{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.9997705106138841, "eval_steps": 500, "global_step": 8714, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.2935779816513764e-07, "loss": 1.6585, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.1467889908256882e-06, "loss": 1.6894, "step": 5 }, { "epoch": 0.0, "learning_rate": 2.2935779816513764e-06, "loss": 1.6975, "step": 10 }, { "epoch": 0.0, "learning_rate": 3.4403669724770644e-06, "loss": 1.6499, "step": 15 }, { "epoch": 0.0, "learning_rate": 4.587155963302753e-06, "loss": 1.67, "step": 20 }, { "epoch": 0.01, "learning_rate": 5.733944954128441e-06, "loss": 1.6319, "step": 25 }, { "epoch": 0.01, "learning_rate": 6.880733944954129e-06, "loss": 1.5944, "step": 30 }, { "epoch": 0.01, "learning_rate": 8.027522935779817e-06, "loss": 1.5345, "step": 35 }, { "epoch": 0.01, "learning_rate": 9.174311926605506e-06, "loss": 1.4475, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.0321100917431194e-05, "loss": 1.4025, "step": 45 }, { "epoch": 0.01, "learning_rate": 1.1467889908256882e-05, "loss": 1.3724, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.2614678899082569e-05, "loss": 1.3035, "step": 55 }, { "epoch": 0.01, "learning_rate": 1.3761467889908258e-05, "loss": 1.3084, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.4908256880733946e-05, "loss": 1.2827, "step": 65 }, { "epoch": 0.02, "learning_rate": 1.6055045871559634e-05, "loss": 1.2762, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.7201834862385323e-05, "loss": 1.2332, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.834862385321101e-05, "loss": 1.259, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.94954128440367e-05, "loss": 1.2198, "step": 85 }, { "epoch": 0.02, "learning_rate": 2.0642201834862388e-05, "loss": 1.1714, "step": 90 }, { "epoch": 0.02, "learning_rate": 2.1788990825688073e-05, "loss": 1.1934, "step": 95 }, { "epoch": 0.02, "learning_rate": 2.2935779816513765e-05, "loss": 1.1567, "step": 100 }, { "epoch": 0.02, "learning_rate": 2.408256880733945e-05, "loss": 1.1338, "step": 105 }, { "epoch": 0.03, "learning_rate": 2.5229357798165138e-05, "loss": 1.1182, "step": 110 }, { "epoch": 0.03, "learning_rate": 2.6376146788990823e-05, "loss": 1.1278, "step": 115 }, { "epoch": 0.03, "learning_rate": 2.7522935779816515e-05, "loss": 1.1047, "step": 120 }, { "epoch": 0.03, "learning_rate": 2.8669724770642203e-05, "loss": 1.1185, "step": 125 }, { "epoch": 0.03, "learning_rate": 2.9816513761467892e-05, "loss": 1.1142, "step": 130 }, { "epoch": 0.03, "learning_rate": 3.096330275229358e-05, "loss": 1.1013, "step": 135 }, { "epoch": 0.03, "learning_rate": 3.211009174311927e-05, "loss": 1.0514, "step": 140 }, { "epoch": 0.03, "learning_rate": 3.325688073394496e-05, "loss": 1.0727, "step": 145 }, { "epoch": 0.03, "learning_rate": 3.4403669724770645e-05, "loss": 1.1009, "step": 150 }, { "epoch": 0.04, "learning_rate": 3.555045871559633e-05, "loss": 1.0711, "step": 155 }, { "epoch": 0.04, "learning_rate": 3.669724770642202e-05, "loss": 1.0761, "step": 160 }, { "epoch": 0.04, "learning_rate": 3.784403669724771e-05, "loss": 1.0391, "step": 165 }, { "epoch": 0.04, "learning_rate": 3.89908256880734e-05, "loss": 1.1219, "step": 170 }, { "epoch": 0.04, "learning_rate": 4.013761467889908e-05, "loss": 1.0719, "step": 175 }, { "epoch": 0.04, "learning_rate": 4.1284403669724776e-05, "loss": 1.0562, "step": 180 }, { "epoch": 0.04, "learning_rate": 4.2431192660550464e-05, "loss": 1.05, "step": 185 }, { "epoch": 0.04, "learning_rate": 4.3577981651376146e-05, "loss": 1.0729, "step": 190 }, { "epoch": 0.04, "learning_rate": 4.4724770642201834e-05, "loss": 1.0563, "step": 195 }, { "epoch": 0.05, "learning_rate": 4.587155963302753e-05, "loss": 1.0261, "step": 200 }, { "epoch": 0.05, "learning_rate": 4.701834862385321e-05, "loss": 1.0511, "step": 205 }, { "epoch": 0.05, "learning_rate": 4.81651376146789e-05, "loss": 1.0172, "step": 210 }, { "epoch": 0.05, "learning_rate": 4.931192660550459e-05, "loss": 1.0347, "step": 215 }, { "epoch": 0.05, "learning_rate": 5.0458715596330276e-05, "loss": 1.0125, "step": 220 }, { "epoch": 0.05, "learning_rate": 5.1605504587155965e-05, "loss": 1.0627, "step": 225 }, { "epoch": 0.05, "learning_rate": 5.2752293577981646e-05, "loss": 1.0431, "step": 230 }, { "epoch": 0.05, "learning_rate": 5.389908256880735e-05, "loss": 1.0538, "step": 235 }, { "epoch": 0.06, "learning_rate": 5.504587155963303e-05, "loss": 1.0462, "step": 240 }, { "epoch": 0.06, "learning_rate": 5.619266055045872e-05, "loss": 0.9942, "step": 245 }, { "epoch": 0.06, "learning_rate": 5.733944954128441e-05, "loss": 1.0193, "step": 250 }, { "epoch": 0.06, "learning_rate": 5.8486238532110095e-05, "loss": 1.0587, "step": 255 }, { "epoch": 0.06, "learning_rate": 5.9633027522935784e-05, "loss": 1.0127, "step": 260 }, { "epoch": 0.06, "learning_rate": 6.0779816513761465e-05, "loss": 0.9959, "step": 265 }, { "epoch": 0.06, "learning_rate": 6.192660550458716e-05, "loss": 1.0016, "step": 270 }, { "epoch": 0.06, "learning_rate": 6.307339449541286e-05, "loss": 0.9981, "step": 275 }, { "epoch": 0.06, "learning_rate": 6.422018348623854e-05, "loss": 1.0489, "step": 280 }, { "epoch": 0.07, "learning_rate": 6.536697247706422e-05, "loss": 1.0379, "step": 285 }, { "epoch": 0.07, "learning_rate": 6.651376146788991e-05, "loss": 1.0224, "step": 290 }, { "epoch": 0.07, "learning_rate": 6.76605504587156e-05, "loss": 0.9877, "step": 295 }, { "epoch": 0.07, "learning_rate": 6.880733944954129e-05, "loss": 0.9851, "step": 300 }, { "epoch": 0.07, "learning_rate": 6.995412844036697e-05, "loss": 0.9941, "step": 305 }, { "epoch": 0.07, "learning_rate": 7.110091743119265e-05, "loss": 0.9819, "step": 310 }, { "epoch": 0.07, "learning_rate": 7.224770642201836e-05, "loss": 1.001, "step": 315 }, { "epoch": 0.07, "learning_rate": 7.339449541284404e-05, "loss": 1.0097, "step": 320 }, { "epoch": 0.07, "learning_rate": 7.454128440366973e-05, "loss": 0.9966, "step": 325 }, { "epoch": 0.08, "learning_rate": 7.568807339449542e-05, "loss": 1.0303, "step": 330 }, { "epoch": 0.08, "learning_rate": 7.68348623853211e-05, "loss": 1.0198, "step": 335 }, { "epoch": 0.08, "learning_rate": 7.79816513761468e-05, "loss": 1.0267, "step": 340 }, { "epoch": 0.08, "learning_rate": 7.912844036697248e-05, "loss": 0.9932, "step": 345 }, { "epoch": 0.08, "learning_rate": 8.027522935779816e-05, "loss": 1.03, "step": 350 }, { "epoch": 0.08, "learning_rate": 8.142201834862386e-05, "loss": 0.9871, "step": 355 }, { "epoch": 0.08, "learning_rate": 8.256880733944955e-05, "loss": 0.9775, "step": 360 }, { "epoch": 0.08, "learning_rate": 8.371559633027523e-05, "loss": 1.0259, "step": 365 }, { "epoch": 0.08, "learning_rate": 8.486238532110093e-05, "loss": 1.0187, "step": 370 }, { "epoch": 0.09, "learning_rate": 8.600917431192661e-05, "loss": 0.9691, "step": 375 }, { "epoch": 0.09, "learning_rate": 8.715596330275229e-05, "loss": 1.0352, "step": 380 }, { "epoch": 0.09, "learning_rate": 8.830275229357799e-05, "loss": 1.0247, "step": 385 }, { "epoch": 0.09, "learning_rate": 8.944954128440367e-05, "loss": 1.0261, "step": 390 }, { "epoch": 0.09, "learning_rate": 9.059633027522936e-05, "loss": 0.9829, "step": 395 }, { "epoch": 0.09, "learning_rate": 9.174311926605506e-05, "loss": 1.0048, "step": 400 }, { "epoch": 0.09, "learning_rate": 9.288990825688074e-05, "loss": 0.9894, "step": 405 }, { "epoch": 0.09, "learning_rate": 9.403669724770642e-05, "loss": 0.9692, "step": 410 }, { "epoch": 0.1, "learning_rate": 9.518348623853212e-05, "loss": 0.9943, "step": 415 }, { "epoch": 0.1, "learning_rate": 9.63302752293578e-05, "loss": 0.9668, "step": 420 }, { "epoch": 0.1, "learning_rate": 9.74770642201835e-05, "loss": 0.9863, "step": 425 }, { "epoch": 0.1, "learning_rate": 9.862385321100918e-05, "loss": 1.0023, "step": 430 }, { "epoch": 0.1, "learning_rate": 9.977064220183486e-05, "loss": 1.0009, "step": 435 }, { "epoch": 0.1, "learning_rate": 0.00010091743119266055, "loss": 0.9834, "step": 440 }, { "epoch": 0.1, "learning_rate": 0.00010206422018348625, "loss": 1.0098, "step": 445 }, { "epoch": 0.1, "learning_rate": 0.00010321100917431193, "loss": 1.0153, "step": 450 }, { "epoch": 0.1, "learning_rate": 0.00010435779816513762, "loss": 1.0075, "step": 455 }, { "epoch": 0.11, "learning_rate": 0.00010550458715596329, "loss": 0.9688, "step": 460 }, { "epoch": 0.11, "learning_rate": 0.000106651376146789, "loss": 0.979, "step": 465 }, { "epoch": 0.11, "learning_rate": 0.0001077981651376147, "loss": 0.9917, "step": 470 }, { "epoch": 0.11, "learning_rate": 0.00010894495412844036, "loss": 0.9693, "step": 475 }, { "epoch": 0.11, "learning_rate": 0.00011009174311926606, "loss": 1.0148, "step": 480 }, { "epoch": 0.11, "learning_rate": 0.00011123853211009174, "loss": 0.9609, "step": 485 }, { "epoch": 0.11, "learning_rate": 0.00011238532110091744, "loss": 1.0058, "step": 490 }, { "epoch": 0.11, "learning_rate": 0.00011353211009174312, "loss": 1.0091, "step": 495 }, { "epoch": 0.11, "learning_rate": 0.00011467889908256881, "loss": 1.0326, "step": 500 }, { "epoch": 0.12, "learning_rate": 0.00011582568807339451, "loss": 1.0317, "step": 505 }, { "epoch": 0.12, "learning_rate": 0.00011697247706422019, "loss": 1.0165, "step": 510 }, { "epoch": 0.12, "learning_rate": 0.00011811926605504589, "loss": 0.9853, "step": 515 }, { "epoch": 0.12, "learning_rate": 0.00011926605504587157, "loss": 0.9794, "step": 520 }, { "epoch": 0.12, "learning_rate": 0.00012041284403669726, "loss": 1.0263, "step": 525 }, { "epoch": 0.12, "learning_rate": 0.00012155963302752293, "loss": 0.9859, "step": 530 }, { "epoch": 0.12, "learning_rate": 0.00012270642201834864, "loss": 1.0099, "step": 535 }, { "epoch": 0.12, "learning_rate": 0.00012385321100917432, "loss": 1.0113, "step": 540 }, { "epoch": 0.13, "learning_rate": 0.000125, "loss": 1.0108, "step": 545 }, { "epoch": 0.13, "learning_rate": 0.0001261467889908257, "loss": 1.019, "step": 550 }, { "epoch": 0.13, "learning_rate": 0.0001272935779816514, "loss": 1.0102, "step": 555 }, { "epoch": 0.13, "learning_rate": 0.00012844036697247707, "loss": 0.9832, "step": 560 }, { "epoch": 0.13, "learning_rate": 0.00012958715596330276, "loss": 0.9758, "step": 565 }, { "epoch": 0.13, "learning_rate": 0.00013073394495412844, "loss": 0.9706, "step": 570 }, { "epoch": 0.13, "learning_rate": 0.00013188073394495412, "loss": 0.9925, "step": 575 }, { "epoch": 0.13, "learning_rate": 0.00013302752293577983, "loss": 0.9824, "step": 580 }, { "epoch": 0.13, "learning_rate": 0.0001341743119266055, "loss": 0.9866, "step": 585 }, { "epoch": 0.14, "learning_rate": 0.0001353211009174312, "loss": 0.9674, "step": 590 }, { "epoch": 0.14, "learning_rate": 0.0001364678899082569, "loss": 0.9735, "step": 595 }, { "epoch": 0.14, "learning_rate": 0.00013761467889908258, "loss": 0.9868, "step": 600 }, { "epoch": 0.14, "learning_rate": 0.00013876146788990826, "loss": 0.9869, "step": 605 }, { "epoch": 0.14, "learning_rate": 0.00013990825688073395, "loss": 0.9869, "step": 610 }, { "epoch": 0.14, "learning_rate": 0.00014105504587155965, "loss": 0.9901, "step": 615 }, { "epoch": 0.14, "learning_rate": 0.0001422018348623853, "loss": 1.0292, "step": 620 }, { "epoch": 0.14, "learning_rate": 0.00014334862385321102, "loss": 0.9755, "step": 625 }, { "epoch": 0.14, "learning_rate": 0.00014449541284403673, "loss": 0.9811, "step": 630 }, { "epoch": 0.15, "learning_rate": 0.00014564220183486238, "loss": 1.01, "step": 635 }, { "epoch": 0.15, "learning_rate": 0.0001467889908256881, "loss": 0.9728, "step": 640 }, { "epoch": 0.15, "learning_rate": 0.00014793577981651377, "loss": 0.9966, "step": 645 }, { "epoch": 0.15, "learning_rate": 0.00014908256880733945, "loss": 1.0176, "step": 650 }, { "epoch": 0.15, "learning_rate": 0.00015022935779816513, "loss": 0.9708, "step": 655 }, { "epoch": 0.15, "learning_rate": 0.00015137614678899084, "loss": 0.9673, "step": 660 }, { "epoch": 0.15, "learning_rate": 0.00015252293577981652, "loss": 0.9735, "step": 665 }, { "epoch": 0.15, "learning_rate": 0.0001536697247706422, "loss": 0.9707, "step": 670 }, { "epoch": 0.15, "learning_rate": 0.00015481651376146791, "loss": 0.9657, "step": 675 }, { "epoch": 0.16, "learning_rate": 0.0001559633027522936, "loss": 0.9753, "step": 680 }, { "epoch": 0.16, "learning_rate": 0.00015711009174311928, "loss": 1.0284, "step": 685 }, { "epoch": 0.16, "learning_rate": 0.00015825688073394496, "loss": 0.9755, "step": 690 }, { "epoch": 0.16, "learning_rate": 0.00015940366972477064, "loss": 0.978, "step": 695 }, { "epoch": 0.16, "learning_rate": 0.00016055045871559632, "loss": 0.9792, "step": 700 }, { "epoch": 0.16, "learning_rate": 0.00016169724770642203, "loss": 0.9706, "step": 705 }, { "epoch": 0.16, "learning_rate": 0.0001628440366972477, "loss": 0.9856, "step": 710 }, { "epoch": 0.16, "learning_rate": 0.0001639908256880734, "loss": 0.9805, "step": 715 }, { "epoch": 0.17, "learning_rate": 0.0001651376146788991, "loss": 0.9914, "step": 720 }, { "epoch": 0.17, "learning_rate": 0.00016628440366972479, "loss": 0.9314, "step": 725 }, { "epoch": 0.17, "learning_rate": 0.00016743119266055047, "loss": 0.9705, "step": 730 }, { "epoch": 0.17, "learning_rate": 0.00016857798165137615, "loss": 0.9763, "step": 735 }, { "epoch": 0.17, "learning_rate": 0.00016972477064220186, "loss": 1.0024, "step": 740 }, { "epoch": 0.17, "learning_rate": 0.0001708715596330275, "loss": 0.9623, "step": 745 }, { "epoch": 0.17, "learning_rate": 0.00017201834862385322, "loss": 0.996, "step": 750 }, { "epoch": 0.17, "learning_rate": 0.00017316513761467893, "loss": 0.9984, "step": 755 }, { "epoch": 0.17, "learning_rate": 0.00017431192660550458, "loss": 0.9738, "step": 760 }, { "epoch": 0.18, "learning_rate": 0.0001754587155963303, "loss": 0.9751, "step": 765 }, { "epoch": 0.18, "learning_rate": 0.00017660550458715597, "loss": 0.9783, "step": 770 }, { "epoch": 0.18, "learning_rate": 0.00017775229357798166, "loss": 1.0108, "step": 775 }, { "epoch": 0.18, "learning_rate": 0.00017889908256880734, "loss": 0.9498, "step": 780 }, { "epoch": 0.18, "learning_rate": 0.00018004587155963305, "loss": 0.9716, "step": 785 }, { "epoch": 0.18, "learning_rate": 0.00018119266055045873, "loss": 1.0429, "step": 790 }, { "epoch": 0.18, "learning_rate": 0.0001823394495412844, "loss": 1.0272, "step": 795 }, { "epoch": 0.18, "learning_rate": 0.00018348623853211012, "loss": 0.9578, "step": 800 }, { "epoch": 0.18, "learning_rate": 0.0001846330275229358, "loss": 0.9731, "step": 805 }, { "epoch": 0.19, "learning_rate": 0.00018577981651376148, "loss": 0.9592, "step": 810 }, { "epoch": 0.19, "learning_rate": 0.00018692660550458716, "loss": 0.9514, "step": 815 }, { "epoch": 0.19, "learning_rate": 0.00018807339449541284, "loss": 0.955, "step": 820 }, { "epoch": 0.19, "learning_rate": 0.00018922018348623853, "loss": 0.9706, "step": 825 }, { "epoch": 0.19, "learning_rate": 0.00019036697247706424, "loss": 0.975, "step": 830 }, { "epoch": 0.19, "learning_rate": 0.00019151376146788992, "loss": 0.9891, "step": 835 }, { "epoch": 0.19, "learning_rate": 0.0001926605504587156, "loss": 1.0004, "step": 840 }, { "epoch": 0.19, "learning_rate": 0.0001938073394495413, "loss": 0.9773, "step": 845 }, { "epoch": 0.2, "learning_rate": 0.000194954128440367, "loss": 0.9578, "step": 850 }, { "epoch": 0.2, "learning_rate": 0.00019610091743119267, "loss": 0.9549, "step": 855 }, { "epoch": 0.2, "learning_rate": 0.00019724770642201835, "loss": 0.9628, "step": 860 }, { "epoch": 0.2, "learning_rate": 0.00019839449541284406, "loss": 0.9863, "step": 865 }, { "epoch": 0.2, "learning_rate": 0.00019954128440366972, "loss": 0.982, "step": 870 }, { "epoch": 0.2, "learning_rate": 0.00019999992777982591, "loss": 1.0301, "step": 875 }, { "epoch": 0.2, "learning_rate": 0.00019999948643469536, "loss": 0.9882, "step": 880 }, { "epoch": 0.2, "learning_rate": 0.00019999864386852187, "loss": 0.9953, "step": 885 }, { "epoch": 0.2, "learning_rate": 0.00019999740008468594, "loss": 0.9656, "step": 890 }, { "epoch": 0.21, "learning_rate": 0.00019999575508817795, "loss": 1.0045, "step": 895 }, { "epoch": 0.21, "learning_rate": 0.00019999370888559804, "loss": 0.9876, "step": 900 }, { "epoch": 0.21, "learning_rate": 0.00019999126148515601, "loss": 0.9823, "step": 905 }, { "epoch": 0.21, "learning_rate": 0.0001999884128966714, "loss": 0.9675, "step": 910 }, { "epoch": 0.21, "learning_rate": 0.0001999851631315734, "loss": 0.9774, "step": 915 }, { "epoch": 0.21, "learning_rate": 0.00019998151220290082, "loss": 0.9583, "step": 920 }, { "epoch": 0.21, "learning_rate": 0.00019997746012530206, "loss": 0.9841, "step": 925 }, { "epoch": 0.21, "learning_rate": 0.00019997300691503497, "loss": 0.9405, "step": 930 }, { "epoch": 0.21, "learning_rate": 0.00019996815258996686, "loss": 0.9624, "step": 935 }, { "epoch": 0.22, "learning_rate": 0.0001999628971695744, "loss": 0.9885, "step": 940 }, { "epoch": 0.22, "learning_rate": 0.00019995724067494353, "loss": 1.0082, "step": 945 }, { "epoch": 0.22, "learning_rate": 0.00019995118312876944, "loss": 0.959, "step": 950 }, { "epoch": 0.22, "learning_rate": 0.0001999447245553564, "loss": 1.0017, "step": 955 }, { "epoch": 0.22, "learning_rate": 0.0001999378649806177, "loss": 0.9921, "step": 960 }, { "epoch": 0.22, "learning_rate": 0.00019993060443207554, "loss": 1.0108, "step": 965 }, { "epoch": 0.22, "learning_rate": 0.00019992294293886095, "loss": 0.9921, "step": 970 }, { "epoch": 0.22, "learning_rate": 0.00019991488053171353, "loss": 0.9854, "step": 975 }, { "epoch": 0.22, "learning_rate": 0.00019990641724298156, "loss": 0.9967, "step": 980 }, { "epoch": 0.23, "learning_rate": 0.00019989755310662174, "loss": 0.9682, "step": 985 }, { "epoch": 0.23, "learning_rate": 0.000199888288158199, "loss": 0.9793, "step": 990 }, { "epoch": 0.23, "learning_rate": 0.00019987862243488647, "loss": 0.963, "step": 995 }, { "epoch": 0.23, "learning_rate": 0.00019986855597546526, "loss": 0.9852, "step": 1000 }, { "epoch": 0.23, "learning_rate": 0.0001998580888203243, "loss": 0.9586, "step": 1005 }, { "epoch": 0.23, "learning_rate": 0.00019984722101146029, "loss": 0.9769, "step": 1010 }, { "epoch": 0.23, "learning_rate": 0.00019983595259247735, "loss": 0.9582, "step": 1015 }, { "epoch": 0.23, "learning_rate": 0.000199824283608587, "loss": 0.9767, "step": 1020 }, { "epoch": 0.24, "learning_rate": 0.00019981221410660788, "loss": 0.9681, "step": 1025 }, { "epoch": 0.24, "learning_rate": 0.00019979974413496566, "loss": 0.9942, "step": 1030 }, { "epoch": 0.24, "learning_rate": 0.00019978687374369271, "loss": 0.982, "step": 1035 }, { "epoch": 0.24, "learning_rate": 0.00019977360298442803, "loss": 0.9758, "step": 1040 }, { "epoch": 0.24, "learning_rate": 0.00019975993191041695, "loss": 1.0581, "step": 1045 }, { "epoch": 0.24, "learning_rate": 0.00019974586057651102, "loss": 0.9499, "step": 1050 }, { "epoch": 0.24, "learning_rate": 0.0001997313890391676, "loss": 0.9758, "step": 1055 }, { "epoch": 0.24, "learning_rate": 0.00019971651735644995, "loss": 0.9742, "step": 1060 }, { "epoch": 0.24, "learning_rate": 0.00019970124558802663, "loss": 0.9327, "step": 1065 }, { "epoch": 0.25, "learning_rate": 0.00019968557379517152, "loss": 0.9624, "step": 1070 }, { "epoch": 0.25, "learning_rate": 0.00019966950204076346, "loss": 0.9812, "step": 1075 }, { "epoch": 0.25, "learning_rate": 0.00019965303038928608, "loss": 0.9728, "step": 1080 }, { "epoch": 0.25, "learning_rate": 0.0001996361589068274, "loss": 0.9561, "step": 1085 }, { "epoch": 0.25, "learning_rate": 0.00019961888766107972, "loss": 0.9741, "step": 1090 }, { "epoch": 0.25, "learning_rate": 