4.350838665894446e-05, + "loss": 1.0642, + "step": 2290 + }, + { + "epoch": 0.71, + "learning_rate": 4.3481287051896954e-05, + "loss": 1.0328, + "step": 2295 + }, + { + "epoch": 0.71, + "learning_rate": 4.345413947563115e-05, + "loss": 1.1278, + "step": 2300 + }, + { + "epoch": 0.71, + "learning_rate": 4.342694400061012e-05, + "loss": 1.0001, + "step": 2305 + }, + { + "epoch": 0.71, + "learning_rate": 4.3399700697421224e-05, + "loss": 1.0696, + "step": 2310 + }, + { + "epoch": 0.71, + "learning_rate": 4.3372409636776e-05, + "loss": 1.0366, + "step": 2315 + }, + { + "epoch": 0.71, + "learning_rate": 4.334507088950992e-05, + "loss": 1.0823, + "step": 2320 + }, + { + "epoch": 0.72, + "learning_rate": 4.3317684526582247e-05, + "loss": 1.0983, + "step": 2325 + }, + { + "epoch": 0.72, + "learning_rate": 4.329025061907582e-05, + "loss": 1.1735, + "step": 2330 + }, + { + "epoch": 0.72, + "learning_rate": 4.326276923819688e-05, + "loss": 1.0041, + "step": 2335 + }, + { + "epoch": 0.72, + "learning_rate": 4.323524045527491e-05, + "loss": 1.1114, + "step": 2340 + }, + { + "epoch": 0.72, + "learning_rate": 4.320766434176239e-05, + "loss": 1.1262, + "step": 2345 + }, + { + "epoch": 0.72, + "learning_rate": 4.318004096923468e-05, + "loss": 1.0595, + "step": 2350 + }, + { + "epoch": 0.72, + "learning_rate": 4.3152370409389795e-05, + "loss": 0.9875, + "step": 2355 + }, + { + "epoch": 0.73, + "learning_rate": 4.3124652734048245e-05, + "loss": 0.9279, + "step": 2360 + }, + { + "epoch": 0.73, + "learning_rate": 4.309688801515278e-05, + "loss": 1.0465, + "step": 2365 + }, + { + "epoch": 0.73, + "learning_rate": 4.306907632476831e-05, + "loss": 1.052, + "step": 2370 + }, + { + "epoch": 0.73, + "learning_rate": 4.304121773508163e-05, + "loss": 0.9938, + "step": 2375 + }, + { + "epoch": 0.73, + "learning_rate": 4.301331231840128e-05, + "loss": 1.0698, + "step": 2380 + }, + { + "epoch": 0.73, + "learning_rate": 4.2985360147157325e-05, + "loss": 1.1129, + "step": 2385 + }, + { + "epoch": 0.74, + "learning_rate": 4.295736129390121e-05, + "loss": 1.0031, + "step": 2390 + }, + { + "epoch": 0.74, + "learning_rate": 4.2929315831305514e-05, + "loss": 1.0514, + "step": 2395 + }, + { + "epoch": 0.74, + "learning_rate": 4.2901223832163824e-05, + "loss": 0.9657, + "step": 2400 + }, + { + "epoch": 0.74, + "learning_rate": 4.28730853693905e-05, + "loss": 1.1062, + "step": 2405 + }, + { + "epoch": 0.74, + "learning_rate": 4.284490051602049e-05, + "loss": 1.019, + "step": 2410 + }, + { + "epoch": 0.74, + "learning_rate": 4.2816669345209174e-05, + "loss": 1.1063, + "step": 2415 + }, + { + "epoch": 0.74, + "learning_rate": 4.278839193023214e-05, + "loss": 1.0027, + "step": 2420 + }, + { + "epoch": 0.75, + "learning_rate": 4.2760068344485e-05, + "loss": 1.1351, + "step": 2425 + }, + { + "epoch": 0.75, + "learning_rate": 4.273169866148322e-05, + "loss": 1.1294, + "step": 2430 + }, + { + "epoch": 0.75, + "learning_rate": 4.27032829548619e-05, + "loss": 1.0139, + "step": 2435 + }, + { + "epoch": 0.75, + "learning_rate": 4.267482129837561e-05, + "loss": 1.095, + "step": 2440 + }, + { + "epoch": 0.75, + "learning_rate": 4.264631376589817e-05, + "loss": 1.0995, + "step": 2445 + }, + { + "epoch": 0.75, + "learning_rate": 4.261776043142247e-05, + "loss": 1.1142, + "step": 2450 + }, + { + "epoch": 0.76, + "learning_rate": 4.2589161369060314e-05, + "loss": 1.0183, + "step": 2455 + }, + { + "epoch": 0.76, + "learning_rate": 4.256051665304217e-05, + "loss": 1.0213, + "step": 2460 + }, + { + "epoch": 0.76, + "learning_rate": 4.2531826357717e-05, + "loss": 0.9383, + "step": 2465 + }, + { + "epoch": 0.76, + "learning_rate": 4.250309055755208e-05, + "loss": 1.1173, + "step": 2470 + }, + { + "epoch": 0.76, + "learning_rate": 4.247430932713278e-05, + "loss": 1.0133, + "step": 2475 + }, + { + "epoch": 0.76, + "learning_rate": 4.2445482741162424e-05, + "loss": 1.027, + "step": 2480 + }, + { + "epoch": 0.76, + "learning_rate": 4.241661087446202e-05, + "loss": 1.018, + "step": 2485 + }, + { + "epoch": 0.77, + "learning_rate": 4.238769380197012e-05, + "loss": 1.1164, + "step": 2490 + }, + { + "epoch": 0.77, + "learning_rate": 4.235873159874261e-05, + "loss": 1.1352, + "step": 2495 + }, + { + "epoch": 0.77, + "learning_rate": 4.232972433995253e-05, + "loss": 1.1062, + "step": 2500 + }, + { + "epoch": 0.77, + "learning_rate": 4.230067210088981e-05, + "loss": 1.1859, + "step": 2505 + }, + { + "epoch": 0.77, + "learning_rate": 4.22715749569612e-05, + "loss": 1.0326, + "step": 2510 + }, + { + "epoch": 0.77, + "learning_rate": 4.224243298368997e-05, + "loss": 0.9888, + "step": 2515 + }, + { + "epoch": 0.78, + "learning_rate": 4.221324625671574e-05, + "loss": 1.0994, + "step": 2520 + }, + { + "epoch": 0.78, + "learning_rate": 4.2184014851794295e-05, + "loss": 0.9874, + "step": 2525 + }, + { + "epoch": 0.78, + "learning_rate": 4.21547388447974e-05, + "loss": 1.0844, + "step": 2530 + }, + { + "epoch": 0.78, + "learning_rate": 4.212541831171256e-05, + "loss": 1.0335, + "step": 2535 + }, + { + "epoch": 0.78, + "learning_rate": 4.2096053328642885e-05, + "loss": 1.0692, + "step": 2540 + }, + { + "epoch": 0.78, + "learning_rate": 4.2066643971806815e-05, + "loss": 1.1647, + "step": 2545 + }, + { + "epoch": 0.78, + "learning_rate": 4.2037190317538e-05, + "loss": 1.0822, + "step": 2550 + }, + { + "epoch": 0.79, + "learning_rate": 4.200769244228505e-05, + "loss": 0.9937, + "step": 2555 + }, + { + "epoch": 0.79, + "learning_rate": 4.197815042261135e-05, + "loss": 0.9944, + "step": 2560 + }, + { + "epoch": 0.79, + "learning_rate": 4.194856433519489e-05, + "loss": 0.9123, + "step": 2565 + }, + { + "epoch": 0.79, + "learning_rate": 4.191893425682801e-05, + "loss": 1.0537, + "step": 2570 + }, + { + "epoch": 0.79, + "learning_rate": 4.188926026441723e-05, + "loss": 1.1063, + "step": 2575 + }, + { + "epoch": 0.79, + "learning_rate": 4.185954243498309e-05, + "loss": 1.0426, + "step": 2580 + }, + { + "epoch": 0.8, + "learning_rate": 4.182978084565986e-05, + "loss": 0.9949, + "step": 2585 + }, + { + "epoch": 0.8, + "learning_rate": 4.1799975573695427e-05, + "loss": 1.111, + "step": 2590 + }, + { + "epoch": 0.8, + "learning_rate": 4.177012669645105e-05, + "loss": 1.0189, + "step": 2595 + }, + { + "epoch": 0.8, + "learning_rate": 4.1740234291401164e-05, + "loss": 1.0445, + "step": 2600 + }, + { + "epoch": 0.8, + "learning_rate": 4.1710298436133186e-05, + "loss": 1.048, + "step": 2605 + }, + { + "epoch": 0.8, + "learning_rate": 4.168031920834731e-05, + "loss": 1.0453, + "step": 2610 + }, + { + "epoch": 0.8, + "learning_rate": 4.165029668585629e-05, + "loss": 0.9769, + "step": 2615 + }, + { + "epoch": 0.81, + "learning_rate": 4.162023094658529e-05, + "loss": 0.9435, + "step": 2620 + }, + { + "epoch": 0.81, + "learning_rate": 4.159012206857162e-05, + "loss": 1.0399, + "step": 2625 + }, + { + "epoch": 0.81, + "learning_rate": 4.1559970129964554e-05, + "loss": 1.115, + "step": 2630 + }, + { + "epoch": 0.81, + "learning_rate": 4.152977520902515e-05, + "loss": 1.0537, + "step": 2635 + }, + { + "epoch": 0.81, + "learning_rate": 4.149953738412602e-05, + "loss": 0.9469, + "step": 2640 + }, + { + "epoch": 0.81, + "learning_rate": 4.1469256733751126e-05, + "loss": 1.0996, + "step": 2645 + }, + { + "epoch": 0.82, + "learning_rate": 4.143893333649561e-05, + "loss": 1.0838, + "step": 2650 + }, + { + "epoch": 0.82, + "learning_rate": 4.140856727106555e-05, + "loss": 0.9408, + "step": 2655 + }, + { + "epoch": 0.82, + "learning_rate": 4.1378158616277775e-05, + "loss": 1.0324, + "step": 2660 + }, + { + "epoch": 0.82, + "learning_rate": 4.1347707451059644e-05, + "loss": 0.9539, + "step": 2665 + }, + { + "epoch": 0.82, + "learning_rate": 4.131721385444888e-05, + "loss": 1.0231, + "step": 2670 + }, + { + "epoch": 0.82, + "learning_rate": 4.128667790559331e-05, + "loss": 1.0918, + "step": 2675 + }, + { + "epoch": 0.82, + "learning_rate": 4.125609968375072e-05, + "loss": 0.9399, + "step": 2680 + }, + { + "epoch": 0.83, + "learning_rate": 4.1225479268288595e-05, + "loss": 0.9065, + "step": 2685 + }, + { + "epoch": 0.83, + "learning_rate": 4.119481673868395e-05, + "loss": 1.0972, + "step": 2690 + }, + { + "epoch": 0.83, + "learning_rate": 4.116411217452308e-05, + "loss": 1.079, + "step": 2695 + }, + { + "epoch": 0.83, + "learning_rate": 4.1133365655501436e-05, + "loss": 0.9836, + "step": 2700 + }, + { + "epoch": 0.83, + "learning_rate": 4.1102577261423315e-05, + "loss": 1.0088, + "step": 2705 + }, + { + "epoch": 0.83, + "learning_rate": 4.107174707220172e-05, + "loss": 1.0253, + "step": 2710 + }, + { + "epoch": 0.84, + "learning_rate": 4.104087516785816e-05, + "loss": 1.0012, + "step": 2715 + }, + { + "epoch": 0.84, + "learning_rate": 4.100996162852236e-05, + "loss": 1.1425, + "step": 2720 + }, + { + "epoch": 0.84, + "learning_rate": 4.0979006534432184e-05, + "loss": 1.1031, + "step": 2725 + }, + { + "epoch": 0.84, + "learning_rate": 4.0948009965933306e-05, + "loss": 0.975, + "step": 2730 + }, + { + "epoch": 0.84, + "learning_rate": 4.091697200347904e-05, + "loss": 0.9988, + "step": 2735 + }, + { + "epoch": 0.84, + "learning_rate": 4.088589272763019e-05, + "loss": 1.073, + "step": 2740 + }, + { + "epoch": 0.84, + "learning_rate": 4.085477221905474e-05, + "loss": 1.0941, + "step": 2745 + }, + { + "epoch": 0.85, + "learning_rate": 4.0823610558527726e-05, + "loss": 1.0501, + "step": 2750 + }, + { + "epoch": 0.85, + "learning_rate": 4.079240782693099e-05, + "loss": 1.0174, + "step": 2755 + }, + { + "epoch": 0.85, + "learning_rate": 4.076116410525297e-05, + "loss": 1.0716, + "step": 2760 + }, + { + "epoch": 0.85, + "learning_rate": 4.0729879474588495e-05, + "loss": 1.1265, + "step": 2765 + }, + { + "epoch": 0.85, + "learning_rate": 4.06985540161386e-05, + "loss": 1.0415, + "step": 2770 + }, + { + "epoch": 0.85, + "learning_rate": 4.0667187811210244e-05, + "loss": 1.0925, + "step": 2775 + }, + { + "epoch": 0.86, + "learning_rate": 4.063578094121619e-05, + "loss": 1.149, + "step": 2780 + }, + { + "epoch": 0.86, + "learning_rate": 4.060433348767473e-05, + "loss": 0.993, + "step": 2785 + }, + { + "epoch": 0.86, + "learning_rate": 4.05728455322095e-05, + "loss": 1.062, + "step": 2790 + }, + { + "epoch": 0.86, + "learning_rate": 4.054131715654924e-05, + "loss": 0.9873, + "step": 2795 + }, + { + "epoch": 0.86, + "learning_rate": 4.050974844252765e-05, + "loss": 1.0654, + "step": 2800 + }, + { + "epoch": 0.86, + "learning_rate": 4.047813947208307e-05, + "loss": 0.9762, + "step": 2805 + }, + { + "epoch": 0.86, + "learning_rate": 4.044649032725836e-05, + "loss": 1.1045, + "step": 2810 + }, + { + "epoch": 0.87, + "learning_rate": 4.041480109020067e-05, + "loss": 0.9958, + "step": 2815 + }, + { + "epoch": 0.87, + "learning_rate": 4.038307184316118e-05, + "loss": 1.1379, + "step": 2820 + }, + { + "epoch": 0.87, + "learning_rate": 4.035130266849494e-05, + "loss": 0.9394, + "step": 2825 + }, + { + "epoch": 0.87, + "learning_rate": 4.0319493648660636e-05, + "loss": 1.0213, + "step": 2830 + }, + { + "epoch": 0.87, + "learning_rate": 4.028764486622033e-05, + "loss": 1.0984, + "step": 2835 + }, + { + "epoch": 0.87, + "learning_rate": 4.0255756403839364e-05, + "loss": 1.166, + "step": 2840 + }, + { + "epoch": 0.88, + "learning_rate": 4.022382834428602e-05, + "loss": 1.0643, + "step": 2845 + }, + { + "epoch": 0.88, + "learning_rate": 4.019186077043136e-05, + "loss": 0.9503, + "step": 2850 + }, + { + "epoch": 0.88, + "learning_rate": 4.015985376524902e-05, + "loss": 0.99, + "step": 2855 + }, + { + "epoch": 0.88, + "learning_rate": 4.012780741181499e-05, + "loss": 1.0513, + "step": 2860 + }, + { + "epoch": 0.88, + "learning_rate": 4.0095721793307376e-05, + "loss": 1.1222, + "step": 2865 + }, + { + "epoch": 0.88, + "learning_rate": 4.006359699300619e-05, + "loss": 1.0624, + "step": 2870 + }, + { + "epoch": 0.88, + "learning_rate": 4.003143309429317e-05, + "loss": 1.1032, + "step": 2875 + }, + { + "epoch": 0.89, + "learning_rate": 3.999923018065151e-05, + "loss": 1.0131, + "step": 2880 + }, + { + "epoch": 0.89, + "learning_rate": 3.996698833566569e-05, + "loss": 1.0246, + "step": 2885 + }, + { + "epoch": 0.89, + "learning_rate": 3.993470764302122e-05, + "loss": 1.1258, + "step": 2890 + }, + { + "epoch": 0.89, + "learning_rate": 3.990238818650446e-05, + "loss": 1.0478, + "step": 2895 + }, + { + "epoch": 0.89, + "learning_rate": 3.9870030050002366e-05, + "loss": 1.0267, + "step": 2900 + }, + { + "epoch": 0.89, + "learning_rate": 3.983763331750231e-05, + "loss": 0.9854, + "step": 2905 + }, + { + "epoch": 0.9, + "learning_rate": 3.980519807309181e-05, + "loss": 1.0284, + "step": 2910 + }, + { + "epoch": 0.9, + "learning_rate": 3.9772724400958386e-05, + "loss": 1.042, + "step": 2915 + }, + { + "epoch": 0.9, + "learning_rate": 3.9740212385389266e-05, + "loss": 0.9958, + "step": 2920 + }, + { + "epoch": 0.9, + "learning_rate": 3.97076621107712e-05, + "loss": 1.0593, + "step": 2925 + }, + { + "epoch": 0.9, + "learning_rate": 3.9675073661590284e-05, + "loss": 0.9493, + "step": 2930 + }, + { + "epoch": 0.9, + "learning_rate": 3.964244712243165e-05, + "loss": 1.058, + "step": 2935 + }, + { + "epoch": 0.9, + "learning_rate": 3.960978257797931e-05, + "loss": 1.142, + "step": 2940 + }, + { + "epoch": 0.91, + "learning_rate": 3.957708011301592e-05, + "loss": 1.0899, + "step": 2945 + }, + { + "epoch": 0.91, + "learning_rate": 3.9544339812422585e-05, + "loss": 1.11, + "step": 2950 + }, + { + "epoch": 0.91, + "learning_rate": 3.9511561761178565e-05, + "loss": 1.0553, + "step": 2955 + }, + { + "epoch": 0.91, + "learning_rate": 3.947874604436115e-05, + "loss": 1.1098, + "step": 2960 + }, + { + "epoch": 0.91, + "learning_rate": 3.944589274714538e-05, + "loss": 1.0096, + "step": 2965 + }, + { + "epoch": 0.91, + "learning_rate": 3.941300195480382e-05, + "loss": 1.0905, + "step": 2970 + }, + { + "epoch": 0.92, + "learning_rate": 3.938007375270637e-05, + "loss": 0.968, + "step": 2975 + }, + { + "epoch": 0.92, + "learning_rate": 3.934710822632004e-05, + "loss": 1.1126, + "step": 2980 + }, + { + "epoch": 0.92, + "learning_rate": 3.931410546120869e-05, + "loss": 0.9895, + "step": 2985 + }, + { + "epoch": 0.92, + "learning_rate": 3.928106554303286e-05, + "loss": 1.0565, + "step": 2990 + }, + { + "epoch": 0.92, + "learning_rate": 3.9247988557549505e-05, + "loss": 1.0716, + "step": 2995 + }, + { + "epoch": 0.92, + "learning_rate": 3.921487459061181e-05, + "loss": 1.1494, + "step": 3000 + }, + { + "epoch": 0.92, + "learning_rate": 3.9181723728168916e-05, + "loss": 1.1171, + "step": 3005 + }, + { + "epoch": 0.93, + "learning_rate": 3.914853605626576e-05, + "loss": 1.0174, + "step": 3010 + }, + { + "epoch": 0.93, + "learning_rate": 3.911531166104282e-05, + "loss": 0.9992, + "step": 3015 + }, + { + "epoch": 0.93, + "learning_rate": 3.908205062873585e-05, + "loss": 1.071, + "step": 3020 + }, + { + "epoch": 0.93, + "learning_rate": 3.904875304567576e-05, + "loss": 1.0802, + "step": 3025 + }, + { + "epoch": 0.93, + "learning_rate": 3.901541899828827e-05, + "loss": 0.9996, + "step": 3030 + }, + { + "epoch": 0.93, + "learning_rate": 3.898204857309379e-05, + "loss": 1.0683, + "step": 3035 + }, + { + "epoch": 0.94, + "learning_rate": 3.8948641856707114e-05, + "loss": 0.9265, + "step": 3040 + }, + { + "epoch": 0.94, + "learning_rate": 3.8915198935837275e-05, + "loss": 1.0768, + "step": 3045 + }, + { + "epoch": 0.94, + "learning_rate": 3.8881719897287225e-05, + "loss": 0.9765, + "step": 3050 + }, + { + "epoch": 0.94, + "learning_rate": 3.88482048279537e-05, + "loss": 1.0002, + "step": 3055 + }, + { + "epoch": 0.94, + "learning_rate": 3.881465381482696e-05, + "loss": 0.993, + "step": 3060 + }, + { + "epoch": 0.94, + "learning_rate": 3.878106694499052e-05, + "loss": 1.0063, + "step": 3065 + }, + { + "epoch": 0.94, + "learning_rate": 3.8747444305621e-05, + "loss": 1.0884, + "step": 3070 + }, + { + "epoch": 0.95, + "learning_rate": 3.8713785983987835e-05, + "loss": 1.0213, + "step": 3075 + }, + { + "epoch": 0.95, + "learning_rate": 3.8680092067453103e-05, + "loss": 1.025, + "step": 3080 + }, + { + "epoch": 0.95, + "learning_rate": 3.864636264347126e-05, + "loss": 1.0172, + "step": 3085 + }, + { + "epoch": 0.95, + "learning_rate": 3.86125977995889e-05, + "loss": 1.1419, + "step": 3090 + }, + { + "epoch": 0.95, + "learning_rate": 3.857879762344458e-05, + "loss": 1.0307, + "step": 3095 + }, + { + "epoch": 0.95, + "learning_rate": 3.854496220276857e-05, + "loss": 1.1093, + "step": 3100 + }, + { + "epoch": 0.96, + "learning_rate": 3.8511091625382585e-05, + "loss": 1.1207, + "step": 3105 + }, + { + "epoch": 0.96, + "learning_rate": 3.847718597919962e-05, + "loss": 1.066, + "step": 3110 + }, + { + "epoch": 0.96, + "learning_rate": 3.844324535222368e-05, + "loss": 0.9794, + "step": 3115 + }, + { + "epoch": 0.96, + "learning_rate": 3.8409269832549565e-05, + "loss": 1.0531, + "step": 3120 + }, + { + "epoch": 0.96, + "learning_rate": 3.8375259508362645e-05, + "loss": 1.0467, + "step": 3125 + }, + { + "epoch": 0.96, + "learning_rate": 3.834121446793864e-05, + "loss": 1.0779, + "step": 3130 + }, + { + "epoch": 0.96, + "learning_rate": 3.830713479964335e-05, + "loss": 1.123, + "step": 3135 + }, + { + "epoch": 0.97, + "learning_rate": 3.827302059193247e-05, + "loss": 0.9617, + "step": 3140 + }, + { + "epoch": 0.97, + "learning_rate": 3.823887193335134e-05, + "loss": 1.0266, + "step": 3145 + }, + { + "epoch": 0.97, + "learning_rate": 3.820468891253473e-05, + "loss": 1.0904, + "step": 3150 + }, + { + "epoch": 0.97, + "learning_rate": 3.817047161820658e-05, + "loss": 1.0528, + "step": 3155 + }, + { + "epoch": 0.97, + "learning_rate": 3.81362201391798e-05, + "loss": 0.9839, + "step": 3160 + }, + { + "epoch": 0.97, + "learning_rate": 3.810193456435603e-05, + "loss": 1.031, + "step": 3165 + }, + { + "epoch": 0.98, + "learning_rate": 3.806761498272541e-05, + "loss": 1.0493, + "step": 3170 + }, + { + "epoch": 0.98, + "learning_rate": 3.803326148336633e-05, + "loss": 1.0416, + "step": 3175 + }, + { + "epoch": 0.98, + "learning_rate": 3.799887415544523e-05, + "loss": 1.075, + "step": 3180 + }, + { + "epoch": 0.98, + "learning_rate": 3.796445308821636e-05, + "loss": 1.0675, + "step": 3185 + }, + { + "epoch": 0.98, + "learning_rate": 3.7929998371021524e-05, + "loss": 1.0411, + "step": 3190 + }, + { + "epoch": 0.98, + "learning_rate": 3.789551009328987e-05, + "loss": 1.0156, + "step": 3195 + }, + { + "epoch": 0.98, + "learning_rate": 3.786098834453766e-05, + "loss": 1.0797, + "step": 3200 + }, + { + "epoch": 0.99, + "learning_rate": 3.782643321436805e-05, + "loss": 0.9287, + "step": 3205 + }, + { + "epoch": 0.99, + "learning_rate": 3.7791844792470805e-05, + "loss": 1.1242, + "step": 3210 + }, + { + "epoch": 0.99, + "learning_rate": 3.7757223168622116e-05, + "loss": 0.9338, + "step": 3215 + }, + { + "epoch": 0.99, + "learning_rate": 3.7722568432684354e-05, + "loss": 1.0109, + "step": 3220 + }, + { + "epoch": 0.99, + "learning_rate": 3.768788067460583e-05, + "loss": 1.1003, + "step": 3225 + }, + { + "epoch": 0.99, + "learning_rate": 3.765315998442058e-05, + "loss": 1.1123, + "step": 3230 + }, + { + "epoch": 1.0, + "learning_rate": 3.761840645224809e-05, + "loss": 0.9972, + "step": 3235 + }, + { + "epoch": 1.0, + "learning_rate": 3.758362016829311e-05, + "loss": 1.0044, + "step": 3240 + }, + { + "epoch": 1.0, + "learning_rate": 3.7548801222845395e-05, + "loss": 1.0647, + "step": 3245 + }, + { + "epoch": 1.0, + "learning_rate": 3.751394970627947e-05, + "loss": 1.0366, + "step": 3250 + }, + { + "epoch": 1.0, + "learning_rate": 3.747906570905441e-05, + "loss": 1.0976, + "step": 3255 + }, + { + "epoch": 1.0, + "learning_rate": 3.7444149321713586e-05, + "loss": 1.0069, + "step": 3260 + }, + { + "epoch": 1.0, + "learning_rate": 3.7409200634884426e-05, + "loss": 0.9342, + "step": 3265 + }, + { + "epoch": 1.01, + "learning_rate": 3.7374219739278224e-05, + "loss": 1.0361, + "step": 3270 + }, + { + "epoch": 1.01, + "learning_rate": 3.733920672568986e-05, + "loss": 0.9792, + "step": 3275 + }, + { + "epoch": 1.01, + "learning_rate": 3.730416168499757e-05, + "loss": 1.1064, + "step": 3280 + }, + { + "epoch": 1.01, + "learning_rate": 3.726908470816273e-05, + "loss": 1.0247, + "step": 3285 + }, + { + "epoch": 1.01, + "learning_rate": 3.723397588622961e-05, + "loss": 1.104, + "step": 3290 + }, + { + "epoch": 1.01, + "learning_rate": 3.719883531032512e-05, + "loss": 1.059, + "step": 3295 + }, + { + "epoch": 1.02, + "learning_rate": 3.716366307165859e-05, + "loss": 1.0369, + "step": 3300 + }, + { + "epoch": 1.02, + "learning_rate": 3.7128459261521554e-05, + "loss": 1.017, + "step": 3305 + }, + { + "epoch": 1.02, + "learning_rate": 3.7093223971287465e-05, + "loss": 1.118, + "step": 3310 + }, + { + "epoch": 1.02, + "learning_rate": 3.7057957292411514e-05, + "loss": 0.9885, + "step": 3315 + }, + { + "epoch": 1.02, + "learning_rate": 3.702265931643033e-05, + "loss": 1.0737, + "step": 3320 + }, + { + "epoch": 1.02, + "learning_rate": 3.6987330134961804e-05, + "loss": 1.0382, + "step": 3325 + }, + { + "epoch": 1.02, + "learning_rate": 3.695196983970481e-05, + "loss": 0.971, + "step": 3330 + }, + { + "epoch": 1.03, + "learning_rate": 3.691657852243896e-05, + "loss": 0.904, + "step": 3335 + }, + { + "epoch": 1.03, + "learning_rate": 3.688115627502444e-05, + "loss": 1.1477, + "step": 3340 + }, + { + "epoch": 1.03, + "learning_rate": 3.684570318940165e-05, + "loss": 1.0211, + "step": 3345 + }, + { + "epoch": 1.03, + "learning_rate": 3.681021935759108e-05, + "loss": 1.0304, + "step": 3350 + }, + { + "epoch": 1.03, + "learning_rate": 3.677470487169301e-05, + "loss": 1.0198, + "step": 3355 + }, + { + "epoch": 1.03, + "learning_rate": 3.673915982388727e-05, + "loss": 0.9647, + "step": 3360 + }, + { + "epoch": 1.04, + "learning_rate": 3.670358430643303e-05, + "loss": 0.9567, + "step": 3365 + }, + { + "epoch": 1.04, + "learning_rate": 3.6667978411668565e-05, + "loss": 1.0099, + "step": 3370 + }, + { + "epoch": 1.04, + "learning_rate": 3.6632342232010943e-05, + "loss": 0.9998, + "step": 3375 + }, + { + "epoch": 1.04, + "learning_rate": 3.6596675859955885e-05, + "loss": 1.0787, + "step": 3380 + }, + { + "epoch": 1.04, + "learning_rate": 3.656097938807746e-05, + "loss": 1.0247, + "step": 3385 + }, + { + "epoch": 1.04, + "learning_rate": 3.6525252909027874e-05, + "loss": 1.1399, + "step": 3390 + }, + { + "epoch": 1.04, + "learning_rate": 3.6489496515537204e-05, + "loss": 1.0802, + "step": 3395 + }, + { + "epoch": 1.05, + "learning_rate": 3.6453710300413164e-05, + "loss": 1.1163, + "step": 3400 + }, + { + "epoch": 1.05, + "learning_rate": 3.64178943565409e-05, + "loss": 1.0174, + "step": 3405 + }, + { + "epoch": 1.05, + "learning_rate": 3.63820487768827e-05, + "loss": 1.0228, + "step": 3410 + }, + { + "epoch": 1.05, + "learning_rate": 3.6346173654477775e-05, + "loss": 1.1291, + "step": 3415 + }, + { + "epoch": 1.05, + "learning_rate": 3.631026908244204e-05, + "loss": 1.1014, + "step": 3420 + }, + { + "epoch": 1.05, + "learning_rate": 3.6274335153967796e-05, + "loss": 1.0247, + "step": 3425 + }, + { + "epoch": 1.06, + "learning_rate": 3.62383719623236e-05, + "loss": 1.0044, + "step": 3430 + }, + { + "epoch": 1.06, + "learning_rate": 3.620237960085392e-05, + "loss": 1.0267, + "step": 3435 + }, + { + "epoch": 1.06, + "learning_rate": 3.616635816297895e-05, + "loss": 1.1004, + "step": 3440 + }, + { + "epoch": 1.06, + "learning_rate": 3.613030774219436e-05, + "loss": 1.0813, + "step": 3445 + }, + { + "epoch": 1.06, + "learning_rate": 3.6094228432071046e-05, + "loss": 0.9223, + "step": 3450 + }, + { + "epoch": 1.06, + "learning_rate": 3.605812032625487e-05, + "loss": 0.9958, + "step": 3455 + }, + { + "epoch": 1.06, + "learning_rate": 3.602198351846647e-05, + "loss": 1.0147, + "step": 3460 + }, + { + "epoch": 1.07, + "learning_rate": 3.5985818102500934e-05, + "loss": 1.0373, + "step": 3465 + }, + { + "epoch": 1.07, + "learning_rate": 3.594962417222765e-05, + "loss": 1.0289, + "step": 3470 + }, + { + "epoch": 1.07, + "learning_rate": 3.591340182158998e-05, + "loss": 1.0383, + "step": 3475 + }, + { + "epoch": 1.07, + "learning_rate": 3.587715114460508e-05, + "loss": 1.0278, + "step": 3480 + }, + { + "epoch": 1.07, + "learning_rate": 3.584087223536362e-05, + "loss": 1.097, + "step": 3485 + }, + { + "epoch": 1.07, + "learning_rate": 3.580456518802954e-05, + "loss": 1.1521, + "step": 3490 + }, + { + "epoch": 1.08, + "learning_rate": 3.576823009683982e-05, + "loss": 0.9666, + "step": 3495 + }, + { + "epoch": 1.08, + "learning_rate": 3.573186705610423e-05, + "loss": 1.0488, + "step": 3500 + }, + { + "epoch": 1.08, + "learning_rate": 3.5695476160205086e-05, + "loss": 1.0732, + "step": 3505 + }, + { + "epoch": 1.08, + "learning_rate": 3.5659057503596996e-05, + "loss": 1.0906, + "step": 3510 + }, + { + "epoch": 1.08, + "learning_rate": 3.562261118080664e-05, + "loss": 1.0228, + "step": 3515 + }, + { + "epoch": 1.08, + "learning_rate": 3.558613728643249e-05, + "loss": 1.0633, + "step": 3520 + }, + { + "epoch": 1.08, + "learning_rate": 3.554963591514457e-05, + "loss": 1.0465, + "step": 3525 + }, + { + "epoch": 1.09, + "learning_rate": 3.551310716168427e-05, + "loss": 1.0691, + "step": 3530 + }, + { + "epoch": 1.09, + "learning_rate": 3.5476551120863996e-05, + "loss": 0.9574, + "step": 3535 + }, + { + "epoch": 1.09, + "learning_rate": 3.5439967887567014e-05, + "loss": 1.0412, + "step": 3540 + }, + { + "epoch": 1.09, + "learning_rate": 3.5403357556747156e-05, + "loss": 0.9939, + "step": 3545 + }, + { + "epoch": 1.09, + "learning_rate": 3.53667202234286e-05, + "loss": 0.9961, + "step": 3550 + }, + { + "epoch": 1.09, + "learning_rate": 3.533005598270559e-05, + "loss": 1.0105, + "step": 3555 + }, + { + "epoch": 1.1, + "learning_rate": 3.5293364929742236e-05, + "loss": 1.0982, + "step": 3560 + }, + { + "epoch": 1.1, + "learning_rate": 3.525664715977222e-05, + "loss": 1.0937, + "step": 3565 + }, + { + "epoch": 1.1, + "learning_rate": 3.521990276809856e-05, + "loss": 0.9601, + "step": 3570 + }, + { + "epoch": 1.1, + "learning_rate": 3.5183131850093416e-05, + "loss": 1.0575, + "step": 3575 + }, + { + "epoch": 1.1, + "learning_rate": 3.5146334501197756e-05, + "loss": 1.0227, + "step": 3580 + }, + { + "epoch": 1.1, + "learning_rate": 3.510951081692116e-05, + "loss": 0.9877, + "step": 3585 + }, + { + "epoch": 1.1, + "learning_rate": 3.507266089284157e-05, + "loss": 1.0541, + "step": 3590 + }, + { + "epoch": 1.11, + "learning_rate": 3.5035784824605037e-05, + "loss": 1.0163, + "step": 3595 + }, + { + "epoch": 1.11, + "learning_rate": 3.499888270792548e-05, + "loss": 0.9479, + "step": 3600 + }, + { + "epoch": 1.11, + "learning_rate": 3.496195463858439e-05, + "loss": 0.9669, + "step": 3605 + }, + { + "epoch": 1.11, + "learning_rate": 3.492500071243066e-05, + "loss": 1.0112, + "step": 3610 + }, + { + "epoch": 1.11, + "learning_rate": 3.488802102538027e-05, + "loss": 0.9666, + "step": 3615 + }, + { + "epoch": 1.11, + "learning_rate": 3.4851015673416086e-05, + "loss": 1.0645, + "step": 3620 + }, + { + "epoch": 1.12, + "learning_rate": 3.481398475258757e-05, + "loss": 1.0369, + "step": 3625 + }, + { + "epoch": 1.12, + "learning_rate": 3.477692835901056e-05, + "loss": 1.0592, + "step": 3630 + }, + { + "epoch": 1.12, + "learning_rate": 3.473984658886701e-05, + "loss": 0.9238, + "step": 3635 + }, + { + "epoch": 1.12, + "learning_rate": 3.4702739538404725e-05, + "loss": 1.0055, + "step": 3640 + }, + { + "epoch": 1.12, + "learning_rate": 3.466560730393714e-05, + "loss": 1.097, + "step": 3645 + }, + { + "epoch": 1.12, + "learning_rate": 3.462844998184307e-05, + "loss": 1.0496, + "step": 3650 + }, + { + "epoch": 1.12, + "learning_rate": 3.459126766856641e-05, + "loss": 1.0514, + "step": 3655 + }, + { + "epoch": 1.13, + "learning_rate": 3.455406046061595e-05, + "loss": 0.8936, + "step": 3660 + }, + { + "epoch": 1.13, + "learning_rate": 3.45168284545651e-05, + "loss": 1.0003, + "step": 3665 + }, + { + "epoch": 1.13, + "learning_rate": 3.4479571747051606e-05, + "loss": 1.0952, + "step": 3670 + }, + { + "epoch": 1.13, + "learning_rate": 3.444229043477735e-05, + "loss": 1.069, + "step": 3675 + }, + { + "epoch": 1.13, + "learning_rate": 3.440498461450806e-05, + "loss": 1.0353, + "step": 3680 + }, + { + "epoch": 1.13, + "learning_rate": 3.436765438307311e-05, + "loss": 0.9623, + "step": 3685 + }, + { + "epoch": 1.14, + "learning_rate": 3.43302998373652e-05, + "loss": 1.0058, + "step": 3690 + }, + { + "epoch": 1.14, + "learning_rate": 3.4292921074340154e-05, + "loss": 1.0656, + "step": 3695 + }, + { + "epoch": 1.14, + "learning_rate": 3.4255518191016636e-05, + "loss": 0.9749, + "step": 3700 + }, + { + "epoch": 1.14, + "learning_rate": 3.421809128447595e-05, + "loss": 1.0356, + "step": 3705 + }, + { + "epoch": 1.14, + "learning_rate": 3.4180640451861725e-05, + "loss": 1.1173, + "step": 3710 + }, + { + "epoch": 1.14, + "learning_rate": 3.41431657903797e-05, + "loss": 0.9666, + "step": 3715 + }, + { + "epoch": 1.14, + "learning_rate": 3.410566739729746e-05, + "loss": 1.0398, + "step": 3720 + }, + { + "epoch": 1.15, + "learning_rate": 3.406814536994419e-05, + "loss": 1.1342, + "step": 3725 + }, + { + "epoch": 1.15, + "learning_rate": 3.403059980571042e-05, + "loss": 1.1129, + "step": 3730 + }, + { + "epoch": 1.15, + "learning_rate": 3.3993030802047786e-05, + "loss": 0.9932, + "step": 3735 + }, + { + "epoch": 1.15, + "learning_rate": 3.3955438456468726e-05, + "loss": 0.9597, + "step": 3740 + }, + { + "epoch": 1.15, + "learning_rate": 3.3917822866546285e-05, + "loss": 1.0323, + "step": 3745 + }, + { + "epoch": 1.15, + "learning_rate": 3.388018412991385e-05, + "loss": 1.0008, + "step": 3750 + }, + { + "epoch": 1.16, + "learning_rate": 3.3842522344264866e-05, + "loss": 0.9892, + "step": 3755 + }, + { + "epoch": 1.16, + "learning_rate": 3.3804837607352615e-05, + "loss": 1.0909, + "step": 3760 + }, + { + "epoch": 1.16, + "learning_rate": 3.376713001698995e-05, + "loss": 0.9187, + "step": 3765 + }, + { + "epoch": 1.16, + "learning_rate": 3.372939967104904e-05, + "loss": 1.0141, + "step": 3770 + }, + { + "epoch": 1.16, + "learning_rate": 3.369164666746111e-05, + "loss": 0.9919, + "step": 3775 + }, + { + "epoch": 1.16, + "learning_rate": 3.365387110421621e-05, + "loss": 1.1429, + "step": 3780 + }, + { + "epoch": 1.16, + "learning_rate": 3.3616073079362926e-05, + "loss": 0.9625, + "step": 3785 + }, + { + "epoch": 1.17, + "learning_rate": 3.357825269100817e-05, + "loss": 1.1452, + "step": 3790 + }, + { + "epoch": 1.17, + "learning_rate": 3.354041003731686e-05, + "loss": 1.0696, + "step": 3795 + }, + { + "epoch": 1.17, + "learning_rate": 3.350254521651174e-05, + "loss": 1.0936, + "step": 3800 + }, + { + "epoch": 1.17, + "learning_rate": 3.346465832687309e-05, + "loss": 1.021, + "step": 3805 + }, + { + "epoch": 1.17, + "learning_rate": 3.342674946673844e-05, + "loss": 1.0667, + "step": 3810 + }, + { + "epoch": 1.17, + "learning_rate": 3.338881873450237e-05, + "loss": 1.0189, + "step": 3815 + }, + { + "epoch": 1.18, + "learning_rate": 3.335086622861624e-05, + "loss": 0.9894, + "step": 3820 + }, + { + "epoch": 1.18, + "learning_rate": 3.331289204758789e-05, + "loss": 1.0653, + "step": 3825 + }, + { + "epoch": 1.18, + "learning_rate": 3.327489628998146e-05, + "loss": 0.986, + "step": 3830 + }, + { + "epoch": 1.18, + "learning_rate": 3.323687905441705e-05, + "loss": 1.0346, + "step": 3835 + }, + { + "epoch": 1.18, + "learning_rate": 3.319884043957055e-05, + "loss": 1.0292, + "step": 3840 + }, + { + "epoch": 1.18, + "learning_rate": 3.31607805441733e-05, + "loss": 1.0066, + "step": 3845 + }, + { + "epoch": 1.18, + "learning_rate": 3.312269946701191e-05, + "loss": 1.0579, + "step": 3850 + }, + { + "epoch": 1.19, + "learning_rate": 3.308459730692796e-05, + "loss": 0.9057, + "step": 3855 + }, + { + "epoch": 1.19, + "learning_rate": 3.304647416281773e-05, + "loss": 1.0818, + "step": 3860 + }, + { + "epoch": 1.19, + "learning_rate": 3.300833013363199e-05, + "loss": 1.0239, + "step": 3865 + }, + { + "epoch": 1.19, + "learning_rate": 3.2970165318375706e-05, + "loss": 1.0086, + "step": 3870 + }, + { + "epoch": 1.19, + "learning_rate": 3.2931979816107807e-05, + "loss": 0.9671, + "step": 3875 + }, + { + "epoch": 1.19, + "learning_rate": 3.289377372594089e-05, + "loss": 1.0023, + "step": 3880 + }, + { + "epoch": 1.2, + "learning_rate": 3.285554714704103e-05, + "loss": 0.9916, + "step": 3885 + }, + { + "epoch": 1.2, + "learning_rate": 3.281730017862744e-05, + "loss": 1.0509, + "step": 3890 + }, + { + "epoch": 1.2, + "learning_rate": 3.277903291997229e-05, + "loss": 1.1151, + "step": 3895 + }, + { + "epoch": 1.2, + "learning_rate": 3.27407454704004e-05, + "loss": 1.0784, + "step": 3900 + }, + { + "epoch": 1.2, + "learning_rate": 3.2702437929288976e-05, + "loss": 1.0823, + "step": 3905 + }, + { + "epoch": 1.2, + "learning_rate": 3.2664110396067394e-05, + "loss": 1.0827, + "step": 3910 + }, + { + "epoch": 1.2, + "learning_rate": 3.262576297021695e-05, + "loss": 1.0185, + "step": 3915 + }, + { + "epoch": 1.21, + "learning_rate": 3.258739575127049e-05, + "loss": 1.076, + "step": 3920 + }, + { + "epoch": 1.21, + "learning_rate": 3.2549008838812335e-05, + "loss": 1.0609, + "step": 3925 + }, + { + "epoch": 1.21, + "learning_rate": 3.251060233247784e-05, + "loss": 1.007, + "step": 3930 + }, + { + "epoch": 1.21, + "learning_rate": 3.247217633195326e-05, + "loss": 1.1076, + "step": 3935 + }, + { + "epoch": 1.21, + "learning_rate": 3.243373093697544e-05, + "loss": 1.0471, + "step": 3940 + }, + { + "epoch": 1.21, + "learning_rate": 3.239526624733156e-05, + "loss": 0.9854, + "step": 3945 + }, + { + "epoch": 1.22, + "learning_rate": 3.2356782362858864e-05, + "loss": 1.0363, + "step": 3950 + }, + { + "epoch": 1.22, + "learning_rate": 3.231827938344447e-05, + "loss": 0.9353, + "step": 3955 + }, + { + "epoch": 1.22, + "learning_rate": 3.227975740902499e-05, + "loss": 1.0451, + "step": 3960 + }, + { + "epoch": 1.22, + "learning_rate": 3.224121653958639e-05, + "loss": 1.0352, + "step": 3965 + }, + { + "epoch": 1.22, + "learning_rate": 3.220265687516366e-05, + "loss": 1.0196, + "step": 3970 + }, + { + "epoch": 1.22, + "learning_rate": 3.2164078515840563e-05, + "loss": 1.1134, + "step": 3975 + }, + { + "epoch": 1.22, + "learning_rate": 3.21254815617494e-05, + "loss": 1.0645, + "step": 3980 + }, + { + "epoch": 1.23, + "learning_rate": 3.2086866113070746e-05, + "loss": 1.0461, + "step": 3985 + }, + { + "epoch": 1.23, + "learning_rate": 3.204823227003316e-05, + "loss": 1.0053, + "step": 3990 + }, + { + "epoch": 1.23, + "learning_rate": 3.200958013291294e-05, + "loss": 1.0568, + "step": 3995 + }, + { + "epoch": 1.23, + "learning_rate": 3.197090980203388e-05, + "loss": 1.0834, + "step": 4000 + }, + { + "epoch": 1.23, + "learning_rate": 3.1932221377767014e-05, + "loss": 1.0918, + "step": 4005 + }, + { + "epoch": 1.23, + "learning_rate": 3.189351496053031e-05, + "loss": 1.0213, + "step": 4010 + }, + { + "epoch": 1.24, + "learning_rate": 3.185479065078842e-05, + "loss": 1.0997, + "step": 4015 + }, + { + "epoch": 1.24, + "learning_rate": 3.1816048549052516e-05, + "loss": 1.045, + "step": 4020 + }, + { + "epoch": 1.24, + "learning_rate": 3.177728875587986e-05, + "loss": 1.0377, + "step": 4025 + }, + { + "epoch": 1.24, + "learning_rate": 3.173851137187368e-05, + "loss": 1.094, + "step": 4030 + }, + { + "epoch": 1.24, + "learning_rate": 3.169971649768286e-05, + "loss": 1.0368, + "step": 4035 + }, + { + "epoch": 1.24, + "learning_rate": 3.166090423400167e-05, + "loss": 0.9715, + "step": 4040 + }, + { + "epoch": 1.24, + "learning_rate": 3.162207468156952e-05, + "loss": 0.9931, + "step": 4045 + }, + { + "epoch": 1.25, + "learning_rate": 3.15832279411707e-05, + "loss": 1.023, + "step": 4050 + }, + { + "epoch": 1.25, + "learning_rate": 3.154436411363409e-05, + "loss": 1.0863, + "step": 4055 + }, + { + "epoch": 1.25, + "learning_rate": 3.150548329983295e-05, + "loss": 1.0059, + "step": 4060 + }, + { + "epoch": 1.25, + "learning_rate": 3.1466585600684616e-05, + "loss": 1.0329, + "step": 4065 + }, + { + "epoch": 1.25, + "learning_rate": 3.142767111715024e-05, + "loss": 1.1314, + "step": 4070 + }, + { + "epoch": 1.25, + "learning_rate": 3.138873995023456e-05, + "loss": 1.0827, + "step": 4075 + }, + { + "epoch": 1.26, + "learning_rate": 3.1349792200985605e-05, + "loss": 1.1005, + "step": 4080 + }, + { + "epoch": 1.26, + "learning_rate": 3.1310827970494436e-05, + "loss": 0.9919, + "step": 4085 + }, + { + "epoch": 1.26, + "learning_rate": 3.127184735989491e-05, + "loss": 1.0082, + "step": 4090 + }, + { + "epoch": 1.26, + "learning_rate": 3.1232850470363386e-05, + "loss": 0.9109, + "step": 4095 + }, + { + "epoch": 1.26, + "learning_rate": 3.1193837403118484e-05, + "loss": 0.9538, + "step": 4100 + }, + { + "epoch": 1.26, + "learning_rate": 3.1154808259420805e-05, + "loss": 1.0285, + "step": 4105 + }, + { + "epoch": 1.26, + "learning_rate": 3.111576314057268e-05, + "loss": 1.083, + "step": 4110 + }, + { + "epoch": 1.27, + "learning_rate": 3.1076702147917925e-05, + "loss": 1.0394, + "step": 4115 + }, + { + "epoch": 1.27, + "learning_rate": 3.103762538284153e-05, + "loss": 1.1115, + "step": 4120 + }, + { + "epoch": 1.27, + "learning_rate": 3.099853294676942e-05, + "loss": 1.0312, + "step": 4125 + }, + { + "epoch": 1.27, + "learning_rate": 3.095942494116824e-05, + "loss": 0.9684, + "step": 4130 + }, + { + "epoch": 1.27, + "learning_rate": 3.0920301467544974e-05, + "loss": 1.0816, + "step": 4135 + }, + { + "epoch": 1.27, + "learning_rate": 3.088116262744682e-05, + "loss": 1.1177, + "step": 4140 + }, + { + "epoch": 1.28, + "learning_rate": 3.084200852246083e-05, + "loss": 0.9315, + "step": 4145 + }, + { + "epoch": 1.28, + "learning_rate": 3.0802839254213686e-05, + "loss": 1.0066, + "step": 4150 + }, + { + "epoch": 1.28, + "learning_rate": 3.076365492437142e-05, + "loss": 1.0698, + "step": 4155 + }, + { + "epoch": 1.28, + "learning_rate": 3.072445563463915e-05, + "loss": 1.13, + "step": 4160 + }, + { + "epoch": 1.28, + "learning_rate": 