| { | |
| "best_metric": 1.6512105464935303, | |
| "best_model_checkpoint": "output/slava-kpss/checkpoint-1922", | |
| "epoch": 2.0, | |
| "global_step": 1922, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001371902778945302, | |
| "loss": 2.663, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013716111433378645, | |
| "loss": 2.4309, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013711251758398495, | |
| "loss": 2.2912, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001370445014195492, | |
| "loss": 2.3384, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013695708511920587, | |
| "loss": 2.2697, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001368502934605203, | |
| "loss": 2.2379, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00013672415671287354, | |
| "loss": 2.1656, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00013657871062888258, | |
| "loss": 2.1628, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013641399643426666, | |
| "loss": 2.1289, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000136230060816162, | |
| "loss": 2.09, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00013602695590988865, | |
| "loss": 2.0049, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0001358047392841732, | |
| "loss": 2.2023, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00013556347392483116, | |
| "loss": 2.1472, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00013530322821691406, | |
| "loss": 2.0605, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00013502407592532636, | |
| "loss": 2.1713, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00013472609617391705, | |
| "loss": 2.0815, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001344093734230526, | |
| "loss": 1.998, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00013407399744567734, | |
| "loss": 1.9623, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00013372006330186772, | |
| "loss": 2.0139, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00013334767131188837, | |
| "loss": 2.0258, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00013295692702775685, | |
| "loss": 1.9965, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013254794120332568, | |
| "loss": 2.0578, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013212082976288994, | |
| "loss": 1.9713, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00013167571376832926, | |
| "loss": 2.0398, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00013121271938479367, | |
| "loss": 1.9906, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00013073197784494285, | |
| "loss": 1.9652, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001302336254117493, | |
| "loss": 1.9658, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00012971780333987523, | |
| "loss": 2.0634, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012918465783563518, | |
| "loss": 2.0384, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012863434001555456, | |
| "loss": 1.9783, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00012806700586353683, | |
| "loss": 2.0149, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00012748281618665092, | |
| "loss": 1.9433, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00012688193656955137, | |
| "loss": 2.0019, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0001262645373275447, | |
| "loss": 2.036, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001256307934583145, | |
| "loss": 1.9862, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00012498088459231957, | |
| "loss": 1.9842, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00012431499494187896, | |
| "loss": 2.0212, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000123633313248958, | |
| "loss": 1.8205, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00012293603273167084, | |
| "loss": 2.0015, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00012222335102951405, | |
| "loss": 1.8653, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00012149547014734692, | |
| "loss": 2.0522, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00012075259639813464, | |
| "loss": 1.8885, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00011999494034447026, | |
| "loss": 1.8864, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00011922271673889206, | |
| "loss": 1.9415, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00011843614446301341, | |
| "loss": 1.9542, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00011763544646548234, | |
| "loss": 1.8528, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00011682084969878809, | |
| "loss": 1.8431, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00011599258505493302, | |
| "loss": 1.9018, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001151508872999878, | |
| "loss": 1.8971, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011429599500754859, | |
| "loss": 1.9842, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011342815049111488, | |
| "loss": 1.8397, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00011254759973540735, | |
| "loss": 1.854, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0001116545923266452, | |
| "loss": 1.9264, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00011074938138180258, | |
| "loss": 1.833, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00010983222347686431, | |
| "loss": 1.8783, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00010890337857410102, | |
| "loss": 1.8777, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00010796310994838476, | |
| "loss": 1.8851, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00010701168411256533, | |
| "loss": 1.8735, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0001060493707419291, | |
| "loss": 1.9696, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00010507644259776136, | |
| "loss": 1.8878, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00010409317545003389, | |
| "loss": 1.8279, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001030998479992398, | |
| "loss": 1.9672, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010209674179739785, | |
| "loss": 1.889, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00010108414116824834, | |