0.0001996012167213393, "loss": 1.0002, "step": 1095 }, { "epoch": 0.25, "learning_rate": 0.00019958314615850598, "loss": 0.955, "step": 1100 }, { "epoch": 0.25, "learning_rate": 0.00019956467604508312, "loss": 0.9826, "step": 1105 }, { "epoch": 0.25, "learning_rate": 0.00019954580645517697, "loss": 0.9633, "step": 1110 }, { "epoch": 0.26, "learning_rate": 0.00019952653746449676, "loss": 0.9641, "step": 1115 }, { "epoch": 0.26, "learning_rate": 0.0001995068691503541, "loss": 0.9907, "step": 1120 }, { "epoch": 0.26, "learning_rate": 0.00019948680159166283, "loss": 0.9567, "step": 1125 }, { "epoch": 0.26, "learning_rate": 0.00019946633486893865, "loss": 0.9908, "step": 1130 }, { "epoch": 0.26, "learning_rate": 0.00019944546906429874, "loss": 0.9769, "step": 1135 }, { "epoch": 0.26, "learning_rate": 0.00019942420426146153, "loss": 1.0025, "step": 1140 }, { "epoch": 0.26, "learning_rate": 0.00019940254054574637, "loss": 0.9797, "step": 1145 }, { "epoch": 0.26, "learning_rate": 0.00019938047800407302, "loss": 0.9944, "step": 1150 }, { "epoch": 0.27, "learning_rate": 0.00019935801672496157, "loss": 0.9826, "step": 1155 }, { "epoch": 0.27, "learning_rate": 0.00019933515679853182, "loss": 0.9898, "step": 1160 }, { "epoch": 0.27, "learning_rate": 0.00019931189831650303, "loss": 0.9531, "step": 1165 }, { "epoch": 0.27, "learning_rate": 0.0001992882413721937, "loss": 0.9755, "step": 1170 }, { "epoch": 0.27, "learning_rate": 0.00019926418606052088, "loss": 0.9976, "step": 1175 }, { "epoch": 0.27, "learning_rate": 0.0001992397324780001, "loss": 0.9403, "step": 1180 }, { "epoch": 0.27, "learning_rate": 0.00019921488072274475, "loss": 0.9594, "step": 1185 }, { "epoch": 0.27, "learning_rate": 0.00019918963089446577, "loss": 0.951, "step": 1190 }, { "epoch": 0.27, "learning_rate": 0.0001991639830944714, "loss": 0.9648, "step": 1195 }, { "epoch": 0.28, "learning_rate": 0.00019913793742566647, "loss": 0.9617, "step": 1200 }, { "epoch": 0.28, "learning_rate": 0.00019911149399255224, "loss": 0.9618, "step": 1205 }, { "epoch": 0.28, "learning_rate": 0.00019908465290122585, "loss": 0.9366, "step": 1210 }, { "epoch": 0.28, "learning_rate": 0.00019905741425938002, "loss": 0.9376, "step": 1215 }, { "epoch": 0.28, "learning_rate": 0.00019902977817630243, "loss": 1.0024, "step": 1220 }, { "epoch": 0.28, "learning_rate": 0.0001990017447628754, "loss": 0.938, "step": 1225 }, { "epoch": 0.28, "learning_rate": 0.00019897331413157548, "loss": 0.9832, "step": 1230 }, { "epoch": 0.28, "learning_rate": 0.00019894448639647304, "loss": 0.9587, "step": 1235 }, { "epoch": 0.28, "learning_rate": 0.00019891526167323145, "loss": 0.9529, "step": 1240 }, { "epoch": 0.29, "learning_rate": 0.00019888564007910714, "loss": 0.9418, "step": 1245 }, { "epoch": 0.29, "learning_rate": 0.0001988556217329488, "loss": 0.9461, "step": 1250 }, { "epoch": 0.29, "learning_rate": 0.0001988252067551969, "loss": 1.0006, "step": 1255 }, { "epoch": 0.29, "learning_rate": 0.00019879439526788341, "loss": 0.9565, "step": 1260 }, { "epoch": 0.29, "learning_rate": 0.00019876318739463114, "loss": 0.9365, "step": 1265 }, { "epoch": 0.29, "learning_rate": 0.00019873158326065327, "loss": 0.9786, "step": 1270 }, { "epoch": 0.29, "learning_rate": 0.00019869958299275289, "loss": 1.0009, "step": 1275 }, { "epoch": 0.29, "learning_rate": 0.00019866718671932249, "loss": 0.9631, "step": 1280 }, { "epoch": 0.29, "learning_rate": 0.00019863439457034333, "loss": 0.9444, "step": 1285 }, { "epoch": 0.3, "learning_rate": 0.00019860120667738516, "loss": 0.955, "step": 1290 }, { "epoch": 0.3, "learning_rate": 0.00019856762317360546, "loss": 0.9722, "step": 1295 }, { "epoch": 0.3, "learning_rate": 0.00019853364419374902, "loss": 0.9711, "step": 1300 }, { "epoch": 0.3, "learning_rate": 0.00019849926987414732, "loss": 0.9807, "step": 1305 }, { "epoch": 0.3, "learning_rate": 0.00019846450035271808, "loss": 0.9659, "step": 1310 }, { "epoch": 0.3, "learning_rate": 0.0001984293357689647, "loss": 0.9607, "step": 1315 }, { "epoch": 0.3, "learning_rate": 0.00019839377626397554, "loss": 1.013, "step": 1320 }, { "epoch": 0.3, "learning_rate": 0.00019835782198042364, "loss": 0.9871, "step": 1325 }, { "epoch": 0.31, "learning_rate": 0.00019832147306256576, "loss": 0.9709, "step": 1330 }, { "epoch": 0.31, "learning_rate": 0.0001982847296562423, "loss": 0.9503, "step": 1335 }, { "epoch": 0.31, "learning_rate": 0.00019824759190887622, "loss": 0.9515, "step": 1340 }, { "epoch": 0.31, "learning_rate": 0.00019821005996947273, "loss": 0.9858, "step": 1345 }, { "epoch": 0.31, "learning_rate": 0.00019817213398861866, "loss": 0.9564, "step": 1350 }, { "epoch": 0.31, "learning_rate": 0.00019813381411848185, "loss": 0.9755, "step": 1355 }, { "epoch": 0.31, "learning_rate": 0.0001980951005128104, "loss": 0.975, "step": 1360 }, { "epoch": 0.31, "learning_rate": 0.0001980559933269323, "loss": 0.9844, "step": 1365 }, { "epoch": 0.31, "learning_rate": 0.00019801649271775459, "loss": 0.9649, "step": 1370 }, { "epoch": 0.32, "learning_rate": 0.00019797659884376282, "loss": 0.9912, "step": 1375 }, { "epoch": 0.32, "learning_rate": 0.00019793631186502047, "loss": 0.9805, "step": 1380 }, { "epoch": 0.32, "learning_rate": 0.00019789563194316817, "loss": 0.9657, "step": 1385 }, { "epoch": 0.32, "learning_rate": 0.00019785455924142318, "loss": 0.9654, "step": 1390 }, { "epoch": 0.32, "learning_rate": 0.00019781309392457867, "loss": 0.9533, "step": 1395 }, { "epoch": 0.32, "learning_rate": 0.0001977712361590031, "loss": 0.9811, "step": 1400 }, { "epoch": 0.32, "learning_rate": 0.00019772898611263942, "loss": 0.9532, "step": 1405 }, { "epoch": 0.32, "learning_rate": 0.00019768634395500465, "loss": 0.9574, "step": 1410 }, { "epoch": 0.32, "learning_rate": 0.00019764330985718896, "loss": 0.9648, "step": 1415 }, { "epoch": 0.33, "learning_rate": 0.00019759988399185505, "loss": 1.0009, "step": 1420 }, { "epoch": 0.33, "learning_rate": 0.0001975560665332376, "loss": 0.981, "step": 1425 }, { "epoch": 0.33, "learning_rate": 0.00019751185765714234, "loss": 0.9837, "step": 1430 }, { "epoch": 0.33, "learning_rate": 0.00019746725754094544, "loss": 0.9704, "step": 1435 }, { "epoch": 0.33, "learning_rate": 0.00019742226636359296, "loss": 0.9908, "step": 1440 }, { "epoch": 0.33, "learning_rate": 0.00019737688430559986, "loss": 0.9459, "step": 1445 }, { "epoch": 0.33, "learning_rate": 0.00019733111154904943, "loss": 0.925, "step": 1450 }, { "epoch": 0.33, "learning_rate": 0.00019728494827759256, "loss": 0.9632, "step": 1455 }, { "epoch": 0.34, "learning_rate": 0.00019723839467644699, "loss": 0.9579, "step": 1460 }, { "epoch": 0.34, "learning_rate": 0.00019719145093239645, "loss": 0.9994, "step": 1465 }, { "epoch": 0.34, "learning_rate": 0.00019714411723379015, "loss": 0.9736, "step": 1470 }, { "epoch": 0.34, "learning_rate": 0.00019709639377054177, "loss": 0.9684, "step": 1475 }, { "epoch": 0.34, "learning_rate": 0.0001970482807341289, "loss": 0.9459, "step": 1480 }, { "epoch": 0.34, "learning_rate": 0.00019699977831759218, "loss": 0.9807, "step": 1485 }, { "epoch": 0.34, "learning_rate": 0.0001969508867155345, "loss": 0.9577, "step": 1490 }, { "epoch": 0.34, "learning_rate": 0.00019690160612412022, "loss": 0.979, "step": 1495 }, { "epoch": 0.34, "learning_rate": 0.00019685193674107452, "loss": 0.9722, "step": 1500 }, { "epoch": 0.35, "learning_rate": 0.0001968018787656824, "loss": 0.9831, "step": 1505 }, { "epoch": 0.35, "learning_rate": 0.00019675143239878805, "loss": 0.9767, "step": 1510 }, { "epoch": 0.35, "learning_rate": 0.00019670059784279394, "loss": 0.9814, "step": 1515 }, { "epoch": 0.35, "learning_rate": 0.00019664937530166002, "loss": 1.0078, "step": 1520 }, { "epoch": 0.35, "learning_rate": 0.000196597764980903, "loss": 1.2574, "step": 1525 }, { "epoch": 0.35, "learning_rate": 0.00019654576708759538, "loss": 1.0538, "step": 1530 }, { "epoch": 0.35, "learning_rate": 0.00019649338183036475, "loss": 1.0472, "step": 1535 }, { "epoch": 0.35, "learning_rate": 0.00019644060941939286, "loss": 5.8167, "step": 1540 }, { "epoch": 0.35, "learning_rate": 0.0001963874500664148, "loss": 8.9193, "step": 1545 }, { "epoch": 0.36, "learning_rate": 0.00019633390398471817, "loss": 9.1751, "step": 1550 }, { "epoch": 0.36, "learning_rate": 0.00019627997138914224, "loss": 7.9315, "step": 1555 }, { "epoch": 0.36, "learning_rate": 0.00019622565249607704, "loss": 7.6649, "step": 1560 }, { "epoch": 0.36, "learning_rate": 0.00019617094752346254, "loss": 7.5712, "step": 1565 }, { "epoch": 0.36, "learning_rate": 0.0001961158566907877, "loss": 7.5273, "step": 1570 }, { "epoch": 0.36, "learning_rate": 0.00019606038021908963, "loss": 7.5129, "step": 1575 }, { "epoch": 0.36, "learning_rate": 0.00019600451833095287, "loss": 7.4763, "step": 1580 }, { "epoch": 0.36, "learning_rate": 0.00019594827125050807, "loss": 7.4253, "step": 1585 }, { "epoch": 0.36, "learning_rate": 0.00019589163920343163, "loss": 7.3907, "step": 1590 }, { "epoch": 0.37, "learning_rate": 0.00019583462241694436, "loss": 7.3799, "step": 1595 }, { "epoch": 0.37, "learning_rate": 0.00019577722111981078, "loss": 7.349, "step": 1600 }, { "epoch": 0.37, "learning_rate": 0.00019571943554233819, "loss": 7.3447, "step": 1605 }, { "epoch": 0.37, "learning_rate": 0.00019566126591637558, "loss": 7.2859, "step": 1610 }, { "epoch": 0.37, "learning_rate": 0.00019560271247531308, "loss": 7.309, "step": 1615 }, { "epoch": 0.37, "learning_rate": 0.0001955437754540805, "loss": 7.3254, "step": 1620 }, { "epoch": 0.37, "learning_rate": 0.00019548445508914681, "loss": 7.2993, "step": 1625 }, { "epoch": 0.37, "learning_rate": 0.00019542475161851906, "loss": 7.3311, "step": 1630 }, { "epoch": 0.38, "learning_rate": 0.00019536466528174133, "loss": 7.3128, "step": 1635 }, { "epoch": 0.38, "learning_rate": 0.00019530419631989392, "loss": 7.316, "step": 1640 }, { "epoch": 0.38, "learning_rate": 0.00019524334497559226, "loss": 7.2758, "step": 1645 }, { "epoch": 0.38, "learning_rate": 0.00019518211149298595, "loss": 7.3082, "step": 1650 }, { "epoch": 0.38, "learning_rate": 0.00019512049611775795, "loss": 7.2448, "step": 1655 }, { "epoch": 0.38, "learning_rate": 0.00019505849909712332, "loss": 7.2816, "step": 1660 }, { "epoch": 0.38, "learning_rate": 0.00019499612067982848, "loss": 7.2852, "step": 1665 }, { "epoch": 0.38, "learning_rate": 0.00019493336111615003, "loss": 7.2445, "step": 1670 }, { "epoch": 0.38, "learning_rate": 0.00019487022065789385, "loss": 7.2607, "step": 1675 }, { "epoch": 0.39, "learning_rate": 0.00019480669955839402, "loss": 7.2453, "step": 1680 }, { "epoch": 0.39, "learning_rate": 0.00019474279807251186, "loss": 7.2839, "step": 1685 }, { "epoch": 0.39, "learning_rate": 0.00019467851645663494, "loss": 7.2569, "step": 1690 }, { "epoch": 0.39, "learning_rate": 0.00019461385496867594, "loss": 7.2633, "step": 1695 }, { "epoch": 0.39, "learning_rate": 0.00019454881386807163, "loss": 7.2762, "step": 1700 }, { "epoch": 0.39, "learning_rate": 0.000194483393415782, "loss": 7.2273, "step": 1705 }, { "epoch": 0.39, "learning_rate": 0.00019441759387428903, "loss": 7.289, "step": 1710 }, { "epoch": 0.39, "learning_rate": 0.0001943514155075956, "loss": 7.2969, "step": 1715 }, { "epoch": 0.39, "learning_rate": 0.00019428485858122472, "loss": 7.2368, "step": 1720 }, { "epoch": 0.4, "learning_rate": 0.00019421792336221807, "loss": 7.2478, "step": 1725 }, { "epoch": 0.4, "learning_rate": 0.00019415061011913523, "loss": 7.2593, "step": 1730 }, { "epoch": 0.4, "learning_rate": 0.00019408291912205246, "loss": 7.2571, "step": 1735 }, { "epoch": 0.4, "learning_rate": 0.00019401485064256176, "loss": 7.2432, "step": 1740 }, { "epoch": 0.4, "learning_rate": 0.00019394640495376954, "loss": 7.2514, "step": 1745 }, { "epoch": 0.4, "learning_rate": 0.0001938775823302957, "loss": 7.2465, "step": 1750 }, { "epoch": 0.4, "learning_rate": 0.00019380838304827256, "loss": 7.2325, "step": 1755 }, { "epoch": 0.4, "learning_rate": 0.00019373880738534358, "loss": 7.2787, "step": 1760 }, { "epoch": 0.41, "learning_rate": 0.0001936688556206624, "loss": 7.2248, "step": 1765 }, { "epoch": 0.41, "learning_rate": 0.00019359852803489168, "loss": 7.2439, "step": 1770 }, { "epoch": 0.41, "learning_rate": 0.00019352782491020195, "loss": 7.2413, "step": 1775 }, { "epoch": 0.41, "learning_rate": 0.0001934567465302704, "loss": 7.2431, "step": 1780 }, { "epoch": 0.41, "learning_rate": 0.00019338529318028, "loss": 7.2493, "step": 1785 }, { "epoch": 0.41, "learning_rate": 0.00019331346514691813, "loss": 7.2592, "step": 1790 }, { "epoch": 0.41, "learning_rate": 0.00019324126271837538, "loss": 7.2195, "step": 1795 }, { "epoch": 0.41, "learning_rate": 0.00019316868618434455, "loss": 7.2303, "step": 1800 }, { "epoch": 0.41, "learning_rate": 0.00019309573583601954, "loss": 7.2496, "step": 1805 }, { "epoch": 0.42, "learning_rate": 0.00019302241196609397, "loss": 7.2521, "step": 1810 }, { "epoch": 0.42, "learning_rate": 0.00019294871486876013, "loss": 7.246, "step": 1815 }, { "epoch": 0.42, "learning_rate": 0.0001928746448397078, "loss": 7.2781, "step": 1820 }, { "epoch": 0.42, "learning_rate": 0.0001928002021761231, "loss": 7.2274, "step": 1825 }, { "epoch": 0.42, "learning_rate": 0.00019272538717668715, "loss": 7.2651, "step": 1830 }, { "epoch": 0.42, "learning_rate": 0.000192650200141575, "loss": 7.2463, "step": 1835 }, { "epoch": 0.42, "learning_rate": 0.00019257464137245446, "loss": 7.2502, "step": 1840 }, { "epoch": 0.42, "learning_rate": 0.0001924987111724847, "loss": 7.2209, "step": 1845 }, { "epoch": 0.42, "learning_rate": 0.0001924224098463153, "loss": 7.2259, "step": 1850 }, { "epoch": 0.43, "learning_rate": 0.00019234573770008474, "loss": 7.2525, "step": 1855 }, { "epoch": 0.43, "learning_rate": 0.00019226869504141943, "loss": 7.2342, "step": 1860 }, { "epoch": 0.43, "learning_rate": 0.00019219128217943233, "loss": 7.2226, "step": 1865 }, { "epoch": 0.43, "learning_rate": 0.00019211349942472165, "loss": 7.2734, "step": 1870 }, { "epoch": 0.43, "learning_rate": 0.00019203534708936985, "loss": 7.237, "step": 1875 }, { "epoch": 0.43, "learning_rate": 0.00019195682548694208, "loss": 7.2262, "step": 1880 }, { "epoch": 0.43, "learning_rate": 0.00019187793493248524, "loss": 7.2182, "step": 1885 }, { "epoch": 0.43, "learning_rate": 0.00019179867574252638, "loss": 7.2421, "step": 1890 }, { "epoch": 0.43, "learning_rate": 0.00019171904823507178, "loss": 7.2366, "step": 1895 }, { "epoch": 0.44, "learning_rate": 0.00019163905272960528, "loss": 7.221, "step": 1900 }, { "epoch": 0.44, "learning_rate": 0.0001915586895470874, "loss": 7.2467, "step": 1905 }, { "epoch": 0.44, "learning_rate": 0.0001914779590099538, "loss": 7.2481, "step": 1910 }, { "epoch": 0.44, "learning_rate": 0.00019139686144211405, "loss": 7.2351, "step": 1915 }, { "epoch": 0.44, "learning_rate": 0.00019131539716895024, "loss": 7.246, "step": 1920 }, { "epoch": 0.44, "learning_rate": 0.00019123356651731593, "loss": 7.2139, "step": 1925 }, { "epoch": 0.44, "learning_rate": 0.00019115136981553464, "loss": 7.2491, "step": 1930 }, { "epoch": 0.44, "learning_rate": 0.00019106880739339844, "loss": 7.334, "step": 1935 }, { "epoch": 0.45, "learning_rate": 0.00019098587958216688, "loss": 7.2713, "step": 1940 }, { "epoch": 0.45, "learning_rate": 0.0001909025867145655, "loss": 7.2806, "step": 1945 }, { "epoch": 0.45, "learning_rate": 0.00019081892912478456, "loss": 7.2481, "step": 1950 }, { "epoch": 0.45, "learning_rate": 0.00019073490714847759, "loss": 7.2035, "step": 1955 }, { "epoch": 0.45, "learning_rate": 0.00019065052112276018, "loss": 7.2353, "step": 1960 }, { "epoch": 0.45, "learning_rate": 0.00019056577138620856, "loss": 7.2377, "step": 1965 }, { "epoch": 0.45, "learning_rate": 0.00019048065827885827, "loss": 7.2368, "step": 1970 }, { "epoch": 0.45, "learning_rate": 0.00019039518214220276, "loss": 7.2461, "step": 1975 }, { "epoch": 0.45, "learning_rate": 0.000190309343319192, "loss": 7.2415, "step": 1980 }, { "epoch": 0.46, "learning_rate": 0.00019022314215423123, "loss": 7.2384, "step": 1985 }, { "epoch": 0.46, "learning_rate": 0.00019013657899317942, "loss": 7.2566, "step": 1990 }, { "epoch": 0.46, "learning_rate": 0.00019004965418334798, "loss": 7.2231, "step": 1995 }, { "epoch": 0.46, "learning_rate": 0.0001899623680734993, "loss": 7.2459, "step": 2000 }, { "epoch": 0.46, "learning_rate": 0.00018987472101384545, "loss": 7.214, "step": 2005 }, { "epoch": 0.46, "learning_rate": 0.00018978671335604665, "loss": 7.221, "step": 2010 }, { "epoch": 0.46, "learning_rate": 0.00018969834545321, "loss": 7.2021, "step": 2015 }, { "epoch": 0.46, "learning_rate": 0.00018960961765988792, "loss": 7.2066, "step": 2020 }, { "epoch": 0.46, "learning_rate": 0.00018952053033207683, "loss": 7.3269, "step": 2025 }, { "epoch": 0.47, "learning_rate": 0.00018943108382721562, "loss": 7.3757, "step": 2030 }, { "epoch": 0.47, "learning_rate": 0.0001893412785041844, "loss": 7.2544, "step": 2035 }, { "epoch": 0.47, "learning_rate": 0.00018925111472330283, "loss": 7.2194, "step": 2040 }, { "epoch": 0.47, "learning_rate": 0.00018916059284632884, "loss": 7.2536, "step": 2045 }, { "epoch": 0.47, "learning_rate": 0.00018906971323645713, "loss": 7.2394, "step": 2050 }, { "epoch": 0.47, "learning_rate": 0.0001889784762583177, "loss": 7.2531, "step": 2055 }, { "epoch": 0.47, "learning_rate": 0.00018888688227797432, "loss": 7.2228, "step": 2060 }, { "epoch": 0.47, "learning_rate": 0.0001887949316629233, "loss": 7.2619, "step": 2065 }, { "epoch": 0.48, "learning_rate": 0.00018870262478209163, "loss": 7.2391, "step": 2070 }, { "epoch": 0.48, "learning_rate": 0.00018860996200583595, "loss": 7.3725, "step": 2075 }, { "epoch": 0.48, "learning_rate": 0.00018851694370594069, "loss": 7.3694, "step": 2080 }, { "epoch": 0.48, "learning_rate": 0.00018842357025561674, "loss": 7.4133, "step": 2085 }, { "epoch": 0.48, "learning_rate": 0.00018832984202949996, "loss": 7.3363, "step": 2090 }, { "epoch": 0.48, "learning_rate": 0.0001882357594036497, "loss": 7.2878, "step": 2095 }, { "epoch": 0.48, "learning_rate": 0.00018814132275554713, "loss": 7.2645, "step": 2100 }, { "epoch": 0.48, "learning_rate": 0.0001880465324640939, "loss": 7.2928, "step": 2105 }, { "epoch": 0.48, "learning_rate": 0.0001879513889096106, "loss": 7.289, "step": 2110 }, { "epoch": 0.49, "learning_rate": 0.00018785589247383513, "loss": 7.2577, "step": 2115 }, { "epoch": 0.49, "learning_rate": 0.00018776004353992124, "loss": 7.2575, "step": 2120 }, { "epoch": 0.49, "learning_rate": 0.00018766384249243703, "loss": 7.2455, "step": 2125 }, { "epoch": 0.49, "learning_rate": 0.00018756728971736327, "loss": 7.1998, "step": 2130 }, { "epoch": 0.49, "learning_rate": 0.00018747038560209208, "loss": 7.1431, "step": 2135 }, { "epoch": 0.49, "learning_rate": 0.00018737313053542512, "loss": 7.1091, "step": 2140 }, { "epoch": 0.49, "learning_rate": 0.00018727552490757217, "loss": 7.1545, "step": 2145 }, { "epoch": 0.49, "learning_rate": 0.0001871775691101496, "loss": 7.1363, "step": 2150 }, { "epoch": 0.49, "learning_rate": 0.00018707926353617865, "loss": 7.0623, "step": 2155 }, { "epoch": 0.5, "learning_rate": 0.00018698060858008403, "loss": 7.0172, "step": 2160 }, { "epoch": 0.5, "learning_rate": 0.00018688160463769222, "loss": 7.0213, "step": 2165 }, { "epoch": 0.5, "learning_rate": 0.00018678225210622986, "loss": 6.9647, "step": 2170 }, { "epoch": 0.5, "learning_rate": 0.00018668255138432237, "loss": 6.9448, "step": 2175 }, { "epoch": 0.5, "learning_rate": 0.00018658250287199196, "loss": 6.9318, "step": 2180 }, { "epoch": 0.5, "learning_rate": 