{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.5, "eval_steps": 500, "global_step": 22355, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.00012989532543055242, "loss": 2.4901, "step": 20 }, { "epoch": 0.0, "learning_rate": 0.0001297790203533885, "loss": 2.4166, "step": 40 }, { "epoch": 0.01, "learning_rate": 0.00012966271527622454, "loss": 2.3887, "step": 60 }, { "epoch": 0.01, "learning_rate": 0.00012954641019906061, "loss": 2.2707, "step": 80 }, { "epoch": 0.01, "learning_rate": 0.00012943592037575487, "loss": 2.4148, "step": 100 }, { "epoch": 0.01, "learning_rate": 0.0001293196152985909, "loss": 2.2246, "step": 120 }, { "epoch": 0.02, "learning_rate": 0.00012920331022142696, "loss": 2.2086, "step": 140 }, { "epoch": 0.02, "learning_rate": 0.000129087005144263, "loss": 2.2448, "step": 160 }, { "epoch": 0.02, "learning_rate": 0.00012897070006709908, "loss": 2.1513, "step": 180 }, { "epoch": 0.02, "learning_rate": 0.00012885439498993512, "loss": 2.1979, "step": 200 }, { "epoch": 0.02, "learning_rate": 0.00012873808991277117, "loss": 2.1967, "step": 220 }, { "epoch": 0.03, "learning_rate": 0.00012862760008946542, "loss": 2.2009, "step": 240 }, { "epoch": 0.03, "learning_rate": 0.0001285112950123015, "loss": 2.2464, "step": 260 }, { "epoch": 0.03, "learning_rate": 0.00012839498993513754, "loss": 2.1378, "step": 280 }, { "epoch": 0.03, "learning_rate": 0.0001282786848579736, "loss": 2.1907, "step": 300 }, { "epoch": 0.04, "learning_rate": 0.00012816237978080966, "loss": 2.1482, "step": 320 }, { "epoch": 0.04, "learning_rate": 0.0001280460747036457, "loss": 2.1959, "step": 340 }, { "epoch": 0.04, "learning_rate": 0.00012792976962648175, "loss": 2.1316, "step": 360 }, { "epoch": 0.04, "learning_rate": 0.00012781346454931783, "loss": 2.1494, "step": 380 }, { "epoch": 0.04, "learning_rate": 0.00012769715947215387, "loss": 2.2006, "step": 400 }, { "epoch": 0.05, "learning_rate": 0.00012758085439498992, "loss": 2.1951, "step": 420 }, { "epoch": 0.05, "learning_rate": 0.00012746454931782597, "loss": 2.16, "step": 440 }, { "epoch": 0.05, "learning_rate": 0.00012734824424066204, "loss": 2.185, "step": 460 }, { "epoch": 0.05, "learning_rate": 0.00012723193916349808, "loss": 2.1613, "step": 480 }, { "epoch": 0.06, "learning_rate": 0.00012711563408633413, "loss": 2.1638, "step": 500 }, { "epoch": 0.06, "learning_rate": 0.0001269993290091702, "loss": 2.1574, "step": 520 }, { "epoch": 0.06, "learning_rate": 0.00012688302393200625, "loss": 2.1476, "step": 540 }, { "epoch": 0.06, "learning_rate": 0.0001267667188548423, "loss": 2.1674, "step": 560 }, { "epoch": 0.06, "learning_rate": 0.00012665041377767837, "loss": 2.1332, "step": 580 }, { "epoch": 0.07, "learning_rate": 0.00012653410870051442, "loss": 2.1414, "step": 600 }, { "epoch": 0.07, "learning_rate": 0.00012641780362335046, "loss": 2.1813, "step": 620 }, { "epoch": 0.07, "learning_rate": 0.00012630149854618653, "loss": 2.1329, "step": 640 }, { "epoch": 0.07, "learning_rate": 0.00012618519346902258, "loss": 2.2409, "step": 660 }, { "epoch": 0.08, "learning_rate": 0.00012606888839185863, "loss": 2.1425, "step": 680 }, { "epoch": 0.08, "learning_rate": 0.0001259525833146947, "loss": 2.1504, "step": 700 }, { "epoch": 0.08, "learning_rate": 0.00012583627823753075, "loss": 2.1707, "step": 720 }, { "epoch": 0.08, "learning_rate": 0.0001257199731603668, "loss": 2.1545, "step": 740 }, { "epoch": 0.08, "learning_rate": 0.00012560366808320286, "loss": 2.158, "step": 760 }, { "epoch": 0.09, "learning_rate": 0.0001254873630060389, "loss": 2.1406, "step": 780 }, { "epoch": 0.09, "learning_rate": 0.00012537105792887496, "loss": 2.1347, "step": 800 }, { "epoch": 0.09, "learning_rate": 0.000125254752851711, "loss": 2.1311, "step": 820 }, { "epoch": 0.09, "learning_rate": 0.00012513844777454708, "loss": 2.1721, "step": 840 }, { "epoch": 0.1, "learning_rate": 0.00012502214269738312, "loss": 2.133, "step": 860 }, { "epoch": 0.1, "learning_rate": 0.00012490583762021917, "loss": 2.1699, "step": 880 }, { "epoch": 0.1, "learning_rate": 0.00012478953254305521, "loss": 2.1301, "step": 900 }, { "epoch": 0.1, "learning_rate": 0.0001246732274658913, "loss": 2.1117, "step": 920 }, { "epoch": 0.11, "learning_rate": 0.00012455692238872733, "loss": 2.1259, "step": 940 }, { "epoch": 0.11, "learning_rate": 0.0001244406173115634, "loss": 2.1486, "step": 960 }, { "epoch": 0.11, "learning_rate": 0.00012432431223439945, "loss": 2.1521, "step": 980 }, { "epoch": 0.11, "learning_rate": 0.0001242080071572355, "loss": 2.1392, "step": 1000 }, { "epoch": 0.11, "learning_rate": 0.00012409170208007157, "loss": 2.1488, "step": 1020 }, { "epoch": 0.12, "learning_rate": 0.00012397539700290762, "loss": 2.1047, "step": 1040 }, { "epoch": 0.12, "learning_rate": 0.00012385909192574366, "loss": 2.1408, "step": 1060 }, { "epoch": 0.12, "learning_rate": 0.00012374278684857974, "loss": 2.1569, "step": 1080 }, { "epoch": 0.12, "learning_rate": 0.00012362648177141578, "loss": 2.1465, "step": 1100 }, { "epoch": 0.13, "learning_rate": 0.00012351017669425183, "loss": 2.1701, "step": 1120 }, { "epoch": 0.13, "learning_rate": 0.0001233938716170879, "loss": 2.1293, "step": 1140 }, { "epoch": 0.13, "learning_rate": 0.00012327756653992395, "loss": 2.1176, "step": 1160 }, { "epoch": 0.13, "learning_rate": 0.00012316126146276, "loss": 2.1423, "step": 1180 }, { "epoch": 0.13, "learning_rate": 0.00012304495638559607, "loss": 2.0949, "step": 1200 }, { "epoch": 0.14, "learning_rate": 0.0001229286513084321, "loss": 2.1161, "step": 1220 }, { "epoch": 0.14, "learning_rate": 0.00012281234623126816, "loss": 2.1423, "step": 1240 }, { "epoch": 0.14, "learning_rate": 0.0001226960411541042, "loss": 2.1358, "step": 1260 }, { "epoch": 0.14, "learning_rate": 0.00012257973607694028, "loss": 2.1357, "step": 1280 }, { "epoch": 0.15, "learning_rate": 0.00012246343099977632, "loss": 2.078, "step": 1300 }, { "epoch": 0.15, "learning_rate": 0.00012234712592261237, "loss": 2.1491, "step": 1320 }, { "epoch": 0.15, "learning_rate": 0.00012223082084544842, "loss": 2.145, "step": 1340 }, { "epoch": 0.15, "learning_rate": 0.0001221145157682845, "loss": 2.1503, "step": 1360 }, { "epoch": 0.15, "learning_rate": 0.00012199821069112054, "loss": 2.1336, "step": 1380 }, { "epoch": 0.16, "learning_rate": 0.00012188190561395661, "loss": 2.1555, "step": 1400 }, { "epoch": 0.16, "learning_rate": 0.00012176560053679265, "loss": 2.1199, "step": 1420 }, { "epoch": 0.16, "learning_rate": 0.0001216492954596287, "loss": 2.1455, "step": 1440 }, { "epoch": 0.16, "learning_rate": 0.00012153299038246477, "loss": 2.1473, "step": 1460 }, { "epoch": 0.17, "learning_rate": 0.00012141668530530082, "loss": 2.0865, "step": 1480 }, { "epoch": 0.17, "learning_rate": 0.00012130038022813687, "loss": 2.1347, "step": 1500 }, { "epoch": 0.17, "learning_rate": 0.00012118407515097293, "loss": 2.1074, "step": 1520 }, { "epoch": 0.17, "learning_rate": 0.00012106777007380899, "loss": 2.1506, "step": 1540 }, { "epoch": 0.17, "learning_rate": 0.00012095146499664503, "loss": 2.1278, "step": 1560 }, { "epoch": 0.18, "learning_rate": 0.00012083515991948109, "loss": 2.1479, "step": 1580 }, { "epoch": 0.18, "learning_rate": 0.00012071885484231714, "loss": 2.0859, "step": 1600 }, { "epoch": 0.18, "learning_rate": 0.00012060254976515321, "loss": 2.1138, "step": 1620 }, { "epoch": 0.18, "learning_rate": 0.00012048624468798926, "loss": 2.1397, "step": 1640 }, { "epoch": 0.19, "learning_rate": 0.0001203699396108253, "loss": 2.0949, "step": 1660 }, { "epoch": 0.19, "learning_rate": 0.00012025363453366136, "loss": 2.1187, "step": 1680 }, { "epoch": 0.19, "learning_rate": 0.00012013732945649742, "loss": 2.1278, "step": 1700 }, { "epoch": 0.19, "learning_rate": 0.00012002102437933347, "loss": 2.1332, "step": 1720 }, { "epoch": 0.19, "learning_rate": 0.00011990471930216953, "loss": 2.1388, "step": 1740 }, { "epoch": 0.2, "learning_rate": 0.00011978841422500559, "loss": 2.0935, "step": 1760 }, { "epoch": 0.2, "learning_rate": 0.00011967210914784163, "loss": 2.1461, "step": 1780 }, { "epoch": 0.2, "learning_rate": 0.00011955580407067769, "loss": 2.1138, "step": 1800 }, { "epoch": 0.2, "learning_rate": 0.00011943949899351374, "loss": 2.0624, "step": 1820 }, { "epoch": 0.21, "learning_rate": 0.00011932319391634981, "loss": 2.0992, "step": 1840 }, { "epoch": 0.21, "learning_rate": 0.00011920688883918586, "loss": 2.1387, "step": 1860 }, { "epoch": 0.21, "learning_rate": 0.0001190905837620219, "loss": 2.1209, "step": 1880 }, { "epoch": 0.21, "learning_rate": 0.00011897427868485796, "loss": 2.1302, "step": 1900 }, { "epoch": 0.21, "learning_rate": 0.00011885797360769402, "loss": 2.1154, "step": 1920 }, { "epoch": 0.22, "learning_rate": 0.00011874166853053007, "loss": 2.1462, "step": 1940 }, { "epoch": 0.22, "learning_rate": 0.00011862536345336613, "loss": 2.1146, "step": 1960 }, { "epoch": 0.22, "learning_rate": 0.00011850905837620217, "loss": 2.1403, "step": 1980 }, { "epoch": 0.22, "learning_rate": 0.00011839275329903823, "loss": 2.0799, "step": 2000 }, { "epoch": 0.23, "learning_rate": 0.00011827644822187429, "loss": 2.0788, "step": 2020 }, { "epoch": 0.23, "learning_rate": 0.00011816014314471034, "loss": 2.132, "step": 2040 }, { "epoch": 0.23, "learning_rate": 0.00011804383806754641, "loss": 2.1361, "step": 2060 }, { "epoch": 0.23, "learning_rate": 0.00011792753299038246, "loss": 2.1065, "step": 2080 }, { "epoch": 0.23, "learning_rate": 0.0001178112279132185, "loss": 2.1411, "step": 2100 }, { "epoch": 0.24, "learning_rate": 0.00011769492283605456, "loss": 2.0615, "step": 2120 }, { "epoch": 0.24, "learning_rate": 0.00011757861775889062, "loss": 2.163, "step": 2140 }, { "epoch": 0.24, "learning_rate": 0.00011746231268172667, "loss": 2.1092, "step": 2160 }, { "epoch": 0.24, "learning_rate": 0.00011734600760456273, "loss": 2.1337, "step": 2180 }, { "epoch": 0.25, "learning_rate": 0.00011722970252739878, "loss": 2.1282, "step": 2200 }, { "epoch": 0.25, "learning_rate": 0.00011711339745023483, "loss": 2.0993, "step": 2220 }, { "epoch": 0.25, "learning_rate": 0.0001169970923730709, "loss": 2.1268, "step": 2240 }, { "epoch": 0.25, "learning_rate": 0.00011688078729590694, "loss": 2.094, "step": 2260 }, { "epoch": 0.25, "learning_rate": 0.00011676448221874301, "loss": 2.1094, "step": 2280 }, { "epoch": 0.26, "learning_rate": 0.00011664817714157906, "loss": 2.0969, "step": 2300 }, { "epoch": 0.26, "learning_rate": 0.0001165318720644151, "loss": 2.1216, "step": 2320 }, { "epoch": 0.26, "learning_rate": 0.00011641556698725117, "loss": 2.1355, "step": 2340 }, { "epoch": 0.26, "learning_rate": 0.00011629926191008722, "loss": 2.117, "step": 2360 }, { "epoch": 0.27, "learning_rate": 0.00011618295683292327, "loss": 2.1401, "step": 2380 }, { "epoch": 0.27, "learning_rate": 0.00011606665175575933, "loss": 2.1409, "step": 2400 }, { "epoch": 0.27, "learning_rate": 0.00011595034667859538, "loss": 2.1422, "step": 2420 }, { "epoch": 0.27, "learning_rate": 0.00011583404160143144, "loss": 2.1285, "step": 2440 }, { "epoch": 0.28, "learning_rate": 0.0001157177365242675, "loss": 2.0899, "step": 2460 }, { "epoch": 0.28, "learning_rate": 0.00011560143144710354, "loss": 2.0972, "step": 2480 }, { "epoch": 0.28, "learning_rate": 0.00011548512636993961, "loss": 2.0817, "step": 2500 }, { "epoch": 0.28, "learning_rate": 0.00011536882129277566, "loss": 2.1663, "step": 2520 }, { "epoch": 0.28, "learning_rate": 0.0001152525162156117, "loss": 2.1295, "step": 2540 }, { "epoch": 0.29, "learning_rate": 0.00011513621113844777, "loss": 2.1089, "step": 2560 }, { "epoch": 0.29, "learning_rate": 0.00011501990606128383, "loss": 2.1504, "step": 2580 }, { "epoch": 0.29, "learning_rate": 0.00011490360098411987, "loss": 2.1026, "step": 2600 }, { "epoch": 0.29, "learning_rate": 0.00011479311116081413, "loss": 2.1709, "step": 2620 }, { "epoch": 0.3, "learning_rate": 0.00011467680608365018, "loss": 2.1135, "step": 2640 }, { "epoch": 0.3, "learning_rate": 0.00011456050100648623, "loss": 2.1048, "step": 2660 }, { "epoch": 0.3, "learning_rate": 0.00011444419592932229, "loss": 2.0994, "step": 2680 }, { "epoch": 0.3, "learning_rate": 0.00011432789085215834, "loss": 2.1139, "step": 2700 }, { "epoch": 0.3, "learning_rate": 0.00011421158577499441, "loss": 2.0648, "step": 2720 }, { "epoch": 0.31, "learning_rate": 0.00011409528069783046, "loss": 2.1184, "step": 2740 }, { "epoch": 0.31, "learning_rate": 0.0001139789756206665, "loss": 2.0812, "step": 2760 }, { "epoch": 0.31, "learning_rate": 0.00011386848579736077, "loss": 2.168, "step": 2780 }, { "epoch": 0.31, "learning_rate": 0.00011375218072019681, "loss": 