| "loss": 1.8617, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00010006233312666341, | |
| "loss": 1.9077, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.90316072972947e-05, | |
| "loss": 1.8912, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.79922558324811e-05, | |
| "loss": 1.7622, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.694457332944009e-05, | |
| "loss": 1.8778, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.588885674676624e-05, | |
| "loss": 1.8761, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.482540532026027e-05, | |
| "loss": 1.8225, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.37545204781125e-05, | |
| "loss": 1.7817, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.26765057554653e-05, | |
| "loss": 1.7633, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.159166670837789e-05, | |
| "loss": 1.8406, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.05003108272186e-05, | |
| "loss": 1.9374, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.940274744950875e-05, | |
| "loss": 1.8444, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.829928767224302e-05, | |
| "loss": 1.8098, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.71902442637111e-05, | |
| "loss": 1.822, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.607593157484563e-05, | |
| "loss": 1.87, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.495666545012144e-05, | |
| "loss": 1.8821, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.383276313803162e-05, | |
| "loss": 1.8619, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.270454320116558e-05, | |
| "loss": 1.896, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.157232542591454e-05, | |
| "loss": 1.776, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.043643073183026e-05, | |
| "loss": 1.8886, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.92971810806626e-05, | |
| "loss": 1.8725, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.815489938510145e-05, | |
| "loss": 1.8305, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.700990941724947e-05, | |
| "loss": 1.8383, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.586253571685095e-05, | |
| "loss": 1.872, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.471310349930326e-05, | |
| "loss": 1.8617, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.356193856347655e-05, | |
| "loss": 1.8118, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.24093671993686e-05, | |
| "loss": 1.8363, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.125571609561963e-05, | |
| "loss": 1.7498, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.010131224691501e-05, | |
| "loss": 1.8902, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.894648286130055e-05, | |
| "loss": 1.9075, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.779155526743765e-05, | |
| "loss": 1.8081, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.66368568218242e-05, | |
| "loss": 1.8239, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.548271481600758e-05, | |
| "loss": 1.8584, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.432945638381598e-05, | |
| "loss": 1.9037, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.317740840863456e-05, | |
| "loss": 1.7551, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.202689743075261e-05, | |
| "loss": 1.8329, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 6.0878249554807756e-05, | |
| "loss": 1.7776, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.9731790357353845e-05, | |
| "loss": 1.8987, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.8587844794578496e-05, | |
| "loss": 1.6097, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 5.744673711019635e-05, | |
| "loss": 1.8081, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.630879074354446e-05, | |
| "loss": 1.8023, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.517432823790546e-05, | |
| "loss": 1.709, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.404367114908498e-05, | |
| "loss": 1.7632, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.291713995426862e-05, | |
| "loss": 1.796, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.179505396118502e-05, | |
| "loss": 1.7405, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.067773121760007e-05, | |
| "loss": 1.8264, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.9565488421168504e-05, | |
| "loss": 1.8083, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.8458640829668e-05, | |
| "loss": 1.7488, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.735750217164156e-05, | |
| "loss": 1.8385, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.6262384557473104e-05, | |
| "loss": 1.8445, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.517359839092207e-05, | |
| "loss": 1.7808, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.409145228114133e-05, | |
| "loss": 1.7932, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.3016252955204197e-05, | |
| "loss": 1.7931, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.1948305171164515e-05, | |
| "loss": 1.7508, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.08879116316751e-05, | |
| "loss": 1.7877, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.98353728981888e-05, | |
| "loss": 1.8233, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.879098730576618e-05, | |
| "loss": 1.8181, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7755050878514536e-05, | |
| "loss": 1.6906, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.672785724568185e-05, | |
| "loss": 1.7465, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.570969755842952e-05, | |
| "loss": 1.7276, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.4700860407307565e-05, | |
| "loss": 1.7561, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.3701631740455454e-05, | |
| "loss": 1.8063, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.271229478255218e-05, | |
| "loss": 1.6319, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.173312995453793e-05, | |
| "loss": 1.7234, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.07644147941308e-05, | |
| "loss": 1.811, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.9806423877160492e-05, | |
| "loss": 1.8344, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.8859428739741754e-05, | |