0.0001864821069706564, "loss": 6.9102, "step": 2185 }, { "epoch": 0.5, "learning_rate": 0.00018638136408312728, "loss": 6.9001, "step": 2190 }, { "epoch": 0.5, "learning_rate": 0.00018628027461360834, "loss": 6.8805, "step": 2195 }, { "epoch": 0.5, "learning_rate": 0.0001861788389676939, "loss": 6.8864, "step": 2200 }, { "epoch": 0.51, "learning_rate": 0.00018607705755236716, "loss": 6.8746, "step": 2205 }, { "epoch": 0.51, "learning_rate": 0.00018597493077599867, "loss": 6.8675, "step": 2210 }, { "epoch": 0.51, "learning_rate": 0.0001858724590483447, "loss": 6.8548, "step": 2215 }, { "epoch": 0.51, "learning_rate": 0.00018576964278054544, "loss": 6.8581, "step": 2220 }, { "epoch": 0.51, "learning_rate": 0.00018566648238512357, "loss": 6.8075, "step": 2225 }, { "epoch": 0.51, "learning_rate": 0.00018556297827598242, "loss": 6.8101, "step": 2230 }, { "epoch": 0.51, "learning_rate": 0.00018545913086840434, "loss": 6.8087, "step": 2235 }, { "epoch": 0.51, "learning_rate": 0.00018535494057904915, "loss": 6.778, "step": 2240 }, { "epoch": 0.52, "learning_rate": 0.00018525040782595235, "loss": 6.7706, "step": 2245 }, { "epoch": 0.52, "learning_rate": 0.00018514553302852356, "loss": 6.723, "step": 2250 }, { "epoch": 0.52, "learning_rate": 0.0001850403166075446, "loss": 6.7416, "step": 2255 }, { "epoch": 0.52, "learning_rate": 0.00018493475898516813, "loss": 6.7556, "step": 2260 }, { "epoch": 0.52, "learning_rate": 0.0001848288605849157, "loss": 6.7387, "step": 2265 }, { "epoch": 0.52, "learning_rate": 0.00018472262183167614, "loss": 6.7505, "step": 2270 }, { "epoch": 0.52, "learning_rate": 0.00018461604315170389, "loss": 6.727, "step": 2275 }, { "epoch": 0.52, "learning_rate": 0.00018450912497261723, "loss": 6.6703, "step": 2280 }, { "epoch": 0.52, "learning_rate": 0.00018440186772339663, "loss": 6.7113, "step": 2285 }, { "epoch": 0.53, "learning_rate": 0.00018429427183438288, "loss": 6.6782, "step": 2290 }, { "epoch": 0.53, "learning_rate": 0.00018418633773727562, "loss": 6.6859, "step": 2295 }, { "epoch": 0.53, "learning_rate": 0.00018407806586513134, "loss": 6.6368, "step": 2300 }, { "epoch": 0.53, "learning_rate": 0.00018396945665236184, "loss": 6.6691, "step": 2305 }, { "epoch": 0.53, "learning_rate": 0.00018386051053473232, "loss": 6.6673, "step": 2310 }, { "epoch": 0.53, "learning_rate": 0.00018375122794935984, "loss": 6.641, "step": 2315 }, { "epoch": 0.53, "learning_rate": 0.00018364160933471134, "loss": 6.6094, "step": 2320 }, { "epoch": 0.53, "learning_rate": 0.00018353165513060202, "loss": 6.5716, "step": 2325 }, { "epoch": 0.53, "learning_rate": 0.0001834213657781936, "loss": 6.6307, "step": 2330 }, { "epoch": 0.54, "learning_rate": 0.00018331074171999237, "loss": 6.5556, "step": 2335 }, { "epoch": 0.54, "learning_rate": 0.00018319978339984767, "loss": 6.5416, "step": 2340 }, { "epoch": 0.54, "learning_rate": 0.0001830884912629499, "loss": 6.5371, "step": 2345 }, { "epoch": 0.54, "learning_rate": 0.0001829768657558288, "loss": 6.52, "step": 2350 }, { "epoch": 0.54, "learning_rate": 0.00018286490732635172, "loss": 6.5359, "step": 2355 }, { "epoch": 0.54, "learning_rate": 0.00018275261642372175, "loss": 6.5164, "step": 2360 }, { "epoch": 0.54, "learning_rate": 0.0001826399934984759, "loss": 6.5095, "step": 2365 }, { "epoch": 0.54, "learning_rate": 0.0001825270390024834, "loss": 6.5213, "step": 2370 }, { "epoch": 0.55, "learning_rate": 0.0001824137533889438, "loss": 6.4413, "step": 2375 }, { "epoch": 0.55, "learning_rate": 0.00018230013711238513, "loss": 6.4896, "step": 2380 }, { "epoch": 0.55, "learning_rate": 0.00018218619062866211, "loss": 6.4467, "step": 2385 }, { "epoch": 0.55, "learning_rate": 0.00018207191439495438, "loss": 6.4661, "step": 2390 }, { "epoch": 0.55, "learning_rate": 0.00018195730886976462, "loss": 6.4062, "step": 2395 }, { "epoch": 0.55, "learning_rate": 0.00018184237451291665, "loss": 6.4019, "step": 2400 }, { "epoch": 0.55, "learning_rate": 0.00018172711178555368, "loss": 6.4315, "step": 2405 }, { "epoch": 0.55, "learning_rate": 0.00018161152115013637, "loss": 6.4537, "step": 2410 }, { "epoch": 0.55, "learning_rate": 0.0001814956030704411, "loss": 6.4277, "step": 2415 }, { "epoch": 0.56, "learning_rate": 0.00018137935801155794, "loss": 6.4079, "step": 2420 }, { "epoch": 0.56, "learning_rate": 0.00018126278643988895, "loss": 6.3882, "step": 2425 }, { "epoch": 0.56, "learning_rate": 0.0001811458888231462, "loss": 6.3683, "step": 2430 }, { "epoch": 0.56, "learning_rate": 0.00018102866563034995, "loss": 6.3062, "step": 2435 }, { "epoch": 0.56, "learning_rate": 0.0001809111173318267, "loss": 6.324, "step": 2440 }, { "epoch": 0.56, "learning_rate": 0.00018079324439920742, "loss": 6.2728, "step": 2445 }, { "epoch": 0.56, "learning_rate": 0.00018067504730542551, "loss": 6.3452, "step": 2450 }, { "epoch": 0.56, "learning_rate": 0.00018055652652471506, "loss": 6.3298, "step": 2455 }, { "epoch": 0.56, "learning_rate": 0.0001804376825326088, "loss": 6.288, "step": 2460 }, { "epoch": 0.57, "learning_rate": 0.00018031851580593627, "loss": 6.2508, "step": 2465 }, { "epoch": 0.57, "learning_rate": 0.00018019902682282193, "loss": 6.2802, "step": 2470 }, { "epoch": 0.57, "learning_rate": 0.0001800792160626832, "loss": 6.2777, "step": 2475 }, { "epoch": 0.57, "learning_rate": 0.0001799590840062285, "loss": 6.2097, "step": 2480 }, { "epoch": 0.57, "learning_rate": 0.0001798386311354554, "loss": 6.2434, "step": 2485 }, { "epoch": 0.57, "learning_rate": 0.00017971785793364866, "loss": 6.2158, "step": 2490 }, { "epoch": 0.57, "learning_rate": 0.00017959676488537825, "loss": 6.1986, "step": 2495 }, { "epoch": 0.57, "learning_rate": 0.0001794753524764975, "loss": 6.1748, "step": 2500 }, { "epoch": 0.57, "learning_rate": 0.00017935362119414098, "loss": 6.1924, "step": 2505 }, { "epoch": 0.58, "learning_rate": 0.00017923157152672278, "loss": 6.3381, "step": 2510 }, { "epoch": 0.58, "learning_rate": 0.0001791092039639343, "loss": 6.255, "step": 2515 }, { "epoch": 0.58, "learning_rate": 0.00017898651899674254, "loss": 6.2656, "step": 2520 }, { "epoch": 0.58, "learning_rate": 0.0001788635171173879, "loss": 6.2003, "step": 2525 }, { "epoch": 0.58, "learning_rate": 0.00017874019881938233, "loss": 6.1718, "step": 2530 }, { "epoch": 0.58, "learning_rate": 0.00017861656459750735, "loss": 6.1582, "step": 2535 }, { "epoch": 0.58, "learning_rate": 0.000178492614947812, "loss": 6.2319, "step": 2540 }, { "epoch": 0.58, "learning_rate": 0.00017836835036761094, "loss": 6.4222, "step": 2545 }, { "epoch": 0.59, "learning_rate": 0.00017824377135548236, "loss": 6.2872, "step": 2550 }, { "epoch": 0.59, "learning_rate": 0.000178118878411266, "loss": 6.2422, "step": 2555 }, { "epoch": 0.59, "learning_rate": 0.00017799367203606128, "loss": 6.233, "step": 2560 }, { "epoch": 0.59, "learning_rate": 0.0001778681527322251, "loss": 6.1516, "step": 2565 }, { "epoch": 0.59, "learning_rate": 0.00017774232100336982, "loss": 6.1454, "step": 2570 }, { "epoch": 0.59, "learning_rate": 0.00017761617735436152, "loss": 6.1147, "step": 2575 }, { "epoch": 0.59, "learning_rate": 0.00017748972229131757, "loss": 6.179, "step": 2580 }, { "epoch": 0.59, "learning_rate": 0.00017736295632160494, "loss": 6.1024, "step": 2585 }, { "epoch": 0.59, "learning_rate": 0.000177235879953838, "loss": 6.1314, "step": 2590 }, { "epoch": 0.6, "learning_rate": 0.00017710849369787646, "loss": 6.1308, "step": 2595 }, { "epoch": 0.6, "learning_rate": 0.00017698079806482343, "loss": 6.0657, "step": 2600 }, { "epoch": 0.6, "learning_rate": 0.0001768527935670233, "loss": 6.1253, "step": 2605 }, { "epoch": 0.6, "learning_rate": 0.0001767244807180597, "loss": 6.0759, "step": 2610 }, { "epoch": 0.6, "learning_rate": 0.0001765958600327534, "loss": 6.078, "step": 2615 }, { "epoch": 0.6, "learning_rate": 0.00017646693202716033, "loss": 6.0314, "step": 2620 }, { "epoch": 0.6, "learning_rate": 0.00017633769721856938, "loss": 6.0668, "step": 2625 }, { "epoch": 0.6, "learning_rate": 0.0001762081561255005, "loss": 6.0417, "step": 2630 }, { "epoch": 0.6, "learning_rate": 0.0001760783092677025, "loss": 6.0718, "step": 2635 }, { "epoch": 0.61, "learning_rate": 0.00017594815716615093, "loss": 6.0442, "step": 2640 }, { "epoch": 0.61, "learning_rate": 0.00017581770034304613, "loss": 6.06, "step": 2645 }, { "epoch": 0.61, "learning_rate": 0.000175686939321811, "loss": 6.0552, "step": 2650 }, { "epoch": 0.61, "learning_rate": 0.00017555587462708894, "loss": 6.0108, "step": 2655 }, { "epoch": 0.61, "learning_rate": 0.00017542450678474184, "loss": 6.0079, "step": 2660 }, { "epoch": 0.61, "learning_rate": 0.00017529283632184784, "loss": 5.9592, "step": 2665 }, { "epoch": 0.61, "learning_rate": 0.00017516086376669917, "loss": 5.9704, "step": 2670 }, { "epoch": 0.61, "learning_rate": 0.00017502858964880028, "loss": 5.9467, "step": 2675 }, { "epoch": 0.62, "learning_rate": 0.00017489601449886547, "loss": 5.992, "step": 2680 }, { "epoch": 0.62, "learning_rate": 0.0001747631388488169, "loss": 6.0237, "step": 2685 }, { "epoch": 0.62, "learning_rate": 0.00017462996323178235, "loss": 5.9693, "step": 2690 }, { "epoch": 0.62, "learning_rate": 0.00017449648818209314, "loss": 5.9799, "step": 2695 }, { "epoch": 0.62, "learning_rate": 0.00017436271423528206, "loss": 5.9773, "step": 2700 }, { "epoch": 0.62, "learning_rate": 0.00017422864192808107, "loss": 5.891, "step": 2705 }, { "epoch": 0.62, "learning_rate": 0.0001740942717984192, "loss": 5.9719, "step": 2710 }, { "epoch": 0.62, "learning_rate": 0.00017395960438542057, "loss": 5.9558, "step": 2715 }, { "epoch": 0.62, "learning_rate": 0.00017382464022940182, "loss": 5.9125, "step": 2720 }, { "epoch": 0.63, "learning_rate": 0.0001736893798718704, "loss": 5.9074, "step": 2725 }, { "epoch": 0.63, "learning_rate": 0.00017355382385552206, "loss": 5.9078, "step": 2730 }, { "epoch": 0.63, "learning_rate": 0.00017341797272423895, "loss": 5.8941, "step": 2735 }, { "epoch": 0.63, "learning_rate": 0.0001732818270230871, "loss": 5.8736, "step": 2740 }, { "epoch": 0.63, "learning_rate": 0.00017314538729831455, "loss": 5.9292, "step": 2745 }, { "epoch": 0.63, "learning_rate": 0.000173008654097349, "loss": 5.898, "step": 2750 }, { "epoch": 0.63, "learning_rate": 0.00017287162796879568, "loss": 5.8592, "step": 2755 }, { "epoch": 0.63, "learning_rate": 0.000172734309462435, "loss": 5.9182, "step": 2760 }, { "epoch": 0.63, "learning_rate": 0.00017259669912922062, "loss": 5.8856, "step": 2765 }, { "epoch": 0.64, "learning_rate": 0.00017245879752127692, "loss": 5.8911, "step": 2770 }, { "epoch": 0.64, "learning_rate": 0.00017232060519189703, "loss": 5.8303, "step": 2775 }, { "epoch": 0.64, "learning_rate": 0.0001721821226955405, "loss": 5.8551, "step": 2780 }, { "epoch": 0.64, "learning_rate": 0.00017204335058783112, "loss": 5.8347, "step": 2785 }, { "epoch": 0.64, "learning_rate": 0.00017190428942555463, "loss": 5.8918, "step": 2790 }, { "epoch": 0.64, "learning_rate": 0.00017176493976665656, "loss": 5.8401, "step": 2795 }, { "epoch": 0.64, "learning_rate": 0.0001716253021702399, "loss": 5.8697, "step": 2800 }, { "epoch": 0.64, "learning_rate": 0.000171485377196563, "loss": 5.807, "step": 2805 }, { "epoch": 0.64, "learning_rate": 0.0001713451654070371, "loss": 5.8286, "step": 2810 }, { "epoch": 0.65, "learning_rate": 0.00017120466736422437, "loss": 5.8396, "step": 2815 }, { "epoch": 0.65, "learning_rate": 0.0001710638836318354, "loss": 5.7998, "step": 2820 }, { "epoch": 0.65, "learning_rate": 0.00017092281477472704, "loss": 5.8392, "step": 2825 }, { "epoch": 0.65, "learning_rate": 0.00017078146135890014, "loss": 5.7698, "step": 2830 }, { "epoch": 0.65, "learning_rate": 0.0001706398239514973, "loss": 5.7643, "step": 2835 }, { "epoch": 0.65, "learning_rate": 0.0001704979031208004, "loss": 5.8052, "step": 2840 }, { "epoch": 0.65, "learning_rate": 0.00017035569943622876, "loss": 5.7825, "step": 2845 }, { "epoch": 0.65, "learning_rate": 0.0001702132134683363, "loss": 5.7984, "step": 2850 }, { "epoch": 0.66, "learning_rate": 0.00017007044578880968, "loss": 5.7616, "step": 2855 }, { "epoch": 0.66, "learning_rate": 0.00016992739697046586, "loss": 5.7678, "step": 2860 }, { "epoch": 0.66, "learning_rate": 0.00016978406758724968, "loss": 5.7437, "step": 2865 }, { "epoch": 0.66, "learning_rate": 0.00016964045821423178, "loss": 5.7978, "step": 2870 }, { "epoch": 0.66, "learning_rate": 0.0001694965694276061, "loss": 5.7463, "step": 2875 }, { "epoch": 0.66, "learning_rate": 0.00016935240180468775, "loss": 5.7477, "step": 2880 }, { "epoch": 0.66, "learning_rate": 0.0001692079559239105, "loss": 5.7617, "step": 2885 }, { "epoch": 0.66, "learning_rate": 0.00016906323236482465, "loss": 5.7463, "step": 2890 }, { "epoch": 0.66, "learning_rate": 0.00016891823170809447, "loss": 5.7469, "step": 2895 }, { "epoch": 0.67, "learning_rate": 0.00016877295453549614, "loss": 5.7635, "step": 2900 }, { "epoch": 0.67, "learning_rate": 0.00016862740142991527, "loss": 5.6833, "step": 2905 }, { "epoch": 0.67, "learning_rate": 0.00016848157297534453, "loss": 5.721, "step": 2910 }, { "epoch": 0.67, "learning_rate": 0.00016833546975688135, "loss": 5.7013, "step": 2915 }, { "epoch": 0.67, "learning_rate": 0.0001681890923607256, "loss": 5.6713, "step": 2920 }, { "epoch": 0.67, "learning_rate": 0.00016804244137417725, "loss": 5.6403, "step": 2925 }, { "epoch": 0.67, "learning_rate": 0.00016789551738563384, "loss": 5.7303, "step": 2930 }, { "epoch": 0.67, "learning_rate": 0.00016774832098458847, "loss": 5.6954, "step": 2935 }, { "epoch": 0.67, "learning_rate": 0.00016760085276162708, "loss": 5.6597, "step": 2940 }, { "epoch": 0.68, "learning_rate": 0.00016745311330842622, "loss": 5.6494, "step": 2945 }, { "epoch": 0.68, "learning_rate": 0.00016730510321775075, "loss": 5.6584, "step": 2950 }, { "epoch": 0.68, "learning_rate": 0.00016715682308345131, "loss": 5.614, "step": 2955 }, { "epoch": 0.68, "learning_rate": 0.00016700827350046206, "loss": 5.6699, "step": 2960 }, { "epoch": 0.68, "learning_rate": 0.00016685945506479829, "loss": 5.702, "step": 2965 }, { "epoch": 0.68, "learning_rate": 0.00016671036837355386, "loss": 5.6822, "step": 2970 }, { "epoch": 0.68, "learning_rate": 0.00016656101402489913, "loss": 5.6072, "step": 2975 }, { "epoch": 0.68, "learning_rate": 0.00016641139261807818, "loss": 5.6516, "step": 2980 }, { "epoch": 0.69, "learning_rate": 0.00016626150475340667, "loss": 5.6156, "step": 2985 }, { "epoch": 0.69, "learning_rate": 0.00016611135103226937, "loss": 5.6864, "step": 2990 }, { "epoch": 0.69, "learning_rate": 0.00016596093205711765, "loss": 5.6564, "step": 2995 }, { "epoch": 0.69, "learning_rate": 0.00016581024843146725, "loss": 5.6612, "step": 3000 }, { "epoch": 0.69, "learning_rate": 0.0001656593007598956, "loss": 5.6673, "step": 3005 }, { "epoch": 0.69, "learning_rate": 0.00016550808964803978, "loss": 5.6044, "step": 3010 }, { "epoch": 0.69, "learning_rate": 0.00016535661570259359, "loss": 5.6465, "step": 3015 }, { "epoch": 0.69, "learning_rate": 0.00016520487953130552, "loss": 5.6889, "step": 3020 }, { "epoch": 0.69, "learning_rate": 0.00016505288174297616, "loss": 5.6237, "step": 3025 }, { "epoch": 0.7, "learning_rate": 0.00016490062294745571, "loss": 5.621, "step": 3030 }, { "epoch": 0.7, "learning_rate": 0.0001647481037556417, "loss": 5.592, "step": 3035 }, { "epoch": 0.7, "learning_rate": 0.00016459532477947634, "loss": 5.6435, "step": 3040 }, { "epoch": 0.7, "learning_rate": 0.00016444228663194417, "loss": 5.6503, "step": 3045 }, { "epoch": 0.7, "learning_rate": 0.00016428898992706955, "loss": 5.5757, "step": 3050 }, { "epoch": 0.7, "learning_rate": 0.0001641354352799144, "loss": 5.6234, "step": 3055 }, { "epoch": 0.7, "learning_rate": 0.00016398162330657533, "loss": 5.6008, "step": 3060 }, { "epoch": 0.7, "learning_rate": 0.00016382755462418153, "loss": 5.5648, "step": 3065 }, { "epoch": 0.7, "learning_rate": 0.0001636732298508922, "loss": 5.5678, "step": 3070 }, { "epoch": 0.71, "learning_rate": 0.00016351864960589393, "loss": 5.6054, "step": 3075 }, { "epoch": 0.71, "learning_rate": 0.0001633638145093984, "loss": 5.5597, "step": 3080 }, { "epoch": 0.71, "learning_rate": 0.00016320872518263982, "loss": 5.5446, "step": 3085 }, { "epoch": 0.71, "learning_rate": 0.00016305338224787235, "loss": 5.5611, "step": 3090 }, { "epoch": 0.71, "learning_rate": 0.00016289778632836774, "loss": 5.5962, "step": 3095 }, { "epoch": 0.71, "learning_rate": 0.0001627419380484128, "loss": 5.6262, "step": 3100 }, { "epoch": 0.71, "learning_rate": 0.00016258583803330681, "loss": 5.5137, "step": 3105 }, { "epoch": 0.71, "learning_rate": 0.00016242948690935912, "loss": 5.5613, "step": 3110 }, { "epoch": 0.71, "learning_rate": 0.0001622728853038865, "loss": 5.5821, "step": 3115 }, { "epoch": 0.72, "learning_rate": 0.00016211603384521083, "loss": 5.5496, "step": 3120 }, { "epoch": 0.72, "learning_rate": 0.0001619589331626564, "loss": 5.5583, "step": 3125 }, { "epoch": 0.72, "learning_rate": 0.00016180158388654742, "loss": 5.553, "step": 3130 }, { "epoch": 0.72, "learning_rate": 0.00016164398664820557, "loss": 5.5156, "step": 3135 }, { "epoch": 0.72, "learning_rate": 0.00016148614207994735, "loss": 5.5404, "step": 3140 }, { "epoch": 0.72, "learning_rate": 0.00016132805081508163, "loss": 5.5206, "step": 3145 }, { "epoch": 0.72, "learning_rate": 0.00016116971348790712, "loss": 5.5238, "step": 3150 }, { "epoch": 0.72, "learning_rate": 0.0001610111307337097, "loss": 5.4944, "step": 3155 }, { "epoch": 0.73, "learning_rate": 0.0001608523031887601, "loss": 5.5935, "step": 3160 }, { "epoch": 0.73, "learning_rate": 0.00016069323149031105, "loss": 5.5346, "step": 3165 }, { "epoch": 0.73, "learning_rate": 0.00016053391627659505, "loss": 5.4831, "step": 3170 }, { "epoch": 0.73, "learning_rate": 0.00016037435818682142, "loss": 5.5352, "step": 3175 }, { "epoch": 0.73, "learning_rate": 0.0001602145578611742, "loss": 5.5195, "step": 3180 }, { "epoch": 0.73, "learning_rate": 0.00016005451594080911, "loss": 5.5148, "step": 3185 }, { "epoch": 0.73, "learning_rate": 0.00015989423306785142, "loss": 5.4785, "step": 3190 }, { "epoch": 0.73, "learning_rate": 0.00015973370988539294, "loss": 5.4914, "step": 3195 }, { "epoch": 0.73, "learning_rate": 0.00015957294703748982, "loss": 5.4942, "step": 3200 }, { "epoch": 0.74, "learning_rate": 0.00015941194516915977, "loss": 5.4827, "step": 3205 }, { "epoch": 0.74, "learning_rate": 0.00015925070492637944, "loss": 5.4518, "step": 3210 }, { "epoch": 0.74, "learning_rate": 0.00015908922695608192, "loss": 5.4863, "step": 3215 }, { "epoch": 0.74, "learning_rate": 0.0001589275119061542, "loss": 5.4688, "step": 3220 }, { "epoch": 0.74, "learning_rate": 0.00015876556042543433, "loss": 5.4822, "step": 3225 }, { "epoch": 0.74, "learning_rate": 0.00015860337316370916, "loss": 5.4418, "step": 3230 }, { "epoch": 0.74, "learning_rate": 0.00015844095077171142, "loss": 5.4795, "step": 3235 }, { "epoch": 0.74, "learning_rate": 0.0001582782939011173, "loss": 5.4708, "step": 3240 }, { "epoch": 0.74, "learning_rate": 0.0001581154032045437, "loss": 5.4705, "step": 3245 }, { "epoch": 0.75, "learning_rate": 0.00015795227933554568, "loss": 5.4587, "step": 3250 }, { "epoch": 0.75, "learning_rate": 0.000157788922948614, "loss": 5.4739, "step": 3255 }, { "epoch": 0.75, "learning_rate": 0.00015762533469917216, "loss": 5.4528, "step": 3260 }, { "epoch": 0.75, "learning_rate": 0.00015746151524357403, "loss": 5.5438, "step": 3265 }, { "epoch": 0.75, "learning_rate": 0.00015729746523910113, "loss": 5.4371, "step": 3270 }, { "epoch": 0.75, "learning_rate": 0.00015713318534395996, "loss": 