2.0971, "step": 2800 }, { "epoch": 0.32, "learning_rate": 0.00011363587564303286, "loss": 2.0863, "step": 2820 }, { "epoch": 0.32, "learning_rate": 0.00011351957056586893, "loss": 2.1173, "step": 2840 }, { "epoch": 0.32, "learning_rate": 0.00011340326548870498, "loss": 2.0671, "step": 2860 }, { "epoch": 0.32, "learning_rate": 0.00011328696041154103, "loss": 2.1363, "step": 2880 }, { "epoch": 0.32, "learning_rate": 0.00011317065533437709, "loss": 2.1232, "step": 2900 }, { "epoch": 0.33, "learning_rate": 0.00011305435025721314, "loss": 2.108, "step": 2920 }, { "epoch": 0.33, "learning_rate": 0.0001129380451800492, "loss": 2.1008, "step": 2940 }, { "epoch": 0.33, "learning_rate": 0.00011282174010288525, "loss": 2.0993, "step": 2960 }, { "epoch": 0.33, "learning_rate": 0.0001127054350257213, "loss": 2.1125, "step": 2980 }, { "epoch": 0.34, "learning_rate": 0.00011258912994855737, "loss": 2.122, "step": 3000 }, { "epoch": 0.34, "learning_rate": 0.00011247282487139342, "loss": 2.1063, "step": 3020 }, { "epoch": 0.34, "learning_rate": 0.00011235651979422946, "loss": 2.0962, "step": 3040 }, { "epoch": 0.34, "learning_rate": 0.00011224021471706552, "loss": 2.1298, "step": 3060 }, { "epoch": 0.34, "learning_rate": 0.00011212390963990158, "loss": 2.1131, "step": 3080 }, { "epoch": 0.35, "learning_rate": 0.00011200760456273763, "loss": 2.1532, "step": 3100 }, { "epoch": 0.35, "learning_rate": 0.00011189129948557369, "loss": 2.0538, "step": 3120 }, { "epoch": 0.35, "learning_rate": 0.00011177499440840975, "loss": 2.1184, "step": 3140 }, { "epoch": 0.35, "learning_rate": 0.0001116586893312458, "loss": 2.0519, "step": 3160 }, { "epoch": 0.36, "learning_rate": 0.00011154238425408185, "loss": 2.1242, "step": 3180 }, { "epoch": 0.36, "learning_rate": 0.0001114260791769179, "loss": 2.1331, "step": 3200 }, { "epoch": 0.36, "learning_rate": 0.00011130977409975397, "loss": 2.108, "step": 3220 }, { "epoch": 0.36, "learning_rate": 0.00011119346902259002, "loss": 2.1189, "step": 3240 }, { "epoch": 0.36, "learning_rate": 0.00011107716394542606, "loss": 2.0963, "step": 3260 }, { "epoch": 0.37, "learning_rate": 0.00011096667412212033, "loss": 2.1936, "step": 3280 }, { "epoch": 0.37, "learning_rate": 0.00011085036904495638, "loss": 2.1254, "step": 3300 }, { "epoch": 0.37, "learning_rate": 0.00011073406396779242, "loss": 2.1475, "step": 3320 }, { "epoch": 0.37, "learning_rate": 0.00011061775889062848, "loss": 2.1016, "step": 3340 }, { "epoch": 0.38, "learning_rate": 0.00011050145381346454, "loss": 2.131, "step": 3360 }, { "epoch": 0.38, "learning_rate": 0.0001103851487363006, "loss": 2.1093, "step": 3380 }, { "epoch": 0.38, "learning_rate": 0.00011026884365913665, "loss": 2.13, "step": 3400 }, { "epoch": 0.38, "learning_rate": 0.00011015253858197269, "loss": 2.1331, "step": 3420 }, { "epoch": 0.38, "learning_rate": 0.00011003623350480877, "loss": 2.1238, "step": 3440 }, { "epoch": 0.39, "learning_rate": 0.00010991992842764481, "loss": 2.1285, "step": 3460 }, { "epoch": 0.39, "learning_rate": 0.00010980362335048086, "loss": 2.0729, "step": 3480 }, { "epoch": 0.39, "learning_rate": 0.00010968731827331693, "loss": 2.1105, "step": 3500 }, { "epoch": 0.39, "learning_rate": 0.00010957101319615298, "loss": 2.0969, "step": 3520 }, { "epoch": 0.4, "learning_rate": 0.00010945470811898902, "loss": 2.1348, "step": 3540 }, { "epoch": 0.4, "learning_rate": 0.00010933840304182508, "loss": 2.0645, "step": 3560 }, { "epoch": 0.4, "learning_rate": 0.00010922209796466114, "loss": 2.0753, "step": 3580 }, { "epoch": 0.4, "learning_rate": 0.0001091057928874972, "loss": 2.1041, "step": 3600 }, { "epoch": 0.4, "learning_rate": 0.00010898948781033325, "loss": 2.0569, "step": 3620 }, { "epoch": 0.41, "learning_rate": 0.0001088731827331693, "loss": 2.0984, "step": 3640 }, { "epoch": 0.41, "learning_rate": 0.00010875687765600537, "loss": 2.1037, "step": 3660 }, { "epoch": 0.41, "learning_rate": 0.00010864057257884141, "loss": 2.0907, "step": 3680 }, { "epoch": 0.41, "learning_rate": 0.00010852426750167746, "loss": 2.1189, "step": 3700 }, { "epoch": 0.42, "learning_rate": 0.00010840796242451353, "loss": 2.0922, "step": 3720 }, { "epoch": 0.42, "learning_rate": 0.00010829165734734958, "loss": 2.0986, "step": 3740 }, { "epoch": 0.42, "learning_rate": 0.00010817535227018562, "loss": 2.1069, "step": 3760 }, { "epoch": 0.42, "learning_rate": 0.00010805904719302168, "loss": 2.1013, "step": 3780 }, { "epoch": 0.42, "learning_rate": 0.00010794274211585774, "loss": 2.1111, "step": 3800 }, { "epoch": 0.43, "learning_rate": 0.0001078264370386938, "loss": 2.1105, "step": 3820 }, { "epoch": 0.43, "learning_rate": 0.00010771013196152985, "loss": 2.0881, "step": 3840 }, { "epoch": 0.43, "learning_rate": 0.0001075938268843659, "loss": 2.102, "step": 3860 }, { "epoch": 0.43, "learning_rate": 0.00010747752180720197, "loss": 2.1201, "step": 3880 }, { "epoch": 0.44, "learning_rate": 0.00010736121673003801, "loss": 2.128, "step": 3900 }, { "epoch": 0.44, "learning_rate": 0.00010724491165287406, "loss": 2.0483, "step": 3920 }, { "epoch": 0.44, "learning_rate": 0.00010712860657571013, "loss": 2.1274, "step": 3940 }, { "epoch": 0.44, "learning_rate": 0.00010701230149854618, "loss": 2.1372, "step": 3960 }, { "epoch": 0.45, "learning_rate": 0.00010689599642138223, "loss": 2.0902, "step": 3980 }, { "epoch": 0.45, "learning_rate": 0.00010677969134421828, "loss": 2.1103, "step": 4000 }, { "epoch": 0.45, "learning_rate": 0.00010666338626705434, "loss": 2.102, "step": 4020 }, { "epoch": 0.45, "learning_rate": 0.0001065470811898904, "loss": 2.118, "step": 4040 }, { "epoch": 0.45, "learning_rate": 0.00010643077611272645, "loss": 2.0548, "step": 4060 }, { "epoch": 0.46, "learning_rate": 0.0001063144710355625, "loss": 2.0805, "step": 4080 }, { "epoch": 0.46, "learning_rate": 0.00010619816595839857, "loss": 2.073, "step": 4100 }, { "epoch": 0.46, "learning_rate": 0.00010608186088123462, "loss": 2.0895, "step": 4120 }, { "epoch": 0.46, "learning_rate": 0.00010596555580407066, "loss": 2.0976, "step": 4140 }, { "epoch": 0.47, "learning_rate": 0.00010584925072690672, "loss": 2.1084, "step": 4160 }, { "epoch": 0.47, "learning_rate": 0.00010573294564974278, "loss": 2.0852, "step": 4180 }, { "epoch": 0.47, "learning_rate": 0.00010561664057257883, "loss": 2.0955, "step": 4200 }, { "epoch": 0.47, "learning_rate": 0.00010550033549541489, "loss": 2.1188, "step": 4220 }, { "epoch": 0.47, "learning_rate": 0.00010538403041825093, "loss": 2.1081, "step": 4240 }, { "epoch": 0.48, "learning_rate": 0.000105267725341087, "loss": 2.0932, "step": 4260 }, { "epoch": 0.48, "learning_rate": 0.00010515142026392305, "loss": 2.0762, "step": 4280 }, { "epoch": 0.48, "learning_rate": 0.0001050351151867591, "loss": 2.1456, "step": 4300 }, { "epoch": 0.48, "learning_rate": 0.00010491881010959517, "loss": 2.0773, "step": 4320 }, { "epoch": 0.49, "learning_rate": 0.00010480250503243122, "loss": 2.1198, "step": 4340 }, { "epoch": 0.49, "learning_rate": 0.00010468619995526726, "loss": 2.0874, "step": 4360 }, { "epoch": 0.49, "learning_rate": 0.00010456989487810332, "loss": 2.1002, "step": 4380 }, { "epoch": 0.49, "learning_rate": 0.00010445358980093938, "loss": 2.1528, "step": 4400 }, { "epoch": 0.49, "learning_rate": 0.00010433728472377543, "loss": 2.1448, "step": 4420 }, { "epoch": 0.5, "learning_rate": 0.00010422097964661149, "loss": 2.1091, "step": 4440 }, { "epoch": 0.5, "learning_rate": 0.00010410467456944753, "loss": 2.098, "step": 4460 }, { "epoch": 0.5, "learning_rate": 0.0001039883694922836, "loss": 2.0977, "step": 4480 }, { "epoch": 0.5, "learning_rate": 0.00010387206441511965, "loss": 2.0686, "step": 4500 }, { "epoch": 0.51, "learning_rate": 0.0001037557593379557, "loss": 2.0873, "step": 4520 }, { "epoch": 0.51, "learning_rate": 0.00010363945426079177, "loss": 2.0965, "step": 4540 }, { "epoch": 0.51, "learning_rate": 0.00010352314918362782, "loss": 2.1141, "step": 4560 }, { "epoch": 0.51, "learning_rate": 0.00010340684410646386, "loss": 2.1144, "step": 4580 }, { "epoch": 0.51, "learning_rate": 0.00010329053902929992, "loss": 2.0991, "step": 4600 }, { "epoch": 0.52, "learning_rate": 0.00010317423395213598, "loss": 2.1212, "step": 4620 }, { "epoch": 0.52, "learning_rate": 0.00010305792887497203, "loss": 2.065, "step": 4640 }, { "epoch": 0.52, "learning_rate": 0.00010294162379780809, "loss": 2.1417, "step": 4660 }, { "epoch": 0.52, "learning_rate": 0.00010282531872064413, "loss": 2.0534, "step": 4680 }, { "epoch": 0.53, "learning_rate": 0.00010270901364348021, "loss": 2.0935, "step": 4700 }, { "epoch": 0.53, "learning_rate": 0.00010259270856631625, "loss": 2.0936, "step": 4720 }, { "epoch": 0.53, "learning_rate": 0.0001024764034891523, "loss": 2.1046, "step": 4740 }, { "epoch": 0.53, "learning_rate": 0.00010236009841198837, "loss": 2.1202, "step": 4760 }, { "epoch": 0.53, "learning_rate": 0.00010224379333482442, "loss": 2.1168, "step": 4780 }, { "epoch": 0.54, "learning_rate": 0.00010212748825766046, "loss": 2.1452, "step": 4800 }, { "epoch": 0.54, "learning_rate": 0.00010201118318049652, "loss": 2.083, "step": 4820 }, { "epoch": 0.54, "learning_rate": 0.00010189487810333258, "loss": 2.0586, "step": 4840 }, { "epoch": 0.54, "learning_rate": 0.00010177857302616863, "loss": 2.0866, "step": 4860 }, { "epoch": 0.55, "learning_rate": 0.00010166226794900469, "loss": 2.1073, "step": 4880 }, { "epoch": 0.55, "learning_rate": 0.00010154596287184074, "loss": 2.085, "step": 4900 }, { "epoch": 0.55, "learning_rate": 0.00010142965779467681, "loss": 2.1288, "step": 4920 }, { "epoch": 0.55, "learning_rate": 0.00010131335271751285, "loss": 2.095, "step": 4940 }, { "epoch": 0.55, "learning_rate": 0.0001011970476403489, "loss": 2.0773, "step": 4960 }, { "epoch": 0.56, "learning_rate": 0.00010108074256318496, "loss": 2.0692, "step": 4980 }, { "epoch": 0.56, "learning_rate": 0.00010096443748602102, "loss": 2.1083, "step": 5000 }, { "epoch": 0.56, "learning_rate": 0.00010084813240885707, "loss": 2.0798, "step": 5020 }, { "epoch": 0.56, "learning_rate": 0.00010073182733169313, "loss": 2.1033, "step": 5040 }, { "epoch": 0.57, "learning_rate": 0.00010061552225452917, "loss": 2.0525, "step": 5060 }, { "epoch": 0.57, "learning_rate": 0.00010049921717736523, "loss": 2.0929, "step": 5080 }, { "epoch": 0.57, "learning_rate": 0.00010038291210020129, "loss": 2.0907, "step": 5100 }, { "epoch": 0.57, "learning_rate": 0.00010026660702303734, "loss": 2.1307, "step": 5120 }, { "epoch": 0.57, "learning_rate": 0.00010015030194587341, "loss": 2.0979, "step": 5140 }, { "epoch": 0.58, "learning_rate": 0.00010003399686870946, "loss": 2.1352, "step": 5160 }, { "epoch": 0.58, "learning_rate": 9.99176917915455e-05, "loss": 2.1116, "step": 5180 }, { "epoch": 0.58, "learning_rate": 9.980138671438156e-05, "loss": 2.0916, "step": 5200 }, { "epoch": 0.58, "learning_rate": 9.968508163721762e-05, "loss": 2.1329, "step": 5220 }, { "epoch": 0.59, "learning_rate": 9.956877656005367e-05, "loss": 2.105, "step": 5240 }, { "epoch": 0.59, "learning_rate": 9.945247148288973e-05, "loss": 2.0596, "step": 5260 }, { "epoch": 0.59, "learning_rate": 9.933616640572577e-05, "loss": 2.1042, "step": 5280 }, { "epoch": 0.59, "learning_rate": 9.921986132856183e-05, "loss": 2.0807, "step": 5300 }, { "epoch": 0.59, "learning_rate": 9.910355625139789e-05, "loss": 2.1058, "step": 5320 }, { "epoch": 0.6, "learning_rate": 9.898725117423394e-05, "loss": 2.0702, "step": 5340 }, { "epoch": 0.6, "learning_rate": 9.887094609707001e-05, "loss": 2.0739, "step": 5360 }, { "epoch": 0.6, "learning_rate": 9.875464101990606e-05, "loss": 2.0708, "step": 5380 }, { "epoch": 0.6, "learning_rate": 9.86383359427421e-05, "loss": 2.0775, "step": 5400 }, { "epoch": 0.61, "learning_rate": 9.852203086557816e-05, "loss": 2.0901, "step": 5420 }, { "epoch": 0.61, "learning_rate": 9.840572578841422e-05, "loss": 2.1068, "step": 5440 }, { "epoch": 0.61, "learning_rate": 9.828942071125027e-05, "loss": 2.0987, "step": 5460 }, { "epoch": 0.61, "learning_rate": 9.817311563408633e-05, "loss": 2.106, "step": 5480 }, { "epoch": 0.62, "learning_rate": 9.805681055692237e-05, "loss": 2.0821, "step": 5500 }, { "epoch": 0.62, "learning_rate": 9.794050547975843e-05, "loss": 2.1149, "step": 5520 }, { "epoch": 0.62, "learning_rate": 9.782420040259449e-05, "loss": 2.1055, "step": 5540 }, { "epoch": 0.62, "learning_rate": 9.770789532543054e-05, "loss": 2.0827, "step": 5560 }, { "epoch": 0.62, "learning_rate": 9.759159024826661e-05, "loss": 2.122, "step": 5580 }, { "epoch": 0.63, "learning_rate": 9.747528517110266e-05, "loss": 2.1003, "step": 5600 }, { "epoch": 