| "loss": 1.7602, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.7923697801309092e-05, | |
| "loss": 1.8365, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.699949628853528e-05, | |
| "loss": 1.7506, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.60870861601545e-05, | |
| "loss": 1.7098, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.518672603271192e-05, | |
| "loss": 1.8215, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.429867110726057e-05, | |
| "loss": 1.8035, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.3423173097026407e-05, | |
| "loss": 1.7455, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.25604801560617e-05, | |
| "loss": 1.8008, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.1710836808907555e-05, | |
| "loss": 1.7584, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.0874483881285084e-05, | |
| "loss": 1.7836, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.0051658431834844e-05, | |
| "loss": 1.7353, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.924259368492425e-05, | |
| "loss": 1.7518, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.844751896454173e-05, | |
| "loss": 1.7968, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.766665962929623e-05, | |
| "loss": 1.751, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6900237008540944e-05, | |
| "loss": 1.7056, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.6148468339638933e-05, | |
| "loss": 1.7457, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.5411566706388707e-05, | |
| "loss": 1.7712, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4689740978626948e-05, | |
| "loss": 1.7609, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.3983195753025887e-05, | |
| "loss": 1.7358, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3292131295101604e-05, | |
| "loss": 1.8011, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.2616743482450217e-05, | |
| "loss": 1.7338, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1957223749227626e-05, | |
| "loss": 1.7185, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.1313759031888791e-05, | |
| "loss": 1.7987, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.0686531716201893e-05, | |
| "loss": 1.7855, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.0075719585552289e-05, | |
| "loss": 1.755, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.481495770550924e-06, | |
| "loss": 1.7749, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.90402869996171e-06, | |
| "loss": 1.7306, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.343482052961487e-06, | |
| "loss": 1.7385, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.800014712746244e-06, | |
| "loss": 1.8382, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.273780721496786e-06, | |
| "loss": 1.7857, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.7649292367164704e-06, | |
| "loss": 1.7224, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.2736044889534784e-06, | |
| "loss": 1.7514, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.799945740919712e-06, | |
| "loss": 1.7887, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 5.344087248017646e-06, | |
| "loss": 1.7605, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.906158220286551e-06, | |
| "loss": 1.7712, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.486282785778806e-06, | |
| "loss": 1.7334, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.084579955376559e-06, | |
| "loss": 1.749, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.7011635890589766e-06, | |
| "loss": 1.7343, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.3361423636293224e-06, | |
| "loss": 1.7663, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.989619741911281e-06, | |
| "loss": 1.759, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.6616939434230985e-06, | |
| "loss": 1.7025, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.352457916537921e-06, | |
| "loss": 1.7564, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.0619993121382247e-06, | |
| "loss": 1.8452, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.7904004587717305e-06, | |
| "loss": 1.7741, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.5377383393159132e-06, | |
| "loss": 1.7832, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.3040845691577635e-06, | |
| "loss": 1.8237, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0895053758948607e-06, | |
| "loss": 1.7458, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.940615805635918e-07, | |
| "loss": 1.7009, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.178085803998752e-07, | |
| "loss": 1.725, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.607963331371593e-07, | |
| "loss": 1.7469, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.2306934284621745e-07, | |
| "loss": 1.6719, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.04666647320803e-07, | |
| "loss": 1.8424, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.0562180701263117e-07, | |
| "loss": 1.7483, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.2596289551889364e-07, | |
| "loss": 1.7528, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.571249162498684e-08, | |
| "loss": 1.8179, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.4887672904708548e-08, | |
| "loss": 1.7763, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.5000108797349717e-09, | |
| "loss": 1.7489, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.7264798879623413, | |
| "eval_runtime": 80.022, | |
| "eval_samples_per_second": 20.494, | |
| "eval_steps_per_second": 2.562, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.476465669200449e-07, | |
| "loss": 1.7548, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.615902325712679e-07, | |
| "loss": 1.7767, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.38184037085626e-08, | |
| "loss": 1.7322, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.434918674879014e-08, | |
| "loss": 1.7019, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.319579824933328e-08, | |
| "loss": 1.6765, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.665613770225118e-10, | |
| "loss": 1.742, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.6864277124404907, | |
| "eval_runtime": 68.8946, | |