3.0685241486760854e-05, + "loss": 1.0464, + "step": 4165 + }, + { + "epoch": 1.28, + "learning_rate": 3.0646012582519036e-05, + "loss": 0.9566, + "step": 4170 + }, + { + "epoch": 1.28, + "learning_rate": 3.0606769023734536e-05, + "loss": 1.0185, + "step": 4175 + }, + { + "epoch": 1.29, + "learning_rate": 3.0567510912266196e-05, + "loss": 1.0402, + "step": 4180 + }, + { + "epoch": 1.29, + "learning_rate": 3.052823835001067e-05, + "loss": 1.004, + "step": 4185 + }, + { + "epoch": 1.29, + "learning_rate": 3.048895143890208e-05, + "loss": 0.9124, + "step": 4190 + }, + { + "epoch": 1.29, + "learning_rate": 3.0449650280911813e-05, + "loss": 1.0156, + "step": 4195 + }, + { + "epoch": 1.29, + "learning_rate": 3.0410334978048245e-05, + "loss": 1.0107, + "step": 4200 + }, + { + "epoch": 1.29, + "learning_rate": 3.0371005632356436e-05, + "loss": 1.1228, + "step": 4205 + }, + { + "epoch": 1.3, + "learning_rate": 3.033166234591792e-05, + "loss": 1.0834, + "step": 4210 + }, + { + "epoch": 1.3, + "learning_rate": 3.0292305220850402e-05, + "loss": 0.9566, + "step": 4215 + }, + { + "epoch": 1.3, + "learning_rate": 3.0252934359307516e-05, + "loss": 0.9992, + "step": 4220 + }, + { + "epoch": 1.3, + "learning_rate": 3.021354986347854e-05, + "loss": 1.0253, + "step": 4225 + }, + { + "epoch": 1.3, + "learning_rate": 3.017415183558814e-05, + "loss": 0.9846, + "step": 4230 + }, + { + "epoch": 1.3, + "learning_rate": 3.0134740377896113e-05, + "loss": 1.026, + "step": 4235 + }, + { + "epoch": 1.3, + "learning_rate": 3.0095315592697126e-05, + "loss": 1.0863, + "step": 4240 + }, + { + "epoch": 1.31, + "learning_rate": 3.0055877582320398e-05, + "loss": 1.0666, + "step": 4245 + }, + { + "epoch": 1.31, + "learning_rate": 3.0016426449129497e-05, + "loss": 1.0011, + "step": 4250 + }, + { + "epoch": 1.31, + "learning_rate": 2.9976962295522088e-05, + "loss": 1.0669, + "step": 4255 + }, + { + "epoch": 1.31, + "learning_rate": 2.9937485223929567e-05, + "loss": 1.083, + "step": 4260 + }, + { + "epoch": 1.31, + "learning_rate": 2.98979953368169e-05, + "loss": 1.0338, + "step": 4265 + }, + { + "epoch": 1.31, + "learning_rate": 2.98584927366823e-05, + "loss": 1.0149, + "step": 4270 + }, + { + "epoch": 1.32, + "learning_rate": 2.9818977526057002e-05, + "loss": 0.9602, + "step": 4275 + }, + { + "epoch": 1.32, + "learning_rate": 2.9779449807504937e-05, + "loss": 1.0372, + "step": 4280 + }, + { + "epoch": 1.32, + "learning_rate": 2.973990968362252e-05, + "loss": 1.0328, + "step": 4285 + }, + { + "epoch": 1.32, + "learning_rate": 2.970035725703837e-05, + "loss": 1.1233, + "step": 4290 + }, + { + "epoch": 1.32, + "learning_rate": 2.966079263041303e-05, + "loss": 0.9428, + "step": 4295 + }, + { + "epoch": 1.32, + "learning_rate": 2.96212159064387e-05, + "loss": 1.0317, + "step": 4300 + }, + { + "epoch": 1.32, + "learning_rate": 2.9581627187838994e-05, + "loss": 1.0305, + "step": 4305 + }, + { + "epoch": 1.33, + "learning_rate": 2.9542026577368666e-05, + "loss": 0.9929, + "step": 4310 + }, + { + "epoch": 1.33, + "learning_rate": 2.950241417781331e-05, + "loss": 1.0647, + "step": 4315 + }, + { + "epoch": 1.33, + "learning_rate": 2.9462790091989128e-05, + "loss": 1.0141, + "step": 4320 + }, + { + "epoch": 1.33, + "learning_rate": 2.942315442274269e-05, + "loss": 0.933, + "step": 4325 + }, + { + "epoch": 1.33, + "learning_rate": 2.9383507272950567e-05, + "loss": 1.0761, + "step": 4330 + }, + { + "epoch": 1.33, + "learning_rate": 2.9343848745519176e-05, + "loss": 1.0689, + "step": 4335 + }, + { + "epoch": 1.34, + "learning_rate": 2.930417894338446e-05, + "loss": 1.0712, + "step": 4340 + }, + { + "epoch": 1.34, + "learning_rate": 2.92644979695116e-05, + "loss": 1.0525, + "step": 4345 + }, + { + "epoch": 1.34, + "learning_rate": 2.9224805926894817e-05, + "loss": 0.9413, + "step": 4350 + }, + { + "epoch": 1.34, + "learning_rate": 2.9185102918557e-05, + "loss": 0.9872, + "step": 4355 + }, + { + "epoch": 1.34, + "learning_rate": 2.914538904754957e-05, + "loss": 1.0017, + "step": 4360 + }, + { + "epoch": 1.34, + "learning_rate": 2.91056644169521e-05, + "loss": 1.0496, + "step": 4365 + }, + { + "epoch": 1.34, + "learning_rate": 2.9065929129872094e-05, + "loss": 0.9497, + "step": 4370 + }, + { + "epoch": 1.35, + "learning_rate": 2.902618328944472e-05, + "loss": 1.0561, + "step": 4375 + }, + { + "epoch": 1.35, + "learning_rate": 2.8986426998832543e-05, + "loss": 0.982, + "step": 4380 + }, + { + "epoch": 1.35, + "learning_rate": 2.8946660361225246e-05, + "loss": 1.0784, + "step": 4385 + }, + { + "epoch": 1.35, + "learning_rate": 2.8906883479839377e-05, + "loss": 1.0599, + "step": 4390 + }, + { + "epoch": 1.35, + "learning_rate": 2.886709645791805e-05, + "loss": 0.9732, + "step": 4395 + }, + { + "epoch": 1.35, + "learning_rate": 2.8827299398730722e-05, + "loss": 1.1354, + "step": 4400 + }, + { + "epoch": 1.36, + "learning_rate": 2.8787492405572904e-05, + "loss": 1.0808, + "step": 4405 + }, + { + "epoch": 1.36, + "learning_rate": 2.8747675581765865e-05, + "loss": 0.9735, + "step": 4410 + }, + { + "epoch": 1.36, + "learning_rate": 2.870784903065642e-05, + "loss": 1.0123, + "step": 4415 + }, + { + "epoch": 1.36, + "learning_rate": 2.8668012855616617e-05, + "loss": 1.0028, + "step": 4420 + }, + { + "epoch": 1.36, + "learning_rate": 2.8628167160043478e-05, + "loss": 1.0489, + "step": 4425 + }, + { + "epoch": 1.36, + "learning_rate": 2.8588312047358752e-05, + "loss": 0.9024, + "step": 4430 + }, + { + "epoch": 1.36, + "learning_rate": 2.854844762100861e-05, + "loss": 1.0294, + "step": 4435 + }, + { + "epoch": 1.37, + "learning_rate": 2.8508573984463426e-05, + "loss": 1.0046, + "step": 4440 + }, + { + "epoch": 1.37, + "learning_rate": 2.846869124121745e-05, + "loss": 1.0123, + "step": 4445 + }, + { + "epoch": 1.37, + "learning_rate": 2.8428799494788582e-05, + "loss": 1.0837, + "step": 4450 + }, + { + "epoch": 1.37, + "learning_rate": 2.8388898848718092e-05, + "loss": 0.9723, + "step": 4455 + }, + { + "epoch": 1.37, + "learning_rate": 2.834898940657036e-05, + "loss": 1.0546, + "step": 4460 + }, + { + "epoch": 1.37, + "learning_rate": 2.830907127193256e-05, + "loss": 1.0112, + "step": 4465 + }, + { + "epoch": 1.38, + "learning_rate": 2.8269144548414467e-05, + "loss": 0.9805, + "step": 4470 + }, + { + "epoch": 1.38, + "learning_rate": 2.8229209339648132e-05, + "loss": 0.9829, + "step": 4475 + }, + { + "epoch": 1.38, + "learning_rate": 2.8189265749287624e-05, + "loss": 1.0461, + "step": 4480 + }, + { + "epoch": 1.38, + "learning_rate": 2.8149313881008795e-05, + "loss": 0.9407, + "step": 4485 + }, + { + "epoch": 1.38, + "learning_rate": 2.810935383850894e-05, + "loss": 1.1101, + "step": 4490 + }, + { + "epoch": 1.38, + "learning_rate": 2.8069385725506604e-05, + "loss": 0.9011, + "step": 4495 + }, + { + "epoch": 1.38, + "learning_rate": 2.8029409645741267e-05, + "loss": 0.9744, + "step": 4500 + }, + { + "epoch": 1.39, + "learning_rate": 2.798942570297309e-05, + "loss": 1.0441, + "step": 4505 + }, + { + "epoch": 1.39, + "learning_rate": 2.7949434000982633e-05, + "loss": 0.9002, + "step": 4510 + }, + { + "epoch": 1.39, + "learning_rate": 2.790943464357062e-05, + "loss": 1.0177, + "step": 4515 + }, + { + "epoch": 1.39, + "learning_rate": 2.7869427734557612e-05, + "loss": 1.0303, + "step": 4520 + }, + { + "epoch": 1.39, + "learning_rate": 2.782941337778381e-05, + "loss": 0.9473, + "step": 4525 + }, + { + "epoch": 1.39, + "learning_rate": 2.77893916771087e-05, + "loss": 1.0222, + "step": 4530 + }, + { + "epoch": 1.4, + "learning_rate": 2.7749362736410856e-05, + "loss": 1.1259, + "step": 4535 + }, + { + "epoch": 1.4, + "learning_rate": 2.770932665958766e-05, + "loss": 1.0501, + "step": 4540 + }, + { + "epoch": 1.4, + "learning_rate": 2.766928355055498e-05, + "loss": 1.006, + "step": 4545 + }, + { + "epoch": 1.4, + "learning_rate": 2.7629233513246966e-05, + "loss": 1.106, + "step": 4550 + }, + { + "epoch": 1.4, + "learning_rate": 2.758917665161573e-05, + "loss": 1.0226, + "step": 4555 + }, + { + "epoch": 1.4, + "learning_rate": 2.7549113069631105e-05, + "loss": 1.0185, + "step": 4560 + }, + { + "epoch": 1.4, + "learning_rate": 2.7509042871280372e-05, + "loss": 1.0228, + "step": 4565 + }, + { + "epoch": 1.41, + "learning_rate": 2.746896616056798e-05, + "loss": 1.0967, + "step": 4570 + }, + { + "epoch": 1.41, + "learning_rate": 2.7428883041515285e-05, + "loss": 1.0473, + "step": 4575 + }, + { + "epoch": 1.41, + "learning_rate": 2.7388793618160273e-05, + "loss": 1.0192, + "step": 4580 + }, + { + "epoch": 1.41, + "learning_rate": 2.7348697994557287e-05, + "loss": 1.0613, + "step": 4585 + }, + { + "epoch": 1.41, + "learning_rate": 2.730859627477677e-05, + "loss": 1.0174, + "step": 4590 + }, + { + "epoch": 1.41, + "learning_rate": 2.7268488562904997e-05, + "loss": 1.0527, + "step": 4595 + }, + { + "epoch": 1.42, + "learning_rate": 2.7228374963043773e-05, + "loss": 1.0404, + "step": 4600 + }, + { + "epoch": 1.42, + "learning_rate": 2.7188255579310212e-05, + "loss": 1.0819, + "step": 4605 + }, + { + "epoch": 1.42, + "learning_rate": 2.7148130515836418e-05, + "loss": 1.0959, + "step": 4610 + }, + { + "epoch": 1.42, + "learning_rate": 2.7107999876769248e-05, + "loss": 0.9383, + "step": 4615 + }, + { + "epoch": 1.42, + "learning_rate": 2.706786376627003e-05, + "loss": 0.954, + "step": 4620 + }, + { + "epoch": 1.42, + "learning_rate": 2.702772228851429e-05, + "loss": 1.0113, + "step": 4625 + }, + { + "epoch": 1.42, + "learning_rate": 2.6987575547691497e-05, + "loss": 1.0002, + "step": 4630 + }, + { + "epoch": 1.43, + "learning_rate": 2.6947423648004767e-05, + "loss": 1.1014, + "step": 4635 + }, + { + "epoch": 1.43, + "learning_rate": 2.6907266693670598e-05, + "loss": 1.0846, + "step": 4640 + }, + { + "epoch": 1.43, + "learning_rate": 2.686710478891863e-05, + "loss": 1.0428, + "step": 4645 + }, + { + "epoch": 1.43, + "learning_rate": 2.682693803799135e-05, + "loss": 1.0345, + "step": 4650 + }, + { + "epoch": 1.43, + "learning_rate": 2.6786766545143804e-05, + "loss": 0.9686, + "step": 4655 + }, + { + "epoch": 1.43, + "learning_rate": 2.674659041464335e-05, + "loss": 1.0616, + "step": 4660 + }, + { + "epoch": 1.44, + "learning_rate": 2.6706409750769418e-05, + "loss": 0.9076, + "step": 4665 + }, + { + "epoch": 1.44, + "learning_rate": 2.666622465781315e-05, + "loss": 1.0459, + "step": 4670 + }, + { + "epoch": 1.44, + "learning_rate": 2.6626035240077223e-05, + "loss": 1.0601, + "step": 4675 + }, + { + "epoch": 1.44, + "learning_rate": 2.6585841601875528e-05, + "loss": 1.0411, + "step": 4680 + }, + { + "epoch": 1.44, + "learning_rate": 2.6545643847532903e-05, + "loss": 1.0923, + "step": 4685 + }, + { + "epoch": 1.44, + "learning_rate": 2.650544208138489e-05, + "loss": 0.978, + "step": 4690 + }, + { + "epoch": 1.44, + "learning_rate": 2.646523640777741e-05, + "loss": 0.9338, + "step": 4695 + }, + { + "epoch": 1.45, + "learning_rate": 2.6425026931066564e-05, + "loss": 1.0308, + "step": 4700 + }, + { + "epoch": 1.45, + "learning_rate": 2.638481375561831e-05, + "loss": 1.0132, + "step": 4705 + }, + { + "epoch": 1.45, + "learning_rate": 2.6344596985808183e-05, + "loss": 1.0031, + "step": 4710 + }, + { + "epoch": 1.45, + "learning_rate": 2.6304376726021087e-05, + "loss": 1.0087, + "step": 4715 + }, + { + "epoch": 1.45, + "learning_rate": 2.6264153080650943e-05, + "loss": 1.0648, + "step": 4720 + }, + { + "epoch": 1.45, + "learning_rate": 2.6223926154100503e-05, + "loss": 1.0524, + "step": 4725 + }, + { + "epoch": 1.46, + "learning_rate": 2.6183696050781005e-05, + "loss": 1.1848, + "step": 4730 + }, + { + "epoch": 1.46, + "learning_rate": 2.6143462875111934e-05, + "loss": 1.0686, + "step": 4735 + }, + { + "epoch": 1.46, + "learning_rate": 2.610322673152077e-05, + "loss": 0.9396, + "step": 4740 + }, + { + "epoch": 1.46, + "learning_rate": 2.6062987724442677e-05, + "loss": 1.0911, + "step": 4745 + }, + { + "epoch": 1.46, + "learning_rate": 2.602274595832025e-05, + "loss": 1.0587, + "step": 4750 + }, + { + "epoch": 1.46, + "learning_rate": 2.598250153760326e-05, + "loss": 0.948, + "step": 4755 + }, + { + "epoch": 1.46, + "learning_rate": 2.594225456674837e-05, + "loss": 0.991, + "step": 4760 + }, + { + "epoch": 1.47, + "learning_rate": 2.5902005150218834e-05, + "loss": 0.8825, + "step": 4765 + }, + { + "epoch": 1.47, + "learning_rate": 2.5861753392484285e-05, + "loss": 0.9453, + "step": 4770 + }, + { + "epoch": 1.47, + "learning_rate": 2.5821499398020417e-05, + "loss": 1.0483, + "step": 4775 + }, + { + "epoch": 1.47, + "learning_rate": 2.578124327130873e-05, + "loss": 1.0822, + "step": 4780 + }, + { + "epoch": 1.47, + "learning_rate": 2.5740985116836264e-05, + "loss": 1.0588, + "step": 4785 + }, + { + "epoch": 1.47, + "learning_rate": 2.570072503909532e-05, + "loss": 1.139, + "step": 4790 + }, + { + "epoch": 1.48, + "learning_rate": 2.5660463142583192e-05, + "loss": 1.0545, + "step": 4795 + }, + { + "epoch": 1.48, + "learning_rate": 2.562019953180189e-05, + "loss": 1.0565, + "step": 4800 + }, + { + "epoch": 1.48, + "learning_rate": 2.557993431125788e-05, + "loss": 1.0684, + "step": 4805 + }, + { + "epoch": 1.48, + "learning_rate": 2.5539667585461802e-05, + "loss": 1.038, + "step": 4810 + }, + { + "epoch": 1.48, + "learning_rate": 2.549939945892821e-05, + "loss": 1.0951, + "step": 4815 + }, + { + "epoch": 1.48, + "learning_rate": 2.545913003617526e-05, + "loss": 1.0513, + "step": 4820 + }, + { + "epoch": 1.48, + "learning_rate": 2.5418859421724538e-05, + "loss": 1.0937, + "step": 4825 + }, + { + "epoch": 1.49, + "learning_rate": 2.537858772010066e-05, + "loss": 1.0582, + "step": 4830 + }, + { + "epoch": 1.49, + "learning_rate": 2.53383150358311e-05, + "loss": 1.0011, + "step": 4835 + }, + { + "epoch": 1.49, + "learning_rate": 2.529804147344586e-05, + "loss": 1.0611, + "step": 4840 + }, + { + "epoch": 1.49, + "learning_rate": 2.5257767137477246e-05, + "loss": 1.0341, + "step": 4845 + }, + { + "epoch": 1.49, + "learning_rate": 2.5217492132459553e-05, + "loss": 1.1068, + "step": 4850 + }, + { + "epoch": 1.49, + "learning_rate": 2.5177216562928814e-05, + "loss": 0.9726, + "step": 4855 + }, + { + "epoch": 1.5, + "learning_rate": 2.513694053342254e-05, + "loss": 1.0733, + "step": 4860 + }, + { + "epoch": 1.5, + "learning_rate": 2.5096664148479422e-05, + "loss": 1.0716, + "step": 4865 + }, + { + "epoch": 1.5, + "learning_rate": 2.5056387512639072e-05, + "loss": 0.938, + "step": 4870 + }, + { + "epoch": 1.5, + "learning_rate": 2.5016110730441766e-05, + "loss": 1.0029, + "step": 4875 + }, + { + "epoch": 1.5, + "learning_rate": 2.497583390642816e-05, + "loss": 1.0458, + "step": 4880 + }, + { + "epoch": 1.5, + "learning_rate": 2.493555714513901e-05, + "loss": 1.003, + "step": 4885 + }, + { + "epoch": 1.5, + "learning_rate": 2.4895280551114907e-05, + "loss": 0.9719, + "step": 4890 + }, + { + "epoch": 1.51, + "learning_rate": 2.4855004228896013e-05, + "loss": 1.0803, + "step": 4895 + }, + { + "epoch": 1.51, + "learning_rate": 2.481472828302178e-05, + "loss": 1.0759, + "step": 4900 + }, + { + "epoch": 1.51, + "learning_rate": 2.4774452818030695e-05, + "loss": 1.0106, + "step": 4905 + }, + { + "epoch": 1.51, + "learning_rate": 2.4734177938459984e-05, + "loss": 1.0443, + "step": 4910 + }, + { + "epoch": 1.51, + "learning_rate": 2.4693903748845365e-05, + "loss": 0.9647, + "step": 4915 + }, + { + "epoch": 1.51, + "learning_rate": 2.4653630353720748e-05, + "loss": 1.1069, + "step": 4920 + }, + { + "epoch": 1.52, + "learning_rate": 2.4613357857617996e-05, + "loss": 1.1304, + "step": 4925 + }, + { + "epoch": 1.52, + "learning_rate": 2.457308636506664e-05, + "loss": 1.0108, + "step": 4930 + }, + { + "epoch": 1.52, + "learning_rate": 2.453281598059359e-05, + "loss": 1.0776, + "step": 4935 + }, + { + "epoch": 1.52, + "learning_rate": 2.4492546808722906e-05, + "loss": 0.9403, + "step": 4940 + }, + { + "epoch": 1.52, + "learning_rate": 2.4452278953975468e-05, + "loss": 0.937, + "step": 4945 + }, + { + "epoch": 1.52, + "learning_rate": 2.4412012520868758e-05, + "loss": 1.0301, + "step": 4950 + }, + { + "epoch": 1.52, + "learning_rate": 2.4371747613916566e-05, + "loss": 0.9402, + "step": 4955 + }, + { + "epoch": 1.53, + "learning_rate": 2.433148433762872e-05, + "loss": 1.1629, + "step": 4960 + }, + { + "epoch": 1.53, + "learning_rate": 2.429122279651081e-05, + "loss": 0.9118, + "step": 4965 + }, + { + "epoch": 1.53, + "learning_rate": 2.425096309506395e-05, + "loss": 0.9551, + "step": 4970 + }, + { + "epoch": 1.53, + "learning_rate": 2.421070533778442e-05, + "loss": 1.028, + "step": 4975 + }, + { + "epoch": 1.53, + "learning_rate": 2.4170449629163502e-05, + "loss": 1.024, + "step": 4980 + }, + { + "epoch": 1.53, + "learning_rate": 2.4130196073687154e-05, + "loss": 1.1382, + "step": 4985 + }, + { + "epoch": 1.54, + "learning_rate": 2.4089944775835744e-05, + "loss": 1.0157, + "step": 4990 + }, + { + "epoch": 1.54, + "learning_rate": 2.404969584008377e-05, + "loss": 1.0194, + "step": 4995 + }, + { + "epoch": 1.54, + "learning_rate": 2.4009449370899602e-05, + "loss": 1.0408, + "step": 5000 + }, + { + "epoch": 1.54, + "learning_rate": 2.3969205472745213e-05, + "loss": 0.9362, + "step": 5005 + }, + { + "epoch": 1.54, + "learning_rate": 2.39289642500759e-05, + "loss": 1.0095, + "step": 5010 + }, + { + "epoch": 1.54, + "learning_rate": 2.3888725807340017e-05, + "loss": 1.1, + "step": 5015 + }, + { + "epoch": 1.54, + "learning_rate": 2.384849024897869e-05, + "loss": 1.0652, + "step": 5020 + }, + { + "epoch": 1.55, + "learning_rate": 2.38082576794256e-05, + "loss": 0.985, + "step": 5025 + }, + { + "epoch": 1.55, + "learning_rate": 2.3768028203106612e-05, + "loss": 1.0362, + "step": 5030 + }, + { + "epoch": 1.55, + "learning_rate": 2.3727801924439598e-05, + "loss": 1.042, + "step": 5035 + }, + { + "epoch": 1.55, + "learning_rate": 2.3687578947834123e-05, + "loss": 1.0412, + "step": 5040 + }, + { + "epoch": 1.55, + "learning_rate": 2.3647359377691186e-05, + "loss": 1.0032, + "step": 5045 + }, + { + "epoch": 1.55, + "learning_rate": 2.3607143318402945e-05, + "loss": 1.0278, + "step": 5050 + }, + { + "epoch": 1.56, + "learning_rate": 2.3566930874352425e-05, + "loss": 1.031, + "step": 5055 + }, + { + "epoch": 1.56, + "learning_rate": 2.3526722149913288e-05, + "loss": 1.0019, + "step": 5060 + }, + { + "epoch": 1.56, + "learning_rate": 2.3486517249449542e-05, + "loss": 1.1348, + "step": 5065 + }, + { + "epoch": 1.56, + "learning_rate": 2.3446316277315256e-05, + "loss": 1.0082, + "step": 5070 + }, + { + "epoch": 1.56, + "learning_rate": 2.340611933785431e-05, + "loss": 1.0049, + "step": 5075 + }, + { + "epoch": 1.56, + "learning_rate": 2.3365926535400135e-05, + "loss": 0.9933, + "step": 5080 + }, + { + "epoch": 1.56, + "learning_rate": 2.332573797427538e-05, + "loss": 0.9231, + "step": 5085 + }, + { + "epoch": 1.57, + "learning_rate": 2.328555375879173e-05, + "loss": 1.1816, + "step": 5090 + }, + { + "epoch": 1.57, + "learning_rate": 2.3245373993249566e-05, + "loss": 1.0803, + "step": 5095 + }, + { + "epoch": 