0.63, "learning_rate": 9.73589800939387e-05, "loss": 2.0936, "step": 5620 }, { "epoch": 0.63, "learning_rate": 9.724267501677476e-05, "loss": 2.1016, "step": 5640 }, { "epoch": 0.63, "learning_rate": 9.712636993961082e-05, "loss": 2.0689, "step": 5660 }, { "epoch": 0.64, "learning_rate": 9.701006486244687e-05, "loss": 2.0836, "step": 5680 }, { "epoch": 0.64, "learning_rate": 9.689375978528293e-05, "loss": 2.0938, "step": 5700 }, { "epoch": 0.64, "learning_rate": 9.678326996197718e-05, "loss": 2.0761, "step": 5720 }, { "epoch": 0.64, "learning_rate": 9.666696488481323e-05, "loss": 2.1049, "step": 5740 }, { "epoch": 0.64, "learning_rate": 9.655065980764929e-05, "loss": 2.0922, "step": 5760 }, { "epoch": 0.65, "learning_rate": 9.643435473048533e-05, "loss": 2.1141, "step": 5780 }, { "epoch": 0.65, "learning_rate": 9.631804965332141e-05, "loss": 2.15, "step": 5800 }, { "epoch": 0.65, "learning_rate": 9.620174457615745e-05, "loss": 2.1099, "step": 5820 }, { "epoch": 0.65, "learning_rate": 9.60854394989935e-05, "loss": 2.0849, "step": 5840 }, { "epoch": 0.66, "learning_rate": 9.596913442182956e-05, "loss": 2.1233, "step": 5860 }, { "epoch": 0.66, "learning_rate": 9.585282934466562e-05, "loss": 2.0993, "step": 5880 }, { "epoch": 0.66, "learning_rate": 9.573652426750166e-05, "loss": 2.064, "step": 5900 }, { "epoch": 0.66, "learning_rate": 9.562021919033772e-05, "loss": 2.0972, "step": 5920 }, { "epoch": 0.66, "learning_rate": 9.550391411317378e-05, "loss": 2.0913, "step": 5940 }, { "epoch": 0.67, "learning_rate": 9.538760903600983e-05, "loss": 2.1062, "step": 5960 }, { "epoch": 0.67, "learning_rate": 9.527130395884589e-05, "loss": 2.0634, "step": 5980 }, { "epoch": 0.67, "learning_rate": 9.515499888168194e-05, "loss": 2.0826, "step": 6000 }, { "epoch": 0.67, "learning_rate": 9.503869380451801e-05, "loss": 2.0542, "step": 6020 }, { "epoch": 0.68, "learning_rate": 9.492238872735405e-05, "loss": 2.1298, "step": 6040 }, { "epoch": 0.68, "learning_rate": 9.48060836501901e-05, "loss": 2.0832, "step": 6060 }, { "epoch": 0.68, "learning_rate": 9.468977857302616e-05, "loss": 2.0809, "step": 6080 }, { "epoch": 0.68, "learning_rate": 9.457347349586222e-05, "loss": 2.0953, "step": 6100 }, { "epoch": 0.68, "learning_rate": 9.445716841869827e-05, "loss": 2.1176, "step": 6120 }, { "epoch": 0.69, "learning_rate": 9.434086334153433e-05, "loss": 2.1116, "step": 6140 }, { "epoch": 0.69, "learning_rate": 9.422455826437037e-05, "loss": 2.0793, "step": 6160 }, { "epoch": 0.69, "learning_rate": 9.410825318720643e-05, "loss": 2.1492, "step": 6180 }, { "epoch": 0.69, "learning_rate": 9.399194811004249e-05, "loss": 2.1025, "step": 6200 }, { "epoch": 0.7, "learning_rate": 9.387564303287854e-05, "loss": 2.1206, "step": 6220 }, { "epoch": 0.7, "learning_rate": 9.375933795571461e-05, "loss": 2.0791, "step": 6240 }, { "epoch": 0.7, "learning_rate": 9.364303287855066e-05, "loss": 2.1058, "step": 6260 }, { "epoch": 0.7, "learning_rate": 9.35267278013867e-05, "loss": 2.0744, "step": 6280 }, { "epoch": 0.7, "learning_rate": 9.341042272422276e-05, "loss": 2.1379, "step": 6300 }, { "epoch": 0.71, "learning_rate": 9.329411764705882e-05, "loss": 2.0771, "step": 6320 }, { "epoch": 0.71, "learning_rate": 9.317781256989487e-05, "loss": 2.0726, "step": 6340 }, { "epoch": 0.71, "learning_rate": 9.306150749273093e-05, "loss": 2.0929, "step": 6360 }, { "epoch": 0.71, "learning_rate": 9.294520241556697e-05, "loss": 2.0805, "step": 6380 }, { "epoch": 0.72, "learning_rate": 9.282889733840303e-05, "loss": 2.0669, "step": 6400 }, { "epoch": 0.72, "learning_rate": 9.271259226123909e-05, "loss": 2.0981, "step": 6420 }, { "epoch": 0.72, "learning_rate": 9.259628718407514e-05, "loss": 2.0965, "step": 6440 }, { "epoch": 0.72, "learning_rate": 9.247998210691121e-05, "loss": 2.0624, "step": 6460 }, { "epoch": 0.72, "learning_rate": 9.236367702974726e-05, "loss": 2.0785, "step": 6480 }, { "epoch": 0.73, "learning_rate": 9.22473719525833e-05, "loss": 2.0725, "step": 6500 }, { "epoch": 0.73, "learning_rate": 9.213106687541936e-05, "loss": 2.1556, "step": 6520 }, { "epoch": 0.73, "learning_rate": 9.201476179825542e-05, "loss": 2.0998, "step": 6540 }, { "epoch": 0.73, "learning_rate": 9.189845672109147e-05, "loss": 2.0678, "step": 6560 }, { "epoch": 0.74, "learning_rate": 9.178215164392753e-05, "loss": 2.0746, "step": 6580 }, { "epoch": 0.74, "learning_rate": 9.166584656676357e-05, "loss": 2.062, "step": 6600 }, { "epoch": 0.74, "learning_rate": 9.154954148959963e-05, "loss": 2.0578, "step": 6620 }, { "epoch": 0.74, "learning_rate": 9.143323641243569e-05, "loss": 2.0694, "step": 6640 }, { "epoch": 0.74, "learning_rate": 9.131693133527174e-05, "loss": 2.1157, "step": 6660 }, { "epoch": 0.75, "learning_rate": 9.120062625810778e-05, "loss": 2.0948, "step": 6680 }, { "epoch": 0.75, "learning_rate": 9.108432118094386e-05, "loss": 2.1122, "step": 6700 }, { "epoch": 0.75, "learning_rate": 9.097964661149629e-05, "loss": 2.0849, "step": 6720 }, { "epoch": 0.75, "learning_rate": 9.086334153433236e-05, "loss": 2.1186, "step": 6740 }, { "epoch": 0.76, "learning_rate": 9.074703645716841e-05, "loss": 2.113, "step": 6760 }, { "epoch": 0.76, "learning_rate": 9.063073138000446e-05, "loss": 2.1269, "step": 6780 }, { "epoch": 0.76, "learning_rate": 9.051442630284053e-05, "loss": 2.0888, "step": 6800 }, { "epoch": 0.76, "learning_rate": 9.039812122567658e-05, "loss": 2.1094, "step": 6820 }, { "epoch": 0.76, "learning_rate": 9.028181614851262e-05, "loss": 2.1165, "step": 6840 }, { "epoch": 0.77, "learning_rate": 9.016551107134868e-05, "loss": 2.0779, "step": 6860 }, { "epoch": 0.77, "learning_rate": 9.004920599418474e-05, "loss": 2.1145, "step": 6880 }, { "epoch": 0.77, "learning_rate": 8.99329009170208e-05, "loss": 2.0862, "step": 6900 }, { "epoch": 0.77, "learning_rate": 8.981659583985685e-05, "loss": 2.079, "step": 6920 }, { "epoch": 0.78, "learning_rate": 8.970029076269289e-05, "loss": 2.0633, "step": 6940 }, { "epoch": 0.78, "learning_rate": 8.958398568552897e-05, "loss": 2.1121, "step": 6960 }, { "epoch": 0.78, "learning_rate": 8.946768060836501e-05, "loss": 2.0693, "step": 6980 }, { "epoch": 0.78, "learning_rate": 8.935137553120106e-05, "loss": 2.1111, "step": 7000 }, { "epoch": 0.79, "learning_rate": 8.923507045403713e-05, "loss": 2.0946, "step": 7020 }, { "epoch": 0.79, "learning_rate": 8.911876537687318e-05, "loss": 2.0797, "step": 7040 }, { "epoch": 0.79, "learning_rate": 8.900246029970922e-05, "loss": 2.0911, "step": 7060 }, { "epoch": 0.79, "learning_rate": 8.888615522254528e-05, "loss": 2.1028, "step": 7080 }, { "epoch": 0.79, "learning_rate": 8.876985014538134e-05, "loss": 2.0305, "step": 7100 }, { "epoch": 0.8, "learning_rate": 8.86535450682174e-05, "loss": 2.1243, "step": 7120 }, { "epoch": 0.8, "learning_rate": 8.853723999105345e-05, "loss": 2.0735, "step": 7140 }, { "epoch": 0.8, "learning_rate": 8.84209349138895e-05, "loss": 2.1153, "step": 7160 }, { "epoch": 0.8, "learning_rate": 8.830462983672557e-05, "loss": 2.123, "step": 7180 }, { "epoch": 0.81, "learning_rate": 8.818832475956161e-05, "loss": 2.1111, "step": 7200 }, { "epoch": 0.81, "learning_rate": 8.807201968239766e-05, "loss": 2.1151, "step": 7220 }, { "epoch": 0.81, "learning_rate": 8.795571460523372e-05, "loss": 2.0732, "step": 7240 }, { "epoch": 0.81, "learning_rate": 8.783940952806978e-05, "loss": 2.0776, "step": 7260 }, { "epoch": 0.81, "learning_rate": 8.772310445090582e-05, "loss": 2.1173, "step": 7280 }, { "epoch": 0.82, "learning_rate": 8.760679937374188e-05, "loss": 2.109, "step": 7300 }, { "epoch": 0.82, "learning_rate": 8.749049429657794e-05, "loss": 2.0947, "step": 7320 }, { "epoch": 0.82, "learning_rate": 8.7374189219414e-05, "loss": 2.1035, "step": 7340 }, { "epoch": 0.82, "learning_rate": 8.725788414225005e-05, "loss": 2.0857, "step": 7360 }, { "epoch": 0.83, "learning_rate": 8.71473943189443e-05, "loss": 2.1273, "step": 7380 }, { "epoch": 0.83, "learning_rate": 8.703108924178036e-05, "loss": 2.083, "step": 7400 }, { "epoch": 0.83, "learning_rate": 8.691478416461641e-05, "loss": 2.0879, "step": 7420 }, { "epoch": 0.83, "learning_rate": 8.679847908745245e-05, "loss": 2.0691, "step": 7440 }, { "epoch": 0.83, "learning_rate": 8.668217401028853e-05, "loss": 2.0915, "step": 7460 }, { "epoch": 0.84, "learning_rate": 8.656586893312457e-05, "loss": 2.0725, "step": 7480 }, { "epoch": 0.84, "learning_rate": 8.644956385596062e-05, "loss": 2.0862, "step": 7500 }, { "epoch": 0.84, "learning_rate": 8.633325877879668e-05, "loss": 2.0728, "step": 7520 }, { "epoch": 0.84, "learning_rate": 8.621695370163274e-05, "loss": 2.0855, "step": 7540 }, { "epoch": 0.85, "learning_rate": 8.61006486244688e-05, "loss": 2.0736, "step": 7560 }, { "epoch": 0.85, "learning_rate": 8.598434354730484e-05, "loss": 2.0767, "step": 7580 }, { "epoch": 0.85, "learning_rate": 8.586803847014089e-05, "loss": 2.0838, "step": 7600 }, { "epoch": 0.85, "learning_rate": 8.575173339297696e-05, "loss": 2.108, "step": 7620 }, { "epoch": 0.85, "learning_rate": 8.563542831581301e-05, "loss": 2.0527, "step": 7640 }, { "epoch": 0.86, "learning_rate": 8.551912323864905e-05, "loss": 2.0648, "step": 7660 }, { "epoch": 0.86, "learning_rate": 8.540281816148513e-05, "loss": 2.0727, "step": 7680 }, { "epoch": 0.86, "learning_rate": 8.528651308432117e-05, "loss": 2.067, "step": 7700 }, { "epoch": 0.86, "learning_rate": 8.517020800715722e-05, "loss": 2.1203, "step": 7720 }, { "epoch": 0.87, "learning_rate": 8.505390292999328e-05, "loss": 2.0996, "step": 7740 }, { "epoch": 0.87, "learning_rate": 8.493759785282934e-05, "loss": 2.0957, "step": 7760 }, { "epoch": 0.87, "learning_rate": 8.48212927756654e-05, "loss": 2.0995, "step": 7780 }, { "epoch": 0.87, "learning_rate": 8.470498769850144e-05, "loss": 2.1292, "step": 7800 }, { "epoch": 0.87, "learning_rate": 8.458868262133749e-05, "loss": 2.0849, "step": 7820 }, { "epoch": 0.88, "learning_rate": 8.447237754417356e-05, "loss": 2.0608, "step": 7840 }, { "epoch": 0.88, "learning_rate": 8.435607246700961e-05, "loss": 2.0869, "step": 7860 }, { "epoch": 0.88, "learning_rate": 8.423976738984566e-05, "loss": 2.1041, "step": 7880 }, { "epoch": 0.88, "learning_rate": 8.412346231268173e-05, "loss": 2.1176, "step": 7900 }, { "epoch": 0.89, "learning_rate": 8.400715723551778e-05, "loss": 2.0637, "step": 7920 }, { "epoch": 0.89, "learning_rate": 8.389085215835382e-05, "loss": 2.0899, "step": 7940 }, { "epoch": 0.89, "learning_rate": 8.377454708118988e-05, "loss": 2.1238, "step": 7960 }, { "epoch": 0.89, "learning_rate": 8.365824200402594e-05, "loss": 2.074, "step": 7980 }, { "epoch": 0.89, "learning_rate": 8.3541936926862e-05, "loss": 2.0879, "step": 8000 }, { "epoch": 0.9, "learning_rate": 8.342563184969805e-05, "loss": 2.054, "step": 8020 }, { "epoch": 0.9, "learning_rate": 8.330932677253409e-05, "loss": 2.099, "step": 8040 }, { "epoch": 0.9, "learning_rate": 8.319302169537017e-05, "loss": 2.0973, "step": 8060 }, { "epoch": 0.9, "learning_rate": 8.307671661820621e-05, "loss": 2.0318, "step": 8080 }, { "epoch": 0.91, "learning_rate": 8.296041154104226e-05, "loss": 2.0945, "step": 8100 }, { "epoch": 0.91, "learning_rate": 8.284410646387832e-05, "loss": 2.0752, "step": 8120 }, { "epoch": 0.91, "learning_rate": 8.272780138671438e-05, "loss": 2.0991, "step": 8140 }, { "epoch": 0.91, "learning_rate": 8.261149630955042e-05, "loss": 2.0894, "step": 8160 }, { "epoch": 0.91, "learning_rate": 8.249519123238648e-05, "loss": 2.1152, "step": 8180 }, { "epoch": 0.92, "learning_rate": 8.237888615522254e-05, "loss": 2.0758, "step": 8200 }, { "epoch": 0.92, "learning_rate": 8.22625810780586e-05, "loss": 2.0666, "step": 8220 }, { "epoch": 0.92, "learning_rate": 8.214627600089465e-05, "loss": 2.1372, "step": 8240 }, { "epoch": 0.92, "learning_rate": 8.20299709237307e-05, "loss": 2.0993, "step": 8260 }, { "epoch": 0.93, "learning_rate": 8.191366584656677e-05, "loss": 2.0746, "step": 8280 }, { "epoch": 0.93, "learning_rate": 8.179736076940281e-05, "loss": 2.0846, "step": 8300 }, { "epoch": 0.93, "learning_rate": 8.168105569223886e-05, "loss": 2.1056, "step": 8320 }, { "epoch": 0.93, "learning_rate": 8.156475061507492e-05, "loss": 2.0845, "step": 8340 }, { "epoch": 0.93, "learning_rate": 8.144844553791098e-05, "loss": 2.0659, "step": 8360 }, { "epoch": 0.94, "learning_rate": 8.133214046074702e-05, "loss": 2.1159, "step": 8380 }, { "epoch": 0.94, "learning_rate": 8.121583538358308e-05, "loss": 2.0651, "step": 8400 }, { "epoch": 0.94, "learning_rate": 8.109953030641913e-05, "loss": 2.0937, "step": 8420 }, { "epoch": 0.94, "learning_rate": 8.09832252292552e-05, "loss": 2.0918, "step": 8440 }, { "epoch": 0.95, "learning_rate": 8.086692015209125e-05, "loss": 2.0971, "step": 8460 }, { "epoch": 0.95, "learning_rate": 8.07506150749273e-05, "loss": 2.1071, "step": 8480 }, { "epoch": 0.95, "learning_rate": 8.063430999776337e-05, "loss": 2.0947, "step": 8500 }, { "epoch": 0.95, "learning_rate": 8.051800492059941e-05, "loss": 2.1176, "step": 8520 }, { "epoch": 0.96, "learning_rate": 8.040169984343546e-05, "loss": 2.0751, "step": 8540 }, { "epoch": 0.96, "learning_rate": 8.028539476627152e-05, "loss": 2.1119, "step": 8560 }, { "epoch": 0.96, "learning_rate": 8.016908968910758e-05, "loss": 2.0632, "step": 8580 }, { "epoch": 0.96, "learning_rate": 8.005278461194362e-05, "loss": 2.0783, "step": 8600 }, { "epoch": 0.96, "learning_rate": 7.993647953477968e-05, "loss": 2.0886, "step": 8620 }, { "epoch": 0.97, "learning_rate": 7.982017445761573e-05, "loss": 2.0987, "step": 8640 }, { "epoch": 0.97, "learning_rate": 7.97038693804518e-05, "loss": 2.1099, "step": 8660 }, { "epoch": 0.97, "learning_rate": 7.958756430328785e-05, "loss": 2.1042, "step": 8680 }, { "epoch": 0.97, "learning_rate": 7.94712592261239e-05, "loss": 2.0701, "step": 8700 }, { "epoch": 0.98, "learning_rate": 7.935495414895997e-05, "loss": 2.095, "step": 8720 }, { "epoch": 0.98, "learning_rate": 7.923864907179601e-05, "loss": 2.082, "step": 8740 }, { "epoch": 0.98, "learning_rate": 7.912234399463206e-05, "loss": 2.0991, "step": 8760 }, { "epoch": 0.98, "learning_rate": 7.901185417132633e-05, "loss": 2.1346, "step": 8780 }, { "epoch": 0.98, "learning_rate": 7.889554909416237e-05, "loss": 2.0537, "step": 8800 }, { "epoch": 0.99, "learning_rate": 7.877924401699842e-05, "loss": 2.0679, "step": 8820 }, { "epoch": 0.99, "learning_rate": 7.866293893983448e-05, "loss": 2.0561, "step": 8840 }, { "epoch": 0.99, "learning_rate": 7.854663386267054e-05, "loss": 2.0824, "step": 8860 }, { "epoch": 0.99, "learning_rate": 7.84303287855066e-05, "loss": 2.0671, "step": 8880 }, { "epoch": 1.0, "learning_rate": 7.831402370834264e-05, "loss": 2.0845, "step": 8900 }, { "epoch": 1.0, "learning_rate": 7.819771863117869e-05, "loss": 2.0639, "step": 8920 }, { "epoch": 1.0, "learning_rate": 7.808141355401476e-05, "loss": 2.0743, "step": 8940 }, { "epoch": 1.0, "learning_rate": 7.796510847685081e-05, "loss": 2.073, "step": 8960 }, { "epoch": 1.0, "learning_rate": 7.784880339968686e-05, "loss": 2.0485, "step": 8980 }, { "epoch": 1.01, "learning_rate": 7.773249832252293e-05, "loss": 2.0882, "step": 9000 }, { "epoch": 1.01, "learning_rate": 7.761619324535897e-05, "loss": 2.0788, "step": 9020 }, { "epoch": 1.01, "learning_rate": 7.749988816819502e-05, "loss": 2.1103, "step": 9040 }, { "epoch": 1.01, "learning_rate": 7.738358309103108e-05, "loss": 2.0259, "step": 9060 }, { "epoch": 1.02, "learning_rate": 7.726727801386714e-05, "loss": 2.0938, "step": 9080 }, { "epoch": 1.02, "learning_rate": 7.71509729367032e-05, "loss": 2.0847, "step": 9100 }, { "epoch": 1.02, "learning_rate": 7.703466785953925e-05, "loss": 2.071, "step": 9120 }, { "epoch": 1.02, "learning_rate": 7.691836278237529e-05, "loss": 2.0918, "step": 9140 }, { "epoch": 1.02, "learning_rate": 7.680205770521136e-05, "loss": 2.0931, "step": 9160 }, { "epoch": 1.03, "learning_rate": 7.668575262804741e-05, "loss": 2.0808, "step": 9180 }, { "epoch": 1.03, "learning_rate": 7.656944755088346e-05, "loss": 2.0443, "step": 9200 }, { "epoch": 1.03, "learning_rate": 7.645314247371952e-05, "loss": 2.0689, "step": 9220 }, { "epoch": 1.03, "learning_rate": 7.633683739655558e-05, "loss": 2.0699, "step": 9240 }, { "epoch": 1.04, "learning_rate": 7.622053231939162e-05, "loss": 2.0754, "step": 9260 }, { "epoch": 1.04, "learning_rate": 7.610422724222768e-05, "loss": 2.0562, "step": 9280 }, { "epoch": 1.04, "learning_rate": 7.598792216506374e-05, "loss": 2.1127, "step": 9300 }, { "epoch": 1.04, "learning_rate": 7.58716170878998e-05, "loss": 2.0445, "step": 9320 }, { "epoch": 1.04, "learning_rate": 7.575531201073585e-05, "loss": 2.0969, "step": 9340 }, { "epoch": 1.05, "learning_rate": 7.563900693357189e-05, "loss": 2.0689, "step": 9360 }, { "epoch": 1.05, "learning_rate": 7.552270185640797e-05, "loss": 2.0759, "step": 9380 }, { "epoch": 1.05, "learning_rate": 7.540639677924401e-05, "loss": 2.0992, "step": 9400 }, { "epoch": 1.05, "learning_rate": 7.529009170208006e-05, "loss": 2.0964, "step": 9420 }, { "epoch": 1.06, "learning_rate": 7.517378662491612e-05, "loss": 2.124, "step": 9440 }, { "epoch": 1.06, "learning_rate": 7.505748154775218e-05, "loss": 2.0637, "step": 9460 }, { "epoch": 1.06, "learning_rate": 7.494117647058822e-05, "loss": 2.0321, "step": 9480 }, { "epoch": 1.06, "learning_rate": 7.482487139342428e-05, "loss": 2.0486, "step": 9500 }, { "epoch": 1.06, "learning_rate": 7.470856631626033e-05, "loss": 2.0797, "step": 9520 }, { "epoch": 1.07, "learning_rate": 7.45922612390964e-05, "loss": 2.0519, "step": 9540 }, { "epoch": 1.07, "learning_rate": 7.447595616193245e-05, "loss": 2.0909, "step": 9560 }, { "epoch": 1.07, "learning_rate": 7.43596510847685e-05, "loss": 2.1264, "step": 9580 }, { "epoch": 1.07, "learning_rate": 7.424334600760457e-05, "loss": 2.0636, "step": 9600 }, { "epoch": 1.08, "learning_rate": 7.412704093044061e-05, "loss": 2.0728, "step": 9620 }, { "epoch": 1.08, "learning_rate": 7.401073585327666e-05, "loss": 2.1126, "step": 9640 }, { "epoch": 1.08, "learning_rate": 7.389443077611272e-05, "loss": 2.0674, "step": 9660 }, { "epoch": 1.08, "learning_rate": 7.377812569894878e-05, "loss": 2.0758, "step": 9680 }, { "epoch": 1.08, "learning_rate": 7.366182062178482e-05, "loss": 2.1052, "step": 9700 }, { "epoch": 1.09, "learning_rate": 7.354551554462088e-05, "loss": 2.1103, "step": 9720 }, { "epoch": 1.09, "learning_rate": 7.342921046745693e-05, "loss": 2.0949, "step": 9740 }, { "epoch": 1.09, "learning_rate": 7.3312905390293e-05, "loss": 2.0553, "step": 9760 }, { "epoch": 1.09, "learning_rate": 7.320823082084544e-05, "loss": 2.1142, "step": 9780 }, { "epoch": 1.1, "learning_rate": 7.30919257436815e-05, "loss": 2.0894, "step": 9800 }, { "epoch": 1.1, "learning_rate": 7.297562066651756e-05, "loss": 2.0715, "step": 9820 }, { "epoch": 1.1, "learning_rate": 7.28593155893536e-05, "loss": 2.0851, "step": 9840 }, { "epoch": 1.1, "learning_rate": 7.274301051218965e-05, "loss": 2.0747, "step": 9860 }, { "epoch": 1.1, "learning_rate": 7.262670543502572e-05, "loss": 2.0983, "step": 9880 }, { "epoch": 1.11, "learning_rate": 7.251040035786177e-05, "loss": 2.0738, "step": 9900 }, { "epoch": 1.11, "learning_rate": 7.239409528069781e-05, "loss": 2.0499, "step": 9920 }, { "epoch": 1.11, "learning_rate": 7.227779020353389e-05, "loss": 2.0457, "step": 9940 }, { "epoch": 1.11, "learning_rate": 7.216148512636993e-05, "loss": 2.1105, "step": 9960 }, { "epoch": 1.12, "learning_rate": 7.204518004920599e-05, "loss": 2.0765, "step": 9980 }, { "epoch": 1.12, "learning_rate": 7.192887497204204e-05, "loss": 2.0242, "step": 10000 }, { "epoch": 1.12, "learning_rate": 7.18125698948781e-05, "loss": 2.0528, "step": 10020 }, { "epoch": 1.12, "learning_rate": 7.169626481771416e-05, "loss": 2.1105, "step": 10040 }, { "epoch": 1.13, "learning_rate": 7.15799597405502e-05, "loss": 2.0939, "step": 10060 }, { "epoch": 1.13, "learning_rate": 7.146365466338625e-05, "loss": 2.0825, "step": 10080 }, { "epoch": 1.13, "learning_rate": 7.134734958622232e-05, "loss": 2.0786, "step": 10100 }, { "epoch": 1.13, "learning_rate": 7.123104450905837e-05, "loss": 2.0824, "step": 10120 }, { "epoch": 1.13, "learning_rate": 7.111473943189441e-05, "loss": 2.0414, "step": 10140 }, { "epoch": 1.14, "learning_rate": 7.099843435473049e-05, "loss": 2.0486, "step": 10160 }, { "epoch": 1.14, "learning_rate": 7.088212927756653e-05, "loss": 2.0729, "step": 10180 }, { "epoch": 1.14, "learning_rate": 7.076582420040259e-05, "loss": 2.0891, "step": 10200 }, { "epoch": 1.14, "learning_rate": 7.064951912323864e-05, "loss": 2.0465, "step": 10220 }, { "epoch": 1.15, "learning_rate": 7.05332140460747e-05, "loss": 2.0474, "step": 10240 }, { "epoch": 1.15, "learning_rate": 7.041690896891076e-05, "loss": 2.1061, "step": 10260 }, { "epoch": 1.15, "learning_rate": 7.03006038917468e-05, "loss": 2.0628, "step": 10280 }, { "epoch": 1.15, "learning_rate": 7.018429881458285e-05, "loss": 2.1327, "step": 10300 }, { "epoch": 1.15, "learning_rate": 7.006799373741892e-05, "loss": 2.0707, "step": 10320 }, { "epoch": 1.16, "learning_rate": 6.995168866025497e-05, "loss": 2.0648, "step": 10340 }, { "epoch": 1.16, "learning_rate": 6.983538358309102e-05, "loss": 2.0567, "step": 10360 }, { "epoch": 1.16, "learning_rate": 6.971907850592709e-05, "loss": 2.0765, "step": 10380 }, { "epoch": 1.16, "learning_rate": 6.960277342876313e-05, "loss": 2.0661, "step": 10400 }, { "epoch": 1.17, "learning_rate": 6.94864683515992e-05, "loss": 2.1056, "step": 10420 }, { "epoch": 1.17, "learning_rate": 6.937016327443524e-05, "loss": 2.0758, "step": 10440 }, { "epoch": 1.17, "learning_rate": 6.92538581972713e-05, "loss": 2.0754, "step": 10460 }, { "epoch": 1.17, "learning_rate": 6.913755312010736e-05, "loss": 2.067, "step": 10480 }, { "epoch": 1.17, "learning_rate": 6.90212480429434e-05, "loss": 2.0442, "step": 10500 }, { "epoch": 1.18, "learning_rate": 6.890494296577945e-05, "loss": 2.1316, "step": 10520 }, { "epoch": 1.18, "learning_rate": 6.878863788861552e-05, "loss": 2.1042, "step": 10540 }, { "epoch": 1.18, "learning_rate": 6.867233281145157e-05, "loss": 2.0781, "step": 10560 }, { "epoch": 1.18, "learning_rate": 6.855602773428762e-05, "loss": 2.0834, "step": 10580 }, { "epoch": 1.19, "learning_rate": 6.843972265712368e-05, "loss": 2.0664, "step": 10600 }, { "epoch": 1.19, "learning_rate": 6.832341757995974e-05, "loss": 2.0874, "step": 10620 }, { "epoch": 1.19, "learning_rate": 6.82071125027958e-05, "loss": 2.068, "step": 10640 }, { "epoch": 1.19, "learning_rate": 6.809080742563184e-05, "loss": 2.1073, "step": 10660 }, { "epoch": 1.19, "learning_rate": 6.797450234846789e-05, "loss": 2.092, "step": 10680 }, { "epoch": 1.2, "learning_rate": 6.785819727130396e-05, "loss": 2.124, "step": 10700 }, { "epoch": 1.2, "learning_rate": 6.774189219414e-05, "loss": 2.0616, "step": 10720 }, { "epoch": 1.2, "learning_rate": 6.762558711697605e-05, "loss": 2.0983, "step": 10740 }, { "epoch": 1.2, "learning_rate": 6.750928203981213e-05, "loss": 2.0409, "step": 10760 }, { "epoch": 1.21, "learning_rate": 6.740460747036456e-05, "loss": 2.081, "step": 10780 }, { "epoch": 1.21, "learning_rate": 6.728830239320063e-05, "loss": 2.0685, "step": 10800 }, { "epoch": 1.21, "learning_rate": 6.717199731603668e-05, "loss": 2.0529, "step": 10820 }, { "epoch": 1.21, "learning_rate": 6.705569223887272e-05, "loss": 2.0319, "step": 10840 }, { "epoch": 1.21, "learning_rate": 6.694520241556699e-05, "loss": 2.0722, "step": 10860 }, { "epoch": 1.22, "learning_rate": 6.682889733840304e-05, "loss": 2.0737, "step": 10880 }, { "epoch": 1.22, "learning_rate": 6.671259226123908e-05, "loss": 2.0686, "step": 10900 }, { "epoch": 1.22, "learning_rate": 6.659628718407514e-05, "loss": 2.0804, "step": 10920 }, { "epoch": 1.22, "learning_rate": 6.64799821069112e-05, "loss": 2.0479, "step": 10940 }, { "epoch": 1.23, "learning_rate": 6.636367702974725e-05, "loss": 2.0999, "step": 10960 }, { "epoch": 1.23, "learning_rate": 6.624737195258331e-05, "loss": 2.0414, "step": 10980 }, { "epoch": 1.23, "learning_rate": 6.613106687541937e-05, "loss": 2.0921, "step": 11000 }, { "epoch": 1.23, "learning_rate": 6.601476179825543e-05, "loss": 2.0881, "step": 11020 }, { "epoch": 1.23, "learning_rate": 6.589845672109147e-05, "loss": 2.0595, "step": 11040 }, { "epoch": 1.24, "learning_rate": 6.578215164392752e-05, "loss": 2.0746, "step": 11060 }, { "epoch": 1.24, "learning_rate": 6.566584656676359e-05, "loss": 2.0711, "step": 11080 }, { "epoch": 1.24, "learning_rate": 6.554954148959964e-05, "loss": 2.0837, "step": 11100 }, { "epoch": 1.24, "learning_rate": 6.543323641243568e-05, "loss": 2.1156, "step": 11120 }, { "epoch": 1.25, "learning_rate": 6.531693133527174e-05, "loss": 2.0635, "step": 11140 }, { "epoch": 1.25, "learning_rate": 6.52006262581078e-05, "loss": 2.1057, "step": 11160 }, { "epoch": 1.25, "learning_rate": 6.508432118094385e-05, "loss": 2.0678, "step": 11180 }, { "epoch": 1.25, "learning_rate": 6.496801610377991e-05, "loss": 2.0829, "step": 11200 }, { "epoch": 1.25, "learning_rate": 6.485171102661596e-05, "loss": 2.0898, "step": 11220 }, { "epoch": 1.26, "learning_rate": 6.473540594945201e-05, "loss": 2.0493, "step": 11240 }, { "epoch": 1.26, "learning_rate": 6.461910087228807e-05, "loss": 2.0973, "step": 11260 }, { "epoch": 