| "eval_samples_per_second": 20.466, | |
| "eval_steps_per_second": 2.569, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.8649036845078714e-09, | |
| "loss": 1.574, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.9689356194480253e-08, | |
| "loss": 1.7872, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 7.183355379217629e-08, | |
| "loss": 1.688, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3228623692592306e-07, | |
| "loss": 1.7384, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.1103125461527332e-07, | |
| "loss": 1.672, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.0804756876609914e-07, | |
| "loss": 1.7357, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.233092597912044e-07, | |
| "loss": 1.7344, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.567855335352604e-07, | |
| "loss": 1.7574, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.084407295019475e-07, | |
| "loss": 1.7031, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.782343303812844e-07, | |
| "loss": 1.6877, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.0661209728745555e-06, | |
| "loss": 1.6856, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.272050459813843e-06, | |
| "loss": 1.7801, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.495967773573164e-06, | |
| "loss": 1.7647, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.7378130907672579e-06, | |
| "loss": 1.7437, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.997521798234569e-06, | |
| "loss": 1.7193, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.275024510299646e-06, | |
| "loss": 1.8438, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.570247087310856e-06, | |
| "loss": 1.7861, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.883110655447913e-06, | |
| "loss": 1.7835, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.213531627794597e-06, | |
| "loss": 1.6692, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.561421726670198e-06, | |
| "loss": 1.673, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.926688007214648e-06, | |
| "loss": 1.7161, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.3092328822202e-06, | |
| "loss": 1.725, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.708954148203643e-06, | |
| "loss": 1.7126, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.1257450127116035e-06, | |
| "loss": 1.7486, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.559494122852188e-06, | |
| "loss": 1.7327, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.0100855950445935e-06, | |
| "loss": 1.7928, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.477399045979628e-06, | |
| "loss": 1.7449, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.961309624782102e-06, | |
| "loss": 1.7278, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.461688046366857e-06, | |
| "loss": 1.7724, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.97840062597962e-06, | |
| "loss": 1.7395, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.511309314913076e-06, | |
| "loss": 1.7612, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.060271737389124e-06, | |
| "loss": 1.7803, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.625141228596819e-06, | |
| "loss": 1.698, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.0205766873876643e-05, | |
| "loss": 1.7627, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.08019935490398e-05, | |
| "loss": 1.615, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.1413661961812419e-05, | |
| "loss": 1.7236, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.2040608694393166e-05, | |
| "loss": 1.732, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2682666247113272e-05, | |
| "loss": 1.7578, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.333966308318674e-05, | |
| "loss": 1.7636, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4011423674539631e-05, | |
| "loss": 1.6698, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.4697768548705208e-05, | |
| "loss": 1.8001, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.5398514336773276e-05, | |
| "loss": 1.7019, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.611347382238021e-05, | |
| "loss": 1.7685, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.684245599172719e-05, | |
| "loss": 1.8088, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.7585266084612767e-05, | |
| "loss": 1.7428, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.8341705646466328e-05, | |
| "loss": 1.712, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.9111572581368946e-05, | |
| "loss": 1.7453, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.9894661206046444e-05, | |
| "loss": 1.6769, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.069076230482155e-05, | |
| "loss": 1.6822, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.1499663185509284e-05, | |
| "loss": 1.7965, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.232114773624152e-05, | |
| "loss": 1.6978, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.3154996483204744e-05, | |
| "loss": 1.7852, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.400098664927671e-05, | |
| "loss": 1.6758, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.4858892213544702e-05, | |
| "loss": 1.742, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.5728483971691396e-05, | |
| "loss": 1.7897, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.660952959723034e-05, | |
| "loss": 1.6749, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.750179370357635e-05, | |
| "loss": 1.7127, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8405037906932926e-05, | |
| "loss": 1.7259, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.9319020889980675e-05, | |
| "loss": 1.6448, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.0243498466349233e-05, | |
| "loss": 1.6744, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.117822364585623e-05, | |
| "loss": 1.6893, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.212294670049457e-05, | |
| "loss": 1.7822, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.3077415231151914e-05, | |
| "loss": 1.7231, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.4041374235043516e-05, | |
| "loss": 1.7438, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.501456617384015e-05, | |