5.4538, "step": 3275 }, { "epoch": 0.75, "learning_rate": 0.00015696867621727942, "loss": 5.4474, "step": 3280 }, { "epoch": 0.75, "learning_rate": 0.00015680393851910811, "loss": 5.4713, "step": 3285 }, { "epoch": 0.76, "learning_rate": 0.00015663897291041175, "loss": 5.4192, "step": 3290 }, { "epoch": 0.76, "learning_rate": 0.0001564737800530705, "loss": 5.4642, "step": 3295 }, { "epoch": 0.76, "learning_rate": 0.00015630836060987624, "loss": 5.4001, "step": 3300 }, { "epoch": 0.76, "learning_rate": 0.00015614271524453, "loss": 5.4734, "step": 3305 }, { "epoch": 0.76, "learning_rate": 0.00015597684462163923, "loss": 5.4254, "step": 3310 }, { "epoch": 0.76, "learning_rate": 0.00015581074940671527, "loss": 5.4866, "step": 3315 }, { "epoch": 0.76, "learning_rate": 0.0001556444302661704, "loss": 5.4851, "step": 3320 }, { "epoch": 0.76, "learning_rate": 0.00015547788786731553, "loss": 5.4155, "step": 3325 }, { "epoch": 0.76, "learning_rate": 0.00015531112287835717, "loss": 5.4763, "step": 3330 }, { "epoch": 0.77, "learning_rate": 0.00015514413596839498, "loss": 5.3893, "step": 3335 }, { "epoch": 0.77, "learning_rate": 0.00015497692780741908, "loss": 5.4356, "step": 3340 }, { "epoch": 0.77, "learning_rate": 0.00015480949906630724, "loss": 5.4113, "step": 3345 }, { "epoch": 0.77, "learning_rate": 0.0001546418504168222, "loss": 5.429, "step": 3350 }, { "epoch": 0.77, "learning_rate": 0.0001544739825316091, "loss": 5.4215, "step": 3355 }, { "epoch": 0.77, "learning_rate": 0.00015430589608419264, "loss": 5.4567, "step": 3360 }, { "epoch": 0.77, "learning_rate": 0.00015413759174897455, "loss": 5.4459, "step": 3365 }, { "epoch": 0.77, "learning_rate": 0.00015396907020123068, "loss": 5.4098, "step": 3370 }, { "epoch": 0.77, "learning_rate": 0.00015380033211710842, "loss": 5.3947, "step": 3375 }, { "epoch": 0.78, "learning_rate": 0.00015363137817362392, "loss": 5.3972, "step": 3380 }, { "epoch": 0.78, "learning_rate": 0.00015346220904865945, "loss": 5.4151, "step": 3385 }, { "epoch": 0.78, "learning_rate": 0.00015329282542096064, "loss": 5.3657, "step": 3390 }, { "epoch": 0.78, "learning_rate": 0.0001531232279701337, "loss": 5.4261, "step": 3395 }, { "epoch": 0.78, "learning_rate": 0.00015295341737664285, "loss": 5.4168, "step": 3400 }, { "epoch": 0.78, "learning_rate": 0.00015278339432180736, "loss": 5.4326, "step": 3405 }, { "epoch": 0.78, "learning_rate": 0.000152613159487799, "loss": 5.3843, "step": 3410 }, { "epoch": 0.78, "learning_rate": 0.00015244271355763933, "loss": 5.3671, "step": 3415 }, { "epoch": 0.78, "learning_rate": 0.00015227205721519675, "loss": 5.3901, "step": 3420 }, { "epoch": 0.79, "learning_rate": 0.0001521011911451839, "loss": 5.321, "step": 3425 }, { "epoch": 0.79, "learning_rate": 0.00015193011603315503, "loss": 5.4188, "step": 3430 }, { "epoch": 0.79, "learning_rate": 0.00015175883256550291, "loss": 5.3682, "step": 3435 }, { "epoch": 0.79, "learning_rate": 0.00015158734142945644, "loss": 5.3324, "step": 3440 }, { "epoch": 0.79, "learning_rate": 0.00015141564331307768, "loss": 5.3596, "step": 3445 }, { "epoch": 0.79, "learning_rate": 0.0001512437389052591, "loss": 5.3984, "step": 3450 }, { "epoch": 0.79, "learning_rate": 0.00015107162889572092, "loss": 5.3599, "step": 3455 }, { "epoch": 0.79, "learning_rate": 0.0001508993139750083, "loss": 5.3632, "step": 3460 }, { "epoch": 0.8, "learning_rate": 0.00015072679483448852, "loss": 5.3456, "step": 3465 }, { "epoch": 0.8, "learning_rate": 0.0001505540721663481, "loss": 5.3457, "step": 3470 }, { "epoch": 0.8, "learning_rate": 0.00015038114666359042, "loss": 5.339, "step": 3475 }, { "epoch": 0.8, "learning_rate": 0.0001502080190200325, "loss": 5.3411, "step": 3480 }, { "epoch": 0.8, "learning_rate": 0.00015003468993030248, "loss": 5.3975, "step": 3485 }, { "epoch": 0.8, "learning_rate": 0.00014986116008983664, "loss": 5.3451, "step": 3490 }, { "epoch": 0.8, "learning_rate": 0.00014968743019487686, "loss": 5.3741, "step": 3495 }, { "epoch": 0.8, "learning_rate": 0.00014951350094246762, "loss": 5.3924, "step": 3500 }, { "epoch": 0.8, "learning_rate": 0.00014933937303045325, "loss": 5.3798, "step": 3505 }, { "epoch": 0.81, "learning_rate": 0.0001491650471574751, "loss": 5.3578, "step": 3510 }, { "epoch": 0.81, "learning_rate": 0.00014899052402296895, "loss": 5.3184, "step": 3515 }, { "epoch": 0.81, "learning_rate": 0.00014881580432716182, "loss": 5.3755, "step": 3520 }, { "epoch": 0.81, "learning_rate": 0.0001486408887710696, "loss": 5.3483, "step": 3525 }, { "epoch": 0.81, "learning_rate": 0.00014846577805649388, "loss": 5.4266, "step": 3530 }, { "epoch": 0.81, "learning_rate": 0.0001482904728860193, "loss": 5.3295, "step": 3535 }, { "epoch": 0.81, "learning_rate": 0.00014811497396301072, "loss": 5.3377, "step": 3540 }, { "epoch": 0.81, "learning_rate": 0.0001479392819916103, "loss": 5.3249, "step": 3545 }, { "epoch": 0.81, "learning_rate": 0.00014776339767673491, "loss": 5.319, "step": 3550 }, { "epoch": 0.82, "learning_rate": 0.00014758732172407302, "loss": 5.3046, "step": 3555 }, { "epoch": 0.82, "learning_rate": 0.000147411054840082, "loss": 5.2972, "step": 3560 }, { "epoch": 0.82, "learning_rate": 0.00014723459773198535, "loss": 5.3475, "step": 3565 }, { "epoch": 0.82, "learning_rate": 0.00014705795110776974, "loss": 5.3537, "step": 3570 }, { "epoch": 0.82, "learning_rate": 0.00014688111567618225, "loss": 5.3175, "step": 3575 }, { "epoch": 0.82, "learning_rate": 0.0001467040921467275, "loss": 5.337, "step": 3580 }, { "epoch": 0.82, "learning_rate": 0.00014652688122966475, "loss": 5.3152, "step": 3585 }, { "epoch": 0.82, "learning_rate": 0.00014634948363600518, "loss": 5.36, "step": 3590 }, { "epoch": 0.83, "learning_rate": 0.0001461719000775089, "loss": 5.3679, "step": 3595 }, { "epoch": 0.83, "learning_rate": 0.00014599413126668213, "loss": 5.3203, "step": 3600 }, { "epoch": 0.83, "learning_rate": 0.00014581617791677452, "loss": 5.3254, "step": 3605 }, { "epoch": 0.83, "learning_rate": 0.00014563804074177588, "loss": 5.3276, "step": 3610 }, { "epoch": 0.83, "learning_rate": 0.0001454597204564139, "loss": 5.339, "step": 3615 }, { "epoch": 0.83, "learning_rate": 0.00014528121777615058, "loss": 5.2971, "step": 3620 }, { "epoch": 0.83, "learning_rate": 0.00014510253341718, "loss": 5.2645, "step": 3625 }, { "epoch": 0.83, "learning_rate": 0.0001449236680964251, "loss": 5.2741, "step": 3630 }, { "epoch": 0.83, "learning_rate": 0.00014474462253153482, "loss": 5.3032, "step": 3635 }, { "epoch": 0.84, "learning_rate": 0.00014456539744088138, "loss": 5.3179, "step": 3640 }, { "epoch": 0.84, "learning_rate": 0.00014438599354355723, "loss": 5.2993, "step": 3645 }, { "epoch": 0.84, "learning_rate": 0.00014420641155937224, "loss": 5.3037, "step": 3650 }, { "epoch": 0.84, "learning_rate": 0.00014402665220885087, "loss": 5.3506, "step": 3655 }, { "epoch": 0.84, "learning_rate": 0.00014384671621322915, "loss": 5.2998, "step": 3660 }, { "epoch": 0.84, "learning_rate": 0.00014366660429445183, "loss": 5.2896, "step": 3665 }, { "epoch": 0.84, "learning_rate": 0.00014348631717516953, "loss": 5.2643, "step": 3670 }, { "epoch": 0.84, "learning_rate": 0.00014330585557873588, "loss": 5.3258, "step": 3675 }, { "epoch": 0.84, "learning_rate": 0.00014312522022920444, "loss": 5.2385, "step": 3680 }, { "epoch": 0.85, "learning_rate": 0.00014294441185132607, "loss": 5.2566, "step": 3685 }, { "epoch": 0.85, "learning_rate": 0.00014276343117054563, "loss": 5.2949, "step": 3690 }, { "epoch": 0.85, "learning_rate": 0.00014258227891299948, "loss": 5.3243, "step": 3695 }, { "epoch": 0.85, "learning_rate": 0.00014240095580551234, "loss": 5.2613, "step": 3700 }, { "epoch": 0.85, "learning_rate": 0.00014221946257559438, "loss": 5.2501, "step": 3705 }, { "epoch": 0.85, "learning_rate": 0.0001420377999514384, "loss": 5.2639, "step": 3710 }, { "epoch": 0.85, "learning_rate": 0.0001418559686619168, "loss": 5.2594, "step": 3715 }, { "epoch": 0.85, "learning_rate": 0.00014167396943657874, "loss": 5.2639, "step": 3720 }, { "epoch": 0.85, "learning_rate": 0.00014149180300564713, "loss": 5.3033, "step": 3725 }, { "epoch": 0.86, "learning_rate": 0.0001413094701000158, "loss": 5.2797, "step": 3730 }, { "epoch": 0.86, "learning_rate": 0.0001411269714512465, "loss": 5.2528, "step": 3735 }, { "epoch": 0.86, "learning_rate": 0.00014094430779156597, "loss": 5.296, "step": 3740 }, { "epoch": 0.86, "learning_rate": 0.00014076147985386296, "loss": 5.2637, "step": 3745 }, { "epoch": 0.86, "learning_rate": 0.00014057848837168547, "loss": 5.291, "step": 3750 }, { "epoch": 0.86, "learning_rate": 0.00014039533407923758, "loss": 5.2754, "step": 3755 }, { "epoch": 0.86, "learning_rate": 0.00014021201771137663, "loss": 5.2591, "step": 3760 }, { "epoch": 0.86, "learning_rate": 0.0001400285400036102, "loss": 5.2479, "step": 3765 }, { "epoch": 0.87, "learning_rate": 0.00013984490169209333, "loss": 5.2711, "step": 3770 }, { "epoch": 0.87, "learning_rate": 0.00013966110351362523, "loss": 5.299, "step": 3775 }, { "epoch": 0.87, "learning_rate": 0.00013947714620564678, "loss": 5.2509, "step": 3780 }, { "epoch": 0.87, "learning_rate": 0.0001392930305062371, "loss": 5.2775, "step": 3785 }, { "epoch": 0.87, "learning_rate": 0.00013910875715411098, "loss": 5.2834, "step": 3790 }, { "epoch": 0.87, "learning_rate": 0.00013892432688861566, "loss": 5.2648, "step": 3795 }, { "epoch": 0.87, "learning_rate": 0.000138739740449728, "loss": 5.2384, "step": 3800 }, { "epoch": 0.87, "learning_rate": 0.00013855499857805137, "loss": 5.272, "step": 3805 }, { "epoch": 0.87, "learning_rate": 0.0001383701020148129, "loss": 5.2573, "step": 3810 }, { "epoch": 0.88, "learning_rate": 0.0001381850515018603, "loss": 5.2959, "step": 3815 }, { "epoch": 0.88, "learning_rate": 0.000137999847781659, "loss": 5.2411, "step": 3820 }, { "epoch": 0.88, "learning_rate": 0.00013781449159728907, "loss": 5.2197, "step": 3825 }, { "epoch": 0.88, "learning_rate": 0.00013762898369244238, "loss": 5.2314, "step": 3830 }, { "epoch": 0.88, "learning_rate": 0.00013744332481141952, "loss": 5.2228, "step": 3835 }, { "epoch": 0.88, "learning_rate": 0.00013725751569912682, "loss": 5.231, "step": 3840 }, { "epoch": 0.88, "learning_rate": 0.00013707155710107326, "loss": 5.1971, "step": 3845 }, { "epoch": 0.88, "learning_rate": 0.00013688544976336783, "loss": 5.2345, "step": 3850 }, { "epoch": 0.88, "learning_rate": 0.0001366991944327161, "loss": 5.238, "step": 3855 }, { "epoch": 0.89, "learning_rate": 0.00013651279185641752, "loss": 5.246, "step": 3860 }, { "epoch": 0.89, "learning_rate": 0.00013632624278236223, "loss": 5.2443, "step": 3865 }, { "epoch": 0.89, "learning_rate": 0.0001361395479590283, "loss": 5.2232, "step": 3870 }, { "epoch": 0.89, "learning_rate": 0.00013595270813547845, "loss": 5.2496, "step": 3875 }, { "epoch": 0.89, "learning_rate": 0.00013576572406135722, "loss": 5.1688, "step": 3880 }, { "epoch": 0.89, "learning_rate": 0.0001355785964868879, "loss": 5.21, "step": 3885 }, { "epoch": 0.89, "learning_rate": 0.00013539132616286956, "loss": 5.2414, "step": 3890 }, { "epoch": 0.89, "learning_rate": 0.000135203913840674, "loss": 5.2139, "step": 3895 }, { "epoch": 0.9, "learning_rate": 0.00013501636027224282, "loss": 5.214, "step": 3900 }, { "epoch": 0.9, "learning_rate": 0.00013482866621008417, "loss": 5.2154, "step": 3905 }, { "epoch": 0.9, "learning_rate": 0.0001346408324072701, "loss": 5.1666, "step": 3910 }, { "epoch": 0.9, "learning_rate": 0.0001344528596174331, "loss": 5.2159, "step": 3915 }, { "epoch": 0.9, "learning_rate": 0.0001342647485947635, "loss": 5.2013, "step": 3920 }, { "epoch": 0.9, "learning_rate": 0.0001340765000940062, "loss": 5.2193, "step": 3925 }, { "epoch": 0.9, "learning_rate": 0.00013388811487045766, "loss": 5.2116, "step": 3930 }, { "epoch": 0.9, "learning_rate": 0.00013369959367996286, "loss": 5.2398, "step": 3935 }, { "epoch": 0.9, "learning_rate": 0.0001335109372789125, "loss": 5.2073, "step": 3940 }, { "epoch": 0.91, "learning_rate": 0.00013332214642423952, "loss": 5.2629, "step": 3945 }, { "epoch": 0.91, "learning_rate": 0.00013313322187341652, "loss": 5.2636, "step": 3950 }, { "epoch": 0.91, "learning_rate": 0.00013294416438445244, "loss": 5.2398, "step": 3955 }, { "epoch": 0.91, "learning_rate": 0.00013275497471588953, "loss": 5.1794, "step": 3960 }, { "epoch": 0.91, "learning_rate": 0.00013256565362680057, "loss": 5.1936, "step": 3965 }, { "epoch": 0.91, "learning_rate": 0.0001323762018767854, "loss": 5.2014, "step": 3970 }, { "epoch": 0.91, "learning_rate": 0.00013218662022596828, "loss": 5.2163, "step": 3975 }, { "epoch": 0.91, "learning_rate": 0.00013199690943499457, "loss": 5.1886, "step": 3980 }, { "epoch": 0.91, "learning_rate": 0.0001318070702650278, "loss": 5.2418, "step": 3985 }, { "epoch": 0.92, "learning_rate": 0.0001316171034777466, "loss": 5.2023, "step": 3990 }, { "epoch": 0.92, "learning_rate": 0.00013142700983534157, "loss": 5.1956, "step": 3995 }, { "epoch": 0.92, "learning_rate": 0.00013123679010051232, "loss": 5.1696, "step": 4000 }, { "epoch": 0.92, "learning_rate": 0.0001310464450364644, "loss": 5.2116, "step": 4005 }, { "epoch": 0.92, "learning_rate": 0.00013085597540690618, "loss": 5.1862, "step": 4010 }, { "epoch": 0.92, "learning_rate": 0.0001306653819760458, "loss": 5.1857, "step": 4015 }, { "epoch": 0.92, "learning_rate": 0.00013047466550858812, "loss": 5.163, "step": 4020 }, { "epoch": 0.92, "learning_rate": 0.00013028382676973178, "loss": 5.1765, "step": 4025 }, { "epoch": 0.92, "learning_rate": 0.00013009286652516575, "loss": 5.199, "step": 4030 }, { "epoch": 0.93, "learning_rate": 0.00012990178554106674, "loss": 5.1622, "step": 4035 }, { "epoch": 0.93, "learning_rate": 0.00012971058458409576, "loss": 5.1589, "step": 4040 }, { "epoch": 0.93, "learning_rate": 0.0001295192644213952, "loss": 5.1641, "step": 4045 }, { "epoch": 0.93, "learning_rate": 0.00012932782582058584, "loss": 5.1361, "step": 4050 }, { "epoch": 0.93, "learning_rate": 0.00012913626954976355, "loss": 5.1973, "step": 4055 }, { "epoch": 0.93, "learning_rate": 0.00012894459637749627, "loss": 5.1735, "step": 4060 }, { "epoch": 0.93, "learning_rate": 0.00012875280707282116, "loss": 5.1678, "step": 4065 }, { "epoch": 0.93, "learning_rate": 0.0001285609024052411, "loss": 5.187, "step": 4070 }, { "epoch": 0.94, "learning_rate": 0.00012836888314472208, "loss": 5.1512, "step": 4075 }, { "epoch": 0.94, "learning_rate": 0.00012817675006168963, "loss": 5.1546, "step": 4080 }, { "epoch": 0.94, "learning_rate": 0.00012798450392702615, "loss": 5.1529, "step": 4085 }, { "epoch": 0.94, "learning_rate": 0.00012779214551206746, "loss": 5.1515, "step": 4090 }, { "epoch": 0.94, "learning_rate": 0.00012759967558860006, "loss": 5.1514, "step": 4095 }, { "epoch": 0.94, "learning_rate": 0.0001274070949288577, "loss": 5.1721, "step": 4100 }, { "epoch": 0.94, "learning_rate": 0.00012721440430551849, "loss": 5.1295, "step": 4105 }, { "epoch": 0.94, "learning_rate": 0.00012702160449170165, "loss": 5.134, "step": 4110 }, { "epoch": 0.94, "learning_rate": 0.0001268286962609647, "loss": 5.1717, "step": 4115 }, { "epoch": 0.95, "learning_rate": 0.0001266356803873, "loss": 5.1103, "step": 4120 }, { "epoch": 0.95, "learning_rate": 0.00012644255764513178, "loss": 5.1361, "step": 4125 }, { "epoch": 0.95, "learning_rate": 0.0001262493288093131, "loss": 5.1331, "step": 4130 }, { "epoch": 0.95, "learning_rate": 0.00012605599465512278, "loss": 5.1673, "step": 4135 }, { "epoch": 0.95, "learning_rate": 0.000125862555958262, "loss": 5.1172, "step": 4140 }, { "epoch": 0.95, "learning_rate": 0.00012566901349485158, "loss": 5.1707, "step": 4145 }, { "epoch": 0.95, "learning_rate": 0.00012547536804142858, "loss": 5.0802, "step": 4150 }, { "epoch": 0.95, "learning_rate": 0.00012528162037494332, "loss": 5.1604, "step": 4155 }, { "epoch": 0.95, "learning_rate": 0.0001250877712727561, "loss": 5.1485, "step": 4160 }, { "epoch": 0.96, "learning_rate": 0.00012489382151263442, "loss": 5.1316, "step": 4165 }, { "epoch": 0.96, "learning_rate": 0.00012469977187274945, "loss": 5.1652, "step": 4170 }, { "epoch": 0.96, "learning_rate": 0.00012450562313167326, "loss": 5.1424, "step": 4175 }, { "epoch": 0.96, "learning_rate": 0.00012431137606837538, "loss": 5.1375, "step": 4180 }, { "epoch": 0.96, "learning_rate": 0.00012411703146221993, "loss": 5.1874, "step": 4185 }, { "epoch": 0.96, "learning_rate": 0.0001239225900929624, "loss": 5.1192, "step": 4190 }, { "epoch": 0.96, "learning_rate": 0.00012372805274074643, "loss": 5.1478, "step": 4195 }, { "epoch": 0.96, "learning_rate": 0.00012353342018610084, "loss": 5.1118, "step": 4200 }, { "epoch": 0.97, "learning_rate": 0.0001233386932099364, "loss": 5.1292, "step": 4205 }, { "epoch": 0.97, "learning_rate": 0.00012314387259354282, "loss": 5.1816, "step": 4210 }, { "epoch": 0.97, "learning_rate": 0.00012294895911858528, "loss": 5.1334, "step": 4215 }, { "epoch": 0.97, "learning_rate": 0.00012275395356710177, "loss": 5.1112, "step": 4220 }, { "epoch": 0.97, "learning_rate": 0.00012255885672149953, "loss": 5.0939, "step": 4225 }, { "epoch": 0.97, "learning_rate": 0.0001223636693645523, "loss": 5.1542, "step": 4230 }, { "epoch": 0.97, "learning_rate": 0.00012216839227939676, "loss": 5.1012, "step": 4235 }, { "epoch": 0.97, "learning_rate": 0.00012197302624952971, "loss": 5.1341, "step": 4240 }, { "epoch": 0.97, "learning_rate": 0.0001217775720588048, "loss": 5.1443, "step": 4245 }, { "epoch": 0.98, "learning_rate": 0.00012158203049142947, "loss": 5.0631, "step": 4250 }, { "epoch": 0.98, "learning_rate": 0.00012138640233196152, "loss": 5.1662, "step": 4255 }, { "epoch": 0.98, "learning_rate": 0.00012119068836530644, "loss": 5.1628, "step": 4260 }, { "epoch": 0.98, "learning_rate": 0.00012099488937671384, "loss": 5.1924, "step": 4265 }, { "epoch": 0.98, "learning_rate": 0.00012079900615177449, "loss": 5.165, "step": 4270 }, { "epoch": 0.98, "learning_rate": 0.00012060303947641715, "loss": 5.1227, "step": 4275 }, { "epoch": 0.98, "learning_rate": 0.00012040699013690543, "loss": 5.0873, "step": 4280 }, { "epoch": 0.98, "learning_rate": 0.00012021085891983456, "loss": 5.1478, "step": 4285 }, { "epoch": 0.98, "learning_rate": 0.00012001464661212827, "loss": 5.1332, "step": 4290 }, { "epoch": 0.99, "learning_rate": 0.00011981835400103568, "loss": 5.1551, "step": 4295 }, { "epoch": 0.99, "learning_rate": 0.0001196219818741281, "loss": 5.0721, "step": 4300 }, { "epoch": 0.99, "learning_rate": 0.00011942553101929585, "loss": 5.1071, "step": 4305 }, { "epoch": 0.99, "learning_rate": 0.00011922900222474523, "loss": 5.1758, "step": 4310 }, { "epoch": 0.99, "learning_rate": 0.00011903239627899503, "loss": 5.109, "step": 4315 }, { "epoch": 0.99, "learning_rate": 0.00011883571397087387, "loss": 5.1134, "step": 4320 }, { "epoch": 0.99, "learning_rate": 0.00011863895608951654, "loss": 5.1427, "step": 4325 }, { "epoch": 0.99, "learning_rate": 0.00011844212342436112, "loss": 5.088, "step": 4330 }, { "epoch": 0.99, "learning_rate": 0.0001182452167651457, "loss": 5.1406, "step": 4335 }, { "epoch": 1.0, "learning_rate": 0.00011804823690190538, "loss": 5.1298, "step": 4340 }, { "epoch": 1.0, "learning_rate": 0.0001178511846249688, "loss": 5.1483, "step": 4345 }, { "epoch": 1.0, "learning_rate": 0.00011765406072495528, "loss": 5.1122, "step": 4350 }, { "epoch": 1.0, "learning_rate": 0.0001174568659927714, "loss": 5.113, "step": 4355 }, { "epoch": 1.0, "eval_loss": 5.059106826782227, "eval_runtime": 1138.4406, "eval_samples_per_second": 13.555, "eval_steps_per_second": 0.424, "step": 4357 }, { "epoch": 1.0, "learning_rate": 