1.26, "learning_rate": 6.450279579512413e-05, "loss": 2.0676, "step": 11280 }, { "epoch": 1.26, "learning_rate": 6.438649071796018e-05, "loss": 2.0913, "step": 11300 }, { "epoch": 1.27, "learning_rate": 6.427018564079624e-05, "loss": 2.0759, "step": 11320 }, { "epoch": 1.27, "learning_rate": 6.41538805636323e-05, "loss": 2.0771, "step": 11340 }, { "epoch": 1.27, "learning_rate": 6.403757548646834e-05, "loss": 2.0751, "step": 11360 }, { "epoch": 1.27, "learning_rate": 6.39212704093044e-05, "loss": 2.0917, "step": 11380 }, { "epoch": 1.27, "learning_rate": 6.380496533214045e-05, "loss": 2.0849, "step": 11400 }, { "epoch": 1.28, "learning_rate": 6.368866025497651e-05, "loss": 2.1137, "step": 11420 }, { "epoch": 1.28, "learning_rate": 6.357235517781256e-05, "loss": 2.0939, "step": 11440 }, { "epoch": 1.28, "learning_rate": 6.345605010064862e-05, "loss": 2.0741, "step": 11460 }, { "epoch": 1.28, "learning_rate": 6.333974502348468e-05, "loss": 2.0774, "step": 11480 }, { "epoch": 1.29, "learning_rate": 6.322343994632073e-05, "loss": 2.0635, "step": 11500 }, { "epoch": 1.29, "learning_rate": 6.310713486915678e-05, "loss": 2.0485, "step": 11520 }, { "epoch": 1.29, "learning_rate": 6.299082979199284e-05, "loss": 2.039, "step": 11540 }, { "epoch": 1.29, "learning_rate": 6.28745247148289e-05, "loss": 2.0711, "step": 11560 }, { "epoch": 1.3, "learning_rate": 6.275821963766495e-05, "loss": 2.0571, "step": 11580 }, { "epoch": 1.3, "learning_rate": 6.2641914560501e-05, "loss": 2.0698, "step": 11600 }, { "epoch": 1.3, "learning_rate": 6.252560948333705e-05, "loss": 2.0765, "step": 11620 }, { "epoch": 1.3, "learning_rate": 6.240930440617311e-05, "loss": 2.0773, "step": 11640 }, { "epoch": 1.3, "learning_rate": 6.229299932900916e-05, "loss": 2.0709, "step": 11660 }, { "epoch": 1.31, "learning_rate": 6.217669425184522e-05, "loss": 2.1052, "step": 11680 }, { "epoch": 1.31, "learning_rate": 6.206038917468128e-05, "loss": 2.1026, "step": 11700 }, { "epoch": 1.31, "learning_rate": 6.194408409751734e-05, "loss": 2.1083, "step": 11720 }, { "epoch": 1.31, "learning_rate": 6.182777902035338e-05, "loss": 2.0498, "step": 11740 }, { "epoch": 1.32, "learning_rate": 6.171147394318944e-05, "loss": 2.0841, "step": 11760 }, { "epoch": 1.32, "learning_rate": 6.159516886602549e-05, "loss": 2.0496, "step": 11780 }, { "epoch": 1.32, "learning_rate": 6.147886378886155e-05, "loss": 2.0787, "step": 11800 }, { "epoch": 1.32, "learning_rate": 6.136255871169761e-05, "loss": 2.0895, "step": 11820 }, { "epoch": 1.32, "learning_rate": 6.124625363453365e-05, "loss": 2.0626, "step": 11840 }, { "epoch": 1.33, "learning_rate": 6.112994855736971e-05, "loss": 2.0783, "step": 11860 }, { "epoch": 1.33, "learning_rate": 6.1013643480205765e-05, "loss": 2.0455, "step": 11880 }, { "epoch": 1.33, "learning_rate": 6.089733840304182e-05, "loss": 2.1121, "step": 11900 }, { "epoch": 1.33, "learning_rate": 6.078103332587788e-05, "loss": 2.0669, "step": 11920 }, { "epoch": 1.34, "learning_rate": 6.066472824871393e-05, "loss": 2.0762, "step": 11940 }, { "epoch": 1.34, "learning_rate": 6.0548423171549983e-05, "loss": 2.0947, "step": 11960 }, { "epoch": 1.34, "learning_rate": 6.0432118094386036e-05, "loss": 2.0841, "step": 11980 }, { "epoch": 1.34, "learning_rate": 6.0315813017222096e-05, "loss": 2.011, "step": 12000 }, { "epoch": 1.34, "learning_rate": 6.019950794005814e-05, "loss": 2.0809, "step": 12020 }, { "epoch": 1.35, "learning_rate": 6.00832028628942e-05, "loss": 2.0777, "step": 12040 }, { "epoch": 1.35, "learning_rate": 5.9966897785730254e-05, "loss": 2.0817, "step": 12060 }, { "epoch": 1.35, "learning_rate": 5.9850592708566314e-05, "loss": 2.0892, "step": 12080 }, { "epoch": 1.35, "learning_rate": 5.9734287631402367e-05, "loss": 2.0662, "step": 12100 }, { "epoch": 1.36, "learning_rate": 5.961798255423842e-05, "loss": 2.0668, "step": 12120 }, { "epoch": 1.36, "learning_rate": 5.950167747707448e-05, "loss": 2.0867, "step": 12140 }, { "epoch": 1.36, "learning_rate": 5.9385372399910525e-05, "loss": 2.0867, "step": 12160 }, { "epoch": 1.36, "learning_rate": 5.9269067322746585e-05, "loss": 2.0552, "step": 12180 }, { "epoch": 1.36, "learning_rate": 5.915276224558264e-05, "loss": 2.0824, "step": 12200 }, { "epoch": 1.37, "learning_rate": 5.90364571684187e-05, "loss": 2.0578, "step": 12220 }, { "epoch": 1.37, "learning_rate": 5.892015209125474e-05, "loss": 2.0963, "step": 12240 }, { "epoch": 1.37, "learning_rate": 5.88038470140908e-05, "loss": 2.0727, "step": 12260 }, { "epoch": 1.37, "learning_rate": 5.8687541936926855e-05, "loss": 2.0608, "step": 12280 }, { "epoch": 1.38, "learning_rate": 5.8571236859762915e-05, "loss": 2.0881, "step": 12300 }, { "epoch": 1.38, "learning_rate": 5.846074703645716e-05, "loss": 2.0888, "step": 12320 }, { "epoch": 1.38, "learning_rate": 5.8344441959293214e-05, "loss": 2.0465, "step": 12340 }, { "epoch": 1.38, "learning_rate": 5.823395213598747e-05, "loss": 2.1094, "step": 12360 }, { "epoch": 1.38, "learning_rate": 5.811764705882353e-05, "loss": 2.0192, "step": 12380 }, { "epoch": 1.39, "learning_rate": 5.800134198165958e-05, "loss": 2.0697, "step": 12400 }, { "epoch": 1.39, "learning_rate": 5.788503690449563e-05, "loss": 2.0691, "step": 12420 }, { "epoch": 1.39, "learning_rate": 5.776873182733169e-05, "loss": 2.124, "step": 12440 }, { "epoch": 1.39, "learning_rate": 5.7652426750167745e-05, "loss": 2.0787, "step": 12460 }, { "epoch": 1.4, "learning_rate": 5.75361216730038e-05, "loss": 2.0961, "step": 12480 }, { "epoch": 1.4, "learning_rate": 5.741981659583985e-05, "loss": 2.0777, "step": 12500 }, { "epoch": 1.4, "learning_rate": 5.730351151867591e-05, "loss": 2.1276, "step": 12520 }, { "epoch": 1.4, "learning_rate": 5.7187206441511956e-05, "loss": 2.0695, "step": 12540 }, { "epoch": 1.4, "learning_rate": 5.7070901364348016e-05, "loss": 2.1114, "step": 12560 }, { "epoch": 1.41, "learning_rate": 5.695459628718407e-05, "loss": 2.0385, "step": 12580 }, { "epoch": 1.41, "learning_rate": 5.683829121002013e-05, "loss": 2.0712, "step": 12600 }, { "epoch": 1.41, "learning_rate": 5.6721986132856174e-05, "loss": 2.0475, "step": 12620 }, { "epoch": 1.41, "learning_rate": 5.6605681055692234e-05, "loss": 2.0444, "step": 12640 }, { "epoch": 1.42, "learning_rate": 5.6489375978528293e-05, "loss": 2.0797, "step": 12660 }, { "epoch": 1.42, "learning_rate": 5.6373070901364346e-05, "loss": 2.0404, "step": 12680 }, { "epoch": 1.42, "learning_rate": 5.62567658242004e-05, "loss": 2.0824, "step": 12700 }, { "epoch": 1.42, "learning_rate": 5.614046074703645e-05, "loss": 2.0749, "step": 12720 }, { "epoch": 1.42, "learning_rate": 5.602415566987251e-05, "loss": 2.0788, "step": 12740 }, { "epoch": 1.43, "learning_rate": 5.590785059270856e-05, "loss": 2.0504, "step": 12760 }, { "epoch": 1.43, "learning_rate": 5.579154551554462e-05, "loss": 2.0855, "step": 12780 }, { "epoch": 1.43, "learning_rate": 5.567524043838067e-05, "loss": 2.054, "step": 12800 }, { "epoch": 1.43, "learning_rate": 5.555893536121673e-05, "loss": 2.0847, "step": 12820 }, { "epoch": 1.44, "learning_rate": 5.5442630284052775e-05, "loss": 2.0575, "step": 12840 }, { "epoch": 1.44, "learning_rate": 5.5326325206888835e-05, "loss": 2.0579, "step": 12860 }, { "epoch": 1.44, "learning_rate": 5.521002012972489e-05, "loss": 2.0777, "step": 12880 }, { "epoch": 1.44, "learning_rate": 5.509371505256094e-05, "loss": 2.048, "step": 12900 }, { "epoch": 1.44, "learning_rate": 5.4977409975396993e-05, "loss": 2.0861, "step": 12920 }, { "epoch": 1.45, "learning_rate": 5.486110489823305e-05, "loss": 2.0703, "step": 12940 }, { "epoch": 1.45, "learning_rate": 5.474479982106911e-05, "loss": 2.0672, "step": 12960 }, { "epoch": 1.45, "learning_rate": 5.462849474390516e-05, "loss": 2.072, "step": 12980 }, { "epoch": 1.45, "learning_rate": 5.451218966674122e-05, "loss": 2.0926, "step": 13000 }, { "epoch": 1.46, "learning_rate": 5.439588458957727e-05, "loss": 2.1006, "step": 13020 }, { "epoch": 1.46, "learning_rate": 5.427957951241333e-05, "loss": 2.069, "step": 13040 }, { "epoch": 1.46, "learning_rate": 5.416327443524938e-05, "loss": 2.0707, "step": 13060 }, { "epoch": 1.46, "learning_rate": 5.4046969358085436e-05, "loss": 2.0648, "step": 13080 }, { "epoch": 1.46, "learning_rate": 5.393066428092149e-05, "loss": 2.083, "step": 13100 }, { "epoch": 1.47, "learning_rate": 5.381435920375754e-05, "loss": 2.0851, "step": 13120 }, { "epoch": 1.47, "learning_rate": 5.3698054126593595e-05, "loss": 2.0987, "step": 13140 }, { "epoch": 1.47, "learning_rate": 5.3581749049429654e-05, "loss": 2.0758, "step": 13160 }, { "epoch": 1.47, "learning_rate": 5.346544397226571e-05, "loss": 2.0657, "step": 13180 }, { "epoch": 1.48, "learning_rate": 5.334913889510176e-05, "loss": 2.0461, "step": 13200 }, { "epoch": 1.48, "learning_rate": 5.323283381793782e-05, "loss": 2.0693, "step": 13220 }, { "epoch": 1.48, "learning_rate": 5.311652874077387e-05, "loss": 2.0812, "step": 13240 }, { "epoch": 1.48, "learning_rate": 5.300022366360993e-05, "loss": 2.0818, "step": 13260 }, { "epoch": 1.49, "learning_rate": 5.288391858644598e-05, "loss": 2.0236, "step": 13280 }, { "epoch": 1.49, "learning_rate": 5.276761350928204e-05, "loss": 2.076, "step": 13300 }, { "epoch": 1.49, "learning_rate": 5.265130843211809e-05, "loss": 2.0957, "step": 13320 }, { "epoch": 1.49, "learning_rate": 5.253500335495414e-05, "loss": 2.129, "step": 13340 }, { "epoch": 1.49, "learning_rate": 5.2418698277790196e-05, "loss": 2.1143, "step": 13360 }, { "epoch": 1.5, "learning_rate": 5.2302393200626255e-05, "loss": 2.0316, "step": 13380 }, { "epoch": 1.5, "learning_rate": 5.218608812346231e-05, "loss": 2.0659, "step": 13400 }, { "epoch": 1.5, "learning_rate": 5.206978304629836e-05, "loss": 2.0619, "step": 13420 }, { "epoch": 1.5, "learning_rate": 5.1953477969134414e-05, "loss": 2.0418, "step": 13440 }, { "epoch": 1.51, "learning_rate": 5.1837172891970473e-05, "loss": 2.0524, "step": 13460 }, { "epoch": 1.51, "learning_rate": 5.172086781480653e-05, "loss": 2.1071, "step": 13480 }, { "epoch": 1.51, "learning_rate": 5.160456273764258e-05, "loss": 2.0871, "step": 13500 }, { "epoch": 1.51, "learning_rate": 5.148825766047864e-05, "loss": 2.0552, "step": 13520 }, { "epoch": 1.51, "learning_rate": 5.137195258331469e-05, "loss": 2.0746, "step": 13540 }, { "epoch": 1.52, "learning_rate": 5.1255647506150744e-05, "loss": 2.1019, "step": 13560 }, { "epoch": 1.52, "learning_rate": 5.11393424289868e-05, "loss": 2.0722, "step": 13580 }, { "epoch": 1.52, "learning_rate": 5.102303735182286e-05, "loss": 2.0436, "step": 13600 }, { "epoch": 1.52, "learning_rate": 5.090673227465891e-05, "loss": 2.0747, "step": 13620 }, { "epoch": 1.53, "learning_rate": 5.079042719749496e-05, "loss": 2.0315, "step": 13640 }, { "epoch": 1.53, "learning_rate": 5.0674122120331015e-05, "loss": 2.0546, "step": 13660 }, { "epoch": 1.53, "learning_rate": 5.0557817043167075e-05, "loss": 2.0612, "step": 13680 }, { "epoch": 1.53, "learning_rate": 5.044151196600313e-05, "loss": 2.0716, "step": 13700 }, { "epoch": 1.53, "learning_rate": 5.032520688883918e-05, "loss": 2.0419, "step": 13720 }, { "epoch": 1.54, "learning_rate": 5.020890181167523e-05, "loss": 2.0704, "step": 13740 }, { "epoch": 1.54, "learning_rate": 5.009259673451129e-05, "loss": 2.0328, "step": 13760 }, { "epoch": 1.54, "learning_rate": 4.9976291657347345e-05, "loss": 2.0596, "step": 13780 }, { "epoch": 1.54, "learning_rate": 4.98599865801834e-05, "loss": 2.0896, "step": 13800 }, { "epoch": 1.55, "learning_rate": 4.974368150301946e-05, "loss": 2.0619, "step": 13820 }, { "epoch": 1.55, "learning_rate": 4.963319167971371e-05, "loss": 2.1666, "step": 13840 }, { "epoch": 1.55, "learning_rate": 4.951688660254976e-05, "loss": 2.0784, "step": 13860 }, { "epoch": 1.55, "learning_rate": 4.940058152538582e-05, "loss": 2.0693, "step": 13880 }, { "epoch": 1.55, "learning_rate": 4.928427644822187e-05, "loss": 2.0624, "step": 13900 }, { "epoch": 1.56, "learning_rate": 4.916797137105793e-05, "loss": 2.0706, "step": 13920 }, { "epoch": 1.56, "learning_rate": 4.9051666293893975e-05, "loss": 2.0472, "step": 13940 }, { "epoch": 1.56, "learning_rate": 4.8935361216730035e-05, "loss": 2.1166, "step": 13960 }, { "epoch": 1.56, "learning_rate": 4.881905613956609e-05, "loss": 2.0927, "step": 13980 }, { "epoch": 1.57, "learning_rate": 4.870275106240214e-05, "loss": 2.0265, "step": 14000 }, { "epoch": 1.57, "learning_rate": 4.858644598523819e-05, "loss": 2.0714, "step": 14020 }, { "epoch": 1.57, "learning_rate": 4.847014090807425e-05, "loss": 2.052, "step": 14040 }, { "epoch": 1.57, "learning_rate": 4.835383583091031e-05, "loss": 2.1046, "step": 