1.57, + "learning_rate": 2.3205198781937727e-05, + "loss": 1.0627, + "step": 5100 + }, + { + "epoch": 1.57, + "learning_rate": 2.316502822913324e-05, + "loss": 1.0248, + "step": 5105 + }, + { + "epoch": 1.57, + "learning_rate": 2.3124862439101015e-05, + "loss": 1.1723, + "step": 5110 + }, + { + "epoch": 1.57, + "learning_rate": 2.308470151609362e-05, + "loss": 1.038, + "step": 5115 + }, + { + "epoch": 1.58, + "learning_rate": 2.3044545564350984e-05, + "loss": 1.0658, + "step": 5120 + }, + { + "epoch": 1.58, + "learning_rate": 2.3004394688100133e-05, + "loss": 1.0005, + "step": 5125 + }, + { + "epoch": 1.58, + "learning_rate": 2.2964248991554936e-05, + "loss": 1.0732, + "step": 5130 + }, + { + "epoch": 1.58, + "learning_rate": 2.29241085789158e-05, + "loss": 1.0938, + "step": 5135 + }, + { + "epoch": 1.58, + "learning_rate": 2.2883973554369392e-05, + "loss": 1.0599, + "step": 5140 + }, + { + "epoch": 1.58, + "learning_rate": 2.284384402208845e-05, + "loss": 1.0179, + "step": 5145 + }, + { + "epoch": 1.58, + "learning_rate": 2.280372008623142e-05, + "loss": 1.0789, + "step": 5150 + }, + { + "epoch": 1.59, + "learning_rate": 2.276360185094223e-05, + "loss": 1.0473, + "step": 5155 + }, + { + "epoch": 1.59, + "learning_rate": 2.272348942035002e-05, + "loss": 1.0889, + "step": 5160 + }, + { + "epoch": 1.59, + "learning_rate": 2.2683382898568845e-05, + "loss": 1.0422, + "step": 5165 + }, + { + "epoch": 1.59, + "learning_rate": 2.2643282389697438e-05, + "loss": 1.1119, + "step": 5170 + }, + { + "epoch": 1.59, + "learning_rate": 2.260318799781892e-05, + "loss": 1.0205, + "step": 5175 + }, + { + "epoch": 1.59, + "learning_rate": 2.2563099827000535e-05, + "loss": 1.1009, + "step": 5180 + }, + { + "epoch": 1.6, + "learning_rate": 2.25230179812934e-05, + "loss": 1.0832, + "step": 5185 + }, + { + "epoch": 1.6, + "learning_rate": 2.248294256473217e-05, + "loss": 1.0892, + "step": 5190 + }, + { + "epoch": 1.6, + "learning_rate": 2.2442873681334845e-05, + "loss": 1.0163, + "step": 5195 + }, + { + "epoch": 1.6, + "learning_rate": 2.240281143510247e-05, + "loss": 1.1435, + "step": 5200 + }, + { + "epoch": 1.6, + "learning_rate": 2.236275593001885e-05, + "loss": 1.0182, + "step": 5205 + }, + { + "epoch": 1.6, + "learning_rate": 2.2322707270050307e-05, + "loss": 1.0538, + "step": 5210 + }, + { + "epoch": 1.6, + "learning_rate": 2.2282665559145378e-05, + "loss": 1.0792, + "step": 5215 + }, + { + "epoch": 1.61, + "learning_rate": 2.2242630901234567e-05, + "loss": 1.029, + "step": 5220 + }, + { + "epoch": 1.61, + "learning_rate": 2.2202603400230086e-05, + "loss": 0.9053, + "step": 5225 + }, + { + "epoch": 1.61, + "learning_rate": 2.2162583160025554e-05, + "loss": 0.9878, + "step": 5230 + }, + { + "epoch": 1.61, + "learning_rate": 2.2122570284495753e-05, + "loss": 1.0592, + "step": 5235 + }, + { + "epoch": 1.61, + "learning_rate": 2.2082564877496368e-05, + "loss": 1.0604, + "step": 5240 + }, + { + "epoch": 1.61, + "learning_rate": 2.2042567042863643e-05, + "loss": 1.0164, + "step": 5245 + }, + { + "epoch": 1.62, + "learning_rate": 2.200257688441423e-05, + "loss": 0.9813, + "step": 5250 + }, + { + "epoch": 1.62, + "learning_rate": 2.1962594505944823e-05, + "loss": 1.0019, + "step": 5255 + }, + { + "epoch": 1.62, + "learning_rate": 2.192262001123193e-05, + "loss": 0.9913, + "step": 5260 + }, + { + "epoch": 1.62, + "learning_rate": 2.1882653504031598e-05, + "loss": 1.147, + "step": 5265 + }, + { + "epoch": 1.62, + "learning_rate": 2.1842695088079144e-05, + "loss": 1.0022, + "step": 5270 + }, + { + "epoch": 1.62, + "learning_rate": 2.1802744867088872e-05, + "loss": 1.0852, + "step": 5275 + }, + { + "epoch": 1.62, + "learning_rate": 2.176280294475383e-05, + "loss": 1.0634, + "step": 5280 + }, + { + "epoch": 1.63, + "learning_rate": 2.1722869424745512e-05, + "loss": 1.0947, + "step": 5285 + }, + { + "epoch": 1.63, + "learning_rate": 2.1682944410713624e-05, + "loss": 1.1165, + "step": 5290 + }, + { + "epoch": 1.63, + "learning_rate": 2.1643028006285787e-05, + "loss": 1.0354, + "step": 5295 + }, + { + "epoch": 1.63, + "learning_rate": 2.1603120315067245e-05, + "loss": 1.0268, + "step": 5300 + }, + { + "epoch": 1.63, + "learning_rate": 2.1563221440640666e-05, + "loss": 0.9327, + "step": 5305 + }, + { + "epoch": 1.63, + "learning_rate": 2.152333148656582e-05, + "loss": 1.0191, + "step": 5310 + }, + { + "epoch": 1.64, + "learning_rate": 2.148345055637932e-05, + "loss": 0.8733, + "step": 5315 + }, + { + "epoch": 1.64, + "learning_rate": 2.1443578753594363e-05, + "loss": 1.0353, + "step": 5320 + }, + { + "epoch": 1.64, + "learning_rate": 2.1403716181700453e-05, + "loss": 1.0335, + "step": 5325 + }, + { + "epoch": 1.64, + "learning_rate": 2.1363862944163127e-05, + "loss": 1.1151, + "step": 5330 + }, + { + "epoch": 1.64, + "learning_rate": 2.13240191444237e-05, + "loss": 0.9734, + "step": 5335 + }, + { + "epoch": 1.64, + "learning_rate": 2.1284184885898995e-05, + "loss": 0.9466, + "step": 5340 + }, + { + "epoch": 1.64, + "learning_rate": 2.1244360271981073e-05, + "loss": 1.0951, + "step": 5345 + }, + { + "epoch": 1.65, + "learning_rate": 2.1204545406036956e-05, + "loss": 0.9825, + "step": 5350 + }, + { + "epoch": 1.65, + "learning_rate": 2.116474039140835e-05, + "loss": 0.9062, + "step": 5355 + }, + { + "epoch": 1.65, + "learning_rate": 2.112494533141141e-05, + "loss": 1.0606, + "step": 5360 + }, + { + "epoch": 1.65, + "learning_rate": 2.108516032933646e-05, + "loss": 0.9909, + "step": 5365 + }, + { + "epoch": 1.65, + "learning_rate": 2.1045385488447698e-05, + "loss": 1.0345, + "step": 5370 + }, + { + "epoch": 1.65, + "learning_rate": 2.1005620911982958e-05, + "loss": 1.0344, + "step": 5375 + }, + { + "epoch": 1.66, + "learning_rate": 2.0965866703153436e-05, + "loss": 0.968, + "step": 5380 + }, + { + "epoch": 1.66, + "learning_rate": 2.09261229651434e-05, + "loss": 1.055, + "step": 5385 + }, + { + "epoch": 1.66, + "learning_rate": 2.0886389801109965e-05, + "loss": 0.9389, + "step": 5390 + }, + { + "epoch": 1.66, + "learning_rate": 2.0846667314182787e-05, + "loss": 0.9148, + "step": 5395 + }, + { + "epoch": 1.66, + "learning_rate": 2.0806955607463813e-05, + "loss": 0.8804, + "step": 5400 + }, + { + "epoch": 1.66, + "learning_rate": 2.0767254784027012e-05, + "loss": 1.0112, + "step": 5405 + }, + { + "epoch": 1.66, + "learning_rate": 2.0727564946918087e-05, + "loss": 1.0573, + "step": 5410 + }, + { + "epoch": 1.67, + "learning_rate": 2.0687886199154248e-05, + "loss": 0.9469, + "step": 5415 + }, + { + "epoch": 1.67, + "learning_rate": 2.064821864372391e-05, + "loss": 1.0492, + "step": 5420 + }, + { + "epoch": 1.67, + "learning_rate": 2.060856238358644e-05, + "loss": 1.0196, + "step": 5425 + }, + { + "epoch": 1.67, + "learning_rate": 2.05689175216719e-05, + "loss": 0.9556, + "step": 5430 + }, + { + "epoch": 1.67, + "learning_rate": 2.052928416088073e-05, + "loss": 1.11, + "step": 5435 + }, + { + "epoch": 1.67, + "learning_rate": 2.0489662404083567e-05, + "loss": 1.1505, + "step": 5440 + }, + { + "epoch": 1.68, + "learning_rate": 2.045005235412089e-05, + "loss": 0.9844, + "step": 5445 + }, + { + "epoch": 1.68, + "learning_rate": 2.0410454113802815e-05, + "loss": 1.1265, + "step": 5450 + }, + { + "epoch": 1.68, + "learning_rate": 2.0370867785908805e-05, + "loss": 0.973, + "step": 5455 + }, + { + "epoch": 1.68, + "learning_rate": 2.0331293473187387e-05, + "loss": 1.0653, + "step": 5460 + }, + { + "epoch": 1.68, + "learning_rate": 2.0291731278355913e-05, + "loss": 1.0825, + "step": 5465 + }, + { + "epoch": 1.68, + "learning_rate": 2.0252181304100285e-05, + "loss": 0.9449, + "step": 5470 + }, + { + "epoch": 1.68, + "learning_rate": 2.021264365307468e-05, + "loss": 1.0501, + "step": 5475 + }, + { + "epoch": 1.69, + "learning_rate": 2.0173118427901286e-05, + "loss": 0.9421, + "step": 5480 + }, + { + "epoch": 1.69, + "learning_rate": 2.013360573117006e-05, + "loss": 1.0641, + "step": 5485 + }, + { + "epoch": 1.69, + "learning_rate": 2.0094105665438402e-05, + "loss": 1.0499, + "step": 5490 + }, + { + "epoch": 1.69, + "learning_rate": 2.005461833323097e-05, + "loss": 1.0024, + "step": 5495 + }, + { + "epoch": 1.69, + "learning_rate": 2.0015143837039336e-05, + "loss": 1.0199, + "step": 5500 + }, + { + "epoch": 1.69, + "learning_rate": 1.9975682279321786e-05, + "loss": 1.0579, + "step": 5505 + }, + { + "epoch": 1.7, + "learning_rate": 1.9936233762503004e-05, + "loss": 1.0438, + "step": 5510 + }, + { + "epoch": 1.7, + "learning_rate": 1.9896798388973837e-05, + "loss": 0.9772, + "step": 5515 + }, + { + "epoch": 1.7, + "learning_rate": 1.9857376261090996e-05, + "loss": 1.1065, + "step": 5520 + }, + { + "epoch": 1.7, + "learning_rate": 1.9817967481176838e-05, + "loss": 0.9751, + "step": 5525 + }, + { + "epoch": 1.7, + "learning_rate": 1.977857215151906e-05, + "loss": 1.0231, + "step": 5530 + }, + { + "epoch": 1.7, + "learning_rate": 1.9739190374370455e-05, + "loss": 1.1194, + "step": 5535 + }, + { + "epoch": 1.7, + "learning_rate": 1.969982225194864e-05, + "loss": 1.0353, + "step": 5540 + }, + { + "epoch": 1.71, + "learning_rate": 1.966046788643578e-05, + "loss": 1.002, + "step": 5545 + }, + { + "epoch": 1.71, + "learning_rate": 1.9621127379978338e-05, + "loss": 1.1341, + "step": 5550 + }, + { + "epoch": 1.71, + "learning_rate": 1.958180083468682e-05, + "loss": 1.0659, + "step": 5555 + }, + { + "epoch": 1.71, + "learning_rate": 1.9542488352635474e-05, + "loss": 1.0098, + "step": 5560 + }, + { + "epoch": 1.71, + "learning_rate": 1.9503190035862056e-05, + "loss": 1.0831, + "step": 5565 + }, + { + "epoch": 1.71, + "learning_rate": 1.946390598636756e-05, + "loss": 1.0214, + "step": 5570 + }, + { + "epoch": 1.72, + "learning_rate": 1.942463630611593e-05, + "loss": 1.0706, + "step": 5575 + }, + { + "epoch": 1.72, + "learning_rate": 1.938538109703384e-05, + "loss": 0.9147, + "step": 5580 + }, + { + "epoch": 1.72, + "learning_rate": 1.9346140461010377e-05, + "loss": 0.9479, + "step": 5585 + }, + { + "epoch": 1.72, + "learning_rate": 1.9306914499896814e-05, + "loss": 1.0038, + "step": 5590 + }, + { + "epoch": 1.72, + "learning_rate": 1.926770331550636e-05, + "loss": 1.0788, + "step": 5595 + }, + { + "epoch": 1.72, + "learning_rate": 1.9228507009613817e-05, + "loss": 1.0741, + "step": 5600 + }, + { + "epoch": 1.72, + "learning_rate": 1.91893256839554e-05, + "loss": 1.0417, + "step": 5605 + }, + { + "epoch": 1.73, + "learning_rate": 1.915015944022846e-05, + "loss": 1.0388, + "step": 5610 + }, + { + "epoch": 1.73, + "learning_rate": 1.9111008380091167e-05, + "loss": 1.0787, + "step": 5615 + }, + { + "epoch": 1.73, + "learning_rate": 1.9071872605162302e-05, + "loss": 1.0506, + "step": 5620 + }, + { + "epoch": 1.73, + "learning_rate": 1.9032752217020965e-05, + "loss": 1.0767, + "step": 5625 + }, + { + "epoch": 1.73, + "learning_rate": 1.899364731720632e-05, + "loss": 1.0827, + "step": 5630 + }, + { + "epoch": 1.73, + "learning_rate": 1.8954558007217325e-05, + "loss": 0.9847, + "step": 5635 + }, + { + "epoch": 1.74, + "learning_rate": 1.8915484388512483e-05, + "loss": 0.9154, + "step": 5640 + }, + { + "epoch": 1.74, + "learning_rate": 1.8876426562509565e-05, + "loss": 0.9668, + "step": 5645 + }, + { + "epoch": 1.74, + "learning_rate": 1.8837384630585363e-05, + "loss": 1.0351, + "step": 5650 + }, + { + "epoch": 1.74, + "learning_rate": 1.8798358694075387e-05, + "loss": 1.0868, + "step": 5655 + }, + { + "epoch": 1.74, + "learning_rate": 1.8759348854273642e-05, + "loss": 0.9459, + "step": 5660 + }, + { + "epoch": 1.74, + "learning_rate": 1.8720355212432375e-05, + "loss": 1.0078, + "step": 5665 + }, + { + "epoch": 1.74, + "learning_rate": 1.868137786976177e-05, + "loss": 0.9809, + "step": 5670 + }, + { + "epoch": 1.75, + "learning_rate": 1.8642416927429703e-05, + "loss": 1.0772, + "step": 5675 + }, + { + "epoch": 1.75, + "learning_rate": 1.8603472486561487e-05, + "loss": 1.0867, + "step": 5680 + }, + { + "epoch": 1.75, + "learning_rate": 1.8564544648239605e-05, + "loss": 1.0038, + "step": 5685 + }, + { + "epoch": 1.75, + "learning_rate": 1.8525633513503453e-05, + "loss": 0.9516, + "step": 5690 + }, + { + "epoch": 1.75, + "learning_rate": 1.8486739183349057e-05, + "loss": 1.0044, + "step": 5695 + }, + { + "epoch": 1.75, + "learning_rate": 1.8447861758728848e-05, + "loss": 0.9407, + "step": 5700 + }, + { + "epoch": 1.76, + "learning_rate": 1.8409001340551364e-05, + "loss": 1.0775, + "step": 5705 + }, + { + "epoch": 1.76, + "learning_rate": 1.8370158029680977e-05, + "loss": 1.0621, + "step": 5710 + }, + { + "epoch": 1.76, + "learning_rate": 1.8331331926937712e-05, + "loss": 1.016, + "step": 5715 + }, + { + "epoch": 1.76, + "learning_rate": 1.829252313309688e-05, + "loss": 0.9907, + "step": 5720 + }, + { + "epoch": 1.76, + "learning_rate": 1.8253731748888896e-05, + "loss": 1.0446, + "step": 5725 + }, + { + "epoch": 1.76, + "learning_rate": 1.8214957874998973e-05, + "loss": 1.0248, + "step": 5730 + }, + { + "epoch": 1.76, + "learning_rate": 1.817620161206687e-05, + "loss": 0.9655, + "step": 5735 + }, + { + "epoch": 1.77, + "learning_rate": 1.8137463060686657e-05, + "loss": 0.9789, + "step": 5740 + }, + { + "epoch": 1.77, + "learning_rate": 1.809874232140641e-05, + "loss": 1.0073, + "step": 5745 + }, + { + "epoch": 1.77, + "learning_rate": 1.8060039494727977e-05, + "loss": 1.0464, + "step": 5750 + }, + { + "epoch": 1.77, + "learning_rate": 1.802135468110673e-05, + "loss": 1.0541, + "step": 5755 + }, + { + "epoch": 1.77, + "learning_rate": 1.798268798095128e-05, + "loss": 1.0394, + "step": 5760 + }, + { + "epoch": 1.77, + "learning_rate": 1.7944039494623195e-05, + "loss": 1.0415, + "step": 5765 + }, + { + "epoch": 1.78, + "learning_rate": 1.7905409322436804e-05, + "loss": 1.0953, + "step": 5770 + }, + { + "epoch": 1.78, + "learning_rate": 1.7866797564658893e-05, + "loss": 1.0372, + "step": 5775 + }, + { + "epoch": 1.78, + "learning_rate": 1.7828204321508446e-05, + "loss": 0.8946, + "step": 5780 + }, + { + "epoch": 1.78, + "learning_rate": 1.7789629693156394e-05, + "loss": 1.0527, + "step": 5785 + }, + { + "epoch": 1.78, + "learning_rate": 1.775107377972534e-05, + "loss": 1.0916, + "step": 5790 + }, + { + "epoch": 1.78, + "learning_rate": 1.7712536681289334e-05, + "loss": 1.0558, + "step": 5795 + }, + { + "epoch": 1.78, + "learning_rate": 1.767401849787357e-05, + "loss": 1.0064, + "step": 5800 + }, + { + "epoch": 1.79, + "learning_rate": 1.7635519329454152e-05, + "loss": 0.9694, + "step": 5805 + }, + { + "epoch": 1.79, + "learning_rate": 1.7597039275957853e-05, + "loss": 0.9636, + "step": 5810 + }, + { + "epoch": 1.79, + "learning_rate": 1.7558578437261807e-05, + "loss": 0.9436, + "step": 5815 + }, + { + "epoch": 1.79, + "learning_rate": 1.7520136913193265e-05, + "loss": 1.0931, + "step": 5820 + }, + { + "epoch": 1.79, + "learning_rate": 1.748171480352937e-05, + "loss": 0.9234, + "step": 5825 + }, + { + "epoch": 1.79, + "learning_rate": 1.7443312207996864e-05, + "loss": 1.0694, + "step": 5830 + }, + { + "epoch": 1.8, + "learning_rate": 1.7404929226271848e-05, + "loss": 0.8943, + "step": 5835 + }, + { + "epoch": 1.8, + "learning_rate": 1.7366565957979503e-05, + "loss": 1.0101, + "step": 5840 + }, + { + "epoch": 1.8, + "learning_rate": 1.732822250269384e-05, + "loss": 1.0712, + "step": 5845 + }, + { + "epoch": 1.8, + "learning_rate": 1.728989895993745e-05, + "loss": 0.9879, + "step": 5850 + }, + { + "epoch": 1.8, + "learning_rate": 1.7251595429181243e-05, + "loss": 1.0048, + "step": 5855 + }, + { + "epoch": 1.8, + "learning_rate": 1.7213312009844186e-05, + "loss": 1.0812, + "step": 5860 + }, + { + "epoch": 1.8, + "learning_rate": 1.717504880129304e-05, + "loss": 1.0002, + "step": 5865 + }, + { + "epoch": 1.81, + "learning_rate": 1.7136805902842127e-05, + "loss": 0.949, + "step": 5870 + }, + { + "epoch": 1.81, + "learning_rate": 1.7098583413753012e-05, + "loss": 0.9801, + "step": 5875 + }, + { + "epoch": 1.81, + "learning_rate": 1.7060381433234325e-05, + "loss": 0.9992, + "step": 5880 + }, + { + "epoch": 1.81, + "learning_rate": 1.702220006044145e-05, + "loss": 0.9444, + "step": 5885 + }, + { + "epoch": 1.81, + "learning_rate": 1.6984039394476287e-05, + "loss": 1.0863, + "step": 5890 + }, + { + "epoch": 1.81, + "learning_rate": 1.6945899534386985e-05, + "loss": 1.0866, + "step": 5895 + }, + { + "epoch": 1.82, + "learning_rate": 1.6907780579167688e-05, + "loss": 0.9463, + "step": 5900 + }, + { + "epoch": 1.82, + "learning_rate": 1.686968262775829e-05, + "loss": 1.0495, + "step": 5905 + }, + { + "epoch": 1.82, + "learning_rate": 1.6831605779044157e-05, + "loss": 1.0122, + "step": 5910 + }, + { + "epoch": 1.82, + "learning_rate": 1.6793550131855896e-05, + "loss": 1.0558, + "step": 5915 + }, + { + "epoch": 1.82, + "learning_rate": 1.675551578496907e-05, + "loss": 1.0185, + "step": 5920 + }, + { + "epoch": 1.82, + "learning_rate": 1.6717502837103975e-05, + "loss": 1.1087, + "step": 5925 + }, + { + "epoch": 1.82, + "learning_rate": 1.6679511386925337e-05, + "loss": 0.9494, + "step": 5930 + }, + { + "epoch": 1.83, + "learning_rate": 1.6641541533042098e-05, + "loss": 1.0056, + "step": 5935 + }, + { + "epoch": 1.83, + "learning_rate": 1.6603593374007153e-05, + "loss": 0.9981, + "step": 5940 + }, + { + "epoch": 1.83, + "learning_rate": 1.656566700831708e-05, + "loss": 1.0714, + "step": 5945 + }, + { + "epoch": 1.83, + "learning_rate": 1.6527762534411888e-05, + "loss": 1.0377, + "step": 5950 + }, + { + "epoch": 1.83, + "learning_rate": 1.6489880050674767e-05, + "loss": 1.0256, + "step": 5955 + }, + { + "epoch": 1.83, + "learning_rate": 1.6452019655431828e-05, + "loss": 1.0665, + "step": 5960 + }, + { + "epoch": 1.84, + "learning_rate": 1.641418144695185e-05, + "loss": 1.0692, + "step": 5965 + }, + { + "epoch": 1.84, + "learning_rate": 1.637636552344604e-05, + "loss": 1.0189, + "step": 5970 + }, + { + "epoch": 1.84, + "learning_rate": 1.6338571983067754e-05, + "loss": 1.1224, + "step": 5975 + }, + { + "epoch": 1.84, + "learning_rate": 1.6300800923912224e-05, + "loss": 1.0142, + "step": 5980 + }, + { + "epoch": 1.84, + "learning_rate": 1.6263052444016374e-05, + "loss": 1.1043, + "step": 5985 + }, + { + "epoch": 1.84, + "learning_rate": 1.62253266413585e-05, + "loss": 1.0403, + "step": 5990 + }, + { + "epoch": 1.84, + "learning_rate": 1.6187623613858038e-05, + "loss": 0.994, + "step": 5995 + }, + { + "epoch": 1.85, + "learning_rate": 1.6149943459375312e-05, + "loss": 1.0363, + "step": 6000 + }, + { + "epoch": 1.85, + "learning_rate": 1.6112286275711298e-05, + "loss": 1.0155, + "step": 6005 + }, + { + "epoch": 1.85, + "learning_rate": 1.6074652160607302e-05, + "loss": 1.0731, + "step": 6010 + }, + { + "epoch": 1.85, + "learning_rate": 1.603704121174479e-05, + "loss": 1.1081, + "step": 6015 + }, + { + "epoch": 1.85, + "learning_rate": 1.5999453526745104e-05, + "loss": 1.0777, + "step": 6020 + }, + { + "epoch": 1.85, + "learning_rate": 1.5961889203169184e-05, + "loss": 1.0508, + "step": 6025 + }, + { + "epoch": 1.86, + "learning_rate": 1.592434833851734e-05, + "loss": 