0.00011725960121960806, "loss": 5.125, "step": 4360 }, { "epoch": 1.0, "learning_rate": 0.00011706226719693698, "loss": 5.0662, "step": 4365 }, { "epoch": 1.0, "learning_rate": 0.00011686486471650798, "loss": 5.1071, "step": 4370 }, { "epoch": 1.0, "learning_rate": 0.00011666739457034532, "loss": 5.1138, "step": 4375 }, { "epoch": 1.01, "learning_rate": 0.0001164698575507449, "loss": 5.0708, "step": 4380 }, { "epoch": 1.01, "learning_rate": 0.00011627225445027086, "loss": 5.0868, "step": 4385 }, { "epoch": 1.01, "learning_rate": 0.00011607458606175253, "loss": 5.0646, "step": 4390 }, { "epoch": 1.01, "learning_rate": 0.00011587685317828112, "loss": 5.0664, "step": 4395 }, { "epoch": 1.01, "learning_rate": 0.00011567905659320663, "loss": 5.1633, "step": 4400 }, { "epoch": 1.01, "learning_rate": 0.0001154811971001347, "loss": 5.1106, "step": 4405 }, { "epoch": 1.01, "learning_rate": 0.00011528327549292326, "loss": 5.056, "step": 4410 }, { "epoch": 1.01, "learning_rate": 0.00011508529256567961, "loss": 5.1278, "step": 4415 }, { "epoch": 1.01, "learning_rate": 0.00011488724911275694, "loss": 5.0478, "step": 4420 }, { "epoch": 1.02, "learning_rate": 0.00011468914592875135, "loss": 5.1035, "step": 4425 }, { "epoch": 1.02, "learning_rate": 0.00011449098380849858, "loss": 5.0953, "step": 4430 }, { "epoch": 1.02, "learning_rate": 0.00011429276354707086, "loss": 5.1338, "step": 4435 }, { "epoch": 1.02, "learning_rate": 0.00011409448593977363, "loss": 5.1327, "step": 4440 }, { "epoch": 1.02, "learning_rate": 0.00011389615178214253, "loss": 5.133, "step": 4445 }, { "epoch": 1.02, "learning_rate": 0.00011369776186993994, "loss": 5.0279, "step": 4450 }, { "epoch": 1.02, "learning_rate": 0.00011349931699915209, "loss": 5.0927, "step": 4455 }, { "epoch": 1.02, "learning_rate": 0.0001133008179659856, "loss": 5.0782, "step": 4460 }, { "epoch": 1.02, "learning_rate": 0.0001131022655668645, "loss": 5.0554, "step": 4465 }, { "epoch": 1.03, "learning_rate": 0.00011290366059842683, "loss": 5.0716, "step": 4470 }, { "epoch": 1.03, "learning_rate": 0.00011270500385752167, "loss": 5.0599, "step": 4475 }, { "epoch": 1.03, "learning_rate": 0.00011250629614120571, "loss": 5.0337, "step": 4480 }, { "epoch": 1.03, "learning_rate": 0.00011230753824674025, "loss": 5.0706, "step": 4485 }, { "epoch": 1.03, "learning_rate": 0.00011210873097158786, "loss": 5.0528, "step": 4490 }, { "epoch": 1.03, "learning_rate": 0.00011190987511340928, "loss": 5.0766, "step": 4495 }, { "epoch": 1.03, "learning_rate": 0.00011171097147006013, "loss": 5.0202, "step": 4500 }, { "epoch": 1.03, "learning_rate": 0.0001115120208395878, "loss": 5.0974, "step": 4505 }, { "epoch": 1.03, "learning_rate": 0.00011131302402022821, "loss": 5.0267, "step": 4510 }, { "epoch": 1.04, "learning_rate": 0.00011111398181040251, "loss": 5.0824, "step": 4515 }, { "epoch": 1.04, "learning_rate": 0.00011091489500871408, "loss": 5.0885, "step": 4520 }, { "epoch": 1.04, "learning_rate": 0.00011071576441394516, "loss": 5.0802, "step": 4525 }, { "epoch": 1.04, "learning_rate": 0.00011051659082505366, "loss": 5.0019, "step": 4530 }, { "epoch": 1.04, "learning_rate": 0.0001103173750411701, "loss": 5.0051, "step": 4535 }, { "epoch": 1.04, "learning_rate": 0.00011011811786159416, "loss": 5.0516, "step": 4540 }, { "epoch": 1.04, "learning_rate": 0.00010991882008579177, "loss": 5.0614, "step": 4545 }, { "epoch": 1.04, "learning_rate": 0.00010971948251339157, "loss": 5.0201, "step": 4550 }, { "epoch": 1.05, "learning_rate": 0.00010952010594418201, "loss": 5.0697, "step": 4555 }, { "epoch": 1.05, "learning_rate": 0.00010932069117810787, "loss": 5.0648, "step": 4560 }, { "epoch": 1.05, "learning_rate": 0.00010912123901526736, "loss": 4.9865, "step": 4565 }, { "epoch": 1.05, "learning_rate": 0.00010892175025590856, "loss": 5.0553, "step": 4570 }, { "epoch": 1.05, "learning_rate": 0.00010872222570042652, "loss": 4.9917, "step": 4575 }, { "epoch": 1.05, "learning_rate": 0.00010852266614935982, "loss": 4.9836, "step": 4580 }, { "epoch": 1.05, "learning_rate": 0.00010832307240338752, "loss": 5.045, "step": 4585 }, { "epoch": 1.05, "learning_rate": 0.00010812344526332578, "loss": 5.0025, "step": 4590 }, { "epoch": 1.05, "learning_rate": 0.00010792378553012488, "loss": 5.0356, "step": 4595 }, { "epoch": 1.06, "learning_rate": 0.00010772409400486571, "loss": 5.062, "step": 4600 }, { "epoch": 1.06, "learning_rate": 0.0001075243714887569, "loss": 5.038, "step": 4605 }, { "epoch": 1.06, "learning_rate": 0.00010732461878313125, "loss": 5.0387, "step": 4610 }, { "epoch": 1.06, "learning_rate": 0.0001071248366894428, "loss": 5.0372, "step": 4615 }, { "epoch": 1.06, "learning_rate": 0.00010692502600926348, "loss": 4.9963, "step": 4620 }, { "epoch": 1.06, "learning_rate": 0.00010672518754427988, "loss": 4.9673, "step": 4625 }, { "epoch": 1.06, "learning_rate": 0.00010652532209629011, "loss": 4.9897, "step": 4630 }, { "epoch": 1.06, "learning_rate": 0.0001063254304672005, "loss": 5.026, "step": 4635 }, { "epoch": 1.06, "learning_rate": 0.00010612551345902244, "loss": 5.0018, "step": 4640 }, { "epoch": 1.07, "learning_rate": 0.0001059255718738692, "loss": 4.9921, "step": 4645 }, { "epoch": 1.07, "learning_rate": 0.00010572560651395258, "loss": 5.0137, "step": 4650 }, { "epoch": 1.07, "learning_rate": 0.0001055256181815798, "loss": 4.9699, "step": 4655 }, { "epoch": 1.07, "learning_rate": 0.0001053256076791503, "loss": 5.0523, "step": 4660 }, { "epoch": 1.07, "learning_rate": 0.00010512557580915235, "loss": 4.9911, "step": 4665 }, { "epoch": 1.07, "learning_rate": 0.00010492552337416007, "loss": 4.9636, "step": 4670 }, { "epoch": 1.07, "learning_rate": 0.00010472545117683004, "loss": 4.9693, "step": 4675 }, { "epoch": 1.07, "learning_rate": 0.00010452536001989815, "loss": 5.0237, "step": 4680 }, { "epoch": 1.08, "learning_rate": 0.00010432525070617633, "loss": 4.9668, "step": 4685 }, { "epoch": 1.08, "learning_rate": 0.00010412512403854942, "loss": 5.0143, "step": 4690 }, { "epoch": 1.08, "learning_rate": 0.00010392498081997182, "loss": 4.9906, "step": 4695 }, { "epoch": 1.08, "learning_rate": 0.00010372482185346435, "loss": 5.0234, "step": 4700 }, { "epoch": 1.08, "learning_rate": 0.00010352464794211104, "loss": 4.9865, "step": 4705 }, { "epoch": 1.08, "learning_rate": 0.00010332445988905586, "loss": 4.974, "step": 4710 }, { "epoch": 1.08, "learning_rate": 0.00010312425849749951, "loss": 4.9991, "step": 4715 }, { "epoch": 1.08, "learning_rate": 0.00010292404457069631, "loss": 4.9621, "step": 4720 }, { "epoch": 1.08, "learning_rate": 0.00010272381891195071, "loss": 4.885, "step": 4725 }, { "epoch": 1.09, "learning_rate": 0.00010252358232461436, "loss": 4.9878, "step": 4730 }, { "epoch": 1.09, "learning_rate": 0.00010232333561208264, "loss": 4.9553, "step": 4735 }, { "epoch": 1.09, "learning_rate": 0.00010212307957779173, "loss": 5.036, "step": 4740 }, { "epoch": 1.09, "learning_rate": 0.00010192281502521499, "loss": 4.9396, "step": 4745 }, { "epoch": 1.09, "learning_rate": 0.00010172254275786017, "loss": 5.0147, "step": 4750 }, { "epoch": 1.09, "learning_rate": 0.00010152226357926582, "loss": 4.9507, "step": 4755 }, { "epoch": 1.09, "learning_rate": 0.0001013219782929983, "loss": 4.9798, "step": 4760 }, { "epoch": 1.09, "learning_rate": 0.00010112168770264843, "loss": 4.961, "step": 4765 }, { "epoch": 1.09, "learning_rate": 0.0001009213926118284, "loss": 4.9977, "step": 4770 }, { "epoch": 1.1, "learning_rate": 0.00010072109382416828, "loss": 4.9127, "step": 4775 }, { "epoch": 1.1, "learning_rate": 0.00010052079214331318, "loss": 4.9549, "step": 4780 }, { "epoch": 1.1, "learning_rate": 0.00010032048837291963, "loss": 5.0015, "step": 4785 }, { "epoch": 1.1, "learning_rate": 0.00010012018331665272, "loss": 5.0122, "step": 4790 }, { "epoch": 1.1, "learning_rate": 9.991987777818258e-05, "loss": 4.995, "step": 4795 }, { "epoch": 1.1, "learning_rate": 9.971957256118129e-05, "loss": 4.9526, "step": 4800 }, { "epoch": 1.1, "learning_rate": 9.951926846931969e-05, "loss": 4.9129, "step": 4805 }, { "epoch": 1.1, "learning_rate": 9.931896630626402e-05, "loss": 4.968, "step": 4810 }, { "epoch": 1.1, "learning_rate": 9.911866687567291e-05, "loss": 4.9704, "step": 4815 }, { "epoch": 1.11, "learning_rate": 9.891837098119389e-05, "loss": 4.95, "step": 4820 }, { "epoch": 1.11, "learning_rate": 9.87180794264604e-05, "loss": 4.9443, "step": 4825 }, { "epoch": 1.11, "learning_rate": 9.851779301508842e-05, "loss": 4.9094, "step": 4830 }, { "epoch": 1.11, "learning_rate": 9.831751255067329e-05, "loss": 5.0039, "step": 4835 }, { "epoch": 1.11, "learning_rate": 9.811723883678654e-05, "loss": 4.9285, "step": 4840 }, { "epoch": 1.11, "learning_rate": 9.791697267697255e-05, "loss": 4.9114, "step": 4845 }, { "epoch": 1.11, "learning_rate": 9.771671487474546e-05, "loss": 4.8915, "step": 4850 }, { "epoch": 1.11, "learning_rate": 9.751646623358576e-05, "loss": 4.9254, "step": 4855 }, { "epoch": 1.12, "learning_rate": 9.731622755693737e-05, "loss": 4.9335, "step": 4860 }, { "epoch": 1.12, "learning_rate": 9.711599964820405e-05, "loss": 4.8724, "step": 4865 }, { "epoch": 1.12, "learning_rate": 9.691578331074643e-05, "loss": 4.9849, "step": 4870 }, { "epoch": 1.12, "learning_rate": 9.671557934787874e-05, "loss": 4.9726, "step": 4875 }, { "epoch": 1.12, "learning_rate": 9.651538856286551e-05, "loss": 4.93, "step": 4880 }, { "epoch": 1.12, "learning_rate": 9.631521175891844e-05, "loss": 4.8794, "step": 4885 }, { "epoch": 1.12, "learning_rate": 9.611504973919311e-05, "loss": 4.8882, "step": 4890 }, { "epoch": 1.12, "learning_rate": 9.591490330678579e-05, "loss": 4.9173, "step": 4895 }, { "epoch": 1.12, "learning_rate": 9.571477326473021e-05, "loss": 4.9371, "step": 4900 }, { "epoch": 1.13, "learning_rate": 9.551466041599432e-05, "loss": 4.8697, "step": 4905 }, { "epoch": 1.13, "learning_rate": 9.531456556347712e-05, "loss": 4.935, "step": 4910 }, { "epoch": 1.13, "learning_rate": 9.511448951000535e-05, "loss": 4.9023, "step": 4915 }, { "epoch": 1.13, "learning_rate": 9.491443305833043e-05, "loss": 4.9414, "step": 4920 }, { "epoch": 1.13, "learning_rate": 9.4714397011125e-05, "loss": 4.9216, "step": 4925 }, { "epoch": 1.13, "learning_rate": 9.451438217097994e-05, "loss": 4.9164, "step": 4930 }, { "epoch": 1.13, "learning_rate": 9.4314389340401e-05, "loss": 4.9352, "step": 4935 }, { "epoch": 1.13, "learning_rate": 9.411441932180554e-05, "loss": 4.8866, "step": 4940 }, { "epoch": 1.13, "learning_rate": 9.391447291751961e-05, "loss": 4.902, "step": 4945 }, { "epoch": 1.14, "learning_rate": 9.371455092977423e-05, "loss": 4.916, "step": 4950 }, { "epoch": 1.14, "learning_rate": 9.351465416070274e-05, "loss": 4.8924, "step": 4955 }, { "epoch": 1.14, "learning_rate": 9.331478341233706e-05, "loss": 4.9241, "step": 4960 }, { "epoch": 1.14, "learning_rate": 9.311493948660488e-05, "loss": 4.9589, "step": 4965 }, { "epoch": 1.14, "learning_rate": 9.291512318532614e-05, "loss": 4.874, "step": 4970 }, { "epoch": 1.14, "learning_rate": 9.271533531021005e-05, "loss": 4.8862, "step": 4975 }, { "epoch": 1.14, "learning_rate": 9.251557666285174e-05, "loss": 4.8859, "step": 4980 }, { "epoch": 1.14, "learning_rate": 9.231584804472898e-05, "loss": 4.8948, "step": 4985 }, { "epoch": 1.15, "learning_rate": 9.211615025719919e-05, "loss": 4.8823, "step": 4990 }, { "epoch": 1.15, "learning_rate": 9.1916484101496e-05, "loss": 4.9055, "step": 4995 }, { "epoch": 1.15, "learning_rate": 9.17168503787262e-05, "loss": 4.9045, "step": 5000 }, { "epoch": 1.15, "learning_rate": 9.151724988986635e-05, "loss": 4.9271, "step": 5005 }, { "epoch": 1.15, "learning_rate": 9.131768343575979e-05, "loss": 4.8845, "step": 5010 }, { "epoch": 1.15, "learning_rate": 9.11181518171132e-05, "loss": 4.9525, "step": 5015 }, { "epoch": 1.15, "learning_rate": 9.091865583449351e-05, "loss": 4.8578, "step": 5020 }, { "epoch": 1.15, "learning_rate": 9.071919628832476e-05, "loss": 4.8773, "step": 5025 }, { "epoch": 1.15, "learning_rate": 9.051977397888464e-05, "loss": 4.8676, "step": 5030 }, { "epoch": 1.16, "learning_rate": 9.032038970630163e-05, "loss": 4.8945, "step": 5035 }, { "epoch": 1.16, "learning_rate": 9.01210442705514e-05, "loss": 4.8338, "step": 5040 }, { "epoch": 1.16, "learning_rate": 8.992173847145401e-05, "loss": 4.8511, "step": 5045 }, { "epoch": 1.16, "learning_rate": 8.972247310867027e-05, "loss": 4.9281, "step": 5050 }, { "epoch": 1.16, "learning_rate": 8.952324898169888e-05, "loss": 4.8169, "step": 5055 }, { "epoch": 1.16, "learning_rate": 8.932406688987309e-05, "loss": 4.8685, "step": 5060 }, { "epoch": 1.16, "learning_rate": 8.912492763235744e-05, "loss": 4.878, "step": 5065 }, { "epoch": 1.16, "learning_rate": 8.892583200814466e-05, "loss": 4.8541, "step": 5070 }, { "epoch": 1.16, "learning_rate": 8.872678081605236e-05, "loss": 4.8263, "step": 5075 }, { "epoch": 1.17, "learning_rate": 8.852777485471997e-05, "loss": 4.908, "step": 5080 }, { "epoch": 1.17, "learning_rate": 8.832881492260535e-05, "loss": 4.8555, "step": 5085 }, { "epoch": 1.17, "learning_rate": 8.81299018179817e-05, "loss": 4.8749, "step": 5090 }, { "epoch": 1.17, "learning_rate": 8.793103633893437e-05, "loss": 4.8645, "step": 5095 }, { "epoch": 1.17, "learning_rate": 8.773221928335759e-05, "loss": 4.8361, "step": 5100 }, { "epoch": 1.17, "learning_rate": 8.753345144895136e-05, "loss": 4.9032, "step": 5105 }, { "epoch": 1.17, "learning_rate": 8.73347336332181e-05, "loss": 4.855, "step": 5110 }, { "epoch": 1.17, "learning_rate": 8.713606663345968e-05, "loss": 4.834, "step": 5115 }, { "epoch": 1.17, "learning_rate": 8.693745124677386e-05, "loss": 4.7977, "step": 5120 }, { "epoch": 1.18, "learning_rate": 8.673888827005164e-05, "loss": 4.8372, "step": 5125 }, { "epoch": 1.18, "learning_rate": 8.654037849997342e-05, "loss": 4.8728, "step": 5130 }, { "epoch": 1.18, "learning_rate": 8.634192273300629e-05, "loss": 4.8253, "step": 5135 }, { "epoch": 1.18, "learning_rate": 8.614352176540067e-05, "loss": 4.8623, "step": 5140 }, { "epoch": 1.18, "learning_rate": 8.594517639318705e-05, "loss": 4.9098, "step": 5145 }, { "epoch": 1.18, "learning_rate": 8.57468874121729e-05, "loss": 4.8379, "step": 5150 }, { "epoch": 1.18, "learning_rate": 8.554865561793942e-05, "loss": 4.8979, "step": 5155 }, { "epoch": 1.18, "learning_rate": 8.535048180583838e-05, "loss": 4.8408, "step": 5160 }, { "epoch": 1.19, "learning_rate": 8.515236677098889e-05, "loss": 4.8904, "step": 5165 }, { "epoch": 1.19, "learning_rate": 8.495431130827422e-05, "loss": 4.8321, "step": 5170 }, { "epoch": 1.19, "learning_rate": 8.475631621233869e-05, "loss": 4.8462, "step": 5175 }, { "epoch": 1.19, "learning_rate": 8.455838227758432e-05, "loss": 4.8631, "step": 5180 }, { "epoch": 1.19, "learning_rate": 8.436051029816783e-05, "loss": 4.8231, "step": 5185 }, { "epoch": 1.19, "learning_rate": 8.416270106799726e-05, "loss": 4.8626, "step": 5190 }, { "epoch": 1.19, "learning_rate": 8.396495538072902e-05, "loss": 4.8349, "step": 5195 }, { "epoch": 1.19, "learning_rate": 8.376727402976447e-05, "loss": 4.83, "step": 5200 }, { "epoch": 1.19, "learning_rate": 8.356965780824677e-05, "loss": 4.9056, "step": 5205 }, { "epoch": 1.2, "learning_rate": 8.3372107509058e-05, "loss": 4.8734, "step": 5210 }, { "epoch": 1.2, "learning_rate": 8.317462392481546e-05, "loss": 4.8278, "step": 5215 }, { "epoch": 1.2, "learning_rate": 8.297720784786906e-05, "loss": 4.7766, "step": 5220 }, { "epoch": 1.2, "learning_rate": 8.277986007029756e-05, "loss": 4.8468, "step": 5225 }, { "epoch": 1.2, "learning_rate": 8.2582581383906e-05, "loss": 4.8094, "step": 5230 }, { "epoch": 1.2, "learning_rate": 8.238537258022194e-05, "loss": 4.8178, "step": 5235 }, { "epoch": 1.2, "learning_rate": 8.218823445049265e-05, "loss": 4.8364, "step": 5240 }, { "epoch": 1.2, "learning_rate": 8.199116778568192e-05, "loss": 4.8331, "step": 5245 }, { "epoch": 1.2, "learning_rate": 8.179417337646669e-05, "loss": 4.8482, "step": 5250 }, { "epoch": 1.21, "learning_rate": 8.159725201323408e-05, "loss": 4.8213, "step": 5255 }, { "epoch": 1.21, "learning_rate": 8.140040448607804e-05, "loss": 4.8293, "step": 5260 }, { "epoch": 1.21, "learning_rate": 8.12036315847964e-05, "loss": 4.8363, "step": 5265 }, { "epoch": 1.21, "learning_rate": 8.100693409888748e-05, "loss": 4.8323, "step": 5270 }, { "epoch": 1.21, "learning_rate": 8.081031281754695e-05, "loss": 4.7994, "step": 5275 }, { "epoch": 1.21, "learning_rate": 8.061376852966495e-05, "loss": 4.8269, "step": 5280 }, { "epoch": 1.21, "learning_rate": 8.041730202382245e-05, "loss": 4.8294, "step": 5285 }, { "epoch": 1.21, "learning_rate": 8.02209140882886e-05, "loss": 4.8326, "step": 5290 }, { "epoch": 1.22, "learning_rate": 8.002460551101702e-05, "loss": 4.8782, "step": 5295 }, { "epoch": 1.22, "learning_rate": 7.982837707964321e-05, "loss": 4.8432, "step": 5300 }, { "epoch": 1.22, "learning_rate": 7.963222958148085e-05, "loss": 4.8019, "step": 5305 }, { "epoch": 1.22, "learning_rate": 7.943616380351913e-05, "loss": 4.7782, "step": 5310 }, { "epoch": 1.22, "learning_rate": 7.92401805324192e-05, "loss": 4.8038, "step": 5315 }, { "epoch": 1.22, "learning_rate": 7.904428055451118e-05, "loss": 4.8114, "step": 5320 }, { "epoch": 1.22, "learning_rate": 7.884846465579113e-05, "loss": 4.7944, "step": 5325 }, { "epoch": 1.22, "learning_rate": 7.865273362191759e-05, "loss": 4.8051, "step": 5330 }, { "epoch": 1.22, "learning_rate": 7.845708823820876e-05, "loss": 4.822, "step": 5335 }, { "epoch": 1.23, "learning_rate": 7.826152928963904e-05, "loss": 4.8269, "step": 5340 }, { "epoch": 1.23, "learning_rate": 7.806605756083622e-05, "loss": 4.8259, "step": 5345 }, { "epoch": 1.23, "learning_rate": 7.787067383607796e-05, "loss": 4.8119, "step": 5350 }, { "epoch": 1.23, "learning_rate": 7.767537889928889e-05, "loss": 4.7746, "step": 5355 }, { "epoch": 1.23, "learning_rate": 7.748017353403748e-05, "loss": 4.8108, "step": 5360 }, { "epoch": 1.23, "learning_rate": 7.72850585235327e-05, "loss": 4.7756, "step": 5365 }, { "epoch": 1.23, "learning_rate": 7.70900346506211e-05, "loss": 4.7291, "step": 5370 }, { "epoch": 1.23, "learning_rate": 7.689510269778347e-05, "loss": 4.7695, "step": 5375 }, { "epoch": 1.23, "learning_rate": 7.670026344713189e-05, "loss": 4.7634, "step": 5380 }, { "epoch": 1.24, "learning_rate": 7.650551768040641e-05, "loss": 4.7772, "step": 5385 }, { "epoch": 1.24, "learning_rate": 7.631086617897203e-05, "loss": 4.7792, "step": 5390 }, { "epoch": 1.24, "learning_rate": 7.611630972381557e-05, "loss": 4.7878, "step": 5395 }, { "epoch": 1.24, "learning_rate": 7.592184909554245e-05, "loss": 4.8142, "step": 5400 }, { "epoch": 1.24, "learning_rate": 7.572748507437368e-05, "loss": 4.8524, "step": 5405 }, { "epoch": 1.24, "learning_rate": 7.553321844014258e-05, "loss": 4.803, "step": 5410 }, { "epoch": 1.24, "learning_rate": 7.533904997229183e-05, "loss": 4.8006, "step": 5415 }, { "epoch": 1.24, "learning_rate": 7.514498044987009e-05, "loss": 4.8241, "step": 5420 }, { "epoch": 1.24, "learning_rate": 7.495101065152917e-05, "loss": 4.8204, "step": 5425 }, { "epoch": 1.25, "learning_rate": 7.475714135552074e-05, "loss": 4.8527, "step": 5430 }, { "epoch": 1.25, "learning_rate": 7.456337333969316e-05, "loss": 4.7756, "step": 5435 }, { "epoch": 1.25, "learning_rate": 7.43697073814885e-05, "loss": 4.7708, "step": 5440 }, { "epoch": 1.25, "learning_rate": 7.417614425793932e-05, "loss": 4.8252, "step": 5445 }, { "epoch": 1.25, "learning_rate": 7.39826847456656e-05, "loss": 4.8143, "step": 5450 }, { "epoch": 1.25, "learning_rate": 7.37893296208716e-05, "loss": 