14060 }, { "epoch": 1.57, "learning_rate": 4.823753075374636e-05, "loss": 2.0489, "step": 14080 }, { "epoch": 1.58, "learning_rate": 4.812122567658242e-05, "loss": 2.0593, "step": 14100 }, { "epoch": 1.58, "learning_rate": 4.800492059941847e-05, "loss": 2.065, "step": 14120 }, { "epoch": 1.58, "learning_rate": 4.788861552225453e-05, "loss": 2.1017, "step": 14140 }, { "epoch": 1.58, "learning_rate": 4.7772310445090576e-05, "loss": 2.0644, "step": 14160 }, { "epoch": 1.59, "learning_rate": 4.7656005367926636e-05, "loss": 2.0813, "step": 14180 }, { "epoch": 1.59, "learning_rate": 4.753970029076269e-05, "loss": 2.0849, "step": 14200 }, { "epoch": 1.59, "learning_rate": 4.742339521359874e-05, "loss": 2.101, "step": 14220 }, { "epoch": 1.59, "learning_rate": 4.7307090136434794e-05, "loss": 2.0588, "step": 14240 }, { "epoch": 1.59, "learning_rate": 4.7190785059270854e-05, "loss": 2.0628, "step": 14260 }, { "epoch": 1.6, "learning_rate": 4.707447998210691e-05, "loss": 2.0487, "step": 14280 }, { "epoch": 1.6, "learning_rate": 4.695817490494296e-05, "loss": 2.1002, "step": 14300 }, { "epoch": 1.6, "learning_rate": 4.684186982777901e-05, "loss": 2.0855, "step": 14320 }, { "epoch": 1.6, "learning_rate": 4.672556475061507e-05, "loss": 2.0458, "step": 14340 }, { "epoch": 1.61, "learning_rate": 4.660925967345113e-05, "loss": 2.082, "step": 14360 }, { "epoch": 1.61, "learning_rate": 4.649295459628718e-05, "loss": 2.1003, "step": 14380 }, { "epoch": 1.61, "learning_rate": 4.637664951912324e-05, "loss": 2.0829, "step": 14400 }, { "epoch": 1.61, "learning_rate": 4.626034444195929e-05, "loss": 2.0158, "step": 14420 }, { "epoch": 1.61, "learning_rate": 4.614403936479534e-05, "loss": 2.0687, "step": 14440 }, { "epoch": 1.62, "learning_rate": 4.6027734287631395e-05, "loss": 2.1068, "step": 14460 }, { "epoch": 1.62, "learning_rate": 4.5911429210467455e-05, "loss": 2.0875, "step": 14480 }, { "epoch": 1.62, "learning_rate": 4.579512413330351e-05, "loss": 2.0998, "step": 14500 }, { "epoch": 1.62, "learning_rate": 4.567881905613956e-05, "loss": 2.0661, "step": 14520 }, { "epoch": 1.63, "learning_rate": 4.5562513978975613e-05, "loss": 2.0509, "step": 14540 }, { "epoch": 1.63, "learning_rate": 4.544620890181167e-05, "loss": 2.0392, "step": 14560 }, { "epoch": 1.63, "learning_rate": 4.5329903824647726e-05, "loss": 2.0608, "step": 14580 }, { "epoch": 1.63, "learning_rate": 4.521359874748378e-05, "loss": 2.0562, "step": 14600 }, { "epoch": 1.63, "learning_rate": 4.509729367031984e-05, "loss": 2.0673, "step": 14620 }, { "epoch": 1.64, "learning_rate": 4.498098859315589e-05, "loss": 2.0618, "step": 14640 }, { "epoch": 1.64, "learning_rate": 4.4864683515991944e-05, "loss": 2.0877, "step": 14660 }, { "epoch": 1.64, "learning_rate": 4.4748378438828e-05, "loss": 2.067, "step": 14680 }, { "epoch": 1.64, "learning_rate": 4.4632073361664056e-05, "loss": 2.0975, "step": 14700 }, { "epoch": 1.65, "learning_rate": 4.451576828450011e-05, "loss": 2.0698, "step": 14720 }, { "epoch": 1.65, "learning_rate": 4.439946320733616e-05, "loss": 2.0984, "step": 14740 }, { "epoch": 1.65, "learning_rate": 4.4283158130172215e-05, "loss": 2.0577, "step": 14760 }, { "epoch": 1.65, "learning_rate": 4.4166853053008274e-05, "loss": 2.1047, "step": 14780 }, { "epoch": 1.66, "learning_rate": 4.405054797584433e-05, "loss": 2.0432, "step": 14800 }, { "epoch": 1.66, "learning_rate": 4.393424289868038e-05, "loss": 2.0983, "step": 14820 }, { "epoch": 1.66, "learning_rate": 4.382956832923283e-05, "loss": 2.0853, "step": 14840 }, { "epoch": 1.66, "learning_rate": 4.3713263252068886e-05, "loss": 2.09, "step": 14860 }, { "epoch": 1.66, "learning_rate": 4.359695817490494e-05, "loss": 2.1009, "step": 14880 }, { "epoch": 1.67, "learning_rate": 4.348065309774099e-05, "loss": 2.0858, "step": 14900 }, { "epoch": 1.67, "learning_rate": 4.336434802057705e-05, "loss": 2.0515, "step": 14920 }, { "epoch": 1.67, "learning_rate": 4.3248042943413104e-05, "loss": 2.0471, "step": 14940 }, { "epoch": 1.67, "learning_rate": 4.313173786624916e-05, "loss": 2.1183, "step": 14960 }, { "epoch": 1.68, "learning_rate": 4.301543278908521e-05, "loss": 2.1014, "step": 14980 }, { "epoch": 1.68, "learning_rate": 4.289912771192127e-05, "loss": 2.0757, "step": 15000 }, { "epoch": 1.68, "learning_rate": 4.278282263475732e-05, "loss": 2.082, "step": 15020 }, { "epoch": 1.68, "learning_rate": 4.2666517557593375e-05, "loss": 2.0863, "step": 15040 }, { "epoch": 1.68, "learning_rate": 4.255021248042943e-05, "loss": 2.0706, "step": 15060 }, { "epoch": 1.69, "learning_rate": 4.243390740326549e-05, "loss": 2.0888, "step": 15080 }, { "epoch": 1.69, "learning_rate": 4.231760232610154e-05, "loss": 2.0727, "step": 15100 }, { "epoch": 1.69, "learning_rate": 4.220129724893759e-05, "loss": 2.0461, "step": 15120 }, { "epoch": 1.69, "learning_rate": 4.2084992171773646e-05, "loss": 2.0925, "step": 15140 }, { "epoch": 1.7, "learning_rate": 4.1968687094609706e-05, "loss": 2.0566, "step": 15160 }, { "epoch": 1.7, "learning_rate": 4.185238201744575e-05, "loss": 2.0723, "step": 15180 }, { "epoch": 1.7, "learning_rate": 4.1741892194140005e-05, "loss": 2.1198, "step": 15200 }, { "epoch": 1.7, "learning_rate": 4.1625587116976064e-05, "loss": 2.0785, "step": 15220 }, { "epoch": 1.7, "learning_rate": 4.1509282039812124e-05, "loss": 2.0565, "step": 15240 }, { "epoch": 1.71, "learning_rate": 4.139297696264817e-05, "loss": 2.0527, "step": 15260 }, { "epoch": 1.71, "learning_rate": 4.127667188548423e-05, "loss": 2.0889, "step": 15280 }, { "epoch": 1.71, "learning_rate": 4.116036680832028e-05, "loss": 2.0971, "step": 15300 }, { "epoch": 1.71, "learning_rate": 4.104406173115634e-05, "loss": 2.115, "step": 15320 }, { "epoch": 1.72, "learning_rate": 4.092775665399239e-05, "loss": 2.0408, "step": 15340 }, { "epoch": 1.72, "learning_rate": 4.081145157682845e-05, "loss": 2.0875, "step": 15360 }, { "epoch": 1.72, "learning_rate": 4.06951464996645e-05, "loss": 2.0583, "step": 15380 }, { "epoch": 1.72, "learning_rate": 4.057884142250055e-05, "loss": 2.0844, "step": 15400 }, { "epoch": 1.72, "learning_rate": 4.0462536345336606e-05, "loss": 2.1288, "step": 15420 }, { "epoch": 1.73, "learning_rate": 4.0346231268172666e-05, "loss": 2.0507, "step": 15440 }, { "epoch": 1.73, "learning_rate": 4.022992619100872e-05, "loss": 2.0735, "step": 15460 }, { "epoch": 1.73, "learning_rate": 4.011362111384477e-05, "loss": 2.1198, "step": 15480 }, { "epoch": 1.73, "learning_rate": 3.999731603668083e-05, "loss": 2.0738, "step": 15500 }, { "epoch": 1.74, "learning_rate": 3.9881010959516884e-05, "loss": 2.0293, "step": 15520 }, { "epoch": 1.74, "learning_rate": 3.976470588235294e-05, "loss": 2.0787, "step": 15540 }, { "epoch": 1.74, "learning_rate": 3.964840080518899e-05, "loss": 2.0727, "step": 15560 }, { "epoch": 1.74, "learning_rate": 3.953209572802505e-05, "loss": 2.0838, "step": 15580 }, { "epoch": 1.74, "learning_rate": 3.94157906508611e-05, "loss": 2.0455, "step": 15600 }, { "epoch": 1.75, "learning_rate": 3.9299485573697154e-05, "loss": 2.0603, "step": 15620 }, { "epoch": 1.75, "learning_rate": 3.918318049653321e-05, "loss": 2.0766, "step": 15640 }, { "epoch": 1.75, "learning_rate": 3.906687541936927e-05, "loss": 2.101, "step": 15660 }, { "epoch": 1.75, "learning_rate": 3.895057034220532e-05, "loss": 2.0211, "step": 15680 }, { "epoch": 1.76, "learning_rate": 3.883426526504137e-05, "loss": 2.0948, "step": 15700 }, { "epoch": 1.76, "learning_rate": 3.8717960187877425e-05, "loss": 2.0982, "step": 15720 }, { "epoch": 1.76, "learning_rate": 3.8601655110713485e-05, "loss": 2.0548, "step": 15740 }, { "epoch": 1.76, "learning_rate": 3.8485350033549544e-05, "loss": 2.0852, "step": 15760 }, { "epoch": 1.76, "learning_rate": 3.836904495638559e-05, "loss": 2.0868, "step": 15780 }, { "epoch": 1.77, "learning_rate": 3.825273987922165e-05, "loss": 2.0769, "step": 15800 }, { "epoch": 1.77, "learning_rate": 3.81364348020577e-05, "loss": 2.0442, "step": 15820 }, { "epoch": 1.77, "learning_rate": 3.8020129724893756e-05, "loss": 2.0668, "step": 15840 }, { "epoch": 1.77, "learning_rate": 3.790382464772981e-05, "loss": 2.0529, "step": 15860 }, { "epoch": 1.78, "learning_rate": 3.778751957056587e-05, "loss": 2.079, "step": 15880 }, { "epoch": 1.78, "learning_rate": 3.767121449340192e-05, "loss": 2.1526, "step": 15900 }, { "epoch": 1.78, "learning_rate": 3.7554909416237974e-05, "loss": 2.0635, "step": 15920 }, { "epoch": 1.78, "learning_rate": 3.7438604339074026e-05, "loss": 2.0773, "step": 15940 }, { "epoch": 1.78, "learning_rate": 3.7322299261910086e-05, "loss": 2.0929, "step": 15960 }, { "epoch": 1.79, "learning_rate": 3.720599418474614e-05, "loss": 2.0554, "step": 15980 }, { "epoch": 1.79, "learning_rate": 3.708968910758219e-05, "loss": 2.1168, "step": 16000 }, { "epoch": 1.79, "learning_rate": 3.6973384030418244e-05, "loss": 2.0428, "step": 16020 }, { "epoch": 1.79, "learning_rate": 3.6857078953254304e-05, "loss": 2.0781, "step": 16040 }, { "epoch": 1.8, "learning_rate": 3.674077387609036e-05, "loss": 2.0755, "step": 16060 }, { "epoch": 1.8, "learning_rate": 3.662446879892641e-05, "loss": 2.0364, "step": 16080 }, { "epoch": 1.8, "learning_rate": 3.650816372176247e-05, "loss": 2.1061, "step": 16100 }, { "epoch": 1.8, "learning_rate": 3.639185864459852e-05, "loss": 2.0723, "step": 16120 }, { "epoch": 1.8, "learning_rate": 3.6275553567434575e-05, "loss": 2.0164, "step": 16140 }, { "epoch": 1.81, "learning_rate": 3.615924849027063e-05, "loss": 2.0676, "step": 16160 }, { "epoch": 1.81, "learning_rate": 3.604294341310669e-05, "loss": 2.0667, "step": 16180 }, { "epoch": 1.81, "learning_rate": 3.592663833594274e-05, "loss": 2.0623, "step": 16200 }, { "epoch": 1.81, "learning_rate": 3.581033325877879e-05, "loss": 2.0801, "step": 16220 }, { "epoch": 1.82, "learning_rate": 3.5694028181614846e-05, "loss": 2.0449, "step": 16240 }, { "epoch": 1.82, "learning_rate": 3.5577723104450905e-05, "loss": 2.0812, "step": 16260 }, { "epoch": 1.82, "learning_rate": 3.546141802728695e-05, "loss": 2.0907, "step": 16280 }, { "epoch": 1.82, "learning_rate": 3.534511295012301e-05, "loss": 2.0436, "step": 16300 }, { "epoch": 1.83, "learning_rate": 3.522880787295907e-05, "loss": 2.111, "step": 16320 }, { "epoch": 1.83, "learning_rate": 3.511250279579512e-05, "loss": 2.0559, "step": 16340 }, { "epoch": 1.83, "learning_rate": 3.4996197718631176e-05, "loss": 2.1178, "step": 16360 }, { "epoch": 1.83, "learning_rate": 3.487989264146723e-05, "loss": 2.0414, "step": 16380 }, { "epoch": 1.83, "learning_rate": 3.476358756430329e-05, "loss": 2.0485, "step": 16400 }, { "epoch": 1.84, "learning_rate": 3.464728248713934e-05, "loss": 2.0785, "step": 16420 }, { "epoch": 1.84, "learning_rate": 3.4530977409975394e-05, "loss": 2.0779, "step": 16440 }, { "epoch": 1.84, "learning_rate": 3.441467233281145e-05, "loss": 2.0431, "step": 16460 }, { "epoch": 1.84, "learning_rate": 3.4298367255647506e-05, "loss": 2.1142, "step": 16480 }, { "epoch": 1.85, "learning_rate": 3.418206217848355e-05, "loss": 2.0406, "step": 16500 }, { "epoch": 1.85, "learning_rate": 3.406575710131961e-05, "loss": 2.0579, "step": 16520 }, { "epoch": 1.85, "learning_rate": 3.3949452024155665e-05, "loss": 2.0933, "step": 16540 }, { "epoch": 1.85, "learning_rate": 3.3833146946991724e-05, "loss": 2.0477, "step": 16560 }, { "epoch": 1.85, "learning_rate": 3.371684186982777e-05, "loss": 2.0948, "step": 16580 }, { "epoch": 1.86, "learning_rate": 3.360053679266383e-05, "loss": 2.0587, "step": 16600 }, { "epoch": 1.86, "learning_rate": 3.348423171549989e-05, "loss": 2.0628, "step": 16620 }, { "epoch": 1.86, "learning_rate": 3.336792663833594e-05, "loss": 2.0993, "step": 16640 }, { "epoch": 1.86, "learning_rate": 3.3251621561171995e-05, "loss": 2.0701, "step": 16660 }, { "epoch": 1.87, "learning_rate": 3.313531648400805e-05, "loss": 2.0582, "step": 16680 }, { "epoch": 1.87, "learning_rate": 3.301901140684411e-05, "loss": 2.044, "step": 16700 }, { "epoch": 1.87, "learning_rate": 3.2902706329680154e-05, "loss": 2.0492, "step": 16720 }, { "epoch": 1.87, "learning_rate": 3.278640125251621e-05, "loss": 2.0256, "step": 16740 }, { "epoch": 1.87, "learning_rate": 3.2670096175352266e-05, "loss": 2.0283, "step": 16760 }, { "epoch": 1.88, "learning_rate": 3.2553791098188326e-05, "loss": 2.0441, "step": 16780 }, { "epoch": 1.88, "learning_rate": 3.243748602102438e-05, "loss": 2.071, "step": 16800 }, { "epoch": 1.88, "learning_rate": 3.232118094386043e-05, "loss": 2.089, "step": 16820 }, { "epoch": 1.88, "learning_rate": 3.220487586669649e-05, "loss": 2.0854, "step": 16840 }, { "epoch": 1.89, "learning_rate": 