| "loss": 1.776, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.599673104247454e-05, | |
| "loss": 1.7459, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.6987606438605696e-05, | |
| "loss": 1.7127, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.798692763272465e-05, | |
| "loss": 1.6762, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.899442763888126e-05, | |
| "loss": 1.7619, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.000983728601432e-05, | |
| "loss": 1.7135, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.103288528986497e-05, | |
| "loss": 1.7328, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.20632983254554e-05, | |
| "loss": 1.7118, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.3100801100111876e-05, | |
| "loss": 1.801, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.4145116427014376e-05, | |
| "loss": 1.7959, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.519596529925159e-05, | |
| "loss": 1.7436, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.625306696436258e-05, | |
| "loss": 1.7913, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.7316138999344416e-05, | |
| "loss": 1.7002, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.838489738610674e-05, | |
| "loss": 1.6758, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.945905658735171e-05, | |
| "loss": 1.7027, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.0538329622860605e-05, | |
| "loss": 1.7378, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.1622428146165326e-05, | |
| "loss": 1.6332, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.271106252158521e-05, | |
| "loss": 1.7335, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.3803941901608256e-05, | |
| "loss": 1.7126, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.4900774304595864e-05, | |
| "loss": 1.7287, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.600126669279115e-05, | |
| "loss": 1.6935, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.710512505060865e-05, | |
| "loss": 1.7214, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.8212054463186014e-05, | |
| "loss": 1.7355, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.9321759195175425e-05, | |
| "loss": 1.7537, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.043394276975451e-05, | |
| "loss": 1.776, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.154830804783502e-05, | |
| "loss": 1.783, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.266455730744911e-05, | |
| "loss": 1.7703, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.378239232329044e-05, | |
| "loss": 1.7259, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 6.490151444639059e-05, | |
| "loss": 1.6852, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.602162468390815e-05, | |
| "loss": 1.675, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.714242377900986e-05, | |
| "loss": 1.7367, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.826361229082211e-05, | |
| "loss": 1.8171, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.938489067443205e-05, | |
| "loss": 1.7388, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.050595936091584e-05, | |
| "loss": 1.7459, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.162651883737402e-05, | |
| "loss": 1.808, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.274626972695132e-05, | |
| "loss": 1.7933, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.386491286882046e-05, | |
| "loss": 1.7711, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.498214939810823e-05, | |
| "loss": 1.7745, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.609768082574206e-05, | |
| "loss": 1.6494, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 7.721120911819707e-05, | |
| "loss": 1.7815, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.83224367771204e-05, | |
| "loss": 1.7829, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.943106691881334e-05, | |
| "loss": 1.7363, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.053680335354889e-05, | |
| "loss": 1.7149, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.163935066470398e-05, | |
| "loss": 1.7134, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.273841428768484e-05, | |
| "loss": 1.7653, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.383370058862563e-05, | |
| "loss": 1.7677, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.492491694283713e-05, | |
| "loss": 1.6578, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.601177181298704e-05, | |
| "loss": 1.7271, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.709397482698893e-05, | |
| "loss": 1.6661, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.817123685558034e-05, | |
| "loss": 1.7347, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.92432700895683e-05, | |
| "loss": 1.7298, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.03097881167231e-05, | |
| "loss": 1.7343, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.137050599829776e-05, | |
| "loss": 1.7651, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.242514034515462e-05, | |
| "loss": 1.7716, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.34734093934778e-05, | |
| "loss": 1.8064, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.451503308005074e-05, | |
| "loss": 1.8239, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.554973311708053e-05, | |
| "loss": 1.7248, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.657723306654681e-05, | |
| "loss": 1.6457, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.759725841405743e-05, | |
| "loss": 1.8182, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.86095366421894e-05, | |
| "loss": 1.6211, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.961379730329683e-05, | |
| "loss": 1.6631, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00010060977209176536, | |
| "loss": 1.7168, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0001015971949156952, | |
| "loss": 1.752, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00010257580196799162, | |
| "loss": 1.7518, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00010354533179684596, | |
| "loss": 1.7591, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010450552537558691, | |