4.7498, "step": 5455 }, { "epoch": 1.25, "learning_rate": 7.359607965934274e-05, "loss": 4.7539, "step": 5460 }, { "epoch": 1.25, "learning_rate": 7.340293563644256e-05, "loss": 4.7292, "step": 5465 }, { "epoch": 1.26, "learning_rate": 7.32098983271094e-05, "loss": 4.8232, "step": 5470 }, { "epoch": 1.26, "learning_rate": 7.301696850585366e-05, "loss": 4.8129, "step": 5475 }, { "epoch": 1.26, "learning_rate": 7.282414694675426e-05, "loss": 4.7475, "step": 5480 }, { "epoch": 1.26, "learning_rate": 7.263143442345592e-05, "loss": 4.7766, "step": 5485 }, { "epoch": 1.26, "learning_rate": 7.243883170916574e-05, "loss": 4.7741, "step": 5490 }, { "epoch": 1.26, "learning_rate": 7.22463395766504e-05, "loss": 4.7999, "step": 5495 }, { "epoch": 1.26, "learning_rate": 7.205395879823271e-05, "loss": 4.7793, "step": 5500 }, { "epoch": 1.26, "learning_rate": 7.186169014578883e-05, "loss": 4.7619, "step": 5505 }, { "epoch": 1.26, "learning_rate": 7.166953439074504e-05, "loss": 4.7841, "step": 5510 }, { "epoch": 1.27, "learning_rate": 7.14774923040746e-05, "loss": 4.7674, "step": 5515 }, { "epoch": 1.27, "learning_rate": 7.128556465629475e-05, "loss": 4.7724, "step": 5520 }, { "epoch": 1.27, "learning_rate": 7.109375221746352e-05, "loss": 4.7496, "step": 5525 }, { "epoch": 1.27, "learning_rate": 7.090205575717678e-05, "loss": 4.7153, "step": 5530 }, { "epoch": 1.27, "learning_rate": 7.071047604456499e-05, "loss": 4.7254, "step": 5535 }, { "epoch": 1.27, "learning_rate": 7.051901384829021e-05, "loss": 4.767, "step": 5540 }, { "epoch": 1.27, "learning_rate": 7.032766993654303e-05, "loss": 4.7913, "step": 5545 }, { "epoch": 1.27, "learning_rate": 7.013644507703937e-05, "loss": 4.8, "step": 5550 }, { "epoch": 1.27, "learning_rate": 6.994534003701765e-05, "loss": 4.7623, "step": 5555 }, { "epoch": 1.28, "learning_rate": 6.975435558323532e-05, "loss": 4.7464, "step": 5560 }, { "epoch": 1.28, "learning_rate": 6.956349248196627e-05, "loss": 4.7224, "step": 5565 }, { "epoch": 1.28, "learning_rate": 6.937275149899725e-05, "loss": 4.6828, "step": 5570 }, { "epoch": 1.28, "learning_rate": 6.918213339962518e-05, "loss": 4.7704, "step": 5575 }, { "epoch": 1.28, "learning_rate": 6.899163894865395e-05, "loss": 4.742, "step": 5580 }, { "epoch": 1.28, "learning_rate": 6.880126891039124e-05, "loss": 4.7355, "step": 5585 }, { "epoch": 1.28, "learning_rate": 6.86110240486457e-05, "loss": 4.7362, "step": 5590 }, { "epoch": 1.28, "learning_rate": 6.842090512672358e-05, "loss": 4.7741, "step": 5595 }, { "epoch": 1.29, "learning_rate": 6.823091290742602e-05, "loss": 4.7723, "step": 5600 }, { "epoch": 1.29, "learning_rate": 6.804104815304566e-05, "loss": 4.7569, "step": 5605 }, { "epoch": 1.29, "learning_rate": 6.785131162536374e-05, "loss": 4.7718, "step": 5610 }, { "epoch": 1.29, "learning_rate": 6.766170408564705e-05, "loss": 4.7562, "step": 5615 }, { "epoch": 1.29, "learning_rate": 6.747222629464484e-05, "loss": 4.7369, "step": 5620 }, { "epoch": 1.29, "learning_rate": 6.728287901258581e-05, "loss": 4.7388, "step": 5625 }, { "epoch": 1.29, "learning_rate": 6.709366299917497e-05, "loss": 4.7534, "step": 5630 }, { "epoch": 1.29, "learning_rate": 6.690457901359073e-05, "loss": 4.7534, "step": 5635 }, { "epoch": 1.29, "learning_rate": 6.671562781448166e-05, "loss": 4.6859, "step": 5640 }, { "epoch": 1.3, "learning_rate": 6.652681015996369e-05, "loss": 4.738, "step": 5645 }, { "epoch": 1.3, "learning_rate": 6.633812680761684e-05, "loss": 4.7085, "step": 5650 }, { "epoch": 1.3, "learning_rate": 6.614957851448227e-05, "loss": 4.7201, "step": 5655 }, { "epoch": 1.3, "learning_rate": 6.59611660370594e-05, "loss": 4.7149, "step": 5660 }, { "epoch": 1.3, "learning_rate": 6.577289013130252e-05, "loss": 4.7387, "step": 5665 }, { "epoch": 1.3, "learning_rate": 6.558475155261811e-05, "loss": 4.6589, "step": 5670 }, { "epoch": 1.3, "learning_rate": 6.539675105586158e-05, "loss": 4.735, "step": 5675 }, { "epoch": 1.3, "learning_rate": 6.52088893953344e-05, "loss": 4.7227, "step": 5680 }, { "epoch": 1.3, "learning_rate": 6.502116732478096e-05, "loss": 4.7259, "step": 5685 }, { "epoch": 1.31, "learning_rate": 6.48335855973855e-05, "loss": 4.687, "step": 5690 }, { "epoch": 1.31, "learning_rate": 6.464614496576935e-05, "loss": 4.7622, "step": 5695 }, { "epoch": 1.31, "learning_rate": 6.445884618198754e-05, "loss": 4.6623, "step": 5700 }, { "epoch": 1.31, "learning_rate": 6.427168999752614e-05, "loss": 4.738, "step": 5705 }, { "epoch": 1.31, "learning_rate": 6.408467716329894e-05, "loss": 4.7262, "step": 5710 }, { "epoch": 1.31, "learning_rate": 6.389780842964468e-05, "loss": 4.7594, "step": 5715 }, { "epoch": 1.31, "learning_rate": 6.371108454632391e-05, "loss": 4.7301, "step": 5720 }, { "epoch": 1.31, "learning_rate": 6.352450626251587e-05, "loss": 4.739, "step": 5725 }, { "epoch": 1.31, "learning_rate": 6.33380743268159e-05, "loss": 4.7733, "step": 5730 }, { "epoch": 1.32, "learning_rate": 6.315178948723186e-05, "loss": 4.7411, "step": 5735 }, { "epoch": 1.32, "learning_rate": 6.29656524911817e-05, "loss": 4.7295, "step": 5740 }, { "epoch": 1.32, "learning_rate": 6.277966408548992e-05, "loss": 4.7167, "step": 5745 }, { "epoch": 1.32, "learning_rate": 6.259382501638509e-05, "loss": 4.7173, "step": 5750 }, { "epoch": 1.32, "learning_rate": 6.240813602949641e-05, "loss": 4.7548, "step": 5755 }, { "epoch": 1.32, "learning_rate": 6.222259786985101e-05, "loss": 4.7193, "step": 5760 }, { "epoch": 1.32, "learning_rate": 6.20372112818709e-05, "loss": 4.7211, "step": 5765 }, { "epoch": 1.32, "learning_rate": 6.185197700936982e-05, "loss": 4.6844, "step": 5770 }, { "epoch": 1.33, "learning_rate": 6.166689579555053e-05, "loss": 4.705, "step": 5775 }, { "epoch": 1.33, "learning_rate": 6.14819683830016e-05, "loss": 4.6324, "step": 5780 }, { "epoch": 1.33, "learning_rate": 6.129719551369456e-05, "loss": 4.7002, "step": 5785 }, { "epoch": 1.33, "learning_rate": 6.111257792898082e-05, "loss": 4.7011, "step": 5790 }, { "epoch": 1.33, "learning_rate": 6.092811636958877e-05, "loss": 4.7764, "step": 5795 }, { "epoch": 1.33, "learning_rate": 6.0743811575620846e-05, "loss": 4.7315, "step": 5800 }, { "epoch": 1.33, "learning_rate": 6.055966428655042e-05, "loss": 4.7281, "step": 5805 }, { "epoch": 1.33, "learning_rate": 6.0375675241219e-05, "loss": 4.7234, "step": 5810 }, { "epoch": 1.33, "learning_rate": 6.019184517783311e-05, "loss": 4.6957, "step": 5815 }, { "epoch": 1.34, "learning_rate": 6.000817483396148e-05, "loss": 4.6906, "step": 5820 }, { "epoch": 1.34, "learning_rate": 5.982466494653187e-05, "loss": 4.7133, "step": 5825 }, { "epoch": 1.34, "learning_rate": 5.96413162518285e-05, "loss": 4.734, "step": 5830 }, { "epoch": 1.34, "learning_rate": 5.945812948548855e-05, "loss": 4.7603, "step": 5835 }, { "epoch": 1.34, "learning_rate": 5.9275105382499694e-05, "loss": 4.7241, "step": 5840 }, { "epoch": 1.34, "learning_rate": 5.909224467719694e-05, "loss": 4.7101, "step": 5845 }, { "epoch": 1.34, "learning_rate": 5.890954810325966e-05, "loss": 4.7671, "step": 5850 }, { "epoch": 1.34, "learning_rate": 5.8727016393708746e-05, "loss": 4.6897, "step": 5855 }, { "epoch": 1.34, "learning_rate": 5.854465028090355e-05, "loss": 4.7117, "step": 5860 }, { "epoch": 1.35, "learning_rate": 5.836245049653908e-05, "loss": 4.6882, "step": 5865 }, { "epoch": 1.35, "learning_rate": 5.8180417771643006e-05, "loss": 4.7032, "step": 5870 }, { "epoch": 1.35, "learning_rate": 5.799855283657254e-05, "loss": 4.7798, "step": 5875 }, { "epoch": 1.35, "learning_rate": 5.781685642101196e-05, "loss": 4.7255, "step": 5880 }, { "epoch": 1.35, "learning_rate": 5.7635329253969195e-05, "loss": 4.684, "step": 5885 }, { "epoch": 1.35, "learning_rate": 5.7453972063773184e-05, "loss": 4.7053, "step": 5890 }, { "epoch": 1.35, "learning_rate": 5.727278557807085e-05, "loss": 4.6871, "step": 5895 }, { "epoch": 1.35, "learning_rate": 5.7091770523824317e-05, "loss": 4.6497, "step": 5900 }, { "epoch": 1.36, "learning_rate": 5.691092762730774e-05, "loss": 4.7039, "step": 5905 }, { "epoch": 1.36, "learning_rate": 5.673025761410462e-05, "loss": 4.6889, "step": 5910 }, { "epoch": 1.36, "learning_rate": 5.654976120910478e-05, "loss": 4.7159, "step": 5915 }, { "epoch": 1.36, "learning_rate": 5.636943913650147e-05, "loss": 4.6792, "step": 5920 }, { "epoch": 1.36, "learning_rate": 5.618929211978857e-05, "loss": 4.692, "step": 5925 }, { "epoch": 1.36, "learning_rate": 5.60093208817575e-05, "loss": 4.6622, "step": 5930 }, { "epoch": 1.36, "learning_rate": 5.582952614449445e-05, "loss": 4.73, "step": 5935 }, { "epoch": 1.36, "learning_rate": 5.564990862937744e-05, "loss": 4.6451, "step": 5940 }, { "epoch": 1.36, "learning_rate": 5.5470469057073404e-05, "loss": 4.6663, "step": 5945 }, { "epoch": 1.37, "learning_rate": 5.5291208147535466e-05, "loss": 4.7326, "step": 5950 }, { "epoch": 1.37, "learning_rate": 5.511212661999967e-05, "loss": 4.6735, "step": 5955 }, { "epoch": 1.37, "learning_rate": 5.4933225192982586e-05, "loss": 4.6278, "step": 5960 }, { "epoch": 1.37, "learning_rate": 5.4754504584278e-05, "loss": 4.6864, "step": 5965 }, { "epoch": 1.37, "learning_rate": 5.457596551095441e-05, "loss": 4.7086, "step": 5970 }, { "epoch": 1.37, "learning_rate": 5.4397608689351656e-05, "loss": 4.7157, "step": 5975 }, { "epoch": 1.37, "learning_rate": 5.421943483507863e-05, "loss": 4.6579, "step": 5980 }, { "epoch": 1.37, "learning_rate": 5.404144466300998e-05, "loss": 4.6555, "step": 5985 }, { "epoch": 1.37, "learning_rate": 5.3863638887283364e-05, "loss": 4.687, "step": 5990 }, { "epoch": 1.38, "learning_rate": 5.3686018221296665e-05, "loss": 4.6402, "step": 5995 }, { "epoch": 1.38, "learning_rate": 5.350858337770498e-05, "loss": 4.6733, "step": 6000 }, { "epoch": 1.38, "learning_rate": 5.333133506841797e-05, "loss": 4.6318, "step": 6005 }, { "epoch": 1.38, "learning_rate": 5.315427400459678e-05, "loss": 4.6787, "step": 6010 }, { "epoch": 1.38, "learning_rate": 5.297740089665129e-05, "loss": 4.683, "step": 6015 }, { "epoch": 1.38, "learning_rate": 5.280071645423726e-05, "loss": 4.6409, "step": 6020 }, { "epoch": 1.38, "learning_rate": 5.262422138625349e-05, "loss": 4.6501, "step": 6025 }, { "epoch": 1.38, "learning_rate": 5.244791640083906e-05, "loss": 4.6639, "step": 6030 }, { "epoch": 1.38, "learning_rate": 5.227180220537016e-05, "loss": 4.6952, "step": 6035 }, { "epoch": 1.39, "learning_rate": 5.2095879506457736e-05, "loss": 4.7301, "step": 6040 }, { "epoch": 1.39, "learning_rate": 5.192014900994423e-05, "loss": 4.7045, "step": 6045 }, { "epoch": 1.39, "learning_rate": 5.174461142090111e-05, "loss": 4.6886, "step": 6050 }, { "epoch": 1.39, "learning_rate": 5.156926744362562e-05, "loss": 4.6499, "step": 6055 }, { "epoch": 1.39, "learning_rate": 5.139411778163832e-05, "loss": 4.6756, "step": 6060 }, { "epoch": 1.39, "learning_rate": 5.121916313768018e-05, "loss": 4.6674, "step": 6065 }, { "epoch": 1.39, "learning_rate": 5.104440421370962e-05, "loss": 4.6264, "step": 6070 }, { "epoch": 1.39, "learning_rate": 5.0869841710899815e-05, "loss": 4.6938, "step": 6075 }, { "epoch": 1.4, "learning_rate": 5.0695476329635825e-05, "loss": 4.6492, "step": 6080 }, { "epoch": 1.4, "learning_rate": 5.052130876951192e-05, "loss": 4.6972, "step": 6085 }, { "epoch": 1.4, "learning_rate": 5.034733972932855e-05, "loss": 4.6605, "step": 6090 }, { "epoch": 1.4, "learning_rate": 5.017356990708969e-05, "loss": 4.668, "step": 6095 }, { "epoch": 1.4, "learning_rate": 5.000000000000002e-05, "loss": 4.623, "step": 6100 }, { "epoch": 1.4, "learning_rate": 4.982663070446207e-05, "loss": 4.6858, "step": 6105 }, { "epoch": 1.4, "learning_rate": 4.9653462716073594e-05, "loss": 4.6676, "step": 6110 }, { "epoch": 1.4, "learning_rate": 4.9480496729624515e-05, "loss": 4.6803, "step": 6115 }, { "epoch": 1.4, "learning_rate": 4.930773343909434e-05, "loss": 4.6284, "step": 6120 }, { "epoch": 1.41, "learning_rate": 4.91351735376493e-05, "loss": 4.6984, "step": 6125 }, { "epoch": 1.41, "learning_rate": 4.8962817717639555e-05, "loss": 4.6631, "step": 6130 }, { "epoch": 1.41, "learning_rate": 4.879066667059659e-05, "loss": 4.6599, "step": 6135 }, { "epoch": 1.41, "learning_rate": 4.8618721087230014e-05, "loss": 4.6954, "step": 6140 }, { "epoch": 1.41, "learning_rate": 4.844698165742536e-05, "loss": 4.6713, "step": 6145 }, { "epoch": 1.41, "learning_rate": 4.8275449070240854e-05, "loss": 4.6837, "step": 6150 }, { "epoch": 1.41, "learning_rate": 4.810412401390487e-05, "loss": 4.7168, "step": 6155 }, { "epoch": 1.41, "learning_rate": 4.793300717581308e-05, "loss": 4.632, "step": 6160 }, { "epoch": 1.41, "learning_rate": 4.7762099242525847e-05, "loss": 4.6783, "step": 6165 }, { "epoch": 1.42, "learning_rate": 4.7591400899765234e-05, "loss": 4.6229, "step": 6170 }, { "epoch": 1.42, "learning_rate": 4.7420912832412445e-05, "loss": 4.6727, "step": 6175 }, { "epoch": 1.42, "learning_rate": 4.7250635724505e-05, "loss": 4.655, "step": 6180 }, { "epoch": 1.42, "learning_rate": 4.7080570259233935e-05, "loss": 4.6868, "step": 6185 }, { "epoch": 1.42, "learning_rate": 4.6910717118941286e-05, "loss": 4.6991, "step": 6190 }, { "epoch": 1.42, "learning_rate": 4.674107698511704e-05, "loss": 4.6705, "step": 6195 }, { "epoch": 1.42, "learning_rate": 4.6571650538396615e-05, "loss": 4.6316, "step": 6200 }, { "epoch": 1.42, "learning_rate": 4.640243845855806e-05, "loss": 4.6259, "step": 6205 }, { "epoch": 1.43, "learning_rate": 4.6233441424519295e-05, "loss": 4.717, "step": 6210 }, { "epoch": 1.43, "learning_rate": 4.606466011433557e-05, "loss": 4.6154, "step": 6215 }, { "epoch": 1.43, "learning_rate": 4.5896095205196356e-05, "loss": 4.6612, "step": 6220 }, { "epoch": 1.43, "learning_rate": 4.5727747373423116e-05, "loss": 4.6936, "step": 6225 }, { "epoch": 1.43, "learning_rate": 4.5559617294466176e-05, "loss": 4.6148, "step": 6230 }, { "epoch": 1.43, "learning_rate": 4.539170564290237e-05, "loss": 4.6156, "step": 6235 }, { "epoch": 1.43, "learning_rate": 4.522401309243193e-05, "loss": 4.6261, "step": 6240 }, { "epoch": 1.43, "learning_rate": 4.50565403158761e-05, "loss": 4.699, "step": 6245 }, { "epoch": 1.43, "learning_rate": 4.488928798517442e-05, "loss": 4.6464, "step": 6250 }, { "epoch": 1.44, "learning_rate": 4.472225677138186e-05, "loss": 4.6681, "step": 6255 }, { "epoch": 1.44, "learning_rate": 4.455544734466624e-05, "loss": 4.6342, "step": 6260 }, { "epoch": 1.44, "learning_rate": 4.4388860374305474e-05, "loss": 4.6214, "step": 6265 }, { "epoch": 1.44, "learning_rate": 4.422249652868506e-05, "loss": 4.6874, "step": 6270 }, { "epoch": 1.44, "learning_rate": 4.405635647529514e-05, "loss": 4.6813, "step": 6275 }, { "epoch": 1.44, "learning_rate": 4.389044088072798e-05, "loss": 4.6827, "step": 6280 }, { "epoch": 1.44, "learning_rate": 4.3724750410675287e-05, "loss": 4.6237, "step": 6285 }, { "epoch": 1.44, "learning_rate": 4.355928572992547e-05, "loss": 4.7075, "step": 6290 }, { "epoch": 1.44, "learning_rate": 4.339404750236117e-05, "loss": 4.6709, "step": 6295 }, { "epoch": 1.45, "learning_rate": 4.322903639095619e-05, "loss": 4.6309, "step": 6300 }, { "epoch": 1.45, "learning_rate": 4.306425305777333e-05, "loss": 4.6, "step": 6305 }, { "epoch": 1.45, "learning_rate": 4.289969816396132e-05, "loss": 4.6527, "step": 6310 }, { "epoch": 1.45, "learning_rate": 4.2735372369752535e-05, "loss": 4.6906, "step": 6315 }, { "epoch": 1.45, "learning_rate": 4.2571276334459895e-05, "loss": 4.6251, "step": 6320 }, { "epoch": 1.45, "learning_rate": 4.240741071647464e-05, "loss": 4.6286, "step": 6325 }, { "epoch": 1.45, "learning_rate": 4.224377617326353e-05, "loss": 4.6605, "step": 6330 }, { "epoch": 1.45, "learning_rate": 4.208037336136612e-05, "loss": 4.6472, "step": 6335 }, { "epoch": 1.45, "learning_rate": 4.1917202936392265e-05, "loss": 4.6194, "step": 6340 }, { "epoch": 1.46, "learning_rate": 4.175426555301937e-05, "loss": 4.6127, "step": 6345 }, { "epoch": 1.46, "learning_rate": 4.15915618649899e-05, "loss": 4.6649, "step": 6350 }, { "epoch": 1.46, "learning_rate": 4.142909252510866e-05, "loss": 4.6137, "step": 6355 }, { "epoch": 1.46, "learning_rate": 4.126685818524013e-05, "loss": 4.6434, "step": 6360 }, { "epoch": 1.46, "learning_rate": 4.1104859496305994e-05, "loss": 4.6443, "step": 6365 }, { "epoch": 1.46, "learning_rate": 4.094309710828236e-05, "loss": 4.6514, "step": 6370 }, { "epoch": 1.46, "learning_rate": 4.078157167019738e-05, "loss": 4.6482, "step": 6375 }, { "epoch": 1.46, "learning_rate": 4.0620283830128414e-05, "loss": 4.6226, "step": 6380 }, { "epoch": 1.47, "learning_rate": 4.0459234235199515e-05, "loss": 4.6699, "step": 6385 }, { "epoch": 1.47, "learning_rate": 4.029842353157888e-05, "loss": 4.6841, "step": 6390 }, { "epoch": 1.47, "learning_rate": 4.013785236447616e-05, "loss": 4.5908, "step": 6395 }, { "epoch": 1.47, "learning_rate": 3.9977521378140084e-05, "loss": 4.6057, "step": 6400 }, { "epoch": 1.47, "learning_rate": 3.981743121585547e-05, "loss": 4.623, "step": 6405 }, { "epoch": 1.47, "learning_rate": 3.965758251994115e-05, "loss": 4.6461, "step": 6410 }, { "epoch": 1.47, "learning_rate": 3.949797593174698e-05, "loss": 4.6257, "step": 6415 }, { "epoch": 1.47, "learning_rate": 3.933861209165146e-05, "loss": 4.626, "step": 6420 }, { "epoch": 1.47, "learning_rate": 3.917949163905914e-05, "loss": 4.6798, "step": 6425 }, { "epoch": 1.48, "learning_rate": 3.9020615212398016e-05, "loss": 4.6667, "step": 6430 }, { "epoch": 1.48, "learning_rate": 3.886198344911707e-05, "loss": 4.64, "step": 6435 }, { "epoch": 1.48, "learning_rate": 3.8703596985683556e-05, "loss": 4.5999, "step": 6440 }, { "epoch": 1.48, "learning_rate": 3.8545456457580566e-05, "loss": 4.6126, "step": 6445 }, { "epoch": 1.48, "learning_rate": 3.838756249930439e-05, "loss": 4.6735, "step": 6450 }, { "epoch": 1.48, "learning_rate": 3.822991574436213e-05, "loss": 4.661, "step": 6455 }, { "epoch": 1.48, "learning_rate": 3.807251682526902e-05, "loss": 4.6394, "step": 6460 }, { "epoch": 1.48, "learning_rate": 3.791536637354576e-05, "loss": 4.593, "step": 6465 }, { "epoch": 1.48, "learning_rate": 3.775846501971636e-05, "loss": 4.5347, "step": 6470 }, { "epoch": 1.49, "learning_rate": 3.760181339330526e-05, "loss": 4.5798, "step": 6475 }, { "epoch": 1.49, "learning_rate": 3.7445412122835077e-05, "loss": 4.6295, "step": 6480 }, { "epoch": 1.49, "learning_rate": 3.7289261835823695e-05, "loss": 4.6271, "step": 6485 }, { "epoch": 1.49, "learning_rate": 3.713336315878224e-05, "loss": 4.6211, "step": 6490 }, { "epoch": 1.49, "learning_rate": 3.6977716717212165e-05, "loss": 4.6093, "step": 6495 }, { "epoch": 1.49, "learning_rate": 3.6822323135603054e-05, "loss": 4.6382, "step": 6500 }, { "epoch": 1.49, "learning_rate": 3.6667183037429756e-05, "loss": 4.6428, "step": 6505 }, { "epoch": 1.49, "learning_rate": 3.651229704515018e-05, "loss": 4.6137, "step": 6510 }, { "epoch": 1.5, "learning_rate": 3.635766578020279e-05, "loss": 4.5749, "step": 6515 }, { "epoch": 1.5, "learning_rate": 3.6203289863003905e-05, "loss": 4.5919, "step": 6520 }, { "epoch": 1.5, "learning_rate": 3.604916991294537e-05, "loss": 4.6232, "step": 6525 }, { "epoch": 1.5, "learning_rate": 3.5895306548392005e-05, "loss": 4.6407, "step": 6530 }, { "epoch": 1.5, "learning_rate": 3.574170038667923e-05, "loss": 4.5777, "step": 6535 }, { "epoch": 1.5, "learning_rate": 3.558835204411044e-05, "loss": 4.6254, "step": 6540 }, { "epoch": 1.5, "learning_rate": 3.543526213595461e-05, "loss": 4.6482, "step": 6545 }, { "epoch": 1.5, "learning_rate": 3.52824312764438e-05, "loss": 4.615, "step": 6550 }, { "epoch": 1.5, "learning_rate": 