3.2088570789532544e-05, "loss": 2.053, "step": 16860 }, { "epoch": 1.89, "learning_rate": 3.1972265712368596e-05, "loss": 2.0928, "step": 16880 }, { "epoch": 1.89, "learning_rate": 3.185596063520465e-05, "loss": 2.0675, "step": 16900 }, { "epoch": 1.89, "learning_rate": 3.17396555580407e-05, "loss": 2.0425, "step": 16920 }, { "epoch": 1.89, "learning_rate": 3.162335048087676e-05, "loss": 2.056, "step": 16940 }, { "epoch": 1.9, "learning_rate": 3.1507045403712814e-05, "loss": 2.0542, "step": 16960 }, { "epoch": 1.9, "learning_rate": 3.139074032654887e-05, "loss": 2.0871, "step": 16980 }, { "epoch": 1.9, "learning_rate": 3.127443524938492e-05, "loss": 2.1014, "step": 17000 }, { "epoch": 1.9, "learning_rate": 3.115813017222098e-05, "loss": 2.0285, "step": 17020 }, { "epoch": 1.91, "learning_rate": 3.104182509505703e-05, "loss": 2.0229, "step": 17040 }, { "epoch": 1.91, "learning_rate": 3.0925520017893085e-05, "loss": 2.0798, "step": 17060 }, { "epoch": 1.91, "learning_rate": 3.080921494072914e-05, "loss": 2.0637, "step": 17080 }, { "epoch": 1.91, "learning_rate": 3.069290986356519e-05, "loss": 2.0567, "step": 17100 }, { "epoch": 1.91, "learning_rate": 3.057660478640125e-05, "loss": 2.0258, "step": 17120 }, { "epoch": 1.92, "learning_rate": 3.0460299709237303e-05, "loss": 2.0988, "step": 17140 }, { "epoch": 1.92, "learning_rate": 3.034399463207336e-05, "loss": 2.0889, "step": 17160 }, { "epoch": 1.92, "learning_rate": 3.0227689554909412e-05, "loss": 2.0466, "step": 17180 }, { "epoch": 1.92, "learning_rate": 3.0123014985461862e-05, "loss": 2.0953, "step": 17200 }, { "epoch": 1.93, "learning_rate": 3.0006709908297915e-05, "loss": 2.0807, "step": 17220 }, { "epoch": 1.93, "learning_rate": 2.989040483113397e-05, "loss": 2.0344, "step": 17240 }, { "epoch": 1.93, "learning_rate": 2.977991500782822e-05, "loss": 2.1382, "step": 17260 }, { "epoch": 1.93, "learning_rate": 2.9663609930664277e-05, "loss": 2.0324, "step": 17280 }, { "epoch": 1.93, "learning_rate": 2.954730485350033e-05, "loss": 2.0842, "step": 17300 }, { "epoch": 1.94, "learning_rate": 2.943099977633639e-05, "loss": 2.0407, "step": 17320 }, { "epoch": 1.94, "learning_rate": 2.9314694699172443e-05, "loss": 2.0312, "step": 17340 }, { "epoch": 1.94, "learning_rate": 2.91983896220085e-05, "loss": 2.0736, "step": 17360 }, { "epoch": 1.94, "learning_rate": 2.908208454484455e-05, "loss": 2.0568, "step": 17380 }, { "epoch": 1.95, "learning_rate": 2.8965779467680608e-05, "loss": 2.0744, "step": 17400 }, { "epoch": 1.95, "learning_rate": 2.884947439051666e-05, "loss": 2.0691, "step": 17420 }, { "epoch": 1.95, "learning_rate": 2.8733169313352713e-05, "loss": 2.1048, "step": 17440 }, { "epoch": 1.95, "learning_rate": 2.861686423618877e-05, "loss": 2.0583, "step": 17460 }, { "epoch": 1.95, "learning_rate": 2.8500559159024822e-05, "loss": 2.0439, "step": 17480 }, { "epoch": 1.96, "learning_rate": 2.838425408186088e-05, "loss": 2.0741, "step": 17500 }, { "epoch": 1.96, "learning_rate": 2.826794900469693e-05, "loss": 2.0847, "step": 17520 }, { "epoch": 1.96, "learning_rate": 2.8151643927532988e-05, "loss": 2.0784, "step": 17540 }, { "epoch": 1.96, "learning_rate": 2.803533885036904e-05, "loss": 2.0681, "step": 17560 }, { "epoch": 1.97, "learning_rate": 2.79190337732051e-05, "loss": 2.0736, "step": 17580 }, { "epoch": 1.97, "learning_rate": 2.7802728696041153e-05, "loss": 2.1105, "step": 17600 }, { "epoch": 1.97, "learning_rate": 2.768642361887721e-05, "loss": 2.0879, "step": 17620 }, { "epoch": 1.97, "learning_rate": 2.7570118541713262e-05, "loss": 2.0742, "step": 17640 }, { "epoch": 1.97, "learning_rate": 2.7453813464549315e-05, "loss": 2.0421, "step": 17660 }, { "epoch": 1.98, "learning_rate": 2.733750838738537e-05, "loss": 2.0922, "step": 17680 }, { "epoch": 1.98, "learning_rate": 2.7221203310221424e-05, "loss": 2.0433, "step": 17700 }, { "epoch": 1.98, "learning_rate": 2.710489823305748e-05, "loss": 2.1093, "step": 17720 }, { "epoch": 1.98, "learning_rate": 2.6988593155893533e-05, "loss": 2.057, "step": 17740 }, { "epoch": 1.99, "learning_rate": 2.687228807872959e-05, "loss": 2.0387, "step": 17760 }, { "epoch": 1.99, "learning_rate": 2.675598300156564e-05, "loss": 2.0967, "step": 17780 }, { "epoch": 1.99, "learning_rate": 2.6639677924401698e-05, "loss": 2.0828, "step": 17800 }, { "epoch": 1.99, "learning_rate": 2.652337284723775e-05, "loss": 2.0244, "step": 17820 }, { "epoch": 2.0, "learning_rate": 2.6407067770073807e-05, "loss": 2.0618, "step": 17840 }, { "epoch": 2.0, "learning_rate": 2.6290762692909863e-05, "loss": 2.1041, "step": 17860 }, { "epoch": 2.0, "learning_rate": 2.6174457615745916e-05, "loss": 2.0283, "step": 17880 }, { "epoch": 2.0, "learning_rate": 2.6058152538581972e-05, "loss": 2.0234, "step": 17900 }, { "epoch": 2.0, "learning_rate": 2.5941847461418025e-05, "loss": 2.0634, "step": 17920 }, { "epoch": 2.01, "learning_rate": 2.582554238425408e-05, "loss": 2.0737, "step": 17940 }, { "epoch": 2.01, "learning_rate": 2.5709237307090134e-05, "loss": 2.0651, "step": 17960 }, { "epoch": 2.01, "learning_rate": 2.559293222992619e-05, "loss": 2.025, "step": 17980 }, { "epoch": 2.01, "learning_rate": 2.5476627152762243e-05, "loss": 2.0662, "step": 18000 }, { "epoch": 2.02, "learning_rate": 2.53603220755983e-05, "loss": 2.0219, "step": 18020 }, { "epoch": 2.02, "learning_rate": 2.5244016998434352e-05, "loss": 2.0623, "step": 18040 }, { "epoch": 2.02, "learning_rate": 2.5127711921270405e-05, "loss": 2.059, "step": 18060 }, { "epoch": 2.02, "learning_rate": 2.501140684410646e-05, "loss": 2.1055, "step": 18080 }, { "epoch": 2.02, "learning_rate": 2.4895101766942514e-05, "loss": 2.1067, "step": 18100 }, { "epoch": 2.03, "learning_rate": 2.477879668977857e-05, "loss": 2.0925, "step": 18120 }, { "epoch": 2.03, "learning_rate": 2.4662491612614626e-05, "loss": 2.0838, "step": 18140 }, { "epoch": 2.03, "learning_rate": 2.4546186535450682e-05, "loss": 2.0771, "step": 18160 }, { "epoch": 2.03, "learning_rate": 2.4429881458286735e-05, "loss": 2.0885, "step": 18180 }, { "epoch": 2.04, "learning_rate": 2.431357638112279e-05, "loss": 2.0648, "step": 18200 }, { "epoch": 2.04, "learning_rate": 2.4197271303958844e-05, "loss": 2.0681, "step": 18220 }, { "epoch": 2.04, "learning_rate": 2.40809662267949e-05, "loss": 2.0553, "step": 18240 }, { "epoch": 2.04, "learning_rate": 2.3964661149630953e-05, "loss": 2.0908, "step": 18260 }, { "epoch": 2.04, "learning_rate": 2.3848356072467006e-05, "loss": 2.0558, "step": 18280 }, { "epoch": 2.05, "learning_rate": 2.3732050995303062e-05, "loss": 2.0142, "step": 18300 }, { "epoch": 2.05, "learning_rate": 2.3615745918139115e-05, "loss": 2.0327, "step": 18320 }, { "epoch": 2.05, "learning_rate": 2.349944084097517e-05, "loss": 2.1027, "step": 18340 }, { "epoch": 2.05, "learning_rate": 2.3383135763811224e-05, "loss": 2.0602, "step": 18360 }, { "epoch": 2.06, "learning_rate": 2.326683068664728e-05, "loss": 2.0362, "step": 18380 }, { "epoch": 2.06, "learning_rate": 2.3150525609483333e-05, "loss": 2.0628, "step": 18400 }, { "epoch": 2.06, "learning_rate": 2.3034220532319392e-05, "loss": 2.0663, "step": 18420 }, { "epoch": 2.06, "learning_rate": 2.2917915455155445e-05, "loss": 2.0567, "step": 18440 }, { "epoch": 2.06, "learning_rate": 2.28016103779915e-05, "loss": 2.0717, "step": 18460 }, { "epoch": 2.07, "learning_rate": 2.2685305300827554e-05, "loss": 2.0677, "step": 18480 }, { "epoch": 2.07, "learning_rate": 2.2569000223663607e-05, "loss": 2.0664, "step": 18500 }, { "epoch": 2.07, "learning_rate": 2.2452695146499663e-05, "loss": 2.0965, "step": 18520 }, { "epoch": 2.07, "learning_rate": 2.2336390069335716e-05, "loss": 2.1145, "step": 18540 }, { "epoch": 2.08, "learning_rate": 2.2220084992171772e-05, "loss": 2.0512, "step": 18560 }, { "epoch": 2.08, "learning_rate": 2.2103779915007825e-05, "loss": 2.0948, "step": 18580 }, { "epoch": 2.08, "learning_rate": 2.198747483784388e-05, "loss": 2.1022, "step": 18600 }, { "epoch": 2.08, "learning_rate": 2.1871169760679934e-05, "loss": 2.0139, "step": 18620 }, { "epoch": 2.08, "learning_rate": 2.175486468351599e-05, "loss": 2.0366, "step": 18640 }, { "epoch": 2.09, "learning_rate": 2.1638559606352043e-05, "loss": 2.0667, "step": 18660 }, { "epoch": 2.09, "learning_rate": 2.1522254529188103e-05, "loss": 2.0617, "step": 18680 }, { "epoch": 2.09, "learning_rate": 2.1405949452024155e-05, "loss": 2.0557, "step": 18700 }, { "epoch": 2.09, "learning_rate": 2.1289644374860208e-05, "loss": 2.0373, "step": 18720 }, { "epoch": 2.1, "learning_rate": 2.1173339297696264e-05, "loss": 2.054, "step": 18740 }, { "epoch": 2.1, "learning_rate": 2.1057034220532317e-05, "loss": 2.0393, "step": 18760 }, { "epoch": 2.1, "learning_rate": 2.0940729143368373e-05, "loss": 2.0526, "step": 18780 }, { "epoch": 2.1, "learning_rate": 2.0824424066204426e-05, "loss": 2.0431, "step": 18800 }, { "epoch": 2.1, "learning_rate": 2.0708118989040482e-05, "loss": 2.0438, "step": 18820 }, { "epoch": 2.11, "learning_rate": 2.0591813911876535e-05, "loss": 2.077, "step": 18840 }, { "epoch": 2.11, "learning_rate": 2.047550883471259e-05, "loss": 2.0744, "step": 18860 }, { "epoch": 2.11, "learning_rate": 2.0359203757548644e-05, "loss": 2.1086, "step": 18880 }, { "epoch": 2.11, "learning_rate": 2.0242898680384697e-05, "loss": 2.1172, "step": 18900 }, { "epoch": 2.12, "learning_rate": 2.0126593603220753e-05, "loss": 2.0531, "step": 18920 }, { "epoch": 2.12, "learning_rate": 2.0010288526056806e-05, "loss": 2.0392, "step": 18940 }, { "epoch": 2.12, "learning_rate": 1.9893983448892866e-05, "loss": 2.093, "step": 18960 }, { "epoch": 2.12, "learning_rate": 1.977767837172892e-05, "loss": 2.0838, "step": 18980 }, { "epoch": 2.12, "learning_rate": 1.9661373294564975e-05, "loss": 2.0741, "step": 19000 }, { "epoch": 2.13, "learning_rate": 1.9545068217401027e-05, "loss": 2.0726, "step": 19020 }, { "epoch": 2.13, "learning_rate": 1.9428763140237084e-05, "loss": 2.0095, "step": 19040 }, { "epoch": 2.13, "learning_rate": 1.9312458063073136e-05, "loss": 2.0646, "step": 19060 }, { "epoch": 2.13, "learning_rate": 1.9196152985909193e-05, "loss": 2.051, "step": 19080 }, { "epoch": 2.14, "learning_rate": 1.9079847908745245e-05, "loss": 2.0164, "step": 19100 }, { "epoch": 2.14, "learning_rate": 1.8963542831581298e-05, "loss": 2.0951, "step": 19120 }, { "epoch": 2.14, "learning_rate": 1.8847237754417354e-05, "loss": 2.0782, "step": 19140 }, { "epoch": 2.14, "learning_rate": 1.8730932677253407e-05, "loss": 2.0496, "step": 19160 }, { "epoch": 2.14, "learning_rate": 1.8614627600089463e-05, "loss": 2.0653, "step": 19180 }, { "epoch": 2.15, "learning_rate": 1.8498322522925516e-05, "loss": 2.0701, "step": 19200 }, { "epoch": 2.15, "learning_rate": 1.8382017445761572e-05, "loss": 2.0712, "step": 19220 }, { "epoch": 2.15, "learning_rate": 1.826571236859763e-05, "loss": 2.051, "step": 19240 }, { "epoch": 2.15, "learning_rate": 1.816103779915008e-05, "loss": 2.0425, "step": 19260 }, { "epoch": 2.16, "learning_rate": 1.804473272198613e-05, "loss": 2.0742, "step": 19280 }, { "epoch": 2.16, "learning_rate": 1.7928427644822188e-05, "loss": 2.066, "step": 19300 }, { "epoch": 2.16, "learning_rate": 1.781212256765824e-05, "loss": 2.0443, "step": 19320 }, { "epoch": 2.16, "learning_rate": 1.7695817490494297e-05, "loss": 2.0841, "step": 19340 }, { "epoch": 2.17, "learning_rate": 1.757951241333035e-05, "loss": 2.0769, "step": 19360 }, { "epoch": 2.17, "learning_rate": 1.7463207336166406e-05, "loss": 2.0911, "step": 19380 }, { "epoch": 2.17, "learning_rate": 1.734690225900246e-05, "loss": 2.032, "step": 19400 }, { "epoch": 2.17, "learning_rate": 1.723059718183851e-05, "loss": 2.056, "step": 19420 }, { "epoch": 2.17, "learning_rate": 1.7114292104674568e-05, "loss": 2.0717, "step": 19440 }, { "epoch": 2.18, "learning_rate": 1.699798702751062e-05, "loss": 2.0455, "step": 19460 }, { "epoch": 2.18, "learning_rate": 1.6881681950346677e-05, "loss": 2.0697, "step": 19480 }, { "epoch": 2.18, "learning_rate": 1.6771192127040926e-05, "loss": 2.0498, "step": 19500 }, { "epoch": 2.18, "learning_rate": 1.6654887049876983e-05, "loss": 2.0415, "step": 19520 }, { "epoch": 2.19, "learning_rate": 1.6538581972713035e-05, "loss": 2.0745, "step": 19540 }, { "epoch": 2.19, "learning_rate": 1.6422276895549095e-05, "loss": 2.0936, "step": 19560 }, { "epoch": 2.19, "learning_rate": 1.6305971818385148e-05, "loss": 2.0495, "step": 19580 }, { "epoch": 2.19, "learning_rate": 1.61896667412212e-05, "loss": 2.0615, "step": 19600 }, { "epoch": 2.19, "learning_rate": 1.6073361664057257e-05, "loss": 2.0566, "step": 19620 }, { "epoch": 2.2, "learning_rate": 1.595705658689331e-05, "loss": 2.064, "step": 19640 }, { "epoch": 2.2, "learning_rate": 1.5840751509729366e-05, "loss": 2.0304, "step": 19660 }, { "epoch": 2.2, "learning_rate": 1.572444643256542e-05, "loss": 2.0897, "step": 19680 }, { "epoch": 2.2, "learning_rate": 1.5608141355401475e-05, "loss": 2.069, "step": 19700 }, { "epoch": 2.21, "learning_rate": 1.549183627823753e-05, "loss": 2.0497, "step": 19720 }, { "epoch": 2.21, "learning_rate": 1.5375531201073584e-05, "loss": 2.0427, "step": 19740 }, { "epoch": 2.21, "learning_rate": 1.525922612390964e-05, "loss": 2.0802, "step": 19760 }, { "epoch": 2.21, "learning_rate": 1.5142921046745693e-05, "loss": 2.0825, "step": 19780 }, { "epoch": 2.21, "learning_rate": 1.5026615969581747e-05, "loss": 2.0549, "step": 19800 }, { "epoch": 2.22, "learning_rate": 1.4910310892417802e-05, "loss": 2.0846, "step": 19820 }, { "epoch": 2.22, "learning_rate": 1.4794005815253856e-05, "loss": 2.0599, "step": 19840 }, { "epoch": 2.22, "learning_rate": 1.4677700738089911e-05, "loss": 2.0198, "step": 19860 }, { "epoch": 2.22, "learning_rate": 1.4561395660925965e-05, "loss": 2.0434, "step": 19880 }, { "epoch": 2.23, "learning_rate": 1.4445090583762022e-05, "loss": 2.0567, "step": 19900 }, { "epoch": 2.23, "learning_rate": 1.4328785506598076e-05, "loss": 2.0546, "step": 19920 }, { "epoch": 2.23, "learning_rate": 1.421248042943413e-05, "loss": 2.0774, "step": 19940 }, { "epoch": 2.23, "learning_rate": 1.4096175352270185e-05, "loss": 2.0797, "step": 19960 }, { "epoch": 2.23, "learning_rate": 1.397987027510624e-05, "loss": 2.0676, "step": 19980 }, { "epoch": 2.24, "learning_rate": 1.3863565197942292e-05, "loss": 2.0514, "step": 20000 }, { "epoch": 2.24, "learning_rate": 1.3747260120778347e-05, "loss": 2.0721, "step": 20020 }, { "epoch": 2.24, "learning_rate": 1.3630955043614403e-05, "loss": 2.0641, "step": 20040 }, { "epoch": 2.24, "learning_rate": 1.3514649966450458e-05, "loss": 2.0452, "step": 20060 }, { "epoch": 2.25, "learning_rate": 1.3398344889286512e-05, "loss": 2.0922, "step": 20080 }, { "epoch": 2.25, "learning_rate": 1.3282039812122567e-05, "loss": 2.0634, "step": 20100 }, { "epoch": 2.25, "learning_rate": 1.3165734734958621e-05, "loss": 2.0843, "step": 20120 }, { "epoch": 2.25, "learning_rate": 1.3049429657794676e-05, "loss": 2.0678, "step": 20140 }, { "epoch": 2.25, "learning_rate": 1.293312458063073e-05, "loss": 2.0736, "step": 20160 }, { "epoch": 2.26, "learning_rate": 1.2816819503466786e-05, "loss": 2.068, "step": 20180 }, { "epoch": 2.26, "learning_rate": 1.2700514426302839e-05, "loss": 2.1375, "step": 20200 }, { "epoch": 2.26, "learning_rate": 1.2584209349138894e-05, "loss": 2.036, "step": 20220 }, { "epoch": 2.26, "learning_rate": 1.2467904271974948e-05, "loss": 2.0958, "step": 20240 }, { "epoch": 2.27, "learning_rate": 1.2351599194811003e-05, "loss": 2.0575, "step": 20260 }, { "epoch": 2.27, "learning_rate": 1.2235294117647057e-05, "loss": 2.022, "step": 20280 }, { "epoch": 2.27, "learning_rate": 1.2118989040483112e-05, "loss": 2.1452, "step": 20300 }, { "epoch": 2.27, "learning_rate": 1.2002683963319168e-05, "loss": 2.0374, "step": 20320 }, { "epoch": 2.27, "learning_rate": 1.1886378886155222e-05, "loss": 2.0504, "step": 20340 }, { "epoch": 2.28, "learning_rate": 1.1770073808991277e-05, "loss": 2.0327, "step": 20360 }, { "epoch": 2.28, "learning_rate": 1.1653768731827331e-05, "loss": 2.0743, "step": 20380 }, { "epoch": 2.28, "learning_rate": 1.1537463654663386e-05, "loss": 2.0258, "step": 20400 }, { "epoch": 2.28, "learning_rate": 1.1421158577499439e-05, "loss": 2.0431, "step": 20420 }, { "epoch": 2.29, "learning_rate": 1.1304853500335493e-05, "loss": 2.0772, "step": 20440 }, { "epoch": 2.29, "learning_rate": 1.118854842317155e-05, "loss": 2.046, "step": 20460 }, { "epoch": 2.29, "learning_rate": 1.1072243346007604e-05, "loss": 2.0498, "step": 20480 }, { "epoch": 2.29, "learning_rate": 1.0955938268843658e-05, "loss": 2.0543, "step": 20500 }, { "epoch": 2.29, "learning_rate": 1.0839633191679713e-05, "loss": 2.0573, "step": 20520 }, { "epoch": 2.3, "learning_rate": 1.0723328114515767e-05, "loss": 2.0651, "step": 20540 }, { "epoch": 2.3, "learning_rate": 1.0607023037351822e-05, "loss": 2.0172, "step": 20560 }, { "epoch": 2.3, "learning_rate": 1.0490717960187878e-05, "loss": 2.0421, "step": 20580 }, { "epoch": 2.3, "learning_rate": 1.0374412883023932e-05, "loss": 2.0252, "step": 20600 }, { "epoch": 2.31, "learning_rate": 1.0258107805859985e-05, "loss": 2.0479, "step": 20620 }, { "epoch": 2.31, "learning_rate": 1.014180272869604e-05, "loss": 2.0661, "step": 20640 }, { "epoch": 2.31, "learning_rate": 1.0025497651532094e-05, "loss": 2.0576, "step": 20660 }, { "epoch": 2.31, "learning_rate": 9.909192574368149e-06, "loss": 2.0088, "step": 20680 }, { "epoch": 2.31, "learning_rate": 9.792887497204203e-06, "loss": 2.0433, "step": 20700 }, { "epoch": 2.32, "learning_rate": 9.67658242004026e-06, "loss": 2.0522, "step": 20720 }, { "epoch": 2.32, "learning_rate": 9.560277342876314e-06, "loss": 2.0749, "step": 20740 }, { "epoch": 2.32, "learning_rate": 9.443972265712368e-06, "loss": 2.062, "step": 20760 }, { "epoch": 2.32, "learning_rate": 9.327667188548423e-06, "loss": 2.076, "step": 20780 }, { "epoch": 2.33, "learning_rate": 9.211362111384477e-06, "loss": 2.0607, "step": 20800 }, { "epoch": 2.33, "learning_rate": 9.095057034220532e-06, "loss": 2.0379, "step": 20820 }, { "epoch": 2.33, "learning_rate": 8.978751957056585e-06, "loss": 2.0791, "step": 20840 }, { "epoch": 2.33, "learning_rate": 8.862446879892641e-06, "loss": 2.0544, "step": 20860 }, { "epoch": 2.34, "learning_rate": 8.746141802728695e-06, "loss": 2.028, "step": 20880 }, { "epoch": 2.34, "learning_rate": 8.62983672556475e-06, "loss": 2.0247, "step": 20900 }, { "epoch": 2.34, "learning_rate": 8.513531648400804e-06, "loss": 2.0644, "step": 20920 }, { "epoch": 2.34, "learning_rate": 8.397226571236859e-06, "loss": 2.0914, "step": 20940 }, { "epoch": 2.34, "learning_rate": 8.280921494072913e-06, "loss": 2.0588, "step": 20960 }, { "epoch": 2.35, "learning_rate": 8.164616416908968e-06, "loss": 2.082, "step": 20980 }, { "epoch": 2.35, "learning_rate": 8.048311339745022e-06, "loss": 2.0454, "step": 21000 }, { "epoch": 2.35, "learning_rate": 7.932006262581077e-06, "loss": 2.0487, "step": 21020 }, { "epoch": 2.35, "learning_rate": 7.815701185417131e-06, "loss": 2.073, "step": 21040 }, { "epoch": 2.36, "learning_rate": 7.699396108253186e-06, "loss": 2.0735, "step": 21060 }, { "epoch": 2.36, "learning_rate": 7.583091031089241e-06, "loss": 2.1009, "step": 21080 }, { "epoch": 2.36, "learning_rate": 7.466785953925296e-06, "loss": 2.0592, "step": 21100 }, { "epoch": 2.36, "learning_rate": 7.35048087676135e-06, "loss": 2.0746, "step": 21120 }, { "epoch": 2.36, "learning_rate": 7.234175799597405e-06, "loss": 2.0362, "step": 21140 }, { "epoch": 2.37, "learning_rate": 7.117870722433459e-06, "loss": 2.0595, "step": 21160 }, { "epoch": 2.37, "learning_rate": 7.001565645269515e-06, "loss": 2.0417, "step": 21180 }, { "epoch": 2.37, "learning_rate": 6.885260568105568e-06, "loss": 2.0312, "step": 21200 }, { "epoch": 2.37, "learning_rate": 6.768955490941623e-06, "loss": 2.0407, "step": 21220 }, { "epoch": 2.38, "learning_rate": 6.652650413777678e-06, "loss": 2.0756, "step": 21240 }, { "epoch": 2.38, "learning_rate": 6.536345336613733e-06, "loss": 2.0649, "step": 21260 }, { "epoch": 2.38, "learning_rate": 6.420040259449787e-06, "loss": 2.0822, "step": 21280 }, { "epoch": 2.38, "learning_rate": 6.303735182285841e-06, "loss": 2.0615, "step": 21300 }, { "epoch": 2.38, "learning_rate": 6.187430105121896e-06, "loss": 2.0602, "step": 21320 }, { "epoch": 2.39, "learning_rate": 6.071125027957951e-06, "loss": 2.0534, "step": 21340 }, { "epoch": 2.39, "learning_rate": 5.954819950794005e-06, "loss": 2.0889, "step": 21360 }, { "epoch": 2.39, "learning_rate": 5.8385148736300605e-06, "loss": 2.0617, "step": 21380 }, { "epoch": 2.39, "learning_rate": 5.722209796466114e-06, "loss": 2.0512, "step": 21400 }, { "epoch": 2.4, "learning_rate": 5.605904719302169e-06, "loss": 2.0741, "step": 21420 }, { "epoch": 2.4, "learning_rate": 5.489599642138223e-06, "loss": 2.1048, "step": 21440 }, { "epoch": 2.4, "learning_rate": 5.3732945649742785e-06, "loss": 2.0448, "step": 21460 }, { "epoch": 2.4, "learning_rate": 5.256989487810333e-06, "loss": 2.0922, "step": 21480 }, { "epoch": 2.4, "learning_rate": 5.152314918362782e-06, "loss": 2.0596, "step": 21500 }, { "epoch": 2.41, "learning_rate": 5.036009841198836e-06, "loss": 2.0856, "step": 21520 }, { "epoch": 2.41, "learning_rate": 4.9197047640348906e-06, "loss": 2.0461, "step": 21540 }, { "epoch": 2.41, "learning_rate": 4.803399686870946e-06, "loss": 2.0339, "step": 21560 }, { "epoch": 2.41, "learning_rate": 4.687094609707e-06, "loss": 2.0895, "step": 21580 }, { "epoch": 2.42, "learning_rate": 4.570789532543055e-06, "loss": 2.0713, "step": 21600 }, { "epoch": 2.42, "learning_rate": 4.454484455379109e-06, "loss": 2.0354, "step": 21620 }, { "epoch": 2.42, "learning_rate": 4.338179378215164e-06, "loss": 2.0733, "step": 21640 }, { "epoch": 2.42, "learning_rate": 4.221874301051218e-06, "loss": 2.041, "step": 21660 }, { "epoch": 2.42, "learning_rate": 4.105569223887274e-06, "loss": 2.0864, "step": 21680 }, { "epoch": 2.43, "learning_rate": 3.989264146723327e-06, "loss": 2.0765, "step": 21700 }, { "epoch": 2.43, "learning_rate": 3.872959069559383e-06, "loss": 2.0272, "step": 21720 }, { "epoch": 2.43, "learning_rate": 3.756653992395437e-06, "loss": 2.1094, "step": 21740 }, { "epoch": 2.43, "learning_rate": 3.6403489152314913e-06, "loss": 2.0012, "step": 21760 }, { "epoch": 2.44, "learning_rate": 3.5240438380675463e-06, "loss": 2.0396, "step": 21780 }, { "epoch": 2.44, "learning_rate": 3.4077387609036008e-06, "loss": 2.0714, "step": 21800 }, { "epoch": 2.44, "learning_rate": 3.2914336837396553e-06, "loss": 2.0882, "step": 21820 }, { "epoch": 2.44, "learning_rate": 3.1751286065757098e-06, "loss": 2.0789, "step": 21840 }, { "epoch": 2.44, "learning_rate": 3.0588235294117643e-06, "loss": 2.0701, "step": 21860 }, { "epoch": 2.45, "learning_rate": 2.9483337061060162e-06, "loss": 2.0979, "step": 21880 }, { "epoch": 2.45, "learning_rate": 2.8320286289420707e-06, "loss": 2.0812, "step": 21900 }, { "epoch": 2.45, "learning_rate": 2.7157235517781256e-06, "loss": 2.1, "step": 21920 }, { "epoch": 2.45, "learning_rate": 2.59941847461418e-06, "loss": 2.0769, "step": 21940 }, { "epoch": 2.46, "learning_rate": 2.4831133974502346e-06, "loss": 2.0691, "step": 21960 }, { "epoch": 2.46, "learning_rate": 2.366808320286289e-06, "loss": 2.0583, "step": 21980 }, { "epoch": 2.46, "learning_rate": 2.2505032431223436e-06, "loss": 2.057, "step": 22000 }, { "epoch": 2.46, "learning_rate": 2.1341981659583986e-06, "loss": 2.1085, "step": 22020 }, { "epoch": 2.46, "learning_rate": 2.017893088794453e-06, "loss": 2.0809, "step": 22040 }, { "epoch": 2.47, "learning_rate": 1.9015880116305076e-06, "loss": 2.0627, "step": 22060 }, { "epoch": 2.47, "learning_rate": 1.7852829344665623e-06, "loss": 2.0509, "step": 22080 }, { "epoch": 2.47, "learning_rate": 1.6689778573026168e-06, "loss": 2.0466, "step": 22100 }, { "epoch": 2.47, "learning_rate": 1.5526727801386713e-06, "loss": 2.0756, "step": 22120 }, { "epoch": 2.48, "learning_rate": 1.4363677029747258e-06, "loss": 2.038, "step": 22140 }, { "epoch": 2.48, "learning_rate": 1.3200626258107805e-06, "loss": 2.0362, "step": 22160 }, { "epoch": 2.48, "learning_rate": 1.203757548646835e-06, "loss": 2.0785, "step": 22180 }, { "epoch": 2.48, "learning_rate": 1.0874524714828897e-06, "loss": 2.0653, "step": 22200 }, { "epoch": 2.48, "learning_rate": 9.711473943189442e-07, "loss": 2.1218, "step": 22220 }, { "epoch": 2.49, "learning_rate": 8.548423171549988e-07, "loss": 2.0656, "step": 22240 }, { "epoch": 2.49, "learning_rate": 7.385372399910534e-07, "loss": 2.0818, "step": 22260 }, { "epoch": 2.49, "learning_rate": 6.222321628271079e-07, "loss": 2.0546, "step": 22280 }, { "epoch": 2.49, "learning_rate": 5.059270856631625e-07, "loss": 2.0465, "step": 22300 }, { "epoch": 2.5, "learning_rate": 3.8962200849921713e-07, "loss": 2.0968, "step": 22320 }, { "epoch": 2.5, "learning_rate": 2.7331693133527173e-07, "loss": 2.038, "step": 22340 }, { "epoch": 2.5, "step": 22355, "total_flos": 2.924979686683443e+17, "train_loss": 2.08698912321708, "train_runtime": 8788.5423, "train_samples_per_second": 15.261, "train_steps_per_second": 2.544 } ], "logging_steps": 20, "max_steps": 22355, "num_train_epochs": 3, "save_steps": 5000, "total_flos": 2.924979686683443e+17, "trial_name": null, "trial_params": null }