1.1067, + "step": 6030 + }, + { + "epoch": 1.86, + "learning_rate": 1.5886831030229e-05, + "loss": 1.0054, + "step": 6035 + }, + { + "epoch": 1.86, + "learning_rate": 1.5849337375682435e-05, + "loss": 0.9554, + "step": 6040 + }, + { + "epoch": 1.86, + "learning_rate": 1.5811867472194535e-05, + "loss": 1.0532, + "step": 6045 + }, + { + "epoch": 1.86, + "learning_rate": 1.577442141702054e-05, + "loss": 0.9215, + "step": 6050 + }, + { + "epoch": 1.86, + "learning_rate": 1.5736999307353785e-05, + "loss": 1.1278, + "step": 6055 + }, + { + "epoch": 1.86, + "learning_rate": 1.5699601240325474e-05, + "loss": 0.977, + "step": 6060 + }, + { + "epoch": 1.87, + "learning_rate": 1.5662227313004364e-05, + "loss": 1.0781, + "step": 6065 + }, + { + "epoch": 1.87, + "learning_rate": 1.5624877622396588e-05, + "loss": 1.0992, + "step": 6070 + }, + { + "epoch": 1.87, + "learning_rate": 1.5587552265445375e-05, + "loss": 1.0338, + "step": 6075 + }, + { + "epoch": 1.87, + "learning_rate": 1.5550251339030783e-05, + "loss": 1.031, + "step": 6080 + }, + { + "epoch": 1.87, + "learning_rate": 1.5512974939969464e-05, + "loss": 0.9519, + "step": 6085 + }, + { + "epoch": 1.87, + "learning_rate": 1.5475723165014393e-05, + "loss": 0.9882, + "step": 6090 + }, + { + "epoch": 1.88, + "learning_rate": 1.543849611085465e-05, + "loss": 0.8164, + "step": 6095 + }, + { + "epoch": 1.88, + "learning_rate": 1.5401293874115147e-05, + "loss": 1.0518, + "step": 6100 + }, + { + "epoch": 1.88, + "learning_rate": 1.5364116551356376e-05, + "loss": 1.0335, + "step": 6105 + }, + { + "epoch": 1.88, + "learning_rate": 1.532696423907416e-05, + "loss": 0.9326, + "step": 6110 + }, + { + "epoch": 1.88, + "learning_rate": 1.528983703369943e-05, + "loss": 0.9397, + "step": 6115 + }, + { + "epoch": 1.88, + "learning_rate": 1.5252735031597915e-05, + "loss": 0.8766, + "step": 6120 + }, + { + "epoch": 1.88, + "learning_rate": 1.521565832906994e-05, + "loss": 0.9799, + "step": 6125 + }, + { + "epoch": 1.89, + "learning_rate": 1.5178607022350186e-05, + "loss": 1.0206, + "step": 6130 + }, + { + "epoch": 1.89, + "learning_rate": 1.5141581207607391e-05, + "loss": 0.9369, + "step": 6135 + }, + { + "epoch": 1.89, + "learning_rate": 1.5104580980944141e-05, + "loss": 1.0031, + "step": 6140 + }, + { + "epoch": 1.89, + "learning_rate": 1.5067606438396595e-05, + "loss": 0.9775, + "step": 6145 + }, + { + "epoch": 1.89, + "learning_rate": 1.5030657675934256e-05, + "loss": 1.0644, + "step": 6150 + }, + { + "epoch": 1.89, + "learning_rate": 1.4993734789459718e-05, + "loss": 1.0738, + "step": 6155 + }, + { + "epoch": 1.9, + "learning_rate": 1.4956837874808391e-05, + "loss": 1.04, + "step": 6160 + }, + { + "epoch": 1.9, + "learning_rate": 1.4919967027748306e-05, + "loss": 1.1169, + "step": 6165 + }, + { + "epoch": 1.9, + "learning_rate": 1.4883122343979822e-05, + "loss": 1.0017, + "step": 6170 + }, + { + "epoch": 1.9, + "learning_rate": 1.4846303919135355e-05, + "loss": 1.1442, + "step": 6175 + }, + { + "epoch": 1.9, + "learning_rate": 1.4809511848779217e-05, + "loss": 1.0189, + "step": 6180 + }, + { + "epoch": 1.9, + "learning_rate": 1.4772746228407289e-05, + "loss": 1.0581, + "step": 6185 + }, + { + "epoch": 1.9, + "learning_rate": 1.4736007153446801e-05, + "loss": 1.0046, + "step": 6190 + }, + { + "epoch": 1.91, + "learning_rate": 1.4699294719256091e-05, + "loss": 1.0746, + "step": 6195 + }, + { + "epoch": 1.91, + "learning_rate": 1.466260902112433e-05, + "loss": 1.0498, + "step": 6200 + }, + { + "epoch": 1.91, + "learning_rate": 1.4625950154271317e-05, + "loss": 1.0003, + "step": 6205 + }, + { + "epoch": 1.91, + "learning_rate": 1.4589318213847197e-05, + "loss": 1.034, + "step": 6210 + }, + { + "epoch": 1.91, + "learning_rate": 1.4552713294932226e-05, + "loss": 1.0697, + "step": 6215 + }, + { + "epoch": 1.91, + "learning_rate": 1.4516135492536539e-05, + "loss": 1.0007, + "step": 6220 + }, + { + "epoch": 1.92, + "learning_rate": 1.447958490159987e-05, + "loss": 1.0996, + "step": 6225 + }, + { + "epoch": 1.92, + "learning_rate": 1.444306161699131e-05, + "loss": 1.0804, + "step": 6230 + }, + { + "epoch": 1.92, + "learning_rate": 1.4406565733509126e-05, + "loss": 1.0114, + "step": 6235 + }, + { + "epoch": 1.92, + "learning_rate": 1.4370097345880407e-05, + "loss": 0.9933, + "step": 6240 + }, + { + "epoch": 1.92, + "learning_rate": 1.433365654876091e-05, + "loss": 1.0244, + "step": 6245 + }, + { + "epoch": 1.92, + "learning_rate": 1.4297243436734797e-05, + "loss": 1.008, + "step": 6250 + }, + { + "epoch": 1.92, + "learning_rate": 1.4260858104314297e-05, + "loss": 1.0374, + "step": 6255 + }, + { + "epoch": 1.93, + "learning_rate": 1.422450064593961e-05, + "loss": 0.9268, + "step": 6260 + }, + { + "epoch": 1.93, + "learning_rate": 1.4188171155978566e-05, + "loss": 1.0202, + "step": 6265 + }, + { + "epoch": 1.93, + "learning_rate": 1.4151869728726378e-05, + "loss": 1.0609, + "step": 6270 + }, + { + "epoch": 1.93, + "learning_rate": 1.4115596458405459e-05, + "loss": 0.9536, + "step": 6275 + }, + { + "epoch": 1.93, + "learning_rate": 1.4079351439165106e-05, + "loss": 1.0779, + "step": 6280 + }, + { + "epoch": 1.93, + "learning_rate": 1.4043134765081297e-05, + "loss": 0.9873, + "step": 6285 + }, + { + "epoch": 1.94, + "learning_rate": 1.4006946530156462e-05, + "loss": 0.9834, + "step": 6290 + }, + { + "epoch": 1.94, + "learning_rate": 1.397078682831917e-05, + "loss": 1.0204, + "step": 6295 + }, + { + "epoch": 1.94, + "learning_rate": 1.3934655753423976e-05, + "loss": 1.0785, + "step": 6300 + }, + { + "epoch": 1.94, + "learning_rate": 1.389855339925113e-05, + "loss": 1.0589, + "step": 6305 + }, + { + "epoch": 1.94, + "learning_rate": 1.386247985950628e-05, + "loss": 1.0557, + "step": 6310 + }, + { + "epoch": 1.94, + "learning_rate": 1.3826435227820344e-05, + "loss": 1.1363, + "step": 6315 + }, + { + "epoch": 1.94, + "learning_rate": 1.3790419597749199e-05, + "loss": 0.9302, + "step": 6320 + }, + { + "epoch": 1.95, + "learning_rate": 1.3754433062773409e-05, + "loss": 1.0221, + "step": 6325 + }, + { + "epoch": 1.95, + "learning_rate": 1.3718475716298073e-05, + "loss": 0.9994, + "step": 6330 + }, + { + "epoch": 1.95, + "learning_rate": 1.368254765165249e-05, + "loss": 1.1109, + "step": 6335 + }, + { + "epoch": 1.95, + "learning_rate": 1.3646648962089965e-05, + "loss": 1.0474, + "step": 6340 + }, + { + "epoch": 1.95, + "learning_rate": 1.3610779740787571e-05, + "loss": 1.0186, + "step": 6345 + }, + { + "epoch": 1.95, + "learning_rate": 1.3574940080845875e-05, + "loss": 1.03, + "step": 6350 + }, + { + "epoch": 1.96, + "learning_rate": 1.3539130075288731e-05, + "loss": 1.0271, + "step": 6355 + }, + { + "epoch": 1.96, + "learning_rate": 1.3503349817063047e-05, + "loss": 1.041, + "step": 6360 + }, + { + "epoch": 1.96, + "learning_rate": 1.3467599399038445e-05, + "loss": 0.9458, + "step": 6365 + }, + { + "epoch": 1.96, + "learning_rate": 1.3431878914007167e-05, + "loss": 0.9934, + "step": 6370 + }, + { + "epoch": 1.96, + "learning_rate": 1.3396188454683745e-05, + "loss": 1.0479, + "step": 6375 + }, + { + "epoch": 1.96, + "learning_rate": 1.3360528113704751e-05, + "loss": 1.0472, + "step": 6380 + }, + { + "epoch": 1.96, + "learning_rate": 1.332489798362862e-05, + "loss": 0.974, + "step": 6385 + }, + { + "epoch": 1.97, + "learning_rate": 1.3289298156935348e-05, + "loss": 1.0098, + "step": 6390 + }, + { + "epoch": 1.97, + "learning_rate": 1.3253728726026276e-05, + "loss": 1.0218, + "step": 6395 + }, + { + "epoch": 1.97, + "learning_rate": 1.321818978322387e-05, + "loss": 1.0304, + "step": 6400 + }, + { + "epoch": 1.97, + "learning_rate": 1.3182681420771453e-05, + "loss": 1.0134, + "step": 6405 + }, + { + "epoch": 1.97, + "learning_rate": 1.3147203730832963e-05, + "loss": 1.0227, + "step": 6410 + }, + { + "epoch": 1.97, + "learning_rate": 1.3111756805492752e-05, + "loss": 0.9703, + "step": 6415 + }, + { + "epoch": 1.98, + "learning_rate": 1.3076340736755293e-05, + "loss": 1.1353, + "step": 6420 + }, + { + "epoch": 1.98, + "learning_rate": 1.304095561654498e-05, + "loss": 0.9482, + "step": 6425 + }, + { + "epoch": 1.98, + "learning_rate": 1.3005601536705889e-05, + "loss": 0.9812, + "step": 6430 + }, + { + "epoch": 1.98, + "learning_rate": 1.2970278589001505e-05, + "loss": 1.0317, + "step": 6435 + }, + { + "epoch": 1.98, + "learning_rate": 1.293498686511454e-05, + "loss": 1.0497, + "step": 6440 + }, + { + "epoch": 1.98, + "learning_rate": 1.2899726456646635e-05, + "loss": 1.0129, + "step": 6445 + }, + { + "epoch": 1.98, + "learning_rate": 1.2864497455118152e-05, + "loss": 1.0313, + "step": 6450 + }, + { + "epoch": 1.99, + "learning_rate": 1.2829299951967954e-05, + "loss": 1.0461, + "step": 6455 + }, + { + "epoch": 1.99, + "learning_rate": 1.2794134038553141e-05, + "loss": 0.9857, + "step": 6460 + }, + { + "epoch": 1.99, + "learning_rate": 1.2758999806148813e-05, + "loss": 1.0559, + "step": 6465 + }, + { + "epoch": 1.99, + "learning_rate": 1.2723897345947828e-05, + "loss": 1.0278, + "step": 6470 + }, + { + "epoch": 1.99, + "learning_rate": 1.2688826749060611e-05, + "loss": 1.0613, + "step": 6475 + }, + { + "epoch": 1.99, + "learning_rate": 1.2653788106514852e-05, + "loss": 0.9064, + "step": 6480 + }, + { + "epoch": 2.0, + "learning_rate": 1.2618781509255332e-05, + "loss": 0.9998, + "step": 6485 + }, + { + "epoch": 2.0, + "learning_rate": 1.2583807048143617e-05, + "loss": 1.0353, + "step": 6490 + }, + { + "epoch": 2.0, + "learning_rate": 1.2548864813957909e-05, + "loss": 0.9836, + "step": 6495 + }, + { + "epoch": 2.0, + "learning_rate": 1.2513954897392727e-05, + "loss": 0.9905, + "step": 6500 + }, + { + "epoch": 2.0, + "learning_rate": 1.2479077389058708e-05, + "loss": 0.9519, + "step": 6505 + }, + { + "epoch": 2.0, + "learning_rate": 1.2444232379482398e-05, + "loss": 1.0415, + "step": 6510 + }, + { + "epoch": 2.0, + "learning_rate": 1.2409419959105981e-05, + "loss": 0.9797, + "step": 6515 + }, + { + "epoch": 2.01, + "learning_rate": 1.237464021828704e-05, + "loss": 1.0352, + "step": 6520 + }, + { + "epoch": 2.01, + "learning_rate": 1.233989324729834e-05, + "loss": 1.1157, + "step": 6525 + }, + { + "epoch": 2.01, + "learning_rate": 1.2305179136327608e-05, + "loss": 0.9363, + "step": 6530 + }, + { + "epoch": 2.01, + "learning_rate": 1.2270497975477253e-05, + "loss": 1.0566, + "step": 6535 + }, + { + "epoch": 2.01, + "learning_rate": 1.2235849854764194e-05, + "loss": 1.0008, + "step": 6540 + }, + { + "epoch": 2.01, + "learning_rate": 1.2201234864119554e-05, + "loss": 0.9184, + "step": 6545 + }, + { + "epoch": 2.02, + "learning_rate": 1.2166653093388506e-05, + "loss": 0.8536, + "step": 6550 + }, + { + "epoch": 2.02, + "learning_rate": 1.2132104632329963e-05, + "loss": 1.0431, + "step": 6555 + }, + { + "epoch": 2.02, + "learning_rate": 1.2097589570616394e-05, + "loss": 0.9783, + "step": 6560 + }, + { + "epoch": 2.02, + "learning_rate": 1.2063107997833581e-05, + "loss": 1.0286, + "step": 6565 + }, + { + "epoch": 2.02, + "learning_rate": 1.2028660003480399e-05, + "loss": 1.0018, + "step": 6570 + }, + { + "epoch": 2.02, + "learning_rate": 1.1994245676968538e-05, + "loss": 1.0455, + "step": 6575 + }, + { + "epoch": 2.02, + "learning_rate": 1.1959865107622307e-05, + "loss": 1.0873, + "step": 6580 + }, + { + "epoch": 2.03, + "learning_rate": 1.1925518384678421e-05, + "loss": 1.0165, + "step": 6585 + }, + { + "epoch": 2.03, + "learning_rate": 1.1891205597285712e-05, + "loss": 0.8837, + "step": 6590 + }, + { + "epoch": 2.03, + "learning_rate": 1.1856926834504963e-05, + "loss": 1.0718, + "step": 6595 + }, + { + "epoch": 2.03, + "learning_rate": 1.1822682185308612e-05, + "loss": 0.9818, + "step": 6600 + }, + { + "epoch": 2.03, + "learning_rate": 1.1788471738580581e-05, + "loss": 1.0055, + "step": 6605 + }, + { + "epoch": 2.03, + "learning_rate": 1.1754295583116004e-05, + "loss": 0.9947, + "step": 6610 + }, + { + "epoch": 2.04, + "learning_rate": 1.1720153807620999e-05, + "loss": 0.9422, + "step": 6615 + }, + { + "epoch": 2.04, + "learning_rate": 1.168604650071247e-05, + "loss": 1.0686, + "step": 6620 + }, + { + "epoch": 2.04, + "learning_rate": 1.1651973750917854e-05, + "loss": 1.0849, + "step": 6625 + }, + { + "epoch": 2.04, + "learning_rate": 1.1617935646674885e-05, + "loss": 0.9938, + "step": 6630 + }, + { + "epoch": 2.04, + "learning_rate": 1.1583932276331358e-05, + "loss": 0.9205, + "step": 6635 + }, + { + "epoch": 2.04, + "learning_rate": 1.154996372814495e-05, + "loss": 1.1418, + "step": 6640 + }, + { + "epoch": 2.04, + "learning_rate": 1.1516030090282914e-05, + "loss": 1.0868, + "step": 6645 + }, + { + "epoch": 2.05, + "learning_rate": 1.1482131450821937e-05, + "loss": 0.9712, + "step": 6650 + }, + { + "epoch": 2.05, + "learning_rate": 1.1448267897747818e-05, + "loss": 1.035, + "step": 6655 + }, + { + "epoch": 2.05, + "learning_rate": 1.1414439518955334e-05, + "loss": 1.0963, + "step": 6660 + }, + { + "epoch": 2.05, + "learning_rate": 1.1380646402247927e-05, + "loss": 1.0696, + "step": 6665 + }, + { + "epoch": 2.05, + "learning_rate": 1.1346888635337522e-05, + "loss": 0.9375, + "step": 6670 + }, + { + "epoch": 2.05, + "learning_rate": 1.1313166305844306e-05, + "loss": 0.9695, + "step": 6675 + }, + { + "epoch": 2.06, + "learning_rate": 1.1279479501296492e-05, + "loss": 1.0901, + "step": 6680 + }, + { + "epoch": 2.06, + "learning_rate": 1.1245828309130061e-05, + "loss": 1.0466, + "step": 6685 + }, + { + "epoch": 2.06, + "learning_rate": 1.1212212816688558e-05, + "loss": 1.04, + "step": 6690 + }, + { + "epoch": 2.06, + "learning_rate": 1.1178633111222909e-05, + "loss": 1.1043, + "step": 6695 + }, + { + "epoch": 2.06, + "learning_rate": 1.1145089279891102e-05, + "loss": 1.0481, + "step": 6700 + }, + { + "epoch": 2.06, + "learning_rate": 1.1111581409758043e-05, + "loss": 1.0453, + "step": 6705 + }, + { + "epoch": 2.06, + "learning_rate": 1.107810958779531e-05, + "loss": 0.9099, + "step": 6710 + }, + { + "epoch": 2.07, + "learning_rate": 1.1044673900880858e-05, + "loss": 0.9629, + "step": 6715 + }, + { + "epoch": 2.07, + "learning_rate": 1.101127443579891e-05, + "loss": 1.0261, + "step": 6720 + }, + { + "epoch": 2.07, + "learning_rate": 1.0977911279239663e-05, + "loss": 1.0423, + "step": 6725 + }, + { + "epoch": 2.07, + "learning_rate": 1.0944584517799045e-05, + "loss": 1.028, + "step": 6730 + }, + { + "epoch": 2.07, + "learning_rate": 1.091129423797855e-05, + "loss": 1.0888, + "step": 6735 + }, + { + "epoch": 2.07, + "learning_rate": 1.0878040526184965e-05, + "loss": 0.9095, + "step": 6740 + }, + { + "epoch": 2.08, + "learning_rate": 1.0844823468730158e-05, + "loss": 0.9639, + "step": 6745 + }, + { + "epoch": 2.08, + "learning_rate": 1.081164315183088e-05, + "loss": 1.0108, + "step": 6750 + }, + { + "epoch": 2.08, + "learning_rate": 1.0778499661608491e-05, + "loss": 0.9688, + "step": 6755 + }, + { + "epoch": 2.08, + "learning_rate": 1.0745393084088789e-05, + "loss": 1.0971, + "step": 6760 + }, + { + "epoch": 2.08, + "learning_rate": 1.0712323505201773e-05, + "loss": 0.9924, + "step": 6765 + }, + { + "epoch": 2.08, + "learning_rate": 1.0679291010781362e-05, + "loss": 1.0124, + "step": 6770 + }, + { + "epoch": 2.08, + "learning_rate": 1.0646295686565259e-05, + "loss": 0.9781, + "step": 6775 + }, + { + "epoch": 2.09, + "learning_rate": 1.0613337618194691e-05, + "loss": 1.0616, + "step": 6780 + }, + { + "epoch": 2.09, + "learning_rate": 1.0580416891214162e-05, + "loss": 1.0148, + "step": 6785 + }, + { + "epoch": 2.09, + "learning_rate": 1.0547533591071285e-05, + "loss": 1.0322, + "step": 6790 + }, + { + "epoch": 2.09, + "learning_rate": 1.0514687803116499e-05, + "loss": 0.9596, + "step": 6795 + }, + { + "epoch": 2.09, + "learning_rate": 1.0481879612602882e-05, + "loss": 0.9821, + "step": 6800 + }, + { + "epoch": 2.09, + "learning_rate": 1.0449109104685958e-05, + "loss": 1.0154, + "step": 6805 + }, + { + "epoch": 2.1, + "learning_rate": 1.0416376364423396e-05, + "loss": 0.988, + "step": 6810 + }, + { + "epoch": 2.1, + "learning_rate": 1.0383681476774876e-05, + "loss": 0.9057, + "step": 6815 + }, + { + "epoch": 2.1, + "learning_rate": 1.035102452660183e-05, + "loss": 1.0241, + "step": 6820 + }, + { + "epoch": 2.1, + "learning_rate": 1.031840559866717e-05, + "loss": 1.0098, + "step": 6825 + }, + { + "epoch": 2.1, + "learning_rate": 1.0285824777635172e-05, + "loss": 1.0052, + "step": 6830 + }, + { + "epoch": 2.1, + "learning_rate": 1.0253282148071198e-05, + "loss": 0.9371, + "step": 6835 + }, + { + "epoch": 2.1, + "learning_rate": 1.022077779444145e-05, + "loss": 1.019, + "step": 6840 + }, + { + "epoch": 2.11, + "learning_rate": 1.0188311801112823e-05, + "loss": 1.15, + "step": 6845 + }, + { + "epoch": 2.11, + "learning_rate": 1.0155884252352616e-05, + "loss": 0.993, + "step": 6850 + }, + { + "epoch": 2.11, + "learning_rate": 1.0123495232328342e-05, + "loss": 1.0949, + "step": 6855 + }, + { + "epoch": 2.11, + "learning_rate": 1.009114482510754e-05, + "loss": 0.9747, + "step": 6860 + }, + { + "epoch": 2.11, + "learning_rate": 1.0058833114657493e-05, + "loss": 1.12, + "step": 6865 + }, + { + "epoch": 2.11, + "learning_rate": 1.0026560184845066e-05, + "loss": 1.0195, + "step": 6870 + }, + { + "epoch": 2.12, + "learning_rate": 9.994326119436478e-06, + "loss": 1.086, + "step": 6875 + }, + { + "epoch": 2.12, + "learning_rate": 9.962131002097022e-06, + "loss": 1.0505, + "step": 6880 + }, + { + "epoch": 2.12, + "learning_rate": 9.929974916390953e-06, + "loss": 1.1073, + "step": 6885 + }, + { + "epoch": 2.12, + "learning_rate": 9.897857945781196e-06, + "loss": 0.9785, + "step": 6890 + }, + { + "epoch": 2.12, + "learning_rate": 9.865780173629147e-06, + "loss": 0.9759, + "step": 6895 + }, + { + "epoch": 2.12, + "learning_rate": 9.833741683194475e-06, + "loss": 1.0394, + "step": 6900 + }, + { + "epoch": 2.12, + "learning_rate": 9.801742557634872e-06, + "loss": 1.1753, + "step": 6905 + }, + { + "epoch": 2.13, + "learning_rate": 9.76978288000586e-06, + "loss": 1.011, + "step": 6910 + }, + { + "epoch": 2.13, + "learning_rate": 9.73786273326059e-06, + "loss": 0.947, + "step": 6915 + }, + { + "epoch": 2.13, + "learning_rate": 9.70598220024958e-06, + "loss": 1.0201, + "step": 6920 + }, + { + "epoch": 2.13, + "learning_rate": 9.674141363720554e-06, + "loss": 1.0308, + "step": 6925 + }, + { + "epoch": 2.13, + "learning_rate": 9.642340306318203e-06, + "loss": 1.0885, + "step": 6930 + }, + { + "epoch": 2.13, + "learning_rate": 9.61057911058393e-06, + "loss": 0.978, + "step": 6935 + }, + { + "epoch": 2.14, + "learning_rate": 9.578857858955715e-06, + "loss": 0.9905, + "step": 6940 + }, + { + "epoch": 2.14, + "learning_rate": 9.547176633767857e-06, + "loss": 1.0224, + "step": 6945 + }, + { + "epoch": 2.14, + "learning_rate": 9.515535517250737e-06, + "loss": 1.0058, + "step": 6950 + }, + { + "epoch": 2.14, + "learning_rate": 9.483934591530668e-06, + "loss": 0.9798, + "step": 6955 + }, + { + "epoch": 2.14, + "learning_rate": 9.452373938629619e-06, + "loss": 1.0367, + "step": 6960 + }, + { + "epoch": 2.14, + "learning_rate": 9.420853640465025e-06, + "loss": 1.0499, + "step": 6965 + }, + { + "epoch": 2.14, + "learning_rate": 9.389373778849612e-06, + "loss": 1.0117, + "step": 6970 + }, + { + "epoch": 2.15, + "learning_rate": 9.357934435491106e-06, + "loss": 1.1245, + "step": 6975 + }, + { + "epoch": 2.15, + "learning_rate": 9.3265356919921e-06, + "loss": 1.0715, + "step": 6980 + }, + { + "epoch": 2.15, + "learning_rate": 9.295177629849802e-06, + "loss": 1.0251, + "step": 6985 + }, + { + "epoch": 2.15, + "learning_rate": 9.26386033045582e-06, + "loss": 1.0146, + "step": 6990 + }, + { + "epoch": 2.15, + "learning_rate": 9.232583875095949e-06, + "loss": 0.9077, + "step": 6995 + }, + { + "epoch": 2.15, + "learning_rate": 9.201348344950001e-06, + "loss": 0.9325, + "step": 7000 + }, + { + "epoch": 2.16, + "learning_rate": 9.170153821091537e-06, + "loss": 0.9624, + "step": 7005 + }, + { + "epoch": 2.16, + "learning_rate": 9.13900038448771e-06, + "loss": 1.0024, + "step": 7010 + }, + { + "epoch": 2.16, + "learning_rate": 9.107888115999002e-06, + "loss": 0.99, + "step": 7015 + }, + { + "epoch": 2.16, + "learning_rate": 9.07681709637905e-06, + "loss": 0.9876, + "step": 7020 + }, + { + "epoch": 2.16, + "learning_rate": 9.045787406274437e-06, + "loss": 0.923, + "step": 7025 + }, + { + "epoch": 2.16, + "learning_rate": 9.014799126224471e-06, + "loss": 1.0948, + "step": 7030 + }, + { + "epoch": 2.16, + "learning_rate": 8.983852336660959e-06, + "loss": 0.9757, + "step": 7035 + }, + { + "epoch": 2.17, + "learning_rate": 8.952947117908047e-06, + "loss": 1.0848, + "step": 7040 + }, + { + "epoch": 2.17, + "learning_rate": 8.922083550181959e-06, + "loss": 1.0356, + "step": 7045 + }, + { + "epoch": 2.17, + "learning_rate": 8.891261713590807e-06, + "loss": 0.9513, + "step": 7050 + }, + { + "epoch": 2.17, + "learning_rate": 8.860481688134417e-06, + "loss": 1.1004, + "step": 7055 + }, + { + "epoch": 2.17, + "learning_rate": 8.829743553704056e-06, + "loss": 1.0715, + "step": 7060 + }, + { + "epoch": 2.17, + "learning_rate": 8.799047390082296e-06, + "loss": 0.9744, + "step": 7065 + }, + { + "epoch": 2.18, + "learning_rate": 8.768393276942743e-06, + "loss": 0.9795, + "step": 7070 + }, + { + "epoch": 2.18, + "learning_rate": 8.737781293849864e-06, + "loss": 1.0169, + "step": 7075 + }, + { + "epoch": 2.18, + "learning_rate": 8.707211520258782e-06, + "loss": 0.9364, + "step": 7080 + }, + { + "epoch": 2.18, + "learning_rate": 8.676684035515076e-06, + "loss": 0.9445, + "step": 7085 + }, + { + "epoch": 2.18, + "learning_rate": 8.646198918854526e-06, + "loss": 0.9883, + "step": 7090 + }, + { + "epoch": 2.18, + "learning_rate": 8.61575624940298e-06, + "loss": 0.9343, + "step": 7095 + }, + { + "epoch": 2.18, + "learning_rate": 8.585356106176094e-06, + "loss": 0.986, + "step": 7100 + }, + { + "epoch": 2.19, + "learning_rate": 8.55499856807913e-06, + "loss": 1.0758, + "step": 7105 + }, + { + "epoch": 2.19, + "learning_rate": 8.524683713906805e-06, + "loss": 0.971, + "step": 7110 + }, + { + "epoch": 2.19, + "learning_rate": 8.49441162234301e-06, + "loss": 0.9793, + "step": 7115 + }, + { + "epoch": 2.19, + "learning_rate": 8.464182371960668e-06, + "loss": 1.0048, + "step": 7120 + }, + { + "epoch": 2.19, + "learning_rate": 8.433996041221492e-06, + "loss": 1.0582, + "step": 7125 + }, + { + "epoch": 2.19, + "learning_rate": 8.403852708475792e-06, + "loss": 1.0345, + "step": 7130 + }, + { + "epoch": 2.2, + "learning_rate": 8.373752451962286e-06, + "loss": 0.9942, + "step": 7135 + }, + { + "epoch": 2.2, + "learning_rate": 8.34369534980789e-06, + "loss": 0.9963, + "step": 7140 + }, + { + "epoch": 2.2, + "learning_rate": 8.31368148002748e-06, + "loss": 1.0166, + "step": 7145 + }, + { + "epoch": 2.2, + "learning_rate": 8.283710920523763e-06, + "loss": 1.0744, + "step": 7150 + }, + { + "epoch": 2.2, + "learning_rate": 8.253783749086993e-06, + "loss": 1.0384, + "step": 7155 + }, + { + "epoch": 2.2, + "learning_rate": 8.223900043394825e-06, + "loss": 1.0744, + "step": 7160 + }, + { + "epoch": 2.2, + "learning_rate": 8.194059881012105e-06, + "loss": 1.0495, + "step": 7165 + }, + { + "epoch": 2.21, + "learning_rate": 8.164263339390635e-06, + "loss": 1.032, + "step": 7170 + }, + { + "epoch": 2.21, + "learning_rate": 8.13451049586903e-06, + "loss": 0.9786, + "step": 7175 + }, + { + "epoch": 2.21, + "learning_rate": 8.104801427672456e-06, + "loss": 1.1122, + "step": 7180 + }, + { + "epoch": 2.21, + "learning_rate": 8.07513621191246e-06, + "loss": 0.9577, + "step": 7185 + }, + { + "epoch": 2.21, + "learning_rate": 8.045514925586784e-06, + "loss": 0.9227, + "step": 7190 + }, + { + "epoch": 2.21, + "learning_rate": 8.015937645579148e-06, + "loss": 1.0198, + "step": 7195 + }, + { + "epoch": 2.22, + "learning_rate": 7.986404448659023e-06, + "loss": 1.1088, + "step": 7200 + }, + { + "epoch": 2.22, + "learning_rate": 7.956915411481505e-06, + "loss": 1.079, + "step": 7205 + }, + { + "epoch": 2.22, + "learning_rate": 7.927470610587028e-06, + "loss": 1.0856, + "step": 7210 + }, + { + "epoch": 2.22, + "learning_rate": 7.898070122401224e-06, + "loss": 1.1321, + "step": 7215 + }, + { + "epoch": 2.22, + "learning_rate": 7.868714023234727e-06, + "loss": 1.0075, + "step": 7220 + }, + { + "epoch": 2.22, + "learning_rate": 7.839402389282924e-06, + "loss": 0.879, + "step": 7225 + }, + { + "epoch": 2.22, + "learning_rate": 7.810135296625818e-06, + "loss": 1.1296, + "step": 7230 + }, + { + "epoch": 2.23, + "learning_rate": 7.78091282122779e-06, + "loss": 0.9341, + "step": 7235 + }, + { + "epoch": 2.23, + "learning_rate": 7.751735038937405e-06, + "loss": 1.0522, + "step": 7240 + }, + { + "epoch": 2.23, + "learning_rate": 7.722602025487243e-06, + "loss": 0.9456, + "step": 7245 + }, + { + "epoch": 2.23, + "learning_rate": 7.693513856493684e-06, + "loss": 0.9456, + "step": 7250 + }, + { + "epoch": 2.23, + "learning_rate": 7.6644706074567e-06, + "loss": 0.9922, + "step": 7255 + }, + { + "epoch": 2.23, + "learning_rate": 7.63547235375966e-06, + "loss": 1.0362, + "step": 7260 + }, + { + "epoch": 2.24, + "learning_rate": 7.6065191706691795e-06, + "loss": 0.9765, + "step": 7265 + }, + { + "epoch": 2.24, + "learning_rate": 7.577611133334858e-06, + "loss": 0.9506, + "step": 7270 + }, + { + "epoch": 2.24, + "learning_rate": 7.54874831678914e-06, + "loss": 1.012, + "step": 7275 + }, + { + "epoch": 2.24, + "learning_rate": 7.519930795947072e-06, + "loss": 0.9268, + "step": 7280 + }, + { + "epoch": 2.24, + "learning_rate": 7.491158645606167e-06, + "loss": 1.0983, + "step": 7285 + }, + { + "epoch": 2.24, + "learning_rate": 7.462431940446135e-06, + "loss": 1.0734, + "step": 7290 + }, + { + "epoch": 2.24, + "learning_rate": 7.433750755028773e-06, + "loss": 0.9611, + "step": 7295 + }, + { + "epoch": 2.25, + "learning_rate": 7.40511516379769e-06, + "loss": 0.9591, + "step": 7300 + }, + { + "epoch": 2.25, + "learning_rate": 7.376525241078189e-06, + "loss": 0.9786, + "step": 7305 + }, + { + "epoch": 2.25, + "learning_rate": 7.347981061077011e-06, + "loss": 0.9826, + "step": 7310 + }, + { + "epoch": 2.25, + "learning_rate": 7.319482697882168e-06, + "loss": 1.1188, + "step": 7315 + }, + { + "epoch": 2.25, + "learning_rate": 7.291030225462781e-06, + "loss": 1.0463, + "step": 7320 + }, + { + "epoch": 2.25, + "learning_rate": 7.262623717668821e-06, + "loss": 1.0801, + "step": 7325 + }, + { + "epoch": 2.26, + "learning_rate": 7.2342632482309825e-06, + "loss": 0.9915, + "step": 7330 + }, + { + "epoch": 2.26, + "learning_rate": 7.205948890760464e-06, + "loss": 0.985, + "step": 7335 + }, + { + "epoch": 2.26, + "learning_rate": 7.177680718748767e-06, + "loss": 1.1126, + "step": 7340 + }, + { + "epoch": 2.26, + "learning_rate": 7.149458805567505e-06, + "loss": 0.954, + "step": 7345 + }, + { + "epoch": 2.26, + "learning_rate": 7.1212832244682585e-06, + "loss": 1.035, + "step": 7350 + }, + { + "epoch": 2.26, + "learning_rate": 7.093154048582313e-06, + "loss": 1.0202, + "step": 7355 + }, + { + "epoch": 2.26, + "learning_rate": 7.065071350920538e-06, + "loss": 0.9362, + "step": 7360 + }, + { + "epoch": 2.27, + "learning_rate": 7.037035204373147e-06, + "loss": 1.0565, + "step": 7365 + }, + { + "epoch": 2.27, + "learning_rate": 7.009045681709522e-06, + "loss": 1.0125, + "step": 7370 + }, + { + "epoch": 2.27, + "learning_rate": 6.981102855578062e-06, + "loss": 1.0774, + "step": 7375 + }, + { + "epoch": 2.27, + "learning_rate": 6.953206798505918e-06, + "loss": 1.1028, + "step": 7380 + }, + { + "epoch": 2.27, + "learning_rate": 6.925357582898886e-06, + "loss": 0.875, + "step": 7385 + }, + { + "epoch": 2.27, + "learning_rate": 6.8975552810411765e-06, + "loss": 0.99, + "step": 7390 + }, + { + "epoch": 2.28, + "learning_rate": 6.869799965095214e-06, + "loss": 1.0323, + "step": 7395 + }, + { + "epoch": 2.28, + "learning_rate": 6.842091707101473e-06, + "loss": 1.0693, + "step": 7400 + }, + { + "epoch": 2.28, + "learning_rate": 6.814430578978309e-06, + "loss": 0.8959, + "step": 7405 + }, + { + "epoch": 2.28, + "learning_rate": 6.786816652521719e-06, + "loss": 0.9174, + "step": 7410 + }, + { + "epoch": 2.28, + "learning_rate": 6.759249999405212e-06, + "loss": 1.0288, + "step": 7415 + }, + { + "epoch": 2.28, + "learning_rate": 6.73173069117958e-06, + "loss": 0.9369, + "step": 7420 + }, + { + "epoch": 2.28, + "learning_rate": 6.704258799272722e-06, + "loss": 0.9579, + "step": 7425 + }, + { + "epoch": 2.29, + "learning_rate": 6.676834394989495e-06, + "loss": 1.0773, + "step": 7430 + }, + { + "epoch": 2.29, + "learning_rate": 6.649457549511459e-06, + "loss": 0.9937, + "step": 7435 + }, + { + "epoch": 2.29, + "learning_rate": 6.622128333896768e-06, + "loss": 1.0075, + "step": 7440 + }, + { + "epoch": 2.29, + "learning_rate": 6.594846819079939e-06, + "loss": 1.0342, + "step": 7445 + }, + { + "epoch": 2.29, + "learning_rate": 6.56761307587167e-06, + "loss": 0.9605, + "step": 7450 + }, + { + "epoch": 2.29, + "learning_rate": 6.540427174958661e-06, + "loss": 1.0853, + "step": 7455 + }, + { + "epoch": 2.3, + "learning_rate": 6.513289186903463e-06, + "loss": 0.9501, + "step": 7460 + }, + { + "epoch": 2.3, + "learning_rate": 6.486199182144229e-06, + "loss": 1.076, + "step": 7465 + }, + { + "epoch": 2.3, + "learning_rate": 6.459157230994603e-06, + "loss": 1.0476, + "step": 7470 + }, + { + "epoch": 2.3, + "learning_rate": 6.432163403643482e-06, + "loss": 0.9046, + "step": 7475 + }, + { + "epoch": 2.3, + "learning_rate": 6.405217770154853e-06, + "loss": 0.9707, + "step": 7480 + }, + { + "epoch": 2.3, + "learning_rate": 6.378320400467636e-06, + "loss": 1.0573, + "step": 7485 + }, + { + "epoch": 2.3, + "learning_rate": 6.3514713643954475e-06, + "loss": 1.0683, + "step": 7490 + }, + { + "epoch": 2.31, + "learning_rate": 6.324670731626478e-06, + "loss": 1.0403, + "step": 7495 + }, + { + "epoch": 2.31, + "learning_rate": 6.297918571723288e-06, + "loss": 1.0485, + "step": 7500 + }, + { + "epoch": 2.31, + "learning_rate": 6.271214954122581e-06, + "loss": 1.0855, + "step": 7505 + }, + { + "epoch": 2.31, + "learning_rate": 6.244559948135109e-06, + "loss": 1.0325, + "step": 7510 + }, + { + "epoch": 2.31, + "learning_rate": 6.217953622945449e-06, + "loss": 1.0469, + "step": 7515 + }, + { + "epoch": 2.31, + "learning_rate": 6.191396047611794e-06, + "loss": 1.0573, + "step": 7520 + }, + { + "epoch": 2.32, + "learning_rate": 6.164887291065838e-06, + "loss": 1.1132, + "step": 7525 + }, + { + "epoch": 2.32, + "learning_rate": 6.138427422112539e-06, + "loss": 0.9797, + "step": 7530 + }, + { + "epoch": 2.32, + "learning_rate": 6.1120165094299655e-06, + "loss": 1.0327, + "step": 7535 + }, + { + "epoch": 2.32, + "learning_rate": 6.085654621569137e-06, + "loss": 1.0003, + "step": 7540 + }, + { + "epoch": 2.32, + "learning_rate": 6.0593418269538045e-06, + "loss": 1.0296, + "step": 7545 + }, + { + "epoch": 2.32, + "learning_rate": 6.0330781938803034e-06, + "loss": 1.1049, + "step": 7550 + }, + { + "epoch": 2.32, + "learning_rate": 6.006863790517392e-06, + "loss": 0.8916, + "step": 7555 + }, + { + "epoch": 2.33, + "learning_rate": 5.980698684905989e-06, + "loss": 1.1291, + "step": 7560 + }, + { + "epoch": 2.33, + "learning_rate": 5.954582944959111e-06, + "loss": 1.0977, + "step": 7565 + }, + { + "epoch": 2.33, + "learning_rate": 5.928516638461639e-06, + "loss": 1.0215, + "step": 7570 + }, + { + "epoch": 2.33, + "learning_rate": 5.902499833070119e-06, + "loss": 0.9418, + "step": 7575 + }, + { + "epoch": 2.33, + "learning_rate": 5.876532596312645e-06, + "loss": 1.0864, + "step": 7580 + }, + { + "epoch": 2.33, + "learning_rate": 5.850614995588627e-06, + "loss": 0.9589, + "step": 7585 + }, + { + "epoch": 2.34, + "learning_rate": 5.824747098168651e-06, + "loss": 1.1321, + "step": 7590 + }, + { + "epoch": 2.34, + "learning_rate": 5.798928971194301e-06, + "loss": 1.0365, + "step": 7595 + }, + { + "epoch": 2.34, + "learning_rate": 5.773160681677983e-06, + "loss": 0.9498, + "step": 7600 + }, + { + "epoch": 2.34, + "learning_rate": 5.747442296502725e-06, + "loss": 0.8712, + "step": 7605 + }, + { + "epoch": 2.34, + "learning_rate": 5.721773882422057e-06, + "loss": 1.0132, + "step": 7610 + }, + { + "epoch": 2.34, + "learning_rate": 5.69615550605978e-06, + "loss": 1.0136, + "step": 7615 + }, + { + "epoch": 2.34, + "learning_rate": 5.6705872339098186e-06, + "loss": 1.0336, + "step": 7620 + }, + { + "epoch": 2.35, + "learning_rate": 5.645069132336078e-06, + "loss": 1.1022, + "step": 7625 + }, + { + "epoch": 2.35, + "learning_rate": 5.6196012675722055e-06, + "loss": 0.9038, + "step": 7630 + }, + { + "epoch": 2.35, + "learning_rate": 5.594183705721484e-06, + "loss": 1.0103, + "step": 7635 + }, + { + "epoch": 2.35, + "learning_rate": 5.568816512756633e-06, + "loss": 1.0627, + "step": 7640 + }, + { + "epoch": 2.35, + "learning_rate": 5.5434997545196015e-06, + "loss": 1.0406, + "step": 7645 + }, + { + "epoch": 2.35, + "learning_rate": 5.5182334967214725e-06, + "loss": 1.0707, + "step": 7650 + }, + { + "epoch": 2.36, + "learning_rate": 5.493017804942238e-06, + "loss": 1.0736, + "step": 7655 + }, + { + "epoch": 2.36, + "learning_rate": 5.467852744630633e-06, + "loss": 0.9608, + "step": 7660 + }, + { + "epoch": 2.36, + "learning_rate": 5.4427383811039985e-06, + "loss": 1.0646, + "step": 7665 + }, + { + "epoch": 2.36, + "learning_rate": 5.417674779548062e-06, + "loss": 0.9619, + "step": 7670 + }, + { + "epoch": 2.36, + "learning_rate": 5.39266200501681e-06, + "loss": 0.9122, + "step": 7675 + }, + { + "epoch": 2.36, + "learning_rate": 5.367700122432315e-06, + "loss": 1.0728, + "step": 7680 + }, + { + "epoch": 2.36, + "learning_rate": 5.342789196584527e-06, + "loss": 1.0001, + "step": 7685 + }, + { + "epoch": 2.37, + "learning_rate": 5.317929292131163e-06, + "loss": 0.9512, + "step": 7690 + }, + { + "epoch": 2.37, + "learning_rate": 5.293120473597515e-06, + "loss": 1.0009, + "step": 7695 + }, + { + "epoch": 2.37, + "learning_rate": 5.268362805376237e-06, + "loss": 1.0789, + "step": 7700 + }, + { + "epoch": 2.37, + "learning_rate": 5.243656351727258e-06, + "loss": 1.0196, + "step": 7705 + }, + { + "epoch": 2.37, + "learning_rate": 5.219001176777574e-06, + "loss": 1.0164, + "step": 7710 + }, + { + "epoch": 2.37, + "learning_rate": 5.194397344521065e-06, + "loss": 1.0528, + "step": 7715 + }, + { + "epoch": 2.38, + "learning_rate": 5.16984491881837e-06, + "loss": 1.0873, + "step": 7720 + }, + { + "epoch": 2.38, + "learning_rate": 5.145343963396682e-06, + "loss": 1.0581, + "step": 7725 + }, + { + "epoch": 2.38, + "learning_rate": 5.120894541849599e-06, + "loss": 1.0026, + "step": 7730 + }, + { + "epoch": 2.38, + "learning_rate": 5.096496717636984e-06, + "loss": 1.073, + "step": 7735 + }, + { + "epoch": 2.38, + "learning_rate": 5.072150554084745e-06, + "loss": 1.0859, + "step": 7740 + }, + { + "epoch": 2.38, + "learning_rate": 5.04785611438473e-06, + "loss": 0.9867, + "step": 7745 + }, + { + "epoch": 2.38, + "learning_rate": 5.023613461594512e-06, + "loss": 0.9478, + "step": 7750 + }, + { + "epoch": 2.39, + "learning_rate": 4.999422658637254e-06, + "loss": 0.9367, + "step": 7755 + }, + { + "epoch": 2.39, + "learning_rate": 4.9752837683015505e-06, + "loss": 1.0198, + "step": 7760 + }, + { + "epoch": 2.39, + "learning_rate": 4.95119685324125e-06, + "loss": 0.849, + "step": 7765 + }, + { + "epoch": 2.39, + "learning_rate": 4.927161975975284e-06, + "loss": 1.0022, + "step": 7770 + }, + { + "epoch": 2.39, + "learning_rate": 4.903179198887536e-06, + "loss": 1.0787, + "step": 7775 + }, + { + "epoch": 2.39, + "learning_rate": 4.879248584226645e-06, + "loss": 0.9632, + "step": 7780 + }, + { + "epoch": 2.4, + "learning_rate": 4.85537019410586e-06, + "loss": 0.9764, + "step": 7785 + }, + { + "epoch": 2.4, + "learning_rate": 4.831544090502896e-06, + "loss": 0.9713, + "step": 7790 + }, + { + "epoch": 2.4, + "learning_rate": 4.807770335259726e-06, + "loss": 1.0219, + "step": 7795 + }, + { + "epoch": 2.4, + "learning_rate": 4.784048990082484e-06, + "loss": 1.0141, + "step": 7800 + }, + { + "epoch": 2.4, + "learning_rate": 4.760380116541246e-06, + "loss": 1.036, + "step": 7805 + }, + { + "epoch": 2.4, + "learning_rate": 4.736763776069897e-06, + "loss": 1.0524, + "step": 7810 + }, + { + "epoch": 2.4, + "learning_rate": 4.713200029965978e-06, + "loss": 1.1222, + "step": 7815 + }, + { + "epoch": 2.41, + "learning_rate": 4.689688939390521e-06, + "loss": 0.9626, + "step": 7820 + }, + { + "epoch": 2.41, + "learning_rate": 4.666230565367874e-06, + "loss": 1.0736, + "step": 7825 + }, + { + "epoch": 2.41, + "learning_rate": 4.642824968785572e-06, + "loss": 1.0749, + "step": 7830 + }, + { + "epoch": 2.41, + "learning_rate": 4.619472210394154e-06, + "loss": 0.9478, + "step": 7835 + }, + { + "epoch": 2.41, + "learning_rate": 4.596172350807004e-06, + "loss": 1.0398, + "step": 7840 + }, + { + "epoch": 2.41, + "learning_rate": 4.572925450500232e-06, + "loss": 0.943, + "step": 7845 + }, + { + "epoch": 2.42, + "learning_rate": 4.549731569812457e-06, + "loss": 1.0054, + "step": 7850 + }, + { + "epoch": 2.42, + "learning_rate": 4.526590768944713e-06, + "loss": 1.041, + "step": 7855 + }, + { + "epoch": 2.42, + "learning_rate": 4.5035031079602445e-06, + "loss": 0.9824, + "step": 7860 + }, + { + "epoch": 2.42, + "learning_rate": 4.480468646784364e-06, + "loss": 0.9981, + "step": 7865 + }, + { + "epoch": 2.42, + "learning_rate": 4.457487445204311e-06, + "loss": 0.9304, + "step": 7870 + }, + { + "epoch": 2.42, + "learning_rate": 4.434559562869098e-06, + "loss": 1.0557, + "step": 7875 + }, + { + "epoch": 2.42, + "learning_rate": 4.411685059289314e-06, + "loss": 0.9529, + "step": 7880 + }, + { + "epoch": 2.43, + "learning_rate": 4.388863993837031e-06, + "loss": 1.0862, + "step": 7885 + }, + { + "epoch": 2.43, + "learning_rate": 4.366096425745597e-06, + "loss": 0.9976, + "step": 7890 + }, + { + "epoch": 2.43, + "learning_rate": 4.343382414109512e-06, + "loss": 1.0031, + "step": 7895 + }, + { + "epoch": 2.43, + "learning_rate": 4.320722017884274e-06, + "loss": 0.9805, + "step": 7900 + }, + { + "epoch": 2.43, + "learning_rate": 4.2981152958862155e-06, + "loss": 1.1002, + "step": 7905 + }, + { + "epoch": 2.43, + "learning_rate": 4.275562306792352e-06, + "loss": 0.9981, + "step": 7910 + }, + { + "epoch": 2.44, + "learning_rate": 4.253063109140224e-06, + "loss": 1.0023, + "step": 7915 + }, + { + "epoch": 2.44, + "learning_rate": 4.2306177613277765e-06, + "loss": 1.1049, + "step": 7920 + }, + { + "epoch": 2.44, + "learning_rate": 4.208226321613154e-06, + "loss": 1.0171, + "step": 7925 + }, + { + "epoch": 2.44, + "learning_rate": 4.185888848114614e-06, + "loss": 0.9592, + "step": 7930 + }, + { + "epoch": 2.44, + "learning_rate": 4.163605398810305e-06, + "loss": 1.0019, + "step": 7935 + }, + { + "epoch": 2.44, + "learning_rate": 4.141376031538186e-06, + "loss": 1.0001, + "step": 7940 + }, + { + "epoch": 2.44, + "learning_rate": 4.1192008039958235e-06, + "loss": 1.0956, + "step": 7945 + }, + { + "epoch": 2.45, + "learning_rate": 4.097079773740256e-06, + "loss": 0.9754, + "step": 7950 + }, + { + "epoch": 2.45, + "learning_rate": 4.075012998187866e-06, + "loss": 0.9853, + "step": 7955 + }, + { + "epoch": 2.45, + "learning_rate": 4.053000534614218e-06, + "loss": 1.0249, + "step": 7960 + }, + { + "epoch": 2.45, + "learning_rate": 4.03104244015389e-06, + "loss": 0.9884, + "step": 7965 + }, + { + "epoch": 2.45, + "learning_rate": 4.0091387718003415e-06, + "loss": 1.0672, + "step": 7970 + }, + { + "epoch": 2.45, + "learning_rate": 3.987289586405785e-06, + "loss": 1.0511, + "step": 7975 + }, + { + "epoch": 2.46, + "learning_rate": 3.9654949406809995e-06, + "loss": 1.001, + "step": 7980 + }, + { + "epoch": 2.46, + "learning_rate": 3.94375489119522e-06, + "loss": 1.1154, + "step": 7985 + }, + { + "epoch": 2.46, + "learning_rate": 3.922069494375963e-06, + "loss": 1.0152, + "step": 7990 + }, + { + "epoch": 2.46, + "learning_rate": 3.900438806508885e-06, + "loss": 0.8957, + "step": 7995 + }, + { + "epoch": 2.46, + "learning_rate": 3.878862883737666e-06, + "loss": 0.9844, + "step": 8000 + }, + { + "epoch": 2.46, + "learning_rate": 3.857341782063812e-06, + "loss": 1.0729, + "step": 8005 + }, + { + "epoch": 2.46, + "learning_rate": 3.835875557346552e-06, + "loss": 1.0639, + "step": 8010 + }, + { + "epoch": 2.47, + "learning_rate": 3.814464265302692e-06, + "loss": 0.9966, + "step": 8015 + }, + { + "epoch": 2.47, + "learning_rate": 3.7931079615064284e-06, + "loss": 1.0796, + "step": 8020 + }, + { + "epoch": 2.47, + "learning_rate": 3.7718067013892465e-06, + "loss": 1.0698, + "step": 8025 + }, + { + "epoch": 2.47, + "learning_rate": 3.7505605402397753e-06, + "loss": 1.0401, + "step": 8030 + }, + { + "epoch": 2.47, + "learning_rate": 3.7293695332036027e-06, + "loss": 1.0912, + "step": 8035 + }, + { + "epoch": 2.47, + "learning_rate": 3.7082337352831923e-06, + "loss": 1.0626, + "step": 8040 + }, + { + "epoch": 2.48, + "learning_rate": 3.6871532013376896e-06, + "loss": 0.9839, + "step": 8045 + }, + { + "epoch": 2.48, + "learning_rate": 3.666127986082796e-06, + "loss": 1.0956, + "step": 8050 + }, + { + "epoch": 2.48, + "learning_rate": 3.645158144090649e-06, + "loss": 1.0138, + "step": 8055 + }, + { + "epoch": 2.48, + "learning_rate": 3.624243729789642e-06, + "loss": 0.8657, + "step": 8060 + }, + { + "epoch": 2.48, + "learning_rate": 3.603384797464318e-06, + "loss": 1.0273, + "step": 8065 + }, + { + "epoch": 2.48, + "learning_rate": 3.582581401255211e-06, + "loss": 1.0973, + "step": 8070 + }, + { + "epoch": 2.48, + "learning_rate": 3.561833595158698e-06, + "loss": 1.0188, + "step": 8075 + }, + { + "epoch": 2.49, + "learning_rate": 3.5411414330268676e-06, + "loss": 1.0025, + "step": 8080 + }, + { + "epoch": 2.49, + "learning_rate": 3.5205049685674035e-06, + "loss": 1.0431, + "step": 8085 + }, + { + "epoch": 2.49, + "learning_rate": 3.4999242553433954e-06, + "loss": 0.981, + "step": 8090 + }, + { + "epoch": 2.49, + "learning_rate": 3.4793993467732518e-06, + "loss": 1.1148, + "step": 8095 + }, + { + "epoch": 2.49, + "learning_rate": 3.458930296130519e-06, + "loss": 1.035, + "step": 8100 + }, + { + "epoch": 2.49, + "learning_rate": 3.4385171565437606e-06, + "loss": 1.0543, + "step": 8105 + }, + { + "epoch": 2.5, + "learning_rate": 3.418159980996441e-06, + "loss": 1.0029, + "step": 8110 + }, + { + "epoch": 2.5, + "learning_rate": 3.3978588223267383e-06, + "loss": 1.012, + "step": 8115 + }, + { + "epoch": 2.5, + "learning_rate": 3.3776137332274553e-06, + "loss": 0.9635, + "step": 8120 + }, + { + "epoch": 2.5, + "learning_rate": 3.3574247662458645e-06, + "loss": 1.0221, + "step": 8125 + }, + { + "epoch": 2.5, + "learning_rate": 3.3372919737835574e-06, + "loss": 1.0333, + "step": 8130 + }, + { + "epoch": 2.5, + "learning_rate": 3.317215408096322e-06, + "loss": 1.0383, + "step": 8135 + }, + { + "epoch": 2.5, + "learning_rate": 3.297195121294022e-06, + "loss": 1.084, + "step": 8140 + }, + { + "epoch": 2.51, + "learning_rate": 3.2772311653404276e-06, + "loss": 1.0093, + "step": 8145 + }, + { + "epoch": 2.51, + "learning_rate": 3.257323592053116e-06, + "loss": 0.977, + "step": 8150 + }, + { + "epoch": 2.51, + "learning_rate": 3.2374724531033044e-06, + "loss": 0.989, + "step": 8155 + }, + { + "epoch": 2.51, + "learning_rate": 3.2176778000157367e-06, + "loss": 0.9689, + "step": 8160 + }, + { + "epoch": 2.51, + "learning_rate": 3.1979396841685577e-06, + "loss": 0.942, + "step": 8165 + }, + { + "epoch": 2.51, + "learning_rate": 3.17825815679314e-06, + "loss": 1.034, + "step": 8170 + }, + { + "epoch": 2.52, + "learning_rate": 3.1586332689740037e-06, + "loss": 1.0066, + "step": 8175 + }, + { + "epoch": 2.52, + "learning_rate": 3.1390650716486474e-06, + "loss": 0.9275, + "step": 8180 + }, + { + "epoch": 2.52, + "learning_rate": 3.119553615607426e-06, + "loss": 0.8665, + "step": 8185 + }, + { + "epoch": 2.52, + "learning_rate": 3.1000989514934105e-06, + "loss": 0.9829, + "step": 8190 + }, + { + "epoch": 2.52, + "learning_rate": 3.0807011298022852e-06, + "loss": 0.9693, + "step": 8195 + }, + { + "epoch": 2.52, + "learning_rate": 3.061360200882174e-06, + "loss": 1.0815, + "step": 8200 + }, + { + "epoch": 2.52, + "learning_rate": 3.0420762149335565e-06, + "loss": 1.0072, + "step": 8205 + }, + { + "epoch": 2.53, + "learning_rate": 3.022849222009097e-06, + "loss": 1.0478, + "step": 8210 + }, + { + "epoch": 2.53, + "learning_rate": 3.0036792720135266e-06, + "loss": 1.0348, + "step": 8215 + }, + { + "epoch": 2.53, + "learning_rate": 2.9845664147035326e-06, + "loss": 0.9642, + "step": 8220 + }, + { + "epoch": 2.53, + "learning_rate": 2.965510699687615e-06, + "loss": 0.9172, + "step": 8225 + }, + { + "epoch": 2.53, + "learning_rate": 2.9465121764259447e-06, + "loss": 1.0912, + "step": 8230 + }, + { + "epoch": 2.53, + "learning_rate": 2.927570894230261e-06, + "loss": 1.0828, + "step": 8235 + }, + { + "epoch": 2.54, + "learning_rate": 2.908686902263724e-06, + "loss": 1.0382, + "step": 8240 + }, + { + "epoch": 2.54, + "learning_rate": 2.889860249540788e-06, + "loss": 1.0386, + "step": 8245 + }, + { + "epoch": 2.54, + "learning_rate": 2.8710909849270994e-06, + "loss": 1.0863, + "step": 8250 + }, + { + "epoch": 2.54, + "learning_rate": 2.852379157139329e-06, + "loss": 0.9634, + "step": 8255 + }, + { + "epoch": 2.54, + "learning_rate": 2.8337248147450757e-06, + "loss": 1.0676, + "step": 8260 + }, + { + "epoch": 2.54, + "learning_rate": 2.815128006162751e-06, + "loss": 0.941, + "step": 8265 + }, + { + "epoch": 2.54, + "learning_rate": 2.7965887796613884e-06, + "loss": 0.9883, + "step": 8270 + }, + { + "epoch": 2.55, + "learning_rate": 2.7781071833606065e-06, + "loss": 0.9771, + "step": 8275 + }, + { + "epoch": 2.55, + "learning_rate": 2.7596832652304283e-06, + "loss": 0.9266, + "step": 8280 + }, + { + "epoch": 2.55, + "learning_rate": 2.7413170730911597e-06, + "loss": 1.0241, + "step": 8285 + }, + { + "epoch": 2.55, + "learning_rate": 2.7230086546132907e-06, + "loss": 0.9386, + "step": 8290 + }, + { + "epoch": 2.55, + "learning_rate": 2.70475805731735e-06, + "loss": 1.032, + "step": 8295 + }, + { + "epoch": 2.55, + "learning_rate": 2.6865653285737757e-06, + "loss": 1.0707, + "step": 8300 + }, + { + "epoch": 2.56, + "learning_rate": 2.668430515602832e-06, + "loss": 0.9743, + "step": 8305 + }, + { + "epoch": 2.56, + "learning_rate": 2.6503536654744338e-06, + "loss": 1.0133, + "step": 8310 + }, + { + "epoch": 2.56, + "learning_rate": 2.6323348251080626e-06, + "loss": 0.9723, + "step": 8315 + }, + { + "epoch": 2.56, + "learning_rate": 2.6143740412726435e-06, + "loss": 0.9986, + "step": 8320 + }, + { + "epoch": 2.56, + "learning_rate": 2.596471360586378e-06, + "loss": 1.052, + "step": 8325 + }, + { + "epoch": 2.56, + "learning_rate": 2.5786268295166892e-06, + "loss": 0.9558, + "step": 8330 + }, + { + "epoch": 2.56, + "learning_rate": 2.5608404943800622e-06, + "loss": 1.0117, + "step": 8335 + }, + { + "epoch": 2.57, + "learning_rate": 2.5431124013419237e-06, + "loss": 1.0103, + "step": 8340 + }, + { + "epoch": 2.57, + "learning_rate": 2.525442596416541e-06, + "loss": 1.0844, + "step": 8345 + }, + { + "epoch": 2.57, + "learning_rate": 2.5078311254668834e-06, + "loss": 1.1498, + "step": 8350 + }, + { + "epoch": 2.57, + "learning_rate": 2.490278034204502e-06, + "loss": 1.0548, + "step": 8355 + }, + { + "epoch": 2.57, + "learning_rate": 2.4727833681894437e-06, + "loss": 0.8574, + "step": 8360 + }, + { + "epoch": 2.57, + "learning_rate": 2.4553471728300885e-06, + "loss": 1.1363, + "step": 8365 + }, + { + "epoch": 2.58, + "learning_rate": 2.4379694933830634e-06, + "loss": 1.0163, + "step": 8370 + }, + { + "epoch": 2.58, + "learning_rate": 2.4206503749531236e-06, + "loss": 1.1199, + "step": 8375 + }, + { + "epoch": 2.58, + "learning_rate": 2.4033898624929884e-06, + "loss": 1.0465, + "step": 8380 + }, + { + "epoch": 2.58, + "learning_rate": 2.386188000803302e-06, + "loss": 1.0598, + "step": 8385 + }, + { + "epoch": 2.58, + "learning_rate": 2.3690448345324634e-06, + "loss": 1.042, + "step": 8390 + }, + { + "epoch": 2.58, + "learning_rate": 2.351960408176518e-06, + "loss": 0.989, + "step": 8395 + }, + { + "epoch": 2.58, + "learning_rate": 2.3349347660790582e-06, + "loss": 0.9844, + "step": 8400 + }, + { + "epoch": 2.59, + "learning_rate": 2.317967952431094e-06, + "loss": 1.0982, + "step": 8405 + }, + { + "epoch": 2.59, + "learning_rate": 2.3010600112709364e-06, + "loss": 0.9751, + "step": 8410 + }, + { + "epoch": 2.59, + "learning_rate": 2.2842109864841034e-06, + "loss": 1.1124, + "step": 8415 + }, + { + "epoch": 2.59, + "learning_rate": 2.2674209218031787e-06, + "loss": 0.9882, + "step": 8420 + }, + { + "epoch": 2.59, + "learning_rate": 2.25068986080772e-06, + "loss": 1.0219, + "step": 8425 + }, + { + "epoch": 2.59, + "learning_rate": 2.2340178469241467e-06, + "loss": 1.0907, + "step": 8430 + }, + { + "epoch": 2.6, + "learning_rate": 2.2174049234255895e-06, + "loss": 1.0381, + "step": 8435 + }, + { + "epoch": 2.6, + "learning_rate": 2.2008511334318306e-06, + "loss": 0.9869, + "step": 8440 + }, + { + "epoch": 2.6, + "learning_rate": 2.184356519909167e-06, + "loss": 0.9677, + "step": 8445 + }, + { + "epoch": 2.6, + "learning_rate": 2.1679211256702884e-06, + "loss": 1.0891, + "step": 8450 + }, + { + "epoch": 2.6, + "learning_rate": 2.1515449933741854e-06, + "loss": 0.9528, + "step": 8455 + }, + { + "epoch": 2.6, + "learning_rate": 2.135228165526032e-06, + "loss": 0.9859, + "step": 8460 + }, + { + "epoch": 2.6, + "learning_rate": 2.118970684477062e-06, + "loss": 1.0853, + "step": 8465 + }, + { + "epoch": 2.61, + "learning_rate": 2.1027725924244903e-06, + "loss": 1.0674, + "step": 8470 + }, + { + "epoch": 2.61, + "learning_rate": 2.0866339314113662e-06, + "loss": 1.028, + "step": 8475 + }, + { + "epoch": 2.61, + "learning_rate": 2.0705547433264943e-06, + "loss": 0.979, + "step": 8480 + }, + { + "epoch": 2.61, + "learning_rate": 2.0545350699043174e-06, + "loss": 0.9093, + "step": 8485 + }, + { + "epoch": 2.61, + "learning_rate": 2.0385749527247837e-06, + "loss": 1.1284, + "step": 8490 + }, + { + "epoch": 2.61, + "learning_rate": 2.0226744332132812e-06, + "loss": 0.9241, + "step": 8495 + }, + { + "epoch": 2.62, + "learning_rate": 2.0068335526405023e-06, + "loss": 0.9496, + "step": 8500 + }, + { + "epoch": 2.62, + "learning_rate": 1.9910523521223355e-06, + "loss": 1.0355, + "step": 8505 + }, + { + "epoch": 2.62, + "learning_rate": 1.975330872619782e-06, + "loss": 0.9732, + "step": 8510 + }, + { + "epoch": 2.62, + "learning_rate": 1.95966915493882e-06, + "loss": 0.9996, + "step": 8515 + }, + { + "epoch": 2.62, + "learning_rate": 1.9440672397303127e-06, + "loss": 0.9817, + "step": 8520 + }, + { + "epoch": 2.62, + "learning_rate": 1.928525167489914e-06, + "loss": 0.9913, + "step": 8525 + }, + { + "epoch": 2.62, + "learning_rate": 1.913042978557944e-06, + "loss": 0.992, + "step": 8530 + }, + { + "epoch": 2.63, + "learning_rate": 1.8976207131192914e-06, + "loss": 0.9756, + "step": 8535 + }, + { + "epoch": 2.63, + "learning_rate": 1.8822584112033082e-06, + "loss": 1.0444, + "step": 8540 + }, + { + "epoch": 2.63, + "learning_rate": 1.8669561126837236e-06, + "loss": 1.0064, + "step": 8545 + }, + { + "epoch": 2.63, + "learning_rate": 1.8517138572784976e-06, + "loss": 0.9934, + "step": 8550 + }, + { + "epoch": 2.63, + "learning_rate": 1.836531684549772e-06, + "loss": 0.9618, + "step": 8555 + }, + { + "epoch": 2.63, + "learning_rate": 1.821409633903723e-06, + "loss": 1.0905, + "step": 8560 + }, + { + "epoch": 2.64, + "learning_rate": 1.8063477445904835e-06, + "loss": 0.9602, + "step": 8565 + }, + { + "epoch": 2.64, + "learning_rate": 1.7913460557040351e-06, + "loss": 1.0898, + "step": 8570 + }, + { + "epoch": 2.64, + "learning_rate": 1.776404606182097e-06, + "loss": 0.9512, + "step": 8575 + }, + { + "epoch": 2.64, + "learning_rate": 1.7615234348060449e-06, + "loss": 0.9865, + "step": 8580 + }, + { + "epoch": 2.64, + "learning_rate": 1.7467025802007987e-06, + "loss": 1.0721, + "step": 8585 + }, + { + "epoch": 2.64, + "learning_rate": 1.7319420808347142e-06, + "loss": 1.0682, + "step": 8590 + }, + { + "epoch": 2.64, + "learning_rate": 1.717241975019493e-06, + "loss": 0.8984, + "step": 8595 + }, + { + "epoch": 2.65, + "learning_rate": 1.7026023009100944e-06, + "loss": 0.9412, + "step": 8600 + }, + { + "epoch": 2.65, + "learning_rate": 1.688023096504604e-06, + "loss": 1.0409, + "step": 8605 + }, + { + "epoch": 2.65, + "learning_rate": 1.673504399644174e-06, + "loss": 0.9981, + "step": 8610 + }, + { + "epoch": 2.65, + "learning_rate": 1.6590462480128882e-06, + "loss": 1.0451, + "step": 8615 + }, + { + "epoch": 2.65, + "learning_rate": 1.644648679137703e-06, + "loss": 1.0811, + "step": 8620 + }, + { + "epoch": 2.65, + "learning_rate": 1.630311730388312e-06, + "loss": 0.9745, + "step": 8625 + }, + { + "epoch": 2.66, + "learning_rate": 1.6160354389770649e-06, + "loss": 0.9891, + "step": 8630 + }, + { + "epoch": 2.66, + "learning_rate": 1.6018198419588793e-06, + "loss": 1.0398, + "step": 8635 + }, + { + "epoch": 2.66, + "learning_rate": 1.5876649762311458e-06, + "loss": 1.0624, + "step": 8640 + }, + { + "epoch": 2.66, + "learning_rate": 1.5735708785336033e-06, + "loss": 1.0608, + "step": 8645 + }, + { + "epoch": 2.66, + "learning_rate": 1.559537585448273e-06, + "loss": 1.1355, + "step": 8650 + }, + { + "epoch": 2.66, + "learning_rate": 1.5455651333993626e-06, + "loss": 0.9244, + "step": 8655 + }, + { + "epoch": 2.66, + "learning_rate": 1.5316535586531483e-06, + "loss": 1.0366, + "step": 8660 + }, + { + "epoch": 2.67, + "learning_rate": 1.5178028973179104e-06, + "loss": 1.0179, + "step": 8665 + }, + { + "epoch": 2.67, + "learning_rate": 1.504013185343811e-06, + "loss": 1.0496, + "step": 8670 + }, + { + "epoch": 2.67, + "learning_rate": 1.4902844585228282e-06, + "loss": 0.9367, + "step": 8675 + }, + { + "epoch": 2.67, + "learning_rate": 1.476616752488641e-06, + "loss": 0.9192, + "step": 8680 + }, + { + "epoch": 2.67, + "learning_rate": 1.4630101027165444e-06, + "loss": 0.9817, + "step": 8685 + }, + { + "epoch": 2.67, + "learning_rate": 1.4494645445233658e-06, + "loss": 1.0383, + "step": 8690 + }, + { + "epoch": 2.68, + "learning_rate": 1.4359801130673616e-06, + "loss": 0.9878, + "step": 8695 + }, + { + "epoch": 2.68, + "learning_rate": 1.4225568433481329e-06, + "loss": 0.9941, + "step": 8700 + }, + { + "epoch": 2.68, + "learning_rate": 1.4091947702065262e-06, + "loss": 0.9614, + "step": 8705 + }, + { + "epoch": 2.68, + "learning_rate": 1.3958939283245543e-06, + "loss": 1.0378, + "step": 8710 + }, + { + "epoch": 2.68, + "learning_rate": 1.382654352225296e-06, + "loss": 0.9675, + "step": 8715 + }, + { + "epoch": 2.68, + "learning_rate": 1.3694760762728215e-06, + "loss": 0.9759, + "step": 8720 + }, + { + "epoch": 2.68, + "learning_rate": 1.3563591346720804e-06, + "loss": 0.9709, + "step": 8725 + }, + { + "epoch": 2.69, + "learning_rate": 1.3433035614688338e-06, + "loss": 0.9597, + "step": 8730 + }, + { + "epoch": 2.69, + "learning_rate": 1.3303093905495528e-06, + "loss": 0.9755, + "step": 8735 + }, + { + "epoch": 2.69, + "learning_rate": 1.3173766556413393e-06, + "loss": 0.967, + "step": 8740 + }, + { + "epoch": 2.69, + "learning_rate": 1.3045053903118303e-06, + "loss": 1.0492, + "step": 8745 + }, + { + "epoch": 2.69, + "learning_rate": 1.2916956279691223e-06, + "loss": 1.0494, + "step": 8750 + }, + { + "epoch": 2.69, + "learning_rate": 1.2789474018616714e-06, + "loss": 1.1267, + "step": 8755 + }, + { + "epoch": 2.7, + "learning_rate": 1.26626074507821e-06, + "loss": 0.9342, + "step": 8760 + }, + { + "epoch": 2.7, + "learning_rate": 1.2536356905476748e-06, + "loss": 1.0399, + "step": 8765 + }, + { + "epoch": 2.7, + "learning_rate": 1.2410722710390954e-06, + "loss": 1.0174, + "step": 8770 + }, + { + "epoch": 2.7, + "learning_rate": 1.2285705191615426e-06, + "loss": 1.0773, + "step": 8775 + }, + { + "epoch": 2.7, + "learning_rate": 1.21613046736401e-06, + "loss": 0.8982, + "step": 8780 + }, + { + "epoch": 2.7, + "learning_rate": 1.203752147935347e-06, + "loss": 1.0125, + "step": 8785 + }, + { + "epoch": 2.7, + "learning_rate": 1.1914355930041837e-06, + "loss": 0.8979, + "step": 8790 + }, + { + "epoch": 2.71, + "learning_rate": 1.179180834538826e-06, + "loss": 1.0597, + "step": 8795 + }, + { + "epoch": 2.71, + "learning_rate": 1.166987904347186e-06, + "loss": 0.8727, + "step": 8800 + }, + { + "epoch": 2.71, + "learning_rate": 1.1548568340767036e-06, + "loss": 1.0429, + "step": 8805 + }, + { + "epoch": 2.71, + "learning_rate": 1.142787655214253e-06, + "loss": 1.1399, + "step": 8810 + }, + { + "epoch": 2.71, + "learning_rate": 1.1307803990860594e-06, + "loss": 1.0295, + "step": 8815 + }, + { + "epoch": 2.71, + "learning_rate": 1.1188350968576372e-06, + "loss": 1.0259, + "step": 8820 + }, + { + "epoch": 2.72, + "learning_rate": 1.1069517795336825e-06, + "loss": 0.9099, + "step": 8825 + }, + { + "epoch": 2.72, + "learning_rate": 1.0951304779580146e-06, + "loss": 1.0031, + "step": 8830 + }, + { + "epoch": 2.72, + "learning_rate": 1.0833712228134952e-06, + "loss": 1.0911, + "step": 8835 + }, + { + "epoch": 2.72, + "learning_rate": 1.0716740446219175e-06, + "loss": 1.0943, + "step": 8840 + }, + { + "epoch": 2.72, + "learning_rate": 1.0600389737439681e-06, + "loss": 0.9858, + "step": 8845 + }, + { + "epoch": 2.72, + "learning_rate": 1.0484660403791314e-06, + "loss": 1.0377, + "step": 8850 + }, + { + "epoch": 2.72, + "learning_rate": 1.0369552745656013e-06, + "loss": 1.0121, + "step": 8855 + }, + { + "epoch": 2.73, + "learning_rate": 1.025506706180221e-06, + "loss": 0.9278, + "step": 8860 + }, + { + "epoch": 2.73, + "learning_rate": 1.0141203649383924e-06, + "loss": 0.9434, + "step": 8865 + }, + { + "epoch": 2.73, + "learning_rate": 1.0027962803939944e-06, + "loss": 0.965, + "step": 8870 + }, + { + "epoch": 2.73, + "learning_rate": 9.91534481939338e-07, + "loss": 1.1013, + "step": 8875 + }, + { + "epoch": 2.73, + "learning_rate": 9.80334998805041e-07, + "loss": 0.9825, + "step": 8880 + }, + { + "epoch": 2.73, + "learning_rate": 9.691978600599977e-07, + "loss": 0.9567, + "step": 8885 + }, + { + "epoch": 2.74, + "learning_rate": 9.581230946112824e-07, + "loss": 0.8988, + "step": 8890 + }, + { + "epoch": 2.74, + "learning_rate": 9.471107312040567e-07, + "loss": 1.0629, + "step": 8895 + }, + { + "epoch": 2.74, + "learning_rate": 9.361607984215342e-07, + "loss": 0.9501, + "step": 8900 + }, + { + "epoch": 2.74, + "learning_rate": 9.25273324684886e-07, + "loss": 0.9838, + "step": 8905 + }, + { + "epoch": 2.74, + "learning_rate": 9.144483382531571e-07, + "loss": 1.0426, + "step": 8910 + }, + { + "epoch": 2.74, + "learning_rate": 9.036858672232057e-07, + "loss": 1.0276, + "step": 8915 + }, + { + "epoch": 2.74, + "learning_rate": 8.929859395296364e-07, + "loss": 1.0565, + "step": 8920 + }, + { + "epoch": 2.75, + "learning_rate": 8.823485829447003e-07, + "loss": 0.9382, + "step": 8925 + }, + { + "epoch": 2.75, + "learning_rate": 8.717738250782675e-07, + "loss": 1.0466, + "step": 8930 + }, + { + "epoch": 2.75, + "learning_rate": 8.612616933777046e-07, + "loss": 1.0048, + "step": 8935 + }, + { + "epoch": 2.75, + "learning_rate": 8.508122151278442e-07, + "loss": 0.8363, + "step": 8940 + }, + { + "epoch": 2.75, + "learning_rate": 8.404254174509019e-07, + "loss": 1.0045, + "step": 8945 + }, + { + "epoch": 2.75, + "learning_rate": 8.301013273063791e-07, + "loss": 1.0165, + "step": 8950 + }, + { + "epoch": 2.76, + "learning_rate": 8.198399714910404e-07, + "loss": 1.0233, + "step": 8955 + }, + { + "epoch": 2.76, + "learning_rate": 8.096413766388117e-07, + "loss": 0.9936, + "step": 8960 + }, + { + "epoch": 2.76, + "learning_rate": 7.995055692207127e-07, + "loss": 0.8983, + "step": 8965 + }, + { + "epoch": 2.76, + "learning_rate": 7.894325755448073e-07, + "loss": 1.1045, + "step": 8970 + }, + { + "epoch": 2.76, + "learning_rate": 7.794224217561152e-07, + "loss": 1.1072, + "step": 8975 + }, + { + "epoch": 2.76, + "learning_rate": 7.694751338365447e-07, + "loss": 0.9562, + "step": 8980 + }, + { + "epoch": 2.76, + "learning_rate": 7.595907376048512e-07, + "loss": 1.0286, + "step": 8985 + }, + { + "epoch": 2.77, + "learning_rate": 7.497692587165345e-07, + "loss": 0.9885, + "step": 8990 + }, + { + "epoch": 2.77, + "learning_rate": 7.40010722663792e-07, + "loss": 0.9136, + "step": 8995 + }, + { + "epoch": 2.77, + "learning_rate": 7.303151547754627e-07, + "loss": 1.0548, + "step": 9000 + }, + { + "epoch": 2.77, + "learning_rate": 7.20682580216922e-07, + "loss": 0.9183, + "step": 9005 + }, + { + "epoch": 2.77, + "learning_rate": 7.111130239900677e-07, + "loss": 1.0677, + "step": 9010 + }, + { + "epoch": 2.77, + "learning_rate": 7.016065109332226e-07, + "loss": 1.0585, + "step": 9015 + }, + { + "epoch": 2.78, + "learning_rate": 6.921630657210659e-07, + "loss": 0.9667, + "step": 9020 + }, + { + "epoch": 2.78, + "learning_rate": 6.827827128645992e-07, + "loss": 1.1483, + "step": 9025 + }, + { + "epoch": 2.78, + "learning_rate": 6.734654767110521e-07, + "loss": 1.0026, + "step": 9030 + }, + { + "epoch": 2.78, + "learning_rate": 6.6421138144383e-07, + "loss": 1.0057, + "step": 9035 + }, + { + "epoch": 2.78, + "learning_rate": 6.550204510824609e-07, + "loss": 0.963, + "step": 9040 + }, + { + "epoch": 2.78, + "learning_rate": 6.458927094825179e-07, + "loss": 1.0234, + "step": 9045 + }, + { + "epoch": 2.78, + "learning_rate": 6.368281803355691e-07, + "loss": 0.9925, + "step": 9050 + }, + { + "epoch": 2.79, + "learning_rate": 6.27826887169114e-07, + "loss": 1.0105, + "step": 9055 + }, + { + "epoch": 2.79, + "learning_rate": 6.188888533465053e-07, + "loss": 1.0088, + "step": 9060 + }, + { + "epoch": 2.79, + "learning_rate": 6.100141020669137e-07, + "loss": 1.0163, + "step": 9065 + }, + { + "epoch": 2.79, + "learning_rate": 6.012026563652573e-07, + "loss": 0.9228, + "step": 9070 + }, + { + "epoch": 2.79, + "learning_rate": 5.924545391121361e-07, + "loss": 0.9818, + "step": 9075 + }, + { + "epoch": 2.79, + "learning_rate": 5.837697730137814e-07, + "loss": 1.1069, + "step": 9080 + }, + { + "epoch": 2.8, + "learning_rate": 5.751483806119923e-07, + "loss": 0.9276, + "step": 9085 + }, + { + "epoch": 2.8, + "learning_rate": 5.665903842840714e-07, + "loss": 1.0318, + "step": 9090 + }, + { + "epoch": 2.8, + "learning_rate": 5.580958062427866e-07, + "loss": 1.0107, + "step": 9095 + }, + { + "epoch": 2.8, + "learning_rate": 5.496646685362844e-07, + "loss": 1.044, + "step": 9100 + }, + { + "epoch": 2.8, + "learning_rate": 5.412969930480599e-07, + "loss": 1.0403, + "step": 9105 + }, + { + "epoch": 2.8, + "learning_rate": 5.329928014968843e-07, + "loss": 0.979, + "step": 9110 + }, + { + "epoch": 2.8, + "learning_rate": 5.247521154367552e-07, + "loss": 0.9666, + "step": 9115 + }, + { + "epoch": 2.81, + "learning_rate": 5.165749562568323e-07, + "loss": 1.0986, + "step": 9120 + }, + { + "epoch": 2.81, + "learning_rate": 5.084613451813935e-07, + "loss": 0.9006, + "step": 9125 + }, + { + "epoch": 2.81, + "learning_rate": 5.00411303269771e-07, + "loss": 1.0061, + "step": 9130 + }, + { + "epoch": 2.81, + "learning_rate": 4.924248514163038e-07, + "loss": 0.9795, + "step": 9135 + }, + { + "epoch": 2.81, + "learning_rate": 4.845020103502712e-07, + "loss": 0.8639, + "step": 9140 + }, + { + "epoch": 2.81, + "learning_rate": 4.766428006358542e-07, + "loss": 1.0554, + "step": 9145 + }, + { + "epoch": 2.82, + "learning_rate": 4.688472426720714e-07, + "loss": 1.0763, + "step": 9150 + }, + { + "epoch": 2.82, + "learning_rate": 4.611153566927373e-07, + "loss": 0.8512, + "step": 9155 + }, + { + "epoch": 2.82, + "learning_rate": 4.534471627663878e-07, + "loss": 0.9362, + "step": 9160 + }, + { + "epoch": 2.82, + "learning_rate": 4.4584268079625735e-07, + "loss": 1.0311, + "step": 9165 + }, + { + "epoch": 2.82, + "learning_rate": 4.3830193052020186e-07, + "loss": 0.9447, + "step": 9170 + }, + { + "epoch": 2.82, + "learning_rate": 4.308249315106649e-07, + "loss": 1.0359, + "step": 9175 + }, + { + "epoch": 2.82, + "learning_rate": 4.234117031746143e-07, + "loss": 1.1703, + "step": 9180 + }, + { + "epoch": 2.83, + "learning_rate": 4.1606226475350287e-07, + "loss": 0.947, + "step": 9185 + }, + { + "epoch": 2.83, + "learning_rate": 4.087766353232103e-07, + "loss": 0.8837, + "step": 9190 + }, + { + "epoch": 2.83, + "learning_rate": 4.015548337939962e-07, + "loss": 1.03, + "step": 9195 + }, + { + "epoch": 2.83, + "learning_rate": 3.943968789104496e-07, + "loss": 1.0136, + "step": 9200 + }, + { + "epoch": 2.83, + "learning_rate": 3.87302789251448e-07, + "loss": 1.0089, + "step": 9205 + }, + { + "epoch": 2.83, + "learning_rate": 3.8027258323010127e-07, + "loss": 1.0196, + "step": 9210 + }, + { + "epoch": 2.84, + "learning_rate": 3.733062790936964e-07, + "loss": 1.0584, + "step": 9215 + }, + { + "epoch": 2.84, + "learning_rate": 3.6640389492367534e-07, + "loss": 1.0279, + "step": 9220 + }, + { + "epoch": 2.84, + "learning_rate": 3.5956544863555983e-07, + "loss": 1.1492, + "step": 9225 + }, + { + "epoch": 2.84, + "learning_rate": 3.5279095797892127e-07, + "loss": 1.0217, + "step": 9230 + }, + { + "epoch": 2.84, + "learning_rate": 3.460804405373302e-07, + "loss": 1.0449, + "step": 9235 + }, + { + "epoch": 2.84, + "learning_rate": 3.394339137283098e-07, + "loss": 1.0486, + "step": 9240 + }, + { + "epoch": 2.84, + "learning_rate": 3.328513948032991e-07, + "loss": 1.1402, + "step": 9245 + }, + { + "epoch": 2.85, + "learning_rate": 3.263329008475924e-07, + "loss": 0.9133, + "step": 9250 + }, + { + "epoch": 2.85, + "learning_rate": 3.1987844878030307e-07, + "loss": 1.1023, + "step": 9255 + }, + { + "epoch": 2.85, + "learning_rate": 3.1348805535432735e-07, + "loss": 0.9518, + "step": 9260 + }, + { + "epoch": 2.85, + "learning_rate": 3.071617371562946e-07, + "loss": 1.009, + "step": 9265 + }, + { + "epoch": 2.85, + "learning_rate": 3.0089951060651156e-07, + "loss": 0.8592, + "step": 9270 + }, + { + "epoch": 2.85, + "learning_rate": 2.947013919589431e-07, + "loss": 1.1002, + "step": 9275 + }, + { + "epoch": 2.86, + "learning_rate": 2.88567397301151e-07, + "loss": 1.0326, + "step": 9280 + }, + { + "epoch": 2.86, + "learning_rate": 2.824975425542664e-07, + "loss": 1.1102, + "step": 9285 + }, + { + "epoch": 2.86, + "learning_rate": 2.764918434729369e-07, + "loss": 1.0381, + "step": 9290 + }, + { + "epoch": 2.86, + "learning_rate": 2.7055031564529043e-07, + "loss": 0.9609, + "step": 9295 + }, + { + "epoch": 2.86, + "learning_rate": 2.646729744928966e-07, + "loss": 1.0629, + "step": 9300 + }, + { + "epoch": 2.86, + "learning_rate": 2.588598352707278e-07, + "loss": 1.0125, + "step": 9305 + }, + { + "epoch": 2.86, + "learning_rate": 2.531109130671061e-07, + "loss": 1.0196, + "step": 9310 + }, + { + "epoch": 2.87, + "learning_rate": 2.474262228036872e-07, + "loss": 0.9708, + "step": 9315 + }, + { + "epoch": 2.87, + "learning_rate": 2.418057792354045e-07, + "loss": 1.0494, + "step": 9320 + }, + { + "epoch": 2.87, + "learning_rate": 2.3624959695043302e-07, + "loss": 0.9855, + "step": 9325 + }, + { + "epoch": 2.87, + "learning_rate": 2.3075769037015638e-07, + "loss": 1.0688, + "step": 9330 + }, + { + "epoch": 2.87, + "learning_rate": 2.2533007374912485e-07, + "loss": 0.981, + "step": 9335 + }, + { + "epoch": 2.87, + "learning_rate": 2.1996676117502224e-07, + "loss": 1.0131, + "step": 9340 + }, + { + "epoch": 2.88, + "learning_rate": 2.146677665686325e-07, + "loss": 0.9826, + "step": 9345 + }, + { + "epoch": 2.88, + "learning_rate": 2.094331036837871e-07, + "loss": 0.9446, + "step": 9350 + }, + { + "epoch": 2.88, + "learning_rate": 2.0426278610735094e-07, + "loss": 1.0536, + "step": 9355 + }, + { + "epoch": 2.88, + "learning_rate": 1.9915682725917262e-07, + "loss": 1.0056, + "step": 9360 + }, + { + "epoch": 2.88, + "learning_rate": 1.9411524039205376e-07, + "loss": 1.0566, + "step": 9365 + }, + { + "epoch": 2.88, + "learning_rate": 1.891380385917213e-07, + "loss": 1.1621, + "step": 9370 + }, + { + "epoch": 2.88, + "learning_rate": 1.842252347767748e-07, + "loss": 1.1044, + "step": 9375 + }, + { + "epoch": 2.89, + "learning_rate": 1.7937684169867797e-07, + "loss": 1.0607, + "step": 9380 + }, + { + "epoch": 2.89, + "learning_rate": 1.7459287194170615e-07, + "loss": 1.0, + "step": 9385 + }, + { + "epoch": 2.89, + "learning_rate": 1.6987333792292115e-07, + "loss": 0.9634, + "step": 9390 + }, + { + "epoch": 2.89, + "learning_rate": 1.6521825189213526e-07, + "loss": 1.0137, + "step": 9395 + }, + { + "epoch": 2.89, + "learning_rate": 1.6062762593188896e-07, + "loss": 1.0239, + "step": 9400 + }, + { + "epoch": 2.89, + "learning_rate": 1.5610147195740943e-07, + "loss": 0.9871, + "step": 9405 + }, + { + "epoch": 2.9, + "learning_rate": 1.5163980171658542e-07, + "loss": 1.0145, + "step": 9410 + }, + { + "epoch": 2.9, + "learning_rate": 1.472426267899285e-07, + "loss": 0.9816, + "step": 9415 + }, + { + "epoch": 2.9, + "learning_rate": 1.4290995859055633e-07, + "loss": 1.0211, + "step": 9420 + }, + { + "epoch": 2.9, + "learning_rate": 1.386418083641483e-07, + "loss": 0.9766, + "step": 9425 + }, + { + "epoch": 2.9, + "learning_rate": 1.3443818718893442e-07, + "loss": 0.9284, + "step": 9430 + }, + { + "epoch": 2.9, + "learning_rate": 1.3029910597564532e-07, + "loss": 1.1221, + "step": 9435 + }, + { + "epoch": 2.9, + "learning_rate": 1.2622457546749567e-07, + "loss": 1.0471, + "step": 9440 + }, + { + "epoch": 2.91, + "learning_rate": 1.2221460624016466e-07, + "loss": 0.9608, + "step": 9445 + }, + { + "epoch": 2.91, + "learning_rate": 1.1826920870174895e-07, + "loss": 1.0131, + "step": 9450 + }, + { + "epoch": 2.91, + "learning_rate": 1.143883930927514e-07, + "loss": 1.0661, + "step": 9455 + }, + { + "epoch": 2.91, + "learning_rate": 1.1057216948604509e-07, + "loss": 1.0361, + "step": 9460 + }, + { + "epoch": 2.91, + "learning_rate": 1.068205477868539e-07, + "loss": 0.9729, + "step": 9465 + }, + { + "epoch": 2.91, + "learning_rate": 1.0313353773271917e-07, + "loss": 0.9241, + "step": 9470 + }, + { + "epoch": 2.92, + "learning_rate": 9.951114889348855e-08, + "loss": 0.9018, + "step": 9475 + }, + { + "epoch": 2.92, + "learning_rate": 9.595339067127174e-08, + "loss": 0.854, + "step": 9480 + }, + { + "epoch": 2.92, + "learning_rate": 9.24602723004292e-08, + "loss": 1.0278, + "step": 9485 + }, + { + "epoch": 2.92, + "learning_rate": 8.903180284755008e-08, + "loss": 0.9484, + "step": 9490 + }, + { + "epoch": 2.92, + "learning_rate": 8.566799121141334e-08, + "loss": 0.993, + "step": 9495 + }, + { + "epoch": 2.92, + "learning_rate": 8.23688461229849e-08, + "loss": 1.0084, + "step": 9500 + }, + { + "epoch": 2.92, + "learning_rate": 7.913437614538166e-08, + "loss": 1.1513, + "step": 9505 + }, + { + "epoch": 2.93, + "learning_rate": 7.596458967384922e-08, + "loss": 1.0486, + "step": 9510 + }, + { + "epoch": 2.93, + "learning_rate": 7.285949493574806e-08, + "loss": 0.9455, + "step": 9515 + }, + { + "epoch": 2.93, + "learning_rate": 6.98190999905285e-08, + "loss": 1.059, + "step": 9520 + }, + { + "epoch": 2.93, + "learning_rate": 6.684341272970018e-08, + "loss": 1.0284, + "step": 9525 + }, + { + "epoch": 2.93, + "learning_rate": 6.393244087683215e-08, + "loss": 1.0326, + "step": 9530 + }, + { + "epoch": 2.93, + "learning_rate": 6.108619198751109e-08, + "loss": 1.0141, + "step": 9535 + }, + { + "epoch": 2.94, + "learning_rate": 5.8304673449338653e-08, + "loss": 1.0587, + "step": 9540 + }, + { + "epoch": 2.94, + "learning_rate": 5.558789248190366e-08, + "loss": 0.934, + "step": 9545 + }, + { + "epoch": 2.94, + "learning_rate": 5.293585613675989e-08, + "loss": 0.8973, + "step": 9550 + }, + { + "epoch": 2.94, + "learning_rate": 5.034857129741777e-08, + "loss": 0.9969, + "step": 9555 + }, + { + "epoch": 2.94, + "learning_rate": 4.782604467931939e-08, + "loss": 0.9948, + "step": 9560 + }, + { + "epoch": 2.94, + "learning_rate": 4.5368282829827415e-08, + "loss": 1.0369, + "step": 9565 + }, + { + "epoch": 2.94, + "learning_rate": 4.2975292128200064e-08, + "loss": 1.0482, + "step": 9570 + }, + { + "epoch": 2.95, + "learning_rate": 4.064707878557728e-08, + "loss": 0.9402, + "step": 9575 + }, + { + "epoch": 2.95, + "learning_rate": 3.838364884496681e-08, + "loss": 1.0316, + "step": 9580 + }, + { + "epoch": 2.95, + "learning_rate": 3.618500818123039e-08, + "loss": 1.0608, + "step": 9585 + }, + { + "epoch": 2.95, + "learning_rate": 3.405116250106144e-08, + "loss": 1.0775, + "step": 9590 + }, + { + "epoch": 2.95, + "learning_rate": 3.1982117342979624e-08, + "loss": 0.8861, + "step": 9595 + }, + { + "epoch": 2.95, + "learning_rate": 2.9977878077305785e-08, + "loss": 1.0457, + "step": 9600 + }, + { + "epoch": 2.96, + "learning_rate": 2.8038449906153673e-08, + "loss": 1.0104, + "step": 9605 + }, + { + "epoch": 2.96, + "learning_rate": 2.6163837863418806e-08, + "loss": 1.0111, + "step": 9610 + }, + { + "epoch": 2.96, + "learning_rate": 2.4354046814764607e-08, + "loss": 1.0486, + "step": 9615 + }, + { + "epoch": 2.96, + "learning_rate": 2.260908145760299e-08, + "loss": 1.1501, + "step": 9620 + }, + { + "epoch": 2.96, + "learning_rate": 2.0928946321091547e-08, + "loss": 0.9465, + "step": 9625 + }, + { + "epoch": 2.96, + "learning_rate": 1.931364576611139e-08, + "loss": 0.9679, + "step": 9630 + }, + { + "epoch": 2.96, + "learning_rate": 1.7763183985269883e-08, + "loss": 1.0199, + "step": 9635 + }, + { + "epoch": 2.97, + "learning_rate": 1.6277565002875696e-08, + "loss": 1.0658, + "step": 9640 + }, + { + "epoch": 2.97, + "learning_rate": 1.4856792674936004e-08, + "loss": 1.0229, + "step": 9645 + }, + { + "epoch": 2.97, + "learning_rate": 1.3500870689145407e-08, + "loss": 0.9942, + "step": 9650 + }, + { + "epoch": 2.97, + "learning_rate": 1.2209802564877582e-08, + "loss": 1.0165, + "step": 9655 + }, + { + "epoch": 2.97, + "learning_rate": 1.0983591653168645e-08, + "loss": 1.0728, + "step": 9660 + }, + { + "epoch": 2.97, + "learning_rate": 9.822241136722699e-09, + "loss": 1.0423, + "step": 9665 + }, + { + "epoch": 2.98, + "learning_rate": 8.72575402988407e-09, + "loss": 1.0226, + "step": 9670 + }, + { + "epoch": 2.98, + "learning_rate": 7.694133178653973e-09, + "loss": 1.0802, + "step": 9675 + }, + { + "epoch": 2.98, + "learning_rate": 6.727381260657195e-09, + "loss": 1.0125, + "step": 9680 + }, + { + "epoch": 2.98, + "learning_rate": 5.825500785150428e-09, + "loss": 0.9788, + "step": 9685 + }, + { + "epoch": 2.98, + "learning_rate": 4.988494093022267e-09, + "loss": 0.8966, + "step": 9690 + }, + { + "epoch": 2.98, + "learning_rate": 4.216363356765452e-09, + "loss": 1.0167, + "step": 9695 + }, + { + "epoch": 2.98, + "learning_rate": 3.5091105804907487e-09, + "loss": 0.9996, + "step": 9700 + }, + { + "epoch": 2.99, + "learning_rate": 2.8667375999102964e-09, + "loss": 0.9538, + "step": 9705 + }, + { + "epoch": 2.99, + "learning_rate": 2.2892460823403794e-09, + "loss": 1.0679, + "step": 9710 + }, + { + "epoch": 2.99, + "learning_rate": 1.7766375266931035e-09, + "loss": 1.0925, + "step": 9715 + }, + { + "epoch": 2.99, + "learning_rate": 1.328913263473619e-09, + "loss": 0.9632, + "step": 9720 + }, + { + "epoch": 2.99, + "learning_rate": 9.460744547745704e-10, + "loss": 1.0169, + "step": 9725 + }, + { + "epoch": 2.99, + "learning_rate": 6.281220942733201e-10, + "loss": 1.0365, + "step": 9730 + }, + { + "epoch": 3.0, + "learning_rate": 3.750570072375004e-10, + "loss": 1.1074, + "step": 9735 + }, + { + "epoch": 3.0, + "learning_rate": 1.8687985050558355e-10, + "loss": 1.0347, + "step": 9740 + }, + { + "epoch": 3.0, + "learning_rate": 6.359111250908711e-11, + "loss": 1.0258, + "step": 9745 + }, + { + "epoch": 3.0, + "learning_rate": 5.191113247593471e-12, + "loss": 0.9925, + "step": 9750 + }, + { + "epoch": 3.0, + "step": 9750, + "total_flos": 1.0529587773729669e+18, + "train_loss": 1.0360454524602645, + "train_runtime": 8496.2923, + "train_samples_per_second": 18.362, + "train_steps_per_second": 1.148 + } + ], + "logging_steps": 5, + "max_steps": 9750, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 100, + "total_flos": 1.0529587773729669e+18, + "train_batch_size": 4, + "trial_name": null, + "trial_params": null +}