3.512986007877072e-05, "loss": 4.646, "step": 6555 }, { "epoch": 1.51, "learning_rate": 3.497754915508632e-05, "loss": 4.6678, "step": 6560 }, { "epoch": 1.51, "learning_rate": 3.4825499116497176e-05, "loss": 4.6985, "step": 6565 }, { "epoch": 1.51, "learning_rate": 3.467371057306318e-05, "loss": 4.6124, "step": 6570 }, { "epoch": 1.51, "learning_rate": 3.452218413379504e-05, "loss": 4.6164, "step": 6575 }, { "epoch": 1.51, "learning_rate": 3.437092040665183e-05, "loss": 4.6304, "step": 6580 }, { "epoch": 1.51, "learning_rate": 3.421991999853857e-05, "loss": 4.625, "step": 6585 }, { "epoch": 1.51, "learning_rate": 3.406918351530376e-05, "loss": 4.5795, "step": 6590 }, { "epoch": 1.51, "learning_rate": 3.3918711561737046e-05, "loss": 4.6565, "step": 6595 }, { "epoch": 1.51, "learning_rate": 3.3768504741566664e-05, "loss": 4.5725, "step": 6600 }, { "epoch": 1.52, "learning_rate": 3.361856365745705e-05, "loss": 4.5355, "step": 6605 }, { "epoch": 1.52, "learning_rate": 3.346888891100649e-05, "loss": 4.62, "step": 6610 }, { "epoch": 1.52, "learning_rate": 3.331948110274462e-05, "loss": 4.6074, "step": 6615 }, { "epoch": 1.52, "learning_rate": 3.3170340832130134e-05, "loss": 4.6097, "step": 6620 }, { "epoch": 1.52, "learning_rate": 3.302146869754823e-05, "loss": 4.6278, "step": 6625 }, { "epoch": 1.52, "learning_rate": 3.287286529630832e-05, "loss": 4.6484, "step": 6630 }, { "epoch": 1.52, "learning_rate": 3.272453122464152e-05, "loss": 4.5728, "step": 6635 }, { "epoch": 1.52, "learning_rate": 3.2576467077698493e-05, "loss": 4.6126, "step": 6640 }, { "epoch": 1.52, "learning_rate": 3.242867344954674e-05, "loss": 4.6321, "step": 6645 }, { "epoch": 1.53, "learning_rate": 3.228115093316848e-05, "loss": 4.627, "step": 6650 }, { "epoch": 1.53, "learning_rate": 3.213390012045811e-05, "loss": 4.6281, "step": 6655 }, { "epoch": 1.53, "learning_rate": 3.198692160221987e-05, "loss": 4.5743, "step": 6660 }, { "epoch": 1.53, "learning_rate": 3.184021596816563e-05, "loss": 4.5734, "step": 6665 }, { "epoch": 1.53, "learning_rate": 3.169378380691218e-05, "loss": 4.5622, "step": 6670 }, { "epoch": 1.53, "learning_rate": 3.1547625705979265e-05, "loss": 4.5935, "step": 6675 }, { "epoch": 1.53, "learning_rate": 3.140174225178692e-05, "loss": 4.612, "step": 6680 }, { "epoch": 1.53, "learning_rate": 3.1256134029653275e-05, "loss": 4.615, "step": 6685 }, { "epoch": 1.54, "learning_rate": 3.111080162379215e-05, "loss": 4.6019, "step": 6690 }, { "epoch": 1.54, "learning_rate": 3.096574561731072e-05, "loss": 4.5816, "step": 6695 }, { "epoch": 1.54, "learning_rate": 3.082096659220722e-05, "loss": 4.5663, "step": 6700 }, { "epoch": 1.54, "learning_rate": 3.0676465129368556e-05, "loss": 4.5518, "step": 6705 }, { "epoch": 1.54, "learning_rate": 3.0532241808567966e-05, "loss": 4.6307, "step": 6710 }, { "epoch": 1.54, "learning_rate": 3.0388297208462703e-05, "loss": 4.6792, "step": 6715 }, { "epoch": 1.54, "learning_rate": 3.0244631906591825e-05, "loss": 4.5948, "step": 6720 }, { "epoch": 1.54, "learning_rate": 3.01012464793737e-05, "loss": 4.5964, "step": 6725 }, { "epoch": 1.54, "learning_rate": 2.9958141502103722e-05, "loss": 4.6444, "step": 6730 }, { "epoch": 1.55, "learning_rate": 2.9815317548952192e-05, "loss": 4.5379, "step": 6735 }, { "epoch": 1.55, "learning_rate": 2.9672775192961756e-05, "loss": 4.5627, "step": 6740 }, { "epoch": 1.55, "learning_rate": 2.9530515006045368e-05, "loss": 4.6281, "step": 6745 }, { "epoch": 1.55, "learning_rate": 2.938853755898364e-05, "loss": 4.5695, "step": 6750 }, { "epoch": 1.55, "learning_rate": 2.9246843421422998e-05, "loss": 4.6154, "step": 6755 }, { "epoch": 1.55, "learning_rate": 2.910543316187301e-05, "loss": 4.5986, "step": 6760 }, { "epoch": 1.55, "learning_rate": 2.896430734770431e-05, "loss": 4.5794, "step": 6765 }, { "epoch": 1.55, "learning_rate": 2.882346654514627e-05, "loss": 4.5925, "step": 6770 }, { "epoch": 1.55, "learning_rate": 2.8682911319284712e-05, "loss": 4.5376, "step": 6775 }, { "epoch": 1.56, "learning_rate": 2.8542642234059725e-05, "loss": 4.6149, "step": 6780 }, { "epoch": 1.56, "learning_rate": 2.8402659852263257e-05, "loss": 4.5777, "step": 6785 }, { "epoch": 1.56, "learning_rate": 2.826296473553697e-05, "loss": 4.532, "step": 6790 }, { "epoch": 1.56, "learning_rate": 2.812355744436993e-05, "loss": 4.5913, "step": 6795 }, { "epoch": 1.56, "learning_rate": 2.7984438538096392e-05, "loss": 4.5839, "step": 6800 }, { "epoch": 1.56, "learning_rate": 2.784560857489358e-05, "loss": 4.6034, "step": 6805 }, { "epoch": 1.56, "learning_rate": 2.7707068111779377e-05, "loss": 4.501, "step": 6810 }, { "epoch": 1.56, "learning_rate": 2.7568817704610116e-05, "loss": 4.5865, "step": 6815 }, { "epoch": 1.57, "learning_rate": 2.7430857908078345e-05, "loss": 4.5889, "step": 6820 }, { "epoch": 1.57, "learning_rate": 2.7293189275710706e-05, "loss": 4.5401, "step": 6825 }, { "epoch": 1.57, "learning_rate": 2.7155812359865517e-05, "loss": 4.595, "step": 6830 }, { "epoch": 1.57, "learning_rate": 2.7018727711730706e-05, "loss": 4.5628, "step": 6835 }, { "epoch": 1.57, "learning_rate": 2.6881935881321563e-05, "loss": 4.5191, "step": 6840 }, { "epoch": 1.57, "learning_rate": 2.6745437417478502e-05, "loss": 4.5842, "step": 6845 }, { "epoch": 1.57, "learning_rate": 2.6609232867864896e-05, "loss": 4.6019, "step": 6850 }, { "epoch": 1.57, "learning_rate": 2.6473322778964847e-05, "loss": 4.5637, "step": 6855 }, { "epoch": 1.57, "learning_rate": 2.6337707696081094e-05, "loss": 4.5541, "step": 6860 }, { "epoch": 1.58, "learning_rate": 2.6202388163332637e-05, "loss": 4.6075, "step": 6865 }, { "epoch": 1.58, "learning_rate": 2.606736472365272e-05, "loss": 4.5411, "step": 6870 }, { "epoch": 1.58, "learning_rate": 2.5932637918786563e-05, "loss": 4.5657, "step": 6875 }, { "epoch": 1.58, "learning_rate": 2.5798208289289204e-05, "loss": 4.5503, "step": 6880 }, { "epoch": 1.58, "learning_rate": 2.566407637452345e-05, "loss": 4.4874, "step": 6885 }, { "epoch": 1.58, "learning_rate": 2.5530242712657492e-05, "loss": 4.5598, "step": 6890 }, { "epoch": 1.58, "learning_rate": 2.5396707840662903e-05, "loss": 4.6072, "step": 6895 }, { "epoch": 1.58, "learning_rate": 2.526347229431242e-05, "loss": 4.544, "step": 6900 }, { "epoch": 1.58, "learning_rate": 2.513053660817788e-05, "loss": 4.6272, "step": 6905 }, { "epoch": 1.59, "learning_rate": 2.499790131562797e-05, "loss": 4.5902, "step": 6910 }, { "epoch": 1.59, "learning_rate": 2.4865566948826048e-05, "loss": 4.5739, "step": 6915 }, { "epoch": 1.59, "learning_rate": 2.4733534038728257e-05, "loss": 4.6072, "step": 6920 }, { "epoch": 1.59, "learning_rate": 2.4601803115081068e-05, "loss": 4.6311, "step": 6925 }, { "epoch": 1.59, "learning_rate": 2.4470374706419485e-05, "loss": 4.558, "step": 6930 }, { "epoch": 1.59, "learning_rate": 2.4339249340064507e-05, "loss": 4.634, "step": 6935 }, { "epoch": 1.59, "learning_rate": 2.4208427542121504e-05, "loss": 4.5723, "step": 6940 }, { "epoch": 1.59, "learning_rate": 2.407790983747773e-05, "loss": 4.5794, "step": 6945 }, { "epoch": 1.59, "learning_rate": 2.394769674980035e-05, "loss": 4.5407, "step": 6950 }, { "epoch": 1.6, "learning_rate": 2.3817788801534367e-05, "loss": 4.5956, "step": 6955 }, { "epoch": 1.6, "learning_rate": 2.3688186513900455e-05, "loss": 4.6563, "step": 6960 }, { "epoch": 1.6, "learning_rate": 2.3558890406892986e-05, "loss": 4.6067, "step": 6965 }, { "epoch": 1.6, "learning_rate": 2.34299009992778e-05, "loss": 4.5887, "step": 6970 }, { "epoch": 1.6, "learning_rate": 2.3301218808590176e-05, "loss": 4.5513, "step": 6975 }, { "epoch": 1.6, "learning_rate": 2.317284435113278e-05, "loss": 4.5791, "step": 6980 }, { "epoch": 1.6, "learning_rate": 2.3044778141973655e-05, "loss": 4.5854, "step": 6985 }, { "epoch": 1.6, "learning_rate": 2.2917020694944023e-05, "loss": 4.5818, "step": 6990 }, { "epoch": 1.61, "learning_rate": 2.278957252263617e-05, "loss": 4.5703, "step": 6995 }, { "epoch": 1.61, "learning_rate": 2.2662434136401722e-05, "loss": 4.5608, "step": 7000 }, { "epoch": 1.61, "learning_rate": 2.2535606046349177e-05, "loss": 4.5799, "step": 7005 }, { "epoch": 1.61, "learning_rate": 2.2409088761342235e-05, "loss": 4.578, "step": 7010 }, { "epoch": 1.61, "learning_rate": 2.228288278899735e-05, "loss": 4.6076, "step": 7015 }, { "epoch": 1.61, "learning_rate": 2.215698863568213e-05, "loss": 4.5601, "step": 7020 }, { "epoch": 1.61, "learning_rate": 2.203140680651298e-05, "loss": 4.6048, "step": 7025 }, { "epoch": 1.61, "learning_rate": 2.1906137805353212e-05, "loss": 4.5915, "step": 7030 }, { "epoch": 1.61, "learning_rate": 2.1781182134810997e-05, "loss": 4.5576, "step": 7035 }, { "epoch": 1.62, "learning_rate": 2.1656540296237316e-05, "loss": 4.5468, "step": 7040 }, { "epoch": 1.62, "learning_rate": 2.1532212789724094e-05, "loss": 4.5916, "step": 7045 }, { "epoch": 1.62, "learning_rate": 2.1408200114101985e-05, "loss": 4.5717, "step": 7050 }, { "epoch": 1.62, "learning_rate": 2.1284502766938475e-05, "loss": 4.6184, "step": 7055 }, { "epoch": 1.62, "learning_rate": 2.116112124453592e-05, "loss": 4.6147, "step": 7060 }, { "epoch": 1.62, "learning_rate": 2.1038056041929456e-05, "loss": 4.5336, "step": 7065 }, { "epoch": 1.62, "learning_rate": 2.0915307652885164e-05, "loss": 4.563, "step": 7070 }, { "epoch": 1.62, "learning_rate": 2.0792876569897912e-05, "loss": 4.5645, "step": 7075 }, { "epoch": 1.62, "learning_rate": 2.067076328418949e-05, "loss": 4.5716, "step": 7080 }, { "epoch": 1.63, "learning_rate": 2.0548968285706593e-05, "loss": 4.544, "step": 7085 }, { "epoch": 1.63, "learning_rate": 2.0427492063118935e-05, "loss": 4.5293, "step": 7090 }, { "epoch": 1.63, "learning_rate": 2.0306335103817208e-05, "loss": 4.5446, "step": 7095 }, { "epoch": 1.63, "learning_rate": 2.018549789391102e-05, "loss": 4.6236, "step": 7100 }, { "epoch": 1.63, "learning_rate": 2.006498091822726e-05, "loss": 4.5056, "step": 7105 }, { "epoch": 1.63, "learning_rate": 1.994478466030787e-05, "loss": 4.5405, "step": 7110 }, { "epoch": 1.63, "learning_rate": 1.982490960240798e-05, "loss": 4.584, "step": 7115 }, { "epoch": 1.63, "learning_rate": 1.970535622549401e-05, "loss": 4.509, "step": 7120 }, { "epoch": 1.64, "learning_rate": 1.9586125009241774e-05, "loss": 4.5719, "step": 7125 }, { "epoch": 1.64, "learning_rate": 1.946721643203443e-05, "loss": 4.5562, "step": 7130 }, { "epoch": 1.64, "learning_rate": 1.934863097096067e-05, "loss": 4.5957, "step": 7135 }, { "epoch": 1.64, "learning_rate": 1.923036910181275e-05, "loss": 4.5781, "step": 7140 }, { "epoch": 1.64, "learning_rate": 1.9112431299084598e-05, "loss": 4.5014, "step": 7145 }, { "epoch": 1.64, "learning_rate": 1.8994818035969975e-05, "loss": 4.573, "step": 7150 }, { "epoch": 1.64, "learning_rate": 1.8877529784360437e-05, "loss": 4.5279, "step": 7155 }, { "epoch": 1.64, "learning_rate": 1.8760567014843545e-05, "loss": 4.5326, "step": 7160 }, { "epoch": 1.64, "learning_rate": 1.864393019670092e-05, "loss": 4.55, "step": 7165 }, { "epoch": 1.65, "learning_rate": 1.8527619797906494e-05, "loss": 4.5847, "step": 7170 }, { "epoch": 1.65, "learning_rate": 1.8411636285124457e-05, "loss": 4.6046, "step": 7175 }, { "epoch": 1.65, "learning_rate": 1.8295980123707357e-05, "loss": 4.5281, "step": 7180 }, { "epoch": 1.65, "learning_rate": 1.8180651777694535e-05, "loss": 4.5239, "step": 7185 }, { "epoch": 1.65, "learning_rate": 1.8065651709809905e-05, "loss": 4.5677, "step": 7190 }, { "epoch": 1.65, "learning_rate": 1.795098038146038e-05, "loss": 4.5765, "step": 7195 }, { "epoch": 1.65, "learning_rate": 1.783663825273372e-05, "loss": 4.579, "step": 7200 }, { "epoch": 1.65, "learning_rate": 1.772262578239704e-05, "loss": 4.4754, "step": 7205 }, { "epoch": 1.65, "learning_rate": 1.7608943427894686e-05, "loss": 4.5839, "step": 7210 }, { "epoch": 1.66, "learning_rate": 1.7495591645346533e-05, "loss": 4.523, "step": 7215 }, { "epoch": 1.66, "learning_rate": 1.7382570889546124e-05, "loss": 4.5437, "step": 7220 }, { "epoch": 1.66, "learning_rate": 1.7269881613958805e-05, "loss": 4.5336, "step": 7225 }, { "epoch": 1.66, "learning_rate": 1.7157524270720036e-05, "loss": 4.5725, "step": 7230 }, { "epoch": 1.66, "learning_rate": 1.7045499310633428e-05, "loss": 4.5616, "step": 7235 }, { "epoch": 1.66, "learning_rate": 1.6933807183168994e-05, "loss": 4.5541, "step": 7240 }, { "epoch": 1.66, "learning_rate": 1.682244833646135e-05, "loss": 4.5655, "step": 7245 }, { "epoch": 1.66, "learning_rate": 1.6711423217307885e-05, "loss": 4.537, "step": 7250 }, { "epoch": 1.66, "learning_rate": 1.6600732271167098e-05, "loss": 4.5957, "step": 7255 }, { "epoch": 1.67, "learning_rate": 1.64903759421566e-05, "loss": 4.5155, "step": 7260 }, { "epoch": 1.67, "learning_rate": 1.638035467305148e-05, "loss": 4.5458, "step": 7265 }, { "epoch": 1.67, "learning_rate": 1.627066890528247e-05, "loss": 4.5722, "step": 7270 }, { "epoch": 1.67, "learning_rate": 1.6161319078934278e-05, "loss": 4.5482, "step": 7275 }, { "epoch": 1.67, "learning_rate": 1.6052305632743592e-05, "loss": 4.547, "step": 7280 }, { "epoch": 1.67, "learning_rate": 1.594362900409756e-05, "loss": 4.5308, "step": 7285 }, { "epoch": 1.67, "learning_rate": 1.583528962903197e-05, "loss": 4.5775, "step": 7290 }, { "epoch": 1.67, "learning_rate": 1.5727287942229387e-05, "loss": 4.6227, "step": 7295 }, { "epoch": 1.68, "learning_rate": 1.5619624377017537e-05, "loss": 4.5775, "step": 7300 }, { "epoch": 1.68, "learning_rate": 1.55122993653675e-05, "loss": 4.572, "step": 7305 }, { "epoch": 1.68, "learning_rate": 1.540531333789207e-05, "loss": 4.5922, "step": 7310 }, { "epoch": 1.68, "learning_rate": 1.5298666723843867e-05, "loss": 4.5424, "step": 7315 }, { "epoch": 1.68, "learning_rate": 1.5192359951113755e-05, "loss": 4.5955, "step": 7320 }, { "epoch": 1.68, "learning_rate": 1.5086393446229063e-05, "loss": 4.5597, "step": 7325 }, { "epoch": 1.68, "learning_rate": 1.4980767634351877e-05, "loss": 4.5101, "step": 7330 }, { "epoch": 1.68, "learning_rate": 1.4875482939277396e-05, "loss": 4.5899, "step": 7335 }, { "epoch": 1.68, "learning_rate": 1.4770539783432113e-05, "loss": 4.5811, "step": 7340 }, { "epoch": 1.69, "learning_rate": 1.466593858787223e-05, "loss": 4.5274, "step": 7345 }, { "epoch": 1.69, "learning_rate": 1.4561679772281877e-05, "loss": 4.5397, "step": 7350 }, { "epoch": 1.69, "learning_rate": 1.4457763754971553e-05, "loss": 4.5231, "step": 7355 }, { "epoch": 1.69, "learning_rate": 1.4354190952876334e-05, "loss": 4.5274, "step": 7360 }, { "epoch": 1.69, "learning_rate": 1.425096178155415e-05, "loss": 4.4804, "step": 7365 }, { "epoch": 1.69, "learning_rate": 1.4148076655184373e-05, "loss": 4.5091, "step": 7370 }, { "epoch": 1.69, "learning_rate": 1.404553598656585e-05, "loss": 4.5516, "step": 7375 }, { "epoch": 1.69, "learning_rate": 1.3943340187115494e-05, "loss": 4.5279, "step": 7380 }, { "epoch": 1.69, "learning_rate": 1.3841489666866369e-05, "loss": 4.5114, "step": 7385 }, { "epoch": 1.7, "learning_rate": 1.373998483446638e-05, "loss": 4.5324, "step": 7390 }, { "epoch": 1.7, "learning_rate": 1.3638826097176328e-05, "loss": 4.5319, "step": 7395 }, { "epoch": 1.7, "learning_rate": 1.3538013860868436e-05, "loss": 4.482, "step": 7400 }, { "epoch": 1.7, "learning_rate": 1.3437548530024691e-05, "loss": 4.5697, "step": 7405 }, { "epoch": 1.7, "learning_rate": 1.3337430507735205e-05, "loss": 4.5513, "step": 7410 }, { "epoch": 1.7, "learning_rate": 1.3237660195696633e-05, "loss": 4.5081, "step": 7415 }, { "epoch": 1.7, "learning_rate": 1.313823799421051e-05, "loss": 4.5395, "step": 7420 }, { "epoch": 1.7, "learning_rate": 1.3039164302181683e-05, "loss": 4.5547, "step": 7425 }, { "epoch": 1.71, "learning_rate": 1.2940439517116676e-05, "loss": 4.5383, "step": 7430 }, { "epoch": 1.71, "learning_rate": 1.2842064035122125e-05, "loss": 4.5505, "step": 7435 }, { "epoch": 1.71, "learning_rate": 1.2744038250903267e-05, "loss": 4.5571, "step": 7440 }, { "epoch": 1.71, "learning_rate": 1.264636255776208e-05, "loss": 4.551, "step": 7445 }, { "epoch": 1.71, "learning_rate": 1.2549037347596115e-05, "loss": 4.5631, "step": 7450 }, { "epoch": 1.71, "learning_rate": 1.245206301089652e-05, "loss": 4.5251, "step": 7455 }, { "epoch": 1.71, "learning_rate": 1.2355439936746827e-05, "loss": 4.5173, "step": 7460 }, { "epoch": 1.71, "learning_rate": 1.2259168512821062e-05, "loss": 4.4905, "step": 7465 }, { "epoch": 1.71, "learning_rate": 1.2163249125382426e-05, "loss": 4.5231, "step": 7470 }, { "epoch": 1.72, "learning_rate": 1.206768215928169e-05, "loss": 4.5016, "step": 7475 }, { "epoch": 1.72, "learning_rate": 1.1972467997955595e-05, "loss": 4.5838, "step": 7480 }, { "epoch": 1.72, "learning_rate": 1.187760702342534e-05, "loss": 4.5593, "step": 7485 }, { "epoch": 1.72, "learning_rate": 1.1783099616295056e-05, "loss": 4.5412, "step": 7490 }, { "epoch": 1.72, "learning_rate": 1.1688946155750347e-05, "loss": 4.5768, "step": 7495 }, { "epoch": 1.72, "learning_rate": 1.1595147019556607e-05, "loss": 4.5214, "step": 7500 }, { "epoch": 1.72, "learning_rate": 1.1501702584057661e-05, "loss": 4.5133, "step": 7505 }, { "epoch": 1.72, "learning_rate": 1.140861322417417e-05, "loss": 4.5332, "step": 7510 }, { "epoch": 1.72, "learning_rate": 1.1315879313402123e-05, "loss": 4.5513, "step": 7515 }, { "epoch": 1.73, "learning_rate": 1.1223501223811451e-05, "loss": 4.5954, "step": 7520 }, { "epoch": 1.73, "learning_rate": 1.1131479326044348e-05, "loss": 4.4921, "step": 7525 }, { "epoch": 1.73, "learning_rate": 1.1039813989313951e-05, "loss": 4.5349, "step": 7530 }, { "epoch": 1.73, "learning_rate": 1.0948505581402735e-05, "loss": 4.5111, "step": 7535 }, { "epoch": 1.73, "learning_rate": 1.085755446866119e-05, "loss": 4.5016, "step": 7540 }, { "epoch": 1.73, "learning_rate": 1.076696101600615e-05, "loss": 4.4748, "step": 7545 }, { "epoch": 1.73, "learning_rate": 1.0676725586919457e-05, "loss": 4.5399, "step": 7550 }, { "epoch": 1.73, "learning_rate": 1.0586848543446537e-05, "loss": 4.5744, "step": 7555 }, { "epoch": 1.73, "learning_rate": 1.0497330246194848e-05, "loss": 4.4877, "step": 7560 }, { "epoch": 1.74, "learning_rate": 1.0408171054332483e-05, "loss": 4.5206, "step": 7565 }, { "epoch": 1.74, "learning_rate": 1.031937132558668e-05, "loss": 4.5463, "step": 7570 }, { "epoch": 1.74, "learning_rate": 1.0230931416242518e-05, "loss": 4.553, "step": 7575 }, { "epoch": 1.74, "learning_rate": 1.014285168114133e-05, "loss": 4.4749, "step": 7580 }, { "epoch": 1.74, "learning_rate": 1.0055132473679363e-05, "loss": 4.4981, "step": 7585 }, { "epoch": 1.74, "learning_rate": 9.96777414580633e-06, "loss": 4.5109, "step": 7590 }, { "epoch": 1.74, "learning_rate": 9.880777048024014e-06, "loss": 4.5291, "step": 7595 }, { "epoch": 1.74, "learning_rate": 9.794141529384915e-06, "loss": 4.5821, "step": 7600 }, { "epoch": 1.75, "learning_rate": 9.707867937490722e-06, "loss": 4.5805, "step": 7605 }, { "epoch": 1.75, "learning_rate": 9.621956618491024e-06, "loss": 4.4649, "step": 7610 }, { "epoch": 1.75, "learning_rate": 9.536407917081869e-06, "loss": 4.5653, "step": 7615 }, { "epoch": 1.75, "learning_rate": 9.451222176504414e-06, "loss": 4.5206, "step": 7620 }, { "epoch": 1.75, "learning_rate": 9.366399738543574e-06, "loss": 4.5356, "step": 7625 }, { "epoch": 1.75, "learning_rate": 9.281940943526491e-06, "loss": 4.5063, "step": 7630 }, { "epoch": 1.75, "learning_rate": 9.197846130321419e-06, "loss": 4.5086, "step": 7635 }, { "epoch": 1.75, "learning_rate": 9.114115636336152e-06, "loss": 4.5602, "step": 7640 }, { "epoch": 1.75, "learning_rate": 9.030749797516825e-06, "loss": 4.4951, "step": 7645 }, { "epoch": 1.76, "learning_rate": 8.947748948346357e-06, "loss": 4.5339, "step": 7650 }, { "epoch": 1.76, "learning_rate": 8.865113421843407e-06, "loss": 4.4816, "step": 7655 }, { "epoch": 1.76, "learning_rate": 8.782843549560771e-06, "loss": 4.5468, "step": 7660 }, { "epoch": 1.76, "learning_rate": 8.700939661584184e-06, "loss": 4.5583, "step": 7665 }, { "epoch": 1.76, "learning_rate": 8.61940208653097e-06, "loss": 4.5727, "step": 7670 }, { "epoch": 1.76, "learning_rate": 8.538231151548693e-06, "loss": 4.4903, "step": 7675 }, { "epoch": 1.76, "learning_rate": 8.457427182313937e-06, "loss": 4.5253, "step": 7680 }, { "epoch": 1.76, "learning_rate": 8.37699050303089e-06, "loss": 4.5662, "step": 7685 }, { "epoch": 1.76, "learning_rate": 8.296921436430071e-06, "loss": 4.5034, "step": 7690 }, { "epoch": 1.77, "learning_rate": 8.217220303767092e-06, "loss": 4.5178, "step": 7695 }, { "epoch": 1.77, "learning_rate": 8.137887424821277e-06, "loss": 4.5397, "step": 7700 }, { "epoch": 1.77, "learning_rate": 8.058923117894534e-06, "loss": 4.5795, "step": 7705 }, { "epoch": 1.77, "learning_rate": 7.980327699809832e-06, "loss": 4.5315, "step": 7710 }, { "epoch": 1.77, "learning_rate": 7.902101485910185e-06, "loss": 4.5321, "step": 7715 }, { "epoch": 1.77, "learning_rate": 7.824244790057223e-06, "loss": 4.5856, "step": 7720 }, { "epoch": 1.77, "learning_rate": 7.746757924630033e-06, "loss": 4.553, "step": 7725 }, { "epoch": 1.77, "learning_rate": 7.66964120052377e-06, "loss": 4.4908, "step": 7730 }, { "epoch": 1.78, "learning_rate": 7.592894927148553e-06, "loss": 4.5423, "step": 7735 }, { "epoch": 1.78, "learning_rate": 7.516519412428203e-06, "loss": 4.533, "step": 7740 }, { "epoch": 1.78, "learning_rate": 7.440514962798905e-06, "loss": 4.498, "step": 7745 }, { "epoch": 1.78, "learning_rate": 7.3648818832080745e-06, "loss": 4.5032, "step": 7750 }, { "epoch": 1.78, "learning_rate": 7.289620477113068e-06, "loss": 4.537, "step": 7755 }, { "epoch": 1.78, "learning_rate": 7.214731046480094e-06, "loss": 4.5323, "step": 7760 }, { "epoch": 1.78, "learning_rate": 7.140213891782821e-06, "loss": 4.4721, "step": 7765 }, { "epoch": 1.78, "learning_rate": 7.066069312001289e-06, "loss": 4.4735, "step": 7770 }, { "epoch": 1.78, "learning_rate": 6.992297604620679e-06, "loss": 4.4806, "step": 7775 }, { "epoch": 1.79, "learning_rate": 6.918899065630113e-06, "loss": 4.5053, "step": 7780 }, { "epoch": 1.79, "learning_rate": 6.845873989521523e-06, "loss": 4.5612, "step": 7785 }, { "epoch": 1.79, "learning_rate": 6.773222669288359e-06, "loss": 4.5158, "step": 7790 }, { "epoch": 1.79, "learning_rate": 6.700945396424518e-06, "loss": 4.506, "step": 7795 }, { "epoch": 1.79, "learning_rate": 6.629042460923096e-06, "loss": 4.474, "step": 7800 }, { "epoch": 1.79, "learning_rate": 6.5575141512753015e-06, "loss": 4.539, "step": 7805 }, { "epoch": 1.79, "learning_rate": 6.486360754469234e-06, "loss": 4.511, "step": 7810 }, { "epoch": 1.79, "learning_rate": 6.415582555988742e-06, "loss": 4.4885, "step": 7815 }, { "epoch": 1.79, "learning_rate": 6.345179839812343e-06, "loss": 4.5612, "step": 7820 }, { "epoch": 1.8, "learning_rate": 6.275152888411984e-06, "loss": 4.5791, "step": 7825 }, { "epoch": 1.8, "learning_rate": 6.205501982751971e-06, "loss": 4.5407, "step": 7830 }, { "epoch": 1.8, "learning_rate": 6.136227402287809e-06, "loss": 4.539, "step": 7835 }, { "epoch": 1.8, "learning_rate": 6.067329424965162e-06, "loss": 4.5695, "step": 7840 }, { "epoch": 1.8, "learning_rate": 5.998808327218619e-06, "loss": 4.5741, "step": 7845 }, { "epoch": 1.8, "learning_rate": 5.930664383970641e-06, "loss": 4.5079, "step": 7850 }, { "epoch": 1.8, "learning_rate": 5.862897868630468e-06, "loss": 4.5102, "step": 7855 }, { "epoch": 1.8, "learning_rate": 5.795509053093029e-06, "loss": 4.4488, "step": 7860 }, { "epoch": 1.8, "learning_rate": 5.72849820773782e-06, "loss": 4.4671, "step": 7865 }, { "epoch": 1.81, "learning_rate": 5.6618656014278406e-06, "loss": 4.5261, "step": 7870 }, { "epoch": 1.81, "learning_rate": 5.595611501508491e-06, "loss": 4.5491, "step": 7875 }, { "epoch": 1.81, "learning_rate": 5.5297361738065325e-06, "loss": 4.5605, "step": 7880 }, { "epoch": 1.81, "learning_rate": 5.464239882628985e-06, "loss": 4.5522, "step": 7885 }, { "epoch": 1.81, "learning_rate": 5.399122890762143e-06, "loss": 4.5584, "step": 7890 }, { "epoch": 1.81, "learning_rate": 5.334385459470359e-06, "loss": 4.521, "step": 7895 }, { "epoch": 1.81, "learning_rate": 5.270027848495207e-06, "loss": 4.5363, "step": 7900 }, { "epoch": 1.81, "learning_rate": 5.2060503160542785e-06, "loss": 4.5035, "step": 7905 }, { "epoch": 1.82, "learning_rate": 5.1424531188402405e-06, "loss": 4.495, "step": 7910 }, { "epoch": 1.82, "learning_rate": 5.079236512019703e-06, "loss": 4.4942, "step": 7915 }, { "epoch": 1.82, "learning_rate": 5.016400749232297e-06, "loss": 4.5203, "step": 7920 }, { "epoch": 1.82, "learning_rate": 4.953946082589655e-06, "loss": 4.5076, "step": 7925 }, { "epoch": 1.82, "learning_rate": 4.89187276267431e-06, "loss": 4.514, "step": 7930 }, { "epoch": 1.82, "learning_rate": 4.83018103853875e-06, "loss": 4.5452, "step": 7935 }, { "epoch": 1.82, "learning_rate": 4.7688711577044354e-06, "loss": 4.4271, "step": 7940 }, { "epoch": 1.82, "learning_rate": 4.707943366160794e-06, "loss": 4.5472, "step": 7945 }, { "epoch": 1.82, "learning_rate": 4.647397908364182e-06, "loss": 4.5236, "step": 7950 }, { "epoch": 1.83, "learning_rate": 4.587235027236958e-06, "loss": 4.5212, "step": 7955 }, { "epoch": 1.83, "learning_rate": 4.5274549641665105e-06, "loss": 4.5034, "step": 7960 }, { "epoch": 1.83, "learning_rate": 4.468057959004246e-06, "loss": 4.5204, "step": 7965 }, { "epoch": 1.83, "learning_rate": 4.40904425006472e-06, "loss": 4.4921, "step": 7970 }, { "epoch": 1.83, "learning_rate": 4.3504140741245095e-06, "loss": 4.5376, "step": 7975 }, { "epoch": 1.83, "learning_rate": 4.2921676664214535e-06, "loss": 4.5167, "step": 7980 }, { "epoch": 1.83, "learning_rate": 4.234305260653604e-06, "loss": 4.4914, "step": 7985 }, { "epoch": 1.83, "learning_rate": 4.176827088978297e-06, "loss": 4.5076, "step": 7990 }, { "epoch": 1.83, "learning_rate": 4.119733382011215e-06, "loss": 4.5555, "step": 7995 }, { "epoch": 1.84, "learning_rate": 4.0630243688255185e-06, "loss": 4.5328, "step": 8000 }, { "epoch": 1.84, "learning_rate": 4.00670027695087e-06, "loss": 4.5124, "step": 8005 }, { "epoch": 1.84, "learning_rate": 3.950761332372543e-06, "loss": 4.533, "step": 8010 }, { "epoch": 1.84, "learning_rate": 3.8952077595305055e-06, "loss": 4.4785, "step": 8015 }, { "epoch": 1.84, "learning_rate": 3.8400397813185054e-06, "loss": 4.4634, "step": 8020 }, { "epoch": 1.84, "learning_rate": 3.785257619083249e-06, "loss": 4.5603, "step": 8025 }, { "epoch": 1.84, "learning_rate": 3.7308614926234165e-06, "loss": 4.5241, "step": 8030 }, { "epoch": 1.84, "learning_rate": 3.676851620188826e-06, "loss": 4.5561, "step": 8035 }, { "epoch": 1.85, "learning_rate": 3.6232282184795794e-06, "loss": 4.4843, "step": 8040 }, { "epoch": 1.85, "learning_rate": 3.56999150264512e-06, "loss": 4.5225, "step": 8045 }, { "epoch": 1.85, "learning_rate": 3.517141686283498e-06, "loss": 4.5195, "step": 8050 }, { "epoch": 1.85, "learning_rate": 3.4646789814403503e-06, "loss": 4.5384, "step": 8055 }, { "epoch": 1.85, "learning_rate": 3.412603598608188e-06, "loss": 4.5282, "step": 8060 }, { "epoch": 1.85, "learning_rate": 3.360915746725479e-06, "loss": 4.4846, "step": 8065 }, { "epoch": 1.85, "learning_rate": 3.3096156331758e-06, "loss": 4.5073, "step": 8070 }, { "epoch": 1.85, "learning_rate": 3.258703463787105e-06, "loss": 4.5283, "step": 8075 }, { "epoch": 1.85, "learning_rate": 3.2081794428307278e-06, "loss": 4.6029, "step": 8080 }, { "epoch": 1.86, "learning_rate": 3.1580437730207578e-06, "loss": 4.5324, "step": 8085 }, { "epoch": 1.86, "learning_rate": 3.1082966555130654e-06, "loss": 4.4754, "step": 8090 }, { "epoch": 1.86, "learning_rate": 3.058938289904578e-06, "loss": 4.4994, "step": 8095 }, { "epoch": 1.86, "learning_rate": 3.0099688742324715e-06, "loss": 4.4886, "step": 8100 }, { "epoch": 1.86, "learning_rate": 2.9613886049733365e-06, "loss": 4.4838, "step": 8105 }, { "epoch": 1.86, "learning_rate": 2.913197677042456e-06, "loss": 4.5552, "step": 8110 }, { "epoch": 1.86, "learning_rate": 2.865396283792965e-06, "loss": 4.4864, "step": 8115 }, { "epoch": 1.86, "learning_rate": 2.8179846170150903e-06, "loss": 4.4679, "step": 8120 }, { "epoch": 1.86, "learning_rate": 2.7709628669353895e-06, "loss": 4.5362, "step": 8125 }, { "epoch": 1.87, "learning_rate": 2.7243312222159924e-06, "loss": 4.5041, "step": 8130 }, { "epoch": 1.87, "learning_rate": 2.6780898699538483e-06, "loss": 4.5783, "step": 8135 }, { "epoch": 1.87, "learning_rate": 2.6322389956799143e-06, "loss": 4.5313, "step": 8140 }, { "epoch": 1.87, "learning_rate": 2.5867787833585124e-06, "loss": 4.51, "step": 8145 }, { "epoch": 1.87, "learning_rate": 2.541709415386495e-06, "loss": 4.5084, "step": 8150 }, { "epoch": 1.87, "learning_rate": 2.4970310725926148e-06, "loss": 4.4684, "step": 8155 }, { "epoch": 1.87, "learning_rate": 2.4527439342366785e-06, "loss": 4.4974, "step": 8160 }, { "epoch": 1.87, "learning_rate": 2.4088481780089267e-06, "loss": 4.528, "step": 8165 }, { "epoch": 1.87, "learning_rate": 2.3653439800292556e-06, "loss": 4.5452, "step": 8170 }, { "epoch": 1.88, "learning_rate": 2.3222315148465956e-06, "loss": 4.4839, "step": 8175 }, { "epoch": 1.88, "learning_rate": 2.2795109554381024e-06, "loss": 4.5123, "step": 8180 }, { "epoch": 1.88, "learning_rate": 2.237182473208499e-06, "loss": 4.5087, "step": 8185 }, { "epoch": 1.88, "learning_rate": 2.195246237989479e-06, "loss": 4.4865, "step": 8190 }, { "epoch": 1.88, "learning_rate": 2.1537024180388834e-06, "loss": 4.4883, "step": 8195 }, { "epoch": 1.88, "learning_rate": 2.1125511800401234e-06, "loss": 4.4671, "step": 8200 }, { "epoch": 1.88, "learning_rate": 2.0717926891014706e-06, "loss": 4.5324, "step": 8205 }, { "epoch": 1.88, "learning_rate": 2.0314271087554126e-06, "loss": 4.4919, "step": 8210 }, { "epoch": 1.89, "learning_rate": 1.991454600957976e-06, "loss": 4.56, "step": 8215 }, { "epoch": 1.89, "learning_rate": 1.951875326088104e-06, "loss": 4.4642, "step": 8220 }, { "epoch": 1.89, "learning_rate": 1.9126894429469912e-06, "loss": 4.5828, "step": 8225 }, { "epoch": 1.89, "learning_rate": 1.8738971087574275e-06, "loss": 4.5374, "step": 8230 }, { "epoch": 1.89, "learning_rate": 1.8354984791632778e-06, "loss": 4.4741, "step": 8235 }, { "epoch": 1.89, "learning_rate": 1.797493708228659e-06, "loss": 4.4987, "step": 8240 }, { "epoch": 1.89, "learning_rate": 1.759882948437519e-06, "loss": 4.4883, "step": 8245 }, { "epoch": 1.89, "learning_rate": 1.7226663506929142e-06, "loss": 4.4875, "step": 8250 }, { "epoch": 1.89, "learning_rate": 1.685844064316433e-06, "loss": 4.5228, "step": 8255 }, { "epoch": 1.9, "learning_rate": 1.6494162370475852e-06, "loss": 4.5291, "step": 8260 }, { "epoch": 1.9, "learning_rate": 1.6133830150432349e-06, "loss": 4.53, "step": 8265 }, { "epoch": 1.9, "learning_rate": 1.5777445428770022e-06, "loss": 4.524, "step": 8270 }, { "epoch": 1.9, "learning_rate": 1.5425009635386622e-06, "loss": 4.5094, "step": 8275 }, { "epoch": 1.9, "learning_rate": 1.5076524184336027e-06, "loss": 4.5161, "step": 8280 }, { "epoch": 1.9, "learning_rate": 1.4731990473822454e-06, "loss": 4.4785, "step": 8285 }, { "epoch": 1.9, "learning_rate": 1.4391409886194474e-06, "loss": 4.5361, "step": 8290 }, { "epoch": 1.9, "learning_rate": 1.405478378794034e-06, "loss": 4.5615, "step": 8295 }, { "epoch": 1.9, "learning_rate": 1.3722113529681668e-06, "loss": 4.5426, "step": 8300 }, { "epoch": 1.91, "learning_rate": 1.3393400446168435e-06, "loss": 4.5407, "step": 8305 }, { "epoch": 1.91, "learning_rate": 1.306864585627332e-06, "loss": 4.5364, "step": 8310 }, { "epoch": 1.91, "learning_rate": 1.2747851062986926e-06, "loss": 4.5093, "step": 8315 }, { "epoch": 1.91, "learning_rate": 1.2431017353412233e-06, "loss": 4.5303, "step": 8320 }, { "epoch": 1.91, "learning_rate": 1.2118145998758824e-06, "loss": 4.4821, "step": 8325 }, { "epoch": 1.91, "learning_rate": 1.1809238254339105e-06, "loss": 4.5601, "step": 8330 }, { "epoch": 1.91, "learning_rate": 1.1504295359562434e-06, "loss": 4.4995, "step": 8335 }, { "epoch": 1.91, "learning_rate": 1.1203318537929996e-06, "loss": 4.5452, "step": 8340 }, { "epoch": 1.92, "learning_rate": 1.090630899703038e-06, "loss": 4.5717, "step": 8345 }, { "epoch": 1.92, "learning_rate": 1.0613267928534453e-06, "loss": 4.5007, "step": 8350 }, { "epoch": 1.92, "learning_rate": 1.0324196508190832e-06, "loss": 4.4803, "step": 8355 }, { "epoch": 1.92, "learning_rate": 1.0039095895820639e-06, "loss": 4.5672, "step": 8360 }, { "epoch": 1.92, "learning_rate": 9.757967235313526e-07, "loss": 4.5471, "step": 8365 }, { "epoch": 1.92, "learning_rate": 9.480811654622557e-07, "loss": 4.5204, "step": 8370 }, { "epoch": 1.92, "learning_rate": 9.207630265760103e-07, "loss": 4.4771, "step": 8375 }, { "epoch": 1.92, "learning_rate": 8.938424164792736e-07, "loss": 4.4564, "step": 8380 }, { "epoch": 1.92, "learning_rate": 8.673194431837784e-07, "loss": 4.514, "step": 8385 }, { "epoch": 1.93, "learning_rate": 8.411942131058115e-07, "loss": 4.5511, "step": 8390 }, { "epoch": 1.93, "learning_rate": 8.154668310658253e-07, "loss": 4.5572, "step": 8395 }, { "epoch": 1.93, "learning_rate": 7.90137400288049e-07, "loss": 4.4605, "step": 8400 }, { "epoch": 1.93, "learning_rate": 7.652060223999669e-07, "loss": 4.5842, "step": 8405 }, { "epoch": 1.93, "learning_rate": 7.406727974320627e-07, "loss": 4.4843, "step": 8410 }, { "epoch": 1.93, "learning_rate": 7.16537823817276e-07, "loss": 4.4995, "step": 8415 }, { "epoch": 1.93, "learning_rate": 6.928011983907245e-07, "loss": 4.5065, "step": 8420 }, { "epoch": 1.93, "learning_rate": 6.694630163892046e-07, "loss": 4.5232, "step": 8425 }, { "epoch": 1.93, "learning_rate": 6.465233714509245e-07, "loss": 4.5572, "step": 8430 }, { "epoch": 1.94, "learning_rate": 6.239823556150159e-07, "loss": 4.4493, "step": 8435 }, { "epoch": 1.94, "learning_rate": 6.01840059321257e-07, "loss": 4.4967, "step": 8440 }, { "epoch": 1.94, "learning_rate": 5.800965714096496e-07, "loss": 4.5267, "step": 8445 }, { "epoch": 1.94, "learning_rate": 5.587519791200869e-07, "loss": 4.5056, "step": 8450 }, { "epoch": 1.94, "learning_rate": 5.378063680920087e-07, "loss": 4.4833, "step": 8455 }, { "epoch": 1.94, "learning_rate": 5.172598223640468e-07, "loss": 4.5048, "step": 8460 }, { "epoch": 1.94, "learning_rate": 4.971124243736913e-07, "loss": 4.503, "step": 8465 }, { "epoch": 1.94, "learning_rate": 4.773642549569579e-07, "loss": 4.5161, "step": 8470 }, { "epoch": 1.94, "learning_rate": 4.5801539334805466e-07, "loss": 4.532, "step": 8475 }, { "epoch": 1.95, "learning_rate": 4.390659171790934e-07, "loss": 4.5352, "step": 8480 }, { "epoch": 1.95, "learning_rate": 4.205159024797456e-07, "loss": 4.497, "step": 8485 }, { "epoch": 1.95, "learning_rate": 4.023654236769647e-07, "loss": 4.54, "step": 8490 }, { "epoch": 1.95, "learning_rate": 3.8461455359466435e-07, "loss": 4.4851, "step": 8495 }, { "epoch": 1.95, "learning_rate": 3.672633634534295e-07, "loss": 4.4889, "step": 8500 }, { "epoch": 1.95, "learning_rate": 3.5031192287023895e-07, "loss": 4.4845, "step": 8505 }, { "epoch": 1.95, "learning_rate": 3.3376029985819903e-07, "loss": 4.5011, "step": 8510 }, { "epoch": 1.95, "learning_rate": 3.176085608262436e-07, "loss": 4.524, "step": 8515 }, { "epoch": 1.96, "learning_rate": 3.0185677057887885e-07, "loss": 4.5358, "step": 8520 }, { "epoch": 1.96, "learning_rate": 2.8650499231591685e-07, "loss": 4.4509, "step": 8525 }, { "epoch": 1.96, "learning_rate": 2.715532876322646e-07, "loss": 4.4907, "step": 8530 }, { "epoch": 1.96, "learning_rate": 2.570017165175909e-07, "loss": 4.5124, "step": 8535 }, { "epoch": 1.96, "learning_rate": 2.428503373561708e-07, "loss": 4.5268, "step": 8540 }, { "epoch": 1.96, "learning_rate": 2.2909920692660847e-07, "loss": 4.5055, "step": 8545 }, { "epoch": 1.96, "learning_rate": 2.1574838040161473e-07, "loss": 4.5079, "step": 8550 }, { "epoch": 1.96, "learning_rate": 2.0279791134778515e-07, "loss": 4.5324, "step": 8555 }, { "epoch": 1.96, "learning_rate": 1.9024785172541136e-07, "loss": 4.5082, "step": 8560 }, { "epoch": 1.97, "learning_rate": 1.780982518882035e-07, "loss": 4.515, "step": 8565 }, { "epoch": 1.97, "learning_rate": 1.6634916058319018e-07, "loss": 4.5209, "step": 8570 }, { "epoch": 1.97, "learning_rate": 1.5500062495041878e-07, "loss": 4.5149, "step": 8575 }, { "epoch": 1.97, "learning_rate": 1.4405269052284455e-07, "loss": 4.4695, "step": 8580 }, { "epoch": 1.97, "learning_rate": 1.3350540122611942e-07, "loss": 4.5516, "step": 8585 }, { "epoch": 1.97, "learning_rate": 1.2335879937839246e-07, "loss": 4.5657, "step": 8590 }, { "epoch": 1.97, "learning_rate": 1.1361292569017635e-07, "loss": 4.5608, "step": 8595 }, { "epoch": 1.97, "learning_rate": 1.0426781926416996e-07, "loss": 4.5156, "step": 8600 }, { "epoch": 1.97, "learning_rate": 9.532351759510283e-08, "loss": 4.5054, "step": 8605 }, { "epoch": 1.98, "learning_rate": 8.678005656957977e-08, "loss": 4.5362, "step": 8610 }, { "epoch": 1.98, "learning_rate": 7.86374704659254e-08, "loss": 4.5645, "step": 8615 }, { "epoch": 1.98, "learning_rate": 7.089579195409534e-08, "loss": 4.5103, "step": 8620 }, { "epoch": 1.98, "learning_rate": 6.355505209548751e-08, "loss": 4.4785, "step": 8625 }, { "epoch": 1.98, "learning_rate": 5.6615280342842135e-08, "loss": 4.5025, "step": 8630 }, { "epoch": 1.98, "learning_rate": 5.00765045401197e-08, "loss": 4.4908, "step": 8635 }, { "epoch": 1.98, "learning_rate": 4.3938750922412064e-08, "loss": 4.5106, "step": 8640 }, { "epoch": 1.98, "learning_rate": 3.82020441158093e-08, "loss": 4.5141, "step": 8645 }, { "epoch": 1.99, "learning_rate": 3.286640713727751e-08, "loss": 4.5108, "step": 8650 }, { "epoch": 1.99, "learning_rate": 2.7931861394658865e-08, "loss": 4.4968, "step": 8655 }, { "epoch": 1.99, "learning_rate": 2.3398426686471743e-08, "loss": 4.5065, "step": 8660 }, { "epoch": 1.99, "learning_rate": 1.9266121201899634e-08, "loss": 4.4638, "step": 8665 }, { "epoch": 1.99, "learning_rate": 1.5534961520724533e-08, "loss": 4.5557, "step": 8670 }, { "epoch": 1.99, "learning_rate": 1.2204962613204807e-08, "loss": 4.5437, "step": 8675 }, { "epoch": 1.99, "learning_rate": 9.276137840075194e-09, "loss": 4.5022, "step": 8680 }, { "epoch": 1.99, "learning_rate": 6.748498952446891e-09, "loss": 4.4982, "step": 8685 }, { "epoch": 1.99, "learning_rate": 4.6220560918075474e-09, "loss": 4.5106, "step": 8690 }, { "epoch": 2.0, "learning_rate": 2.8968177899213465e-09, "loss": 4.5028, "step": 8695 }, { "epoch": 2.0, "learning_rate": 1.572790968851212e-09, "loss": 4.5242, "step": 8700 }, { "epoch": 2.0, "learning_rate": 6.49980940892192e-10, "loss": 4.4505, "step": 8705 }, { "epoch": 2.0, "learning_rate": 1.2839140858256215e-10, "loss": 4.5056, "step": 8710 }, { "epoch": 2.0, "eval_loss": 4.468899726867676, "eval_runtime": 1139.6781, "eval_samples_per_second": 13.54, "eval_steps_per_second": 0.424, "step": 8714 }, { "epoch": 2.0, "step": 8714, "total_flos": 6.917398977105101e+16, "train_loss": 4.463997247680382, "train_runtime": 72530.8249, "train_samples_per_second": 3.845, "train_steps_per_second": 0.12 } ], "logging_steps": 5, "max_steps": 8714, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 100, "total_flos": 6.917398977105101e+16, "train_batch_size": 4, "trial_name": null, "trial_params": null }