| "loss": 1.7579, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010545612617188394, | |
| "loss": 1.7364, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00010639688021628421, | |
| "loss": 1.7379, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00010732753617006524, | |
| "loss": 1.7443, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00010824784539238402, | |
| "loss": 1.6989, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00010915756200670606, | |
| "loss": 1.6362, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011005644296649529, | |
| "loss": 1.7417, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011094424812014832, | |
| "loss": 1.635, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00011182074027515503, | |
| "loss": 1.6471, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0001126856852614686, | |
| "loss": 1.6803, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011353885199406824, | |
| "loss": 1.6863, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011438001253469733, | |
| "loss": 1.7498, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011520894215276136, | |
| "loss": 1.7037, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00011602541938536831, | |
| "loss": 1.6918, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00011682922609649652, | |
| "loss": 1.7, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0001176201475352733, | |
| "loss": 1.7662, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011839797239334955, | |
| "loss": 1.8041, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011916249286135422, | |
| "loss": 1.7271, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00011991350468441439, | |
| "loss": 1.7033, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00012065080721672542, | |
| "loss": 1.6855, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00012137420347515691, | |
| "loss": 1.6554, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00012208350019188007, | |
| "loss": 1.7365, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00012277850786600282, | |
| "loss": 1.7106, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00012345904081419794, | |
| "loss": 1.735, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00012412491722031186, | |
| "loss": 1.6991, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00012477595918393978, | |
| "loss": 1.6981, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0001254119927679549, | |
| "loss": 1.7559, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00012603284804497882, | |
| "loss": 1.7479, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00012663835914278047, | |
| "loss": 1.7712, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.0001272283642885918, | |
| "loss": 1.7758, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00012780270585232808, | |
| "loss": 1.7353, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00012836123038870173, | |
| "loss": 1.7297, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00012890378867821763, | |
| "loss": 1.8349, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.0001294302357670399, | |
| "loss": 1.8035, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00012994043100571866, | |
| "loss": 1.7481, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.000130434238086767, | |
| "loss": 1.6094, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00013091152508107798, | |
| "loss": 1.7314, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00013137216447317167, | |
| "loss": 1.8012, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0001318160331952632, | |
| "loss": 1.7435, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0001322430126601424, | |
| "loss": 1.7132, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013265298879285635, | |
| "loss": 1.6928, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013304585206118667, | |
| "loss": 1.6412, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00013342149750491278, | |
| "loss": 1.7068, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0001337798247638538, | |
| "loss": 1.7385, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0001341207381046819, | |
| "loss": 1.612, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013444414644649843, | |
| "loss": 1.6917, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.00013474996338516847, | |
| "loss": 1.7257, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00013503810721640465, | |
| "loss": 1.7391, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00013530850095759623, | |
| "loss": 1.7311, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00013556107236837607, | |
| "loss": 1.671, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013579575396992113, | |
| "loss": 1.716, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013601248306298038, | |
| "loss": 1.7233, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00013621120174462615, | |
| "loss": 1.6546, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00013639185692372385, | |
| "loss": 1.7717, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00013655440033511618, | |
| "loss": 1.7035, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00013669878855251797, | |
| "loss": 1.7345, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013682498300011836, | |
| "loss": 1.7678, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013693294996288687, | |
| "loss": 1.6938, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013702266059558108, | |
| "loss": 1.7334, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00013709409093045299, | |
| "loss": 1.6722, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00013714722188365257, | |
| "loss": 1.7148, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00013718203926032623, | |
| "loss": 1.6982, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00013719853375840932, | |
| "loss": 1.7054, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.6512105464935303, | |
| "eval_runtime": 68.7702, | |
| "eval_samples_per_second": 20.503, | |
| "eval_steps_per_second": 2.574, | |
| "step": 1922 | |
| } | |
| ], | |
| "max_steps": 1922, | |
| "num_train_epochs": 2, | |
| "total_flos": 2008029265920000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |