| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 97.62050030506407, | |
| "eval_steps": 500, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.09762050030506407, | |
| "grad_norm": 0.5707364082336426, | |
| "learning_rate": 4e-05, | |
| "loss": 2.2566, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19524100061012814, | |
| "grad_norm": 0.7393651008605957, | |
| "learning_rate": 8e-05, | |
| "loss": 2.203, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2928615009151922, | |
| "grad_norm": 0.8320680856704712, | |
| "learning_rate": 0.00012, | |
| "loss": 1.9533, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3904820012202563, | |
| "grad_norm": 1.2113606929779053, | |
| "learning_rate": 0.00016, | |
| "loss": 1.8878, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4881025015253203, | |
| "grad_norm": 1.4215683937072754, | |
| "learning_rate": 0.0002, | |
| "loss": 1.7688, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5857230018303844, | |
| "grad_norm": 1.8242409229278564, | |
| "learning_rate": 0.0001998688524590164, | |
| "loss": 1.7308, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6833435021354485, | |
| "grad_norm": 1.6889506578445435, | |
| "learning_rate": 0.0001997377049180328, | |
| "loss": 1.6553, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.7809640024405126, | |
| "grad_norm": 2.1998350620269775, | |
| "learning_rate": 0.00019960655737704918, | |
| "loss": 1.5954, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.8785845027455765, | |
| "grad_norm": 1.5626238584518433, | |
| "learning_rate": 0.0001994754098360656, | |
| "loss": 1.5732, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.9762050030506406, | |
| "grad_norm": 2.350252389907837, | |
| "learning_rate": 0.00019934426229508198, | |
| "loss": 1.5001, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.0738255033557047, | |
| "grad_norm": 1.7499576807022095, | |
| "learning_rate": 0.00019921311475409837, | |
| "loss": 1.5056, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.1714460036607688, | |
| "grad_norm": 2.5555903911590576, | |
| "learning_rate": 0.00019908196721311476, | |
| "loss": 1.3946, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.2690665039658329, | |
| "grad_norm": 2.44691801071167, | |
| "learning_rate": 0.00019895081967213115, | |
| "loss": 1.4672, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.366687004270897, | |
| "grad_norm": 1.75604248046875, | |
| "learning_rate": 0.00019881967213114757, | |
| "loss": 1.4255, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.4643075045759608, | |
| "grad_norm": 2.170323610305786, | |
| "learning_rate": 0.00019868852459016393, | |
| "loss": 1.3473, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.561928004881025, | |
| "grad_norm": 2.3095695972442627, | |
| "learning_rate": 0.00019855737704918035, | |
| "loss": 1.3471, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.659548505186089, | |
| "grad_norm": 2.668290138244629, | |
| "learning_rate": 0.00019842622950819674, | |
| "loss": 1.2861, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.757169005491153, | |
| "grad_norm": 2.395897150039673, | |
| "learning_rate": 0.00019829508196721313, | |
| "loss": 1.2914, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.8547895057962172, | |
| "grad_norm": 2.7138261795043945, | |
| "learning_rate": 0.00019816393442622951, | |
| "loss": 1.3455, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.9524100061012812, | |
| "grad_norm": 2.28128981590271, | |
| "learning_rate": 0.0001980327868852459, | |
| "loss": 1.2907, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.0500305064063453, | |
| "grad_norm": 2.284174919128418, | |
| "learning_rate": 0.00019790163934426232, | |
| "loss": 1.2074, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.1476510067114094, | |
| "grad_norm": 2.7497994899749756, | |
| "learning_rate": 0.00019777049180327868, | |
| "loss": 1.2135, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.2452715070164735, | |
| "grad_norm": 3.0772790908813477, | |
| "learning_rate": 0.0001976393442622951, | |
| "loss": 1.1622, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.3428920073215376, | |
| "grad_norm": 2.3680825233459473, | |
| "learning_rate": 0.0001975081967213115, | |
| "loss": 1.2219, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.4405125076266017, | |
| "grad_norm": 3.132049798965454, | |
| "learning_rate": 0.00019737704918032788, | |
| "loss": 1.2523, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.5381330079316657, | |
| "grad_norm": 2.8542871475219727, | |
| "learning_rate": 0.00019724590163934427, | |
| "loss": 1.1163, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.63575350823673, | |
| "grad_norm": 2.5308499336242676, | |
| "learning_rate": 0.00019711475409836066, | |
| "loss": 1.1624, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.733374008541794, | |
| "grad_norm": 2.3678996562957764, | |
| "learning_rate": 0.00019698360655737707, | |
| "loss": 1.1856, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.830994508846858, | |
| "grad_norm": 3.0295815467834473, | |
| "learning_rate": 0.00019685245901639344, | |
| "loss": 1.1193, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.9286150091519216, | |
| "grad_norm": 3.7163286209106445, | |
| "learning_rate": 0.00019672131147540985, | |
| "loss": 1.1479, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.026235509456986, | |
| "grad_norm": 2.840359926223755, | |
| "learning_rate": 0.00019659016393442624, | |
| "loss": 1.1554, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.1238560097620502, | |
| "grad_norm": 2.579162836074829, | |
| "learning_rate": 0.00019645901639344263, | |
| "loss": 1.0088, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.221476510067114, | |
| "grad_norm": 2.772268295288086, | |
| "learning_rate": 0.00019632786885245902, | |
| "loss": 1.0412, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.319097010372178, | |
| "grad_norm": 3.2339484691619873, | |
| "learning_rate": 0.0001961967213114754, | |
| "loss": 1.043, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.416717510677242, | |
| "grad_norm": 3.7858259677886963, | |
| "learning_rate": 0.00019606557377049183, | |
| "loss": 1.1187, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.514338010982306, | |
| "grad_norm": 3.17150616645813, | |
| "learning_rate": 0.00019593442622950822, | |
| "loss": 1.0731, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.61195851128737, | |
| "grad_norm": 2.9254567623138428, | |
| "learning_rate": 0.0001958032786885246, | |
| "loss": 1.0242, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.7095790115924343, | |
| "grad_norm": 2.4400126934051514, | |
| "learning_rate": 0.000195672131147541, | |
| "loss": 1.0911, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.8071995118974984, | |
| "grad_norm": 2.7202625274658203, | |
| "learning_rate": 0.00019554098360655738, | |
| "loss": 1.0451, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.9048200122025625, | |
| "grad_norm": 3.0309741497039795, | |
| "learning_rate": 0.00019540983606557377, | |
| "loss": 1.0188, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.002440512507627, | |
| "grad_norm": 4.291023254394531, | |
| "learning_rate": 0.00019527868852459016, | |
| "loss": 1.0858, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 4.100061012812691, | |
| "grad_norm": 3.4548487663269043, | |
| "learning_rate": 0.00019514754098360658, | |
| "loss": 0.9553, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 4.197681513117755, | |
| "grad_norm": 2.882427453994751, | |
| "learning_rate": 0.00019501639344262297, | |
| "loss": 0.9491, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 4.295302013422819, | |
| "grad_norm": 3.2026467323303223, | |
| "learning_rate": 0.00019488524590163936, | |
| "loss": 0.9529, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 4.392922513727883, | |
| "grad_norm": 2.676392078399658, | |
| "learning_rate": 0.00019475409836065575, | |
| "loss": 0.9296, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.490543014032947, | |
| "grad_norm": 2.8527650833129883, | |
| "learning_rate": 0.00019462295081967214, | |
| "loss": 0.9826, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.588163514338011, | |
| "grad_norm": 2.832995891571045, | |
| "learning_rate": 0.00019449180327868855, | |
| "loss": 0.9337, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.685784014643075, | |
| "grad_norm": 2.9931800365448, | |
| "learning_rate": 0.00019436065573770491, | |
| "loss": 0.9455, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.783404514948139, | |
| "grad_norm": 3.4157402515411377, | |
| "learning_rate": 0.00019422950819672133, | |
| "loss": 1.0286, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 4.881025015253203, | |
| "grad_norm": 2.846972942352295, | |
| "learning_rate": 0.00019409836065573772, | |
| "loss": 0.9536, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.978645515558267, | |
| "grad_norm": 3.4363763332366943, | |
| "learning_rate": 0.0001939672131147541, | |
| "loss": 0.967, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 5.0762660158633315, | |
| "grad_norm": 3.347905397415161, | |
| "learning_rate": 0.0001938360655737705, | |
| "loss": 0.9291, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 5.173886516168396, | |
| "grad_norm": 3.016284704208374, | |
| "learning_rate": 0.0001937049180327869, | |
| "loss": 0.8722, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 5.27150701647346, | |
| "grad_norm": 3.5366976261138916, | |
| "learning_rate": 0.0001935737704918033, | |
| "loss": 0.8839, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 5.369127516778524, | |
| "grad_norm": 3.550899028778076, | |
| "learning_rate": 0.00019344262295081967, | |
| "loss": 0.8386, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.466748017083588, | |
| "grad_norm": 3.651362180709839, | |
| "learning_rate": 0.00019331147540983608, | |
| "loss": 0.8998, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 5.564368517388652, | |
| "grad_norm": 2.9918806552886963, | |
| "learning_rate": 0.00019318032786885247, | |
| "loss": 0.8472, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.661989017693716, | |
| "grad_norm": 3.7066636085510254, | |
| "learning_rate": 0.00019304918032786886, | |
| "loss": 0.8966, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 5.75960951799878, | |
| "grad_norm": 3.1231799125671387, | |
| "learning_rate": 0.00019291803278688525, | |
| "loss": 0.8989, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 5.857230018303844, | |
| "grad_norm": 3.0318100452423096, | |
| "learning_rate": 0.00019278688524590164, | |
| "loss": 0.9222, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.954850518608908, | |
| "grad_norm": 3.3124337196350098, | |
| "learning_rate": 0.00019265573770491806, | |
| "loss": 0.8827, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 6.052471018913972, | |
| "grad_norm": 3.1748123168945312, | |
| "learning_rate": 0.00019252459016393442, | |
| "loss": 0.845, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 6.150091519219036, | |
| "grad_norm": 3.320768356323242, | |
| "learning_rate": 0.00019239344262295084, | |
| "loss": 0.7315, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 6.2477120195241005, | |
| "grad_norm": 4.226164817810059, | |
| "learning_rate": 0.00019226229508196723, | |
| "loss": 0.7999, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 6.345332519829164, | |
| "grad_norm": 3.12298583984375, | |
| "learning_rate": 0.00019213114754098362, | |
| "loss": 0.8337, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.442953020134228, | |
| "grad_norm": 3.4633209705352783, | |
| "learning_rate": 0.000192, | |
| "loss": 0.7834, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 6.540573520439292, | |
| "grad_norm": 3.116576671600342, | |
| "learning_rate": 0.0001918688524590164, | |
| "loss": 0.8386, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 6.638194020744356, | |
| "grad_norm": 3.255690336227417, | |
| "learning_rate": 0.0001917377049180328, | |
| "loss": 0.902, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 6.73581452104942, | |
| "grad_norm": 4.117364406585693, | |
| "learning_rate": 0.00019160655737704917, | |
| "loss": 0.7897, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 6.833435021354484, | |
| "grad_norm": 3.813042163848877, | |
| "learning_rate": 0.0001914754098360656, | |
| "loss": 0.7975, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.931055521659548, | |
| "grad_norm": 3.2562716007232666, | |
| "learning_rate": 0.00019134426229508198, | |
| "loss": 0.885, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 7.028676021964612, | |
| "grad_norm": 3.0337297916412354, | |
| "learning_rate": 0.00019121311475409837, | |
| "loss": 0.7571, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 7.126296522269676, | |
| "grad_norm": 3.4187958240509033, | |
| "learning_rate": 0.00019108196721311476, | |
| "loss": 0.7317, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 7.22391702257474, | |
| "grad_norm": 3.2497494220733643, | |
| "learning_rate": 0.00019095081967213115, | |
| "loss": 0.6953, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 7.3215375228798045, | |
| "grad_norm": 3.452319383621216, | |
| "learning_rate": 0.00019081967213114756, | |
| "loss": 0.7026, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.419158023184869, | |
| "grad_norm": 4.487509250640869, | |
| "learning_rate": 0.00019068852459016395, | |
| "loss": 0.7552, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 7.516778523489933, | |
| "grad_norm": 3.4719717502593994, | |
| "learning_rate": 0.00019055737704918034, | |
| "loss": 0.7426, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 7.614399023794997, | |
| "grad_norm": 3.9266843795776367, | |
| "learning_rate": 0.00019042622950819673, | |
| "loss": 0.762, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 7.712019524100061, | |
| "grad_norm": 3.0717546939849854, | |
| "learning_rate": 0.00019029508196721312, | |
| "loss": 0.7574, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 7.809640024405125, | |
| "grad_norm": 3.793149948120117, | |
| "learning_rate": 0.0001901639344262295, | |
| "loss": 0.7817, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 7.907260524710189, | |
| "grad_norm": 3.288499593734741, | |
| "learning_rate": 0.0001900327868852459, | |
| "loss": 0.7759, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 8.004881025015253, | |
| "grad_norm": 2.6891720294952393, | |
| "learning_rate": 0.00018990163934426232, | |
| "loss": 0.8212, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 8.102501525320317, | |
| "grad_norm": 3.331376075744629, | |
| "learning_rate": 0.0001897704918032787, | |
| "loss": 0.6659, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 8.200122025625381, | |
| "grad_norm": 3.3468194007873535, | |
| "learning_rate": 0.0001896393442622951, | |
| "loss": 0.6458, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 8.297742525930445, | |
| "grad_norm": 4.916980266571045, | |
| "learning_rate": 0.00018950819672131148, | |
| "loss": 0.6748, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 8.39536302623551, | |
| "grad_norm": 3.4678823947906494, | |
| "learning_rate": 0.00018937704918032787, | |
| "loss": 0.6349, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 8.492983526540574, | |
| "grad_norm": 4.540040493011475, | |
| "learning_rate": 0.0001892459016393443, | |
| "loss": 0.7256, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 8.590604026845638, | |
| "grad_norm": 3.6092071533203125, | |
| "learning_rate": 0.00018911475409836065, | |
| "loss": 0.6708, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 8.688224527150702, | |
| "grad_norm": 4.159832954406738, | |
| "learning_rate": 0.00018898360655737707, | |
| "loss": 0.6974, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 8.785845027455766, | |
| "grad_norm": 3.980943202972412, | |
| "learning_rate": 0.00018885245901639346, | |
| "loss": 0.7268, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 8.88346552776083, | |
| "grad_norm": 3.106861114501953, | |
| "learning_rate": 0.00018872131147540985, | |
| "loss": 0.7583, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 8.981086028065894, | |
| "grad_norm": 2.672461748123169, | |
| "learning_rate": 0.00018859016393442624, | |
| "loss": 0.7199, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 9.078706528370958, | |
| "grad_norm": 3.1480796337127686, | |
| "learning_rate": 0.00018845901639344263, | |
| "loss": 0.6583, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 9.176327028676022, | |
| "grad_norm": 3.3770973682403564, | |
| "learning_rate": 0.00018832786885245904, | |
| "loss": 0.6317, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 9.273947528981086, | |
| "grad_norm": 3.452083110809326, | |
| "learning_rate": 0.0001881967213114754, | |
| "loss": 0.6166, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 9.37156802928615, | |
| "grad_norm": 3.6544485092163086, | |
| "learning_rate": 0.00018806557377049182, | |
| "loss": 0.6128, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 9.469188529591214, | |
| "grad_norm": 3.462320566177368, | |
| "learning_rate": 0.0001879344262295082, | |
| "loss": 0.6442, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 9.566809029896278, | |
| "grad_norm": 3.0532991886138916, | |
| "learning_rate": 0.0001878032786885246, | |
| "loss": 0.6631, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 9.664429530201343, | |
| "grad_norm": 4.109907627105713, | |
| "learning_rate": 0.000187672131147541, | |
| "loss": 0.6634, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 9.762050030506407, | |
| "grad_norm": 3.7715537548065186, | |
| "learning_rate": 0.00018754098360655738, | |
| "loss": 0.6004, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 9.85967053081147, | |
| "grad_norm": 4.588443756103516, | |
| "learning_rate": 0.0001874098360655738, | |
| "loss": 0.6816, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 9.957291031116535, | |
| "grad_norm": 3.2902603149414062, | |
| "learning_rate": 0.00018727868852459016, | |
| "loss": 0.682, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 10.054911531421599, | |
| "grad_norm": 3.070920705795288, | |
| "learning_rate": 0.00018714754098360657, | |
| "loss": 0.6007, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 10.152532031726663, | |
| "grad_norm": 3.761850357055664, | |
| "learning_rate": 0.00018701639344262296, | |
| "loss": 0.5478, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 10.250152532031727, | |
| "grad_norm": 2.9650540351867676, | |
| "learning_rate": 0.00018688524590163935, | |
| "loss": 0.5792, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 10.347773032336791, | |
| "grad_norm": 3.3296761512756348, | |
| "learning_rate": 0.00018675409836065574, | |
| "loss": 0.5849, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 10.445393532641855, | |
| "grad_norm": 4.055027008056641, | |
| "learning_rate": 0.00018662295081967213, | |
| "loss": 0.5417, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 10.54301403294692, | |
| "grad_norm": 3.581878662109375, | |
| "learning_rate": 0.00018649180327868855, | |
| "loss": 0.6285, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 10.640634533251983, | |
| "grad_norm": 3.667356491088867, | |
| "learning_rate": 0.00018636065573770494, | |
| "loss": 0.5767, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 10.738255033557047, | |
| "grad_norm": 4.004351615905762, | |
| "learning_rate": 0.00018622950819672133, | |
| "loss": 0.6339, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 10.835875533862112, | |
| "grad_norm": 3.4327890872955322, | |
| "learning_rate": 0.00018609836065573772, | |
| "loss": 0.6258, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 10.933496034167176, | |
| "grad_norm": 3.3382508754730225, | |
| "learning_rate": 0.0001859672131147541, | |
| "loss": 0.641, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 11.03111653447224, | |
| "grad_norm": 2.9254095554351807, | |
| "learning_rate": 0.0001858360655737705, | |
| "loss": 0.5932, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 11.128737034777304, | |
| "grad_norm": 4.12520694732666, | |
| "learning_rate": 0.00018570491803278688, | |
| "loss": 0.5007, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 11.226357535082368, | |
| "grad_norm": 3.348695993423462, | |
| "learning_rate": 0.0001855737704918033, | |
| "loss": 0.4967, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 11.323978035387432, | |
| "grad_norm": 3.62343430519104, | |
| "learning_rate": 0.0001854426229508197, | |
| "loss": 0.5312, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 11.421598535692496, | |
| "grad_norm": 3.4885001182556152, | |
| "learning_rate": 0.00018531147540983608, | |
| "loss": 0.5675, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 11.51921903599756, | |
| "grad_norm": 4.001424789428711, | |
| "learning_rate": 0.00018518032786885247, | |
| "loss": 0.5654, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 11.616839536302624, | |
| "grad_norm": 3.180741310119629, | |
| "learning_rate": 0.00018504918032786886, | |
| "loss": 0.5473, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 11.714460036607688, | |
| "grad_norm": 3.3385472297668457, | |
| "learning_rate": 0.00018491803278688527, | |
| "loss": 0.601, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 11.812080536912752, | |
| "grad_norm": 3.1975746154785156, | |
| "learning_rate": 0.00018478688524590164, | |
| "loss": 0.559, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 11.909701037217816, | |
| "grad_norm": 4.126131057739258, | |
| "learning_rate": 0.00018465573770491805, | |
| "loss": 0.517, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 12.00732153752288, | |
| "grad_norm": 3.1047425270080566, | |
| "learning_rate": 0.00018452459016393444, | |
| "loss": 0.5853, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 12.104942037827945, | |
| "grad_norm": 4.153341293334961, | |
| "learning_rate": 0.0001843934426229508, | |
| "loss": 0.4683, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 12.202562538133009, | |
| "grad_norm": 3.550233840942383, | |
| "learning_rate": 0.00018426229508196722, | |
| "loss": 0.4802, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 12.300183038438073, | |
| "grad_norm": 3.3238728046417236, | |
| "learning_rate": 0.0001841311475409836, | |
| "loss": 0.5219, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 12.397803538743137, | |
| "grad_norm": 4.092185974121094, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 0.4783, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 12.495424039048201, | |
| "grad_norm": 3.4403531551361084, | |
| "learning_rate": 0.0001838688524590164, | |
| "loss": 0.5247, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 12.593044539353265, | |
| "grad_norm": 4.026747226715088, | |
| "learning_rate": 0.0001837377049180328, | |
| "loss": 0.4838, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 12.690665039658327, | |
| "grad_norm": 2.7140889167785645, | |
| "learning_rate": 0.0001836065573770492, | |
| "loss": 0.5417, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 12.788285539963393, | |
| "grad_norm": 4.172866344451904, | |
| "learning_rate": 0.00018347540983606558, | |
| "loss": 0.5133, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 12.885906040268456, | |
| "grad_norm": 4.674078464508057, | |
| "learning_rate": 0.00018334426229508197, | |
| "loss": 0.5093, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 12.98352654057352, | |
| "grad_norm": 3.6748011112213135, | |
| "learning_rate": 0.00018321311475409836, | |
| "loss": 0.5365, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 13.081147040878584, | |
| "grad_norm": 5.05755615234375, | |
| "learning_rate": 0.00018308196721311478, | |
| "loss": 0.4356, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 13.178767541183648, | |
| "grad_norm": 3.9897286891937256, | |
| "learning_rate": 0.00018295081967213114, | |
| "loss": 0.4228, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 13.276388041488712, | |
| "grad_norm": 3.8581581115722656, | |
| "learning_rate": 0.00018281967213114756, | |
| "loss": 0.4019, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 13.374008541793776, | |
| "grad_norm": 3.3325164318084717, | |
| "learning_rate": 0.00018268852459016395, | |
| "loss": 0.4715, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 13.47162904209884, | |
| "grad_norm": 3.3978488445281982, | |
| "learning_rate": 0.00018255737704918034, | |
| "loss": 0.4634, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 13.569249542403904, | |
| "grad_norm": 3.491314172744751, | |
| "learning_rate": 0.00018242622950819673, | |
| "loss": 0.4942, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 13.666870042708968, | |
| "grad_norm": 3.275783061981201, | |
| "learning_rate": 0.00018229508196721312, | |
| "loss": 0.4807, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 13.764490543014032, | |
| "grad_norm": 4.8541717529296875, | |
| "learning_rate": 0.00018216393442622953, | |
| "loss": 0.4684, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 13.862111043319096, | |
| "grad_norm": 4.203775882720947, | |
| "learning_rate": 0.00018203278688524592, | |
| "loss": 0.515, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 13.95973154362416, | |
| "grad_norm": 3.885730504989624, | |
| "learning_rate": 0.0001819016393442623, | |
| "loss": 0.5263, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 14.057352043929225, | |
| "grad_norm": 4.247331619262695, | |
| "learning_rate": 0.0001817704918032787, | |
| "loss": 0.4553, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 14.154972544234289, | |
| "grad_norm": 4.659804344177246, | |
| "learning_rate": 0.0001816393442622951, | |
| "loss": 0.403, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 14.252593044539353, | |
| "grad_norm": 3.525756359100342, | |
| "learning_rate": 0.00018150819672131148, | |
| "loss": 0.4128, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 14.350213544844417, | |
| "grad_norm": 3.7535457611083984, | |
| "learning_rate": 0.00018137704918032787, | |
| "loss": 0.3969, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 14.44783404514948, | |
| "grad_norm": 3.828665256500244, | |
| "learning_rate": 0.00018124590163934429, | |
| "loss": 0.4321, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 14.545454545454545, | |
| "grad_norm": 3.7969393730163574, | |
| "learning_rate": 0.00018111475409836067, | |
| "loss": 0.4382, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 14.643075045759609, | |
| "grad_norm": 4.983769416809082, | |
| "learning_rate": 0.00018098360655737704, | |
| "loss": 0.4656, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 14.740695546064673, | |
| "grad_norm": 3.585157632827759, | |
| "learning_rate": 0.00018085245901639345, | |
| "loss": 0.4576, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 14.838316046369737, | |
| "grad_norm": 3.859992265701294, | |
| "learning_rate": 0.00018072131147540984, | |
| "loss": 0.4712, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 14.935936546674801, | |
| "grad_norm": 3.2431323528289795, | |
| "learning_rate": 0.00018059016393442626, | |
| "loss": 0.4526, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 15.033557046979865, | |
| "grad_norm": 3.1578800678253174, | |
| "learning_rate": 0.00018045901639344262, | |
| "loss": 0.4129, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 15.13117754728493, | |
| "grad_norm": 4.049219608306885, | |
| "learning_rate": 0.00018032786885245904, | |
| "loss": 0.3485, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 15.228798047589994, | |
| "grad_norm": 3.642855167388916, | |
| "learning_rate": 0.00018019672131147543, | |
| "loss": 0.3607, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 15.326418547895058, | |
| "grad_norm": 3.5239570140838623, | |
| "learning_rate": 0.0001800655737704918, | |
| "loss": 0.3844, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 15.424039048200122, | |
| "grad_norm": 2.894887685775757, | |
| "learning_rate": 0.0001799344262295082, | |
| "loss": 0.4111, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 15.521659548505186, | |
| "grad_norm": 3.1862661838531494, | |
| "learning_rate": 0.0001798032786885246, | |
| "loss": 0.4, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 15.61928004881025, | |
| "grad_norm": 3.169221878051758, | |
| "learning_rate": 0.000179672131147541, | |
| "loss": 0.3914, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 15.716900549115314, | |
| "grad_norm": 4.413252830505371, | |
| "learning_rate": 0.00017954098360655737, | |
| "loss": 0.4268, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 15.814521049420378, | |
| "grad_norm": 4.1161208152771, | |
| "learning_rate": 0.0001794098360655738, | |
| "loss": 0.4261, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 15.912141549725442, | |
| "grad_norm": 3.044705867767334, | |
| "learning_rate": 0.00017927868852459018, | |
| "loss": 0.4628, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 16.009762050030506, | |
| "grad_norm": 3.2134783267974854, | |
| "learning_rate": 0.00017914754098360657, | |
| "loss": 0.4368, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 16.107382550335572, | |
| "grad_norm": 3.8068933486938477, | |
| "learning_rate": 0.00017901639344262296, | |
| "loss": 0.2993, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 16.205003050640634, | |
| "grad_norm": 4.097984313964844, | |
| "learning_rate": 0.00017888524590163935, | |
| "loss": 0.3458, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 16.3026235509457, | |
| "grad_norm": 3.5351762771606445, | |
| "learning_rate": 0.00017875409836065576, | |
| "loss": 0.3449, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 16.400244051250763, | |
| "grad_norm": 3.3672287464141846, | |
| "learning_rate": 0.00017862295081967213, | |
| "loss": 0.3708, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 16.49786455155583, | |
| "grad_norm": 3.6430435180664062, | |
| "learning_rate": 0.00017849180327868852, | |
| "loss": 0.369, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 16.59548505186089, | |
| "grad_norm": 4.193872928619385, | |
| "learning_rate": 0.00017836065573770493, | |
| "loss": 0.4063, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 16.693105552165953, | |
| "grad_norm": 3.818737506866455, | |
| "learning_rate": 0.00017822950819672132, | |
| "loss": 0.3777, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 16.79072605247102, | |
| "grad_norm": 3.7047250270843506, | |
| "learning_rate": 0.0001780983606557377, | |
| "loss": 0.3939, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 16.888346552776085, | |
| "grad_norm": 4.005129337310791, | |
| "learning_rate": 0.0001779672131147541, | |
| "loss": 0.4125, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 16.985967053081147, | |
| "grad_norm": 4.845798015594482, | |
| "learning_rate": 0.00017783606557377052, | |
| "loss": 0.428, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 17.08358755338621, | |
| "grad_norm": 4.9272894859313965, | |
| "learning_rate": 0.0001777049180327869, | |
| "loss": 0.3169, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 17.181208053691275, | |
| "grad_norm": 3.334076166152954, | |
| "learning_rate": 0.00017757377049180327, | |
| "loss": 0.336, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 17.278828553996338, | |
| "grad_norm": 3.2157669067382812, | |
| "learning_rate": 0.00017744262295081969, | |
| "loss": 0.3612, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 17.376449054301403, | |
| "grad_norm": 3.595738410949707, | |
| "learning_rate": 0.00017731147540983607, | |
| "loss": 0.3627, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 17.474069554606466, | |
| "grad_norm": 4.483953475952148, | |
| "learning_rate": 0.00017718032786885246, | |
| "loss": 0.316, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 17.57169005491153, | |
| "grad_norm": 3.636593818664551, | |
| "learning_rate": 0.00017704918032786885, | |
| "loss": 0.3751, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 17.669310555216594, | |
| "grad_norm": 3.6285219192504883, | |
| "learning_rate": 0.00017691803278688527, | |
| "loss": 0.3695, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 17.76693105552166, | |
| "grad_norm": 3.8501439094543457, | |
| "learning_rate": 0.00017678688524590166, | |
| "loss": 0.3525, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 17.864551555826722, | |
| "grad_norm": 3.410341739654541, | |
| "learning_rate": 0.00017665573770491802, | |
| "loss": 0.3672, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 17.962172056131788, | |
| "grad_norm": 3.972015380859375, | |
| "learning_rate": 0.00017652459016393444, | |
| "loss": 0.3551, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 18.05979255643685, | |
| "grad_norm": 4.821686744689941, | |
| "learning_rate": 0.00017639344262295083, | |
| "loss": 0.334, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 18.157413056741916, | |
| "grad_norm": 3.1545865535736084, | |
| "learning_rate": 0.00017626229508196724, | |
| "loss": 0.307, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 18.25503355704698, | |
| "grad_norm": 3.6867642402648926, | |
| "learning_rate": 0.0001761311475409836, | |
| "loss": 0.2859, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 18.352654057352044, | |
| "grad_norm": 3.3141942024230957, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 0.341, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 18.450274557657107, | |
| "grad_norm": 3.5395824909210205, | |
| "learning_rate": 0.0001758688524590164, | |
| "loss": 0.3412, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 18.547895057962172, | |
| "grad_norm": 3.7669973373413086, | |
| "learning_rate": 0.00017573770491803277, | |
| "loss": 0.3216, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 18.645515558267235, | |
| "grad_norm": 3.6290788650512695, | |
| "learning_rate": 0.0001756065573770492, | |
| "loss": 0.3222, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 18.7431360585723, | |
| "grad_norm": 3.3772177696228027, | |
| "learning_rate": 0.00017547540983606558, | |
| "loss": 0.3301, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 18.840756558877363, | |
| "grad_norm": 4.57468843460083, | |
| "learning_rate": 0.000175344262295082, | |
| "loss": 0.3289, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 18.93837705918243, | |
| "grad_norm": 4.921297073364258, | |
| "learning_rate": 0.00017521311475409836, | |
| "loss": 0.3456, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 19.03599755948749, | |
| "grad_norm": 3.4319562911987305, | |
| "learning_rate": 0.00017508196721311475, | |
| "loss": 0.3273, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 19.133618059792557, | |
| "grad_norm": 3.177849054336548, | |
| "learning_rate": 0.00017495081967213116, | |
| "loss": 0.2475, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 19.23123856009762, | |
| "grad_norm": 3.6201579570770264, | |
| "learning_rate": 0.00017481967213114753, | |
| "loss": 0.2825, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 19.328859060402685, | |
| "grad_norm": 3.9348137378692627, | |
| "learning_rate": 0.00017468852459016394, | |
| "loss": 0.2999, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 19.426479560707747, | |
| "grad_norm": 3.8892483711242676, | |
| "learning_rate": 0.00017455737704918033, | |
| "loss": 0.2946, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 19.524100061012813, | |
| "grad_norm": 3.8046462535858154, | |
| "learning_rate": 0.00017442622950819675, | |
| "loss": 0.3273, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 19.621720561317876, | |
| "grad_norm": 3.8215136528015137, | |
| "learning_rate": 0.0001742950819672131, | |
| "loss": 0.3057, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 19.71934106162294, | |
| "grad_norm": 4.213250637054443, | |
| "learning_rate": 0.0001741639344262295, | |
| "loss": 0.3168, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 19.816961561928004, | |
| "grad_norm": 4.45851469039917, | |
| "learning_rate": 0.00017403278688524592, | |
| "loss": 0.3483, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 19.91458206223307, | |
| "grad_norm": 3.3840818405151367, | |
| "learning_rate": 0.0001739016393442623, | |
| "loss": 0.316, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 20.012202562538132, | |
| "grad_norm": 2.586810350418091, | |
| "learning_rate": 0.0001737704918032787, | |
| "loss": 0.3167, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 20.109823062843198, | |
| "grad_norm": 2.4695241451263428, | |
| "learning_rate": 0.00017363934426229509, | |
| "loss": 0.2693, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 20.20744356314826, | |
| "grad_norm": 3.0270965099334717, | |
| "learning_rate": 0.0001735081967213115, | |
| "loss": 0.2699, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 20.305064063453326, | |
| "grad_norm": 3.3233821392059326, | |
| "learning_rate": 0.00017337704918032786, | |
| "loss": 0.261, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 20.40268456375839, | |
| "grad_norm": 3.30206561088562, | |
| "learning_rate": 0.00017324590163934425, | |
| "loss": 0.2843, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 20.500305064063454, | |
| "grad_norm": 3.157876968383789, | |
| "learning_rate": 0.00017311475409836067, | |
| "loss": 0.2794, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 20.597925564368516, | |
| "grad_norm": 2.9424281120300293, | |
| "learning_rate": 0.00017298360655737706, | |
| "loss": 0.3006, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 20.695546064673582, | |
| "grad_norm": 4.470980167388916, | |
| "learning_rate": 0.00017285245901639345, | |
| "loss": 0.3001, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 20.793166564978645, | |
| "grad_norm": 5.169465065002441, | |
| "learning_rate": 0.00017272131147540984, | |
| "loss": 0.2855, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 20.89078706528371, | |
| "grad_norm": 4.266078472137451, | |
| "learning_rate": 0.00017259016393442625, | |
| "loss": 0.3039, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 20.988407565588773, | |
| "grad_norm": 3.450538158416748, | |
| "learning_rate": 0.00017245901639344264, | |
| "loss": 0.3151, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 21.08602806589384, | |
| "grad_norm": 3.0616092681884766, | |
| "learning_rate": 0.000172327868852459, | |
| "loss": 0.2414, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 21.1836485661989, | |
| "grad_norm": 2.623687267303467, | |
| "learning_rate": 0.00017219672131147542, | |
| "loss": 0.2257, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 21.281269066503967, | |
| "grad_norm": 3.4922776222229004, | |
| "learning_rate": 0.0001720655737704918, | |
| "loss": 0.2405, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 21.37888956680903, | |
| "grad_norm": 3.7959911823272705, | |
| "learning_rate": 0.0001719344262295082, | |
| "loss": 0.2732, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 21.476510067114095, | |
| "grad_norm": 3.0746493339538574, | |
| "learning_rate": 0.0001718032786885246, | |
| "loss": 0.2592, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 21.574130567419157, | |
| "grad_norm": 4.570422649383545, | |
| "learning_rate": 0.00017167213114754098, | |
| "loss": 0.2742, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 21.671751067724223, | |
| "grad_norm": 4.301755905151367, | |
| "learning_rate": 0.0001715409836065574, | |
| "loss": 0.2843, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 21.769371568029285, | |
| "grad_norm": 3.110278367996216, | |
| "learning_rate": 0.00017140983606557376, | |
| "loss": 0.2863, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 21.86699206833435, | |
| "grad_norm": 4.253392219543457, | |
| "learning_rate": 0.00017127868852459018, | |
| "loss": 0.2968, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 21.964612568639414, | |
| "grad_norm": 3.6689677238464355, | |
| "learning_rate": 0.00017114754098360656, | |
| "loss": 0.2918, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 22.06223306894448, | |
| "grad_norm": 3.50288462638855, | |
| "learning_rate": 0.00017101639344262298, | |
| "loss": 0.2567, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 22.15985356924954, | |
| "grad_norm": 3.2581892013549805, | |
| "learning_rate": 0.00017088524590163934, | |
| "loss": 0.2029, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 22.257474069554608, | |
| "grad_norm": 3.4691686630249023, | |
| "learning_rate": 0.00017075409836065573, | |
| "loss": 0.2299, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 22.35509456985967, | |
| "grad_norm": 4.3092427253723145, | |
| "learning_rate": 0.00017062295081967215, | |
| "loss": 0.2578, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 22.452715070164736, | |
| "grad_norm": 3.8213038444519043, | |
| "learning_rate": 0.0001704918032786885, | |
| "loss": 0.2766, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 22.550335570469798, | |
| "grad_norm": 3.121110200881958, | |
| "learning_rate": 0.00017036065573770493, | |
| "loss": 0.2612, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 22.647956070774864, | |
| "grad_norm": 2.7411513328552246, | |
| "learning_rate": 0.00017022950819672132, | |
| "loss": 0.2459, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 22.745576571079926, | |
| "grad_norm": 3.85884428024292, | |
| "learning_rate": 0.00017009836065573773, | |
| "loss": 0.2757, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 22.843197071384992, | |
| "grad_norm": 3.128300189971924, | |
| "learning_rate": 0.0001699672131147541, | |
| "loss": 0.2498, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 22.940817571690054, | |
| "grad_norm": 2.7686431407928467, | |
| "learning_rate": 0.00016983606557377049, | |
| "loss": 0.2608, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 23.03843807199512, | |
| "grad_norm": 3.427732229232788, | |
| "learning_rate": 0.0001697049180327869, | |
| "loss": 0.2449, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 23.136058572300183, | |
| "grad_norm": 3.1296660900115967, | |
| "learning_rate": 0.0001695737704918033, | |
| "loss": 0.2177, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 23.23367907260525, | |
| "grad_norm": 3.503549575805664, | |
| "learning_rate": 0.00016944262295081968, | |
| "loss": 0.2174, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 23.33129957291031, | |
| "grad_norm": 3.4126474857330322, | |
| "learning_rate": 0.00016931147540983607, | |
| "loss": 0.2354, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 23.428920073215377, | |
| "grad_norm": 3.743910074234009, | |
| "learning_rate": 0.00016918032786885249, | |
| "loss": 0.2471, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 23.52654057352044, | |
| "grad_norm": 3.852505683898926, | |
| "learning_rate": 0.00016904918032786885, | |
| "loss": 0.2558, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 23.624161073825505, | |
| "grad_norm": 3.120332956314087, | |
| "learning_rate": 0.00016891803278688524, | |
| "loss": 0.2273, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 23.721781574130567, | |
| "grad_norm": 3.634793758392334, | |
| "learning_rate": 0.00016878688524590165, | |
| "loss": 0.2452, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 23.819402074435633, | |
| "grad_norm": 3.3393948078155518, | |
| "learning_rate": 0.00016865573770491804, | |
| "loss": 0.2503, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 23.917022574740695, | |
| "grad_norm": 2.6347219944000244, | |
| "learning_rate": 0.00016852459016393443, | |
| "loss": 0.2315, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 24.01464307504576, | |
| "grad_norm": 2.7050418853759766, | |
| "learning_rate": 0.00016839344262295082, | |
| "loss": 0.2638, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 24.112263575350823, | |
| "grad_norm": 2.6256654262542725, | |
| "learning_rate": 0.0001682622950819672, | |
| "loss": 0.2156, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 24.20988407565589, | |
| "grad_norm": 3.150125741958618, | |
| "learning_rate": 0.00016813114754098363, | |
| "loss": 0.2109, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 24.30750457596095, | |
| "grad_norm": 2.8489086627960205, | |
| "learning_rate": 0.000168, | |
| "loss": 0.2092, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 24.405125076266017, | |
| "grad_norm": 2.6634905338287354, | |
| "learning_rate": 0.0001678688524590164, | |
| "loss": 0.2152, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 24.50274557657108, | |
| "grad_norm": 3.6262013912200928, | |
| "learning_rate": 0.0001677377049180328, | |
| "loss": 0.2023, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 24.600366076876146, | |
| "grad_norm": 3.2124197483062744, | |
| "learning_rate": 0.00016760655737704919, | |
| "loss": 0.231, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 24.697986577181208, | |
| "grad_norm": 3.997265100479126, | |
| "learning_rate": 0.00016747540983606558, | |
| "loss": 0.2249, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 24.795607077486274, | |
| "grad_norm": 3.5564634799957275, | |
| "learning_rate": 0.00016734426229508196, | |
| "loss": 0.2337, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 24.893227577791336, | |
| "grad_norm": 3.023794651031494, | |
| "learning_rate": 0.00016721311475409838, | |
| "loss": 0.2523, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 24.990848078096402, | |
| "grad_norm": 3.3614046573638916, | |
| "learning_rate": 0.00016708196721311474, | |
| "loss": 0.2524, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 25.088468578401464, | |
| "grad_norm": 2.63189435005188, | |
| "learning_rate": 0.00016695081967213116, | |
| "loss": 0.1942, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 25.18608907870653, | |
| "grad_norm": 3.4099018573760986, | |
| "learning_rate": 0.00016681967213114755, | |
| "loss": 0.1751, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 25.283709579011592, | |
| "grad_norm": 3.533768653869629, | |
| "learning_rate": 0.00016668852459016397, | |
| "loss": 0.1953, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 25.381330079316655, | |
| "grad_norm": 3.330538511276245, | |
| "learning_rate": 0.00016655737704918033, | |
| "loss": 0.2124, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 25.47895057962172, | |
| "grad_norm": 3.8602449893951416, | |
| "learning_rate": 0.00016642622950819672, | |
| "loss": 0.209, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 25.576571079926783, | |
| "grad_norm": 3.290445566177368, | |
| "learning_rate": 0.00016629508196721313, | |
| "loss": 0.2002, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 25.67419158023185, | |
| "grad_norm": 3.143129825592041, | |
| "learning_rate": 0.0001661639344262295, | |
| "loss": 0.2376, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 25.77181208053691, | |
| "grad_norm": 4.425613880157471, | |
| "learning_rate": 0.0001660327868852459, | |
| "loss": 0.2255, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 25.869432580841977, | |
| "grad_norm": 5.267921447753906, | |
| "learning_rate": 0.0001659016393442623, | |
| "loss": 0.2334, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 25.96705308114704, | |
| "grad_norm": 3.738940715789795, | |
| "learning_rate": 0.0001657704918032787, | |
| "loss": 0.2354, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 26.064673581452105, | |
| "grad_norm": 2.8138153553009033, | |
| "learning_rate": 0.00016563934426229508, | |
| "loss": 0.1706, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 26.162294081757167, | |
| "grad_norm": 3.58404278755188, | |
| "learning_rate": 0.00016550819672131147, | |
| "loss": 0.199, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 26.259914582062233, | |
| "grad_norm": 3.536153554916382, | |
| "learning_rate": 0.00016537704918032789, | |
| "loss": 0.1991, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 26.357535082367296, | |
| "grad_norm": 2.7656655311584473, | |
| "learning_rate": 0.00016524590163934428, | |
| "loss": 0.179, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 26.45515558267236, | |
| "grad_norm": 3.2351813316345215, | |
| "learning_rate": 0.00016511475409836067, | |
| "loss": 0.2142, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 26.552776082977424, | |
| "grad_norm": 4.154785633087158, | |
| "learning_rate": 0.00016498360655737705, | |
| "loss": 0.2007, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 26.65039658328249, | |
| "grad_norm": 3.327596664428711, | |
| "learning_rate": 0.00016485245901639344, | |
| "loss": 0.1934, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 26.748017083587552, | |
| "grad_norm": 3.7125210762023926, | |
| "learning_rate": 0.00016472131147540983, | |
| "loss": 0.2013, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 26.845637583892618, | |
| "grad_norm": 3.177676200866699, | |
| "learning_rate": 0.00016459016393442622, | |
| "loss": 0.2156, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 26.94325808419768, | |
| "grad_norm": 5.258444786071777, | |
| "learning_rate": 0.00016445901639344264, | |
| "loss": 0.2068, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 27.040878584502746, | |
| "grad_norm": 2.750441074371338, | |
| "learning_rate": 0.00016432786885245903, | |
| "loss": 0.2059, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 27.13849908480781, | |
| "grad_norm": 2.5060107707977295, | |
| "learning_rate": 0.00016419672131147542, | |
| "loss": 0.1696, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 27.236119585112874, | |
| "grad_norm": 3.026069164276123, | |
| "learning_rate": 0.0001640655737704918, | |
| "loss": 0.1874, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 27.333740085417936, | |
| "grad_norm": 3.3759682178497314, | |
| "learning_rate": 0.0001639344262295082, | |
| "loss": 0.1936, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 27.431360585723002, | |
| "grad_norm": 4.374515533447266, | |
| "learning_rate": 0.0001638032786885246, | |
| "loss": 0.1839, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 27.528981086028065, | |
| "grad_norm": 3.855091094970703, | |
| "learning_rate": 0.00016367213114754098, | |
| "loss": 0.1998, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 27.62660158633313, | |
| "grad_norm": 3.3890366554260254, | |
| "learning_rate": 0.0001635409836065574, | |
| "loss": 0.1956, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 27.724222086638193, | |
| "grad_norm": 3.516590118408203, | |
| "learning_rate": 0.00016340983606557378, | |
| "loss": 0.1784, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 27.82184258694326, | |
| "grad_norm": 3.740670680999756, | |
| "learning_rate": 0.00016327868852459017, | |
| "loss": 0.1957, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 27.91946308724832, | |
| "grad_norm": 5.360738754272461, | |
| "learning_rate": 0.00016314754098360656, | |
| "loss": 0.219, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 28.017083587553387, | |
| "grad_norm": 2.9260716438293457, | |
| "learning_rate": 0.00016301639344262295, | |
| "loss": 0.1956, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 28.11470408785845, | |
| "grad_norm": 2.962049722671509, | |
| "learning_rate": 0.00016288524590163937, | |
| "loss": 0.1514, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 28.212324588163515, | |
| "grad_norm": 3.02143931388855, | |
| "learning_rate": 0.00016275409836065573, | |
| "loss": 0.1586, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 28.309945088468577, | |
| "grad_norm": 3.138089895248413, | |
| "learning_rate": 0.00016262295081967214, | |
| "loss": 0.1706, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 28.407565588773643, | |
| "grad_norm": 3.2356488704681396, | |
| "learning_rate": 0.00016249180327868853, | |
| "loss": 0.1794, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 28.505186089078705, | |
| "grad_norm": 4.481720447540283, | |
| "learning_rate": 0.00016236065573770492, | |
| "loss": 0.1871, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 28.60280658938377, | |
| "grad_norm": 3.6736230850219727, | |
| "learning_rate": 0.0001622295081967213, | |
| "loss": 0.1649, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 28.700427089688834, | |
| "grad_norm": 4.480273723602295, | |
| "learning_rate": 0.0001620983606557377, | |
| "loss": 0.1837, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 28.7980475899939, | |
| "grad_norm": 3.7047078609466553, | |
| "learning_rate": 0.00016196721311475412, | |
| "loss": 0.2099, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 28.89566809029896, | |
| "grad_norm": 3.0603742599487305, | |
| "learning_rate": 0.00016183606557377048, | |
| "loss": 0.2015, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 28.993288590604028, | |
| "grad_norm": 3.1472814083099365, | |
| "learning_rate": 0.0001617049180327869, | |
| "loss": 0.2016, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 29.09090909090909, | |
| "grad_norm": 2.6735455989837646, | |
| "learning_rate": 0.0001615737704918033, | |
| "loss": 0.1586, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 29.188529591214156, | |
| "grad_norm": 3.0757932662963867, | |
| "learning_rate": 0.00016144262295081968, | |
| "loss": 0.1466, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 29.286150091519218, | |
| "grad_norm": 3.696460008621216, | |
| "learning_rate": 0.00016131147540983607, | |
| "loss": 0.1411, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 29.383770591824284, | |
| "grad_norm": 3.981374979019165, | |
| "learning_rate": 0.00016118032786885245, | |
| "loss": 0.1703, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 29.481391092129346, | |
| "grad_norm": 3.1289327144622803, | |
| "learning_rate": 0.00016104918032786887, | |
| "loss": 0.1708, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 29.579011592434412, | |
| "grad_norm": 2.825981855392456, | |
| "learning_rate": 0.00016091803278688526, | |
| "loss": 0.1731, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 29.676632092739474, | |
| "grad_norm": 3.0257232189178467, | |
| "learning_rate": 0.00016078688524590165, | |
| "loss": 0.1799, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 29.77425259304454, | |
| "grad_norm": 3.0449230670928955, | |
| "learning_rate": 0.00016065573770491804, | |
| "loss": 0.1776, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 29.871873093349603, | |
| "grad_norm": 3.108060121536255, | |
| "learning_rate": 0.00016052459016393443, | |
| "loss": 0.1839, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 29.96949359365467, | |
| "grad_norm": 3.475794792175293, | |
| "learning_rate": 0.00016039344262295082, | |
| "loss": 0.2044, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 30.06711409395973, | |
| "grad_norm": 2.7312111854553223, | |
| "learning_rate": 0.0001602622950819672, | |
| "loss": 0.1509, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 30.164734594264797, | |
| "grad_norm": 2.8779571056365967, | |
| "learning_rate": 0.00016013114754098362, | |
| "loss": 0.1387, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 30.26235509456986, | |
| "grad_norm": 4.596887111663818, | |
| "learning_rate": 0.00016, | |
| "loss": 0.1625, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 30.359975594874925, | |
| "grad_norm": 2.2877578735351562, | |
| "learning_rate": 0.0001598688524590164, | |
| "loss": 0.1611, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 30.457596095179987, | |
| "grad_norm": 2.9951751232147217, | |
| "learning_rate": 0.0001597377049180328, | |
| "loss": 0.1579, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 30.555216595485053, | |
| "grad_norm": 2.9729604721069336, | |
| "learning_rate": 0.00015960655737704918, | |
| "loss": 0.1522, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 30.652837095790115, | |
| "grad_norm": 3.4268481731414795, | |
| "learning_rate": 0.0001594754098360656, | |
| "loss": 0.1673, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 30.75045759609518, | |
| "grad_norm": 3.1322598457336426, | |
| "learning_rate": 0.00015934426229508196, | |
| "loss": 0.1769, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 30.848078096400243, | |
| "grad_norm": 3.1607165336608887, | |
| "learning_rate": 0.00015921311475409838, | |
| "loss": 0.1795, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 30.94569859670531, | |
| "grad_norm": 3.9670820236206055, | |
| "learning_rate": 0.00015908196721311477, | |
| "loss": 0.1714, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 31.04331909701037, | |
| "grad_norm": 3.0311009883880615, | |
| "learning_rate": 0.00015895081967213116, | |
| "loss": 0.1546, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 31.140939597315437, | |
| "grad_norm": 2.023759126663208, | |
| "learning_rate": 0.00015881967213114754, | |
| "loss": 0.1351, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 31.2385600976205, | |
| "grad_norm": 2.746744155883789, | |
| "learning_rate": 0.00015868852459016393, | |
| "loss": 0.1446, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 31.336180597925566, | |
| "grad_norm": 2.9396300315856934, | |
| "learning_rate": 0.00015855737704918035, | |
| "loss": 0.1505, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 31.433801098230628, | |
| "grad_norm": 4.066432476043701, | |
| "learning_rate": 0.0001584262295081967, | |
| "loss": 0.1576, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 31.531421598535694, | |
| "grad_norm": 3.7003040313720703, | |
| "learning_rate": 0.00015829508196721313, | |
| "loss": 0.1629, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 31.629042098840756, | |
| "grad_norm": 3.1914453506469727, | |
| "learning_rate": 0.00015816393442622952, | |
| "loss": 0.1497, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 31.726662599145822, | |
| "grad_norm": 3.582771062850952, | |
| "learning_rate": 0.0001580327868852459, | |
| "loss": 0.1535, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 31.824283099450884, | |
| "grad_norm": 2.601606845855713, | |
| "learning_rate": 0.0001579016393442623, | |
| "loss": 0.1692, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 31.92190359975595, | |
| "grad_norm": 3.206125020980835, | |
| "learning_rate": 0.0001577704918032787, | |
| "loss": 0.1529, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 32.01952410006101, | |
| "grad_norm": 2.8698012828826904, | |
| "learning_rate": 0.0001576393442622951, | |
| "loss": 0.1804, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 32.117144600366075, | |
| "grad_norm": 2.53930401802063, | |
| "learning_rate": 0.00015750819672131147, | |
| "loss": 0.1227, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 32.214765100671144, | |
| "grad_norm": 2.9354329109191895, | |
| "learning_rate": 0.00015737704918032788, | |
| "loss": 0.1299, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 32.31238560097621, | |
| "grad_norm": 3.0723047256469727, | |
| "learning_rate": 0.00015724590163934427, | |
| "loss": 0.1467, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 32.41000610128127, | |
| "grad_norm": 2.8421099185943604, | |
| "learning_rate": 0.00015711475409836066, | |
| "loss": 0.1468, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 32.50762660158633, | |
| "grad_norm": 3.185232400894165, | |
| "learning_rate": 0.00015698360655737705, | |
| "loss": 0.1601, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 32.6052471018914, | |
| "grad_norm": 3.1354596614837646, | |
| "learning_rate": 0.00015685245901639344, | |
| "loss": 0.1466, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 32.70286760219646, | |
| "grad_norm": 3.695125102996826, | |
| "learning_rate": 0.00015672131147540986, | |
| "loss": 0.1535, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 32.800488102501525, | |
| "grad_norm": 3.092844009399414, | |
| "learning_rate": 0.00015659016393442622, | |
| "loss": 0.1625, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 32.89810860280659, | |
| "grad_norm": 3.8023808002471924, | |
| "learning_rate": 0.00015645901639344263, | |
| "loss": 0.1623, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 32.99572910311166, | |
| "grad_norm": 2.934077739715576, | |
| "learning_rate": 0.00015632786885245902, | |
| "loss": 0.1629, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 33.09334960341672, | |
| "grad_norm": 2.633199453353882, | |
| "learning_rate": 0.0001561967213114754, | |
| "loss": 0.1267, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 33.19097010372178, | |
| "grad_norm": 2.3645660877227783, | |
| "learning_rate": 0.0001560655737704918, | |
| "loss": 0.124, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 33.288590604026844, | |
| "grad_norm": 3.1801674365997314, | |
| "learning_rate": 0.0001559344262295082, | |
| "loss": 0.1207, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 33.38621110433191, | |
| "grad_norm": 3.998974323272705, | |
| "learning_rate": 0.0001558032786885246, | |
| "loss": 0.1397, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 33.483831604636975, | |
| "grad_norm": 2.8653316497802734, | |
| "learning_rate": 0.000155672131147541, | |
| "loss": 0.1366, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 33.58145210494204, | |
| "grad_norm": 3.3324031829833984, | |
| "learning_rate": 0.0001555409836065574, | |
| "loss": 0.1418, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 33.6790726052471, | |
| "grad_norm": 2.5941219329833984, | |
| "learning_rate": 0.00015540983606557378, | |
| "loss": 0.1431, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 33.77669310555217, | |
| "grad_norm": 3.388183116912842, | |
| "learning_rate": 0.00015527868852459017, | |
| "loss": 0.15, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 33.87431360585723, | |
| "grad_norm": 2.9687812328338623, | |
| "learning_rate": 0.00015514754098360656, | |
| "loss": 0.1471, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 33.971934106162294, | |
| "grad_norm": 3.783984899520874, | |
| "learning_rate": 0.00015501639344262294, | |
| "loss": 0.1822, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 34.06955460646736, | |
| "grad_norm": 2.6898393630981445, | |
| "learning_rate": 0.00015488524590163936, | |
| "loss": 0.1217, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 34.16717510677242, | |
| "grad_norm": 2.337120532989502, | |
| "learning_rate": 0.00015475409836065575, | |
| "loss": 0.1261, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 34.26479560707749, | |
| "grad_norm": 2.8725621700286865, | |
| "learning_rate": 0.00015462295081967214, | |
| "loss": 0.1376, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 34.36241610738255, | |
| "grad_norm": 2.8058671951293945, | |
| "learning_rate": 0.00015449180327868853, | |
| "loss": 0.1307, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 34.46003660768761, | |
| "grad_norm": 2.90006685256958, | |
| "learning_rate": 0.00015436065573770492, | |
| "loss": 0.1364, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 34.557657107992675, | |
| "grad_norm": 4.001739978790283, | |
| "learning_rate": 0.00015422950819672133, | |
| "loss": 0.1416, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 34.655277608297745, | |
| "grad_norm": 3.0421645641326904, | |
| "learning_rate": 0.0001540983606557377, | |
| "loss": 0.137, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 34.75289810860281, | |
| "grad_norm": 3.429624080657959, | |
| "learning_rate": 0.00015396721311475411, | |
| "loss": 0.1524, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 34.85051860890787, | |
| "grad_norm": 2.7907767295837402, | |
| "learning_rate": 0.0001538360655737705, | |
| "loss": 0.1344, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 34.94813910921293, | |
| "grad_norm": 2.5428404808044434, | |
| "learning_rate": 0.0001537049180327869, | |
| "loss": 0.1329, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 35.045759609518, | |
| "grad_norm": 3.0310592651367188, | |
| "learning_rate": 0.00015357377049180328, | |
| "loss": 0.1289, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 35.14338010982306, | |
| "grad_norm": 2.757962226867676, | |
| "learning_rate": 0.00015344262295081967, | |
| "loss": 0.1148, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 35.241000610128125, | |
| "grad_norm": 2.47044038772583, | |
| "learning_rate": 0.0001533114754098361, | |
| "loss": 0.1234, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 35.33862111043319, | |
| "grad_norm": 2.766308546066284, | |
| "learning_rate": 0.00015318032786885245, | |
| "loss": 0.1154, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 35.43624161073826, | |
| "grad_norm": 3.6302130222320557, | |
| "learning_rate": 0.00015304918032786887, | |
| "loss": 0.1323, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 35.53386211104332, | |
| "grad_norm": 3.4185805320739746, | |
| "learning_rate": 0.00015291803278688526, | |
| "loss": 0.1409, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 35.63148261134838, | |
| "grad_norm": 2.238168954849243, | |
| "learning_rate": 0.00015278688524590165, | |
| "loss": 0.1216, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 35.729103111653444, | |
| "grad_norm": 3.636718988418579, | |
| "learning_rate": 0.00015265573770491803, | |
| "loss": 0.1431, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 35.82672361195851, | |
| "grad_norm": 3.1300854682922363, | |
| "learning_rate": 0.00015252459016393442, | |
| "loss": 0.1472, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 35.924344112263576, | |
| "grad_norm": 3.463369846343994, | |
| "learning_rate": 0.00015239344262295084, | |
| "loss": 0.1461, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 36.02196461256864, | |
| "grad_norm": 4.020668983459473, | |
| "learning_rate": 0.0001522622950819672, | |
| "loss": 0.1276, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 36.1195851128737, | |
| "grad_norm": 3.0597307682037354, | |
| "learning_rate": 0.00015213114754098362, | |
| "loss": 0.1137, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 36.21720561317877, | |
| "grad_norm": 2.3329412937164307, | |
| "learning_rate": 0.000152, | |
| "loss": 0.1271, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 36.31482611348383, | |
| "grad_norm": 3.5883936882019043, | |
| "learning_rate": 0.0001518688524590164, | |
| "loss": 0.1112, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 36.412446613788894, | |
| "grad_norm": 2.7755954265594482, | |
| "learning_rate": 0.0001517377049180328, | |
| "loss": 0.1123, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 36.51006711409396, | |
| "grad_norm": 3.594489336013794, | |
| "learning_rate": 0.00015160655737704918, | |
| "loss": 0.1338, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 36.607687614399026, | |
| "grad_norm": 2.8571972846984863, | |
| "learning_rate": 0.0001514754098360656, | |
| "loss": 0.1256, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 36.70530811470409, | |
| "grad_norm": 3.561603307723999, | |
| "learning_rate": 0.00015134426229508198, | |
| "loss": 0.1239, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 36.80292861500915, | |
| "grad_norm": 3.265953302383423, | |
| "learning_rate": 0.00015121311475409837, | |
| "loss": 0.1348, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 36.90054911531421, | |
| "grad_norm": 3.5381383895874023, | |
| "learning_rate": 0.00015108196721311476, | |
| "loss": 0.1402, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 36.99816961561928, | |
| "grad_norm": 3.6995930671691895, | |
| "learning_rate": 0.00015095081967213115, | |
| "loss": 0.1323, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 37.095790115924345, | |
| "grad_norm": 2.1127350330352783, | |
| "learning_rate": 0.00015081967213114754, | |
| "loss": 0.1045, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 37.19341061622941, | |
| "grad_norm": 2.4780166149139404, | |
| "learning_rate": 0.00015068852459016393, | |
| "loss": 0.1173, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 37.29103111653447, | |
| "grad_norm": 2.4735465049743652, | |
| "learning_rate": 0.00015055737704918035, | |
| "loss": 0.1158, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 37.38865161683954, | |
| "grad_norm": 4.070529937744141, | |
| "learning_rate": 0.00015042622950819673, | |
| "loss": 0.1222, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 37.4862721171446, | |
| "grad_norm": 2.87009596824646, | |
| "learning_rate": 0.00015029508196721312, | |
| "loss": 0.1237, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 37.58389261744966, | |
| "grad_norm": 1.9312514066696167, | |
| "learning_rate": 0.00015016393442622951, | |
| "loss": 0.1277, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 37.681513117754726, | |
| "grad_norm": 2.3069088459014893, | |
| "learning_rate": 0.0001500327868852459, | |
| "loss": 0.1204, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 37.779133618059795, | |
| "grad_norm": 3.6415116786956787, | |
| "learning_rate": 0.00014990163934426232, | |
| "loss": 0.1251, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 37.87675411836486, | |
| "grad_norm": 2.912848472595215, | |
| "learning_rate": 0.00014977049180327868, | |
| "loss": 0.1277, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 37.97437461866992, | |
| "grad_norm": 3.3791749477386475, | |
| "learning_rate": 0.0001496393442622951, | |
| "loss": 0.1215, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 38.07199511897498, | |
| "grad_norm": 3.6682169437408447, | |
| "learning_rate": 0.0001495081967213115, | |
| "loss": 0.1131, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 38.16961561928005, | |
| "grad_norm": 2.891191005706787, | |
| "learning_rate": 0.00014937704918032788, | |
| "loss": 0.1105, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 38.267236119585114, | |
| "grad_norm": 2.662431478500366, | |
| "learning_rate": 0.00014924590163934427, | |
| "loss": 0.1144, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 38.364856619890176, | |
| "grad_norm": 3.193325996398926, | |
| "learning_rate": 0.00014911475409836066, | |
| "loss": 0.1102, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 38.46247712019524, | |
| "grad_norm": 2.497185230255127, | |
| "learning_rate": 0.00014898360655737707, | |
| "loss": 0.101, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 38.56009762050031, | |
| "grad_norm": 2.6551053524017334, | |
| "learning_rate": 0.00014885245901639343, | |
| "loss": 0.1178, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 38.65771812080537, | |
| "grad_norm": 3.329808473587036, | |
| "learning_rate": 0.00014872131147540985, | |
| "loss": 0.11, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 38.75533862111043, | |
| "grad_norm": 2.581897258758545, | |
| "learning_rate": 0.00014859016393442624, | |
| "loss": 0.1228, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 38.852959121415495, | |
| "grad_norm": 2.805368423461914, | |
| "learning_rate": 0.00014845901639344263, | |
| "loss": 0.1249, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 38.950579621720564, | |
| "grad_norm": 2.7007319927215576, | |
| "learning_rate": 0.00014832786885245902, | |
| "loss": 0.1319, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 39.04820012202563, | |
| "grad_norm": 2.029555082321167, | |
| "learning_rate": 0.0001481967213114754, | |
| "loss": 0.1132, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 39.14582062233069, | |
| "grad_norm": 1.6910994052886963, | |
| "learning_rate": 0.00014806557377049182, | |
| "loss": 0.0968, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 39.24344112263575, | |
| "grad_norm": 2.706334114074707, | |
| "learning_rate": 0.0001479344262295082, | |
| "loss": 0.1014, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 39.34106162294082, | |
| "grad_norm": 2.848043441772461, | |
| "learning_rate": 0.0001478032786885246, | |
| "loss": 0.114, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 39.43868212324588, | |
| "grad_norm": 2.6262383460998535, | |
| "learning_rate": 0.000147672131147541, | |
| "loss": 0.1144, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 39.536302623550945, | |
| "grad_norm": 2.2859811782836914, | |
| "learning_rate": 0.00014754098360655738, | |
| "loss": 0.1151, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 39.63392312385601, | |
| "grad_norm": 2.8703114986419678, | |
| "learning_rate": 0.00014740983606557377, | |
| "loss": 0.1084, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 39.73154362416108, | |
| "grad_norm": 2.1996986865997314, | |
| "learning_rate": 0.00014727868852459016, | |
| "loss": 0.1268, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 39.82916412446614, | |
| "grad_norm": 2.9079976081848145, | |
| "learning_rate": 0.00014714754098360658, | |
| "loss": 0.1184, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 39.9267846247712, | |
| "grad_norm": 3.584907293319702, | |
| "learning_rate": 0.00014701639344262297, | |
| "loss": 0.131, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 40.024405125076264, | |
| "grad_norm": 2.2893424034118652, | |
| "learning_rate": 0.00014688524590163936, | |
| "loss": 0.1111, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 40.12202562538133, | |
| "grad_norm": 3.5768001079559326, | |
| "learning_rate": 0.00014675409836065575, | |
| "loss": 0.0993, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 40.219646125686396, | |
| "grad_norm": 4.004862308502197, | |
| "learning_rate": 0.00014662295081967214, | |
| "loss": 0.1048, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 40.31726662599146, | |
| "grad_norm": 2.0014472007751465, | |
| "learning_rate": 0.00014649180327868852, | |
| "loss": 0.1003, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 40.41488712629652, | |
| "grad_norm": 2.492799997329712, | |
| "learning_rate": 0.00014636065573770491, | |
| "loss": 0.102, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 40.51250762660159, | |
| "grad_norm": 2.779287099838257, | |
| "learning_rate": 0.00014622950819672133, | |
| "loss": 0.1035, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 40.61012812690665, | |
| "grad_norm": 3.0162174701690674, | |
| "learning_rate": 0.00014609836065573772, | |
| "loss": 0.1112, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 40.707748627211714, | |
| "grad_norm": 3.1944477558135986, | |
| "learning_rate": 0.0001459672131147541, | |
| "loss": 0.1167, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 40.80536912751678, | |
| "grad_norm": 2.849069833755493, | |
| "learning_rate": 0.0001458360655737705, | |
| "loss": 0.1111, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 40.902989627821846, | |
| "grad_norm": 3.3523714542388916, | |
| "learning_rate": 0.0001457049180327869, | |
| "loss": 0.1157, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 41.00061012812691, | |
| "grad_norm": 3.1365504264831543, | |
| "learning_rate": 0.0001455737704918033, | |
| "loss": 0.1228, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 41.09823062843197, | |
| "grad_norm": 2.1375391483306885, | |
| "learning_rate": 0.00014544262295081967, | |
| "loss": 0.0856, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 41.19585112873703, | |
| "grad_norm": 2.6286661624908447, | |
| "learning_rate": 0.00014531147540983608, | |
| "loss": 0.0944, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 41.2934716290421, | |
| "grad_norm": 1.7328729629516602, | |
| "learning_rate": 0.00014518032786885247, | |
| "loss": 0.0933, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 41.391092129347165, | |
| "grad_norm": 2.313880205154419, | |
| "learning_rate": 0.00014504918032786886, | |
| "loss": 0.1063, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 41.48871262965223, | |
| "grad_norm": 2.3604514598846436, | |
| "learning_rate": 0.00014491803278688525, | |
| "loss": 0.0963, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 41.58633312995729, | |
| "grad_norm": 2.1344621181488037, | |
| "learning_rate": 0.00014478688524590164, | |
| "loss": 0.1084, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 41.68395363026236, | |
| "grad_norm": 3.828547239303589, | |
| "learning_rate": 0.00014465573770491806, | |
| "loss": 0.112, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 41.78157413056742, | |
| "grad_norm": 2.653571367263794, | |
| "learning_rate": 0.00014452459016393442, | |
| "loss": 0.1241, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 41.87919463087248, | |
| "grad_norm": 2.482323408126831, | |
| "learning_rate": 0.00014439344262295084, | |
| "loss": 0.1096, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 41.976815131177545, | |
| "grad_norm": 2.309413433074951, | |
| "learning_rate": 0.00014426229508196722, | |
| "loss": 0.1191, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 42.074435631482615, | |
| "grad_norm": 2.5040862560272217, | |
| "learning_rate": 0.00014413114754098361, | |
| "loss": 0.0932, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 42.17205613178768, | |
| "grad_norm": 2.5942790508270264, | |
| "learning_rate": 0.000144, | |
| "loss": 0.0934, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 42.26967663209274, | |
| "grad_norm": 2.0345799922943115, | |
| "learning_rate": 0.0001438688524590164, | |
| "loss": 0.0913, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 42.3672971323978, | |
| "grad_norm": 2.216151714324951, | |
| "learning_rate": 0.0001437377049180328, | |
| "loss": 0.0966, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 42.464917632702864, | |
| "grad_norm": 2.8628721237182617, | |
| "learning_rate": 0.00014360655737704917, | |
| "loss": 0.102, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 42.56253813300793, | |
| "grad_norm": 3.280900239944458, | |
| "learning_rate": 0.0001434754098360656, | |
| "loss": 0.0922, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 42.660158633312996, | |
| "grad_norm": 2.90677809715271, | |
| "learning_rate": 0.00014334426229508198, | |
| "loss": 0.1027, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 42.75777913361806, | |
| "grad_norm": 3.169517755508423, | |
| "learning_rate": 0.00014321311475409837, | |
| "loss": 0.112, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 42.85539963392312, | |
| "grad_norm": 2.617112159729004, | |
| "learning_rate": 0.00014308196721311476, | |
| "loss": 0.1063, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 42.95302013422819, | |
| "grad_norm": 2.3260226249694824, | |
| "learning_rate": 0.00014295081967213115, | |
| "loss": 0.1129, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 43.05064063453325, | |
| "grad_norm": 2.255563735961914, | |
| "learning_rate": 0.00014281967213114756, | |
| "loss": 0.0943, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 43.148261134838314, | |
| "grad_norm": 2.7513527870178223, | |
| "learning_rate": 0.00014268852459016395, | |
| "loss": 0.0929, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 43.24588163514338, | |
| "grad_norm": 1.92955482006073, | |
| "learning_rate": 0.00014255737704918034, | |
| "loss": 0.0871, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 43.343502135448446, | |
| "grad_norm": 2.5168724060058594, | |
| "learning_rate": 0.00014242622950819673, | |
| "loss": 0.0915, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 43.44112263575351, | |
| "grad_norm": 2.497450351715088, | |
| "learning_rate": 0.00014229508196721312, | |
| "loss": 0.094, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 43.53874313605857, | |
| "grad_norm": 2.645716667175293, | |
| "learning_rate": 0.0001421639344262295, | |
| "loss": 0.1039, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 43.63636363636363, | |
| "grad_norm": 2.7844698429107666, | |
| "learning_rate": 0.0001420327868852459, | |
| "loss": 0.0979, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 43.7339841366687, | |
| "grad_norm": 2.3445496559143066, | |
| "learning_rate": 0.00014190163934426231, | |
| "loss": 0.0936, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 43.831604636973765, | |
| "grad_norm": 3.3879520893096924, | |
| "learning_rate": 0.0001417704918032787, | |
| "loss": 0.1147, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 43.92922513727883, | |
| "grad_norm": 2.5144670009613037, | |
| "learning_rate": 0.0001416393442622951, | |
| "loss": 0.1064, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 44.02684563758389, | |
| "grad_norm": 1.741097092628479, | |
| "learning_rate": 0.00014150819672131148, | |
| "loss": 0.0966, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 44.12446613788896, | |
| "grad_norm": 2.3388848304748535, | |
| "learning_rate": 0.00014137704918032787, | |
| "loss": 0.0877, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 44.22208663819402, | |
| "grad_norm": 1.8241106271743774, | |
| "learning_rate": 0.00014124590163934426, | |
| "loss": 0.0937, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 44.31970713849908, | |
| "grad_norm": 2.50146222114563, | |
| "learning_rate": 0.00014111475409836065, | |
| "loss": 0.0877, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 44.417327638804146, | |
| "grad_norm": 2.029155969619751, | |
| "learning_rate": 0.00014098360655737707, | |
| "loss": 0.0841, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 44.514948139109215, | |
| "grad_norm": 3.042222738265991, | |
| "learning_rate": 0.00014085245901639346, | |
| "loss": 0.0956, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 44.61256863941428, | |
| "grad_norm": 3.1099298000335693, | |
| "learning_rate": 0.00014072131147540985, | |
| "loss": 0.1026, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 44.71018913971934, | |
| "grad_norm": 2.7761776447296143, | |
| "learning_rate": 0.00014059016393442624, | |
| "loss": 0.0994, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 44.8078096400244, | |
| "grad_norm": 4.281795501708984, | |
| "learning_rate": 0.00014045901639344262, | |
| "loss": 0.0995, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 44.90543014032947, | |
| "grad_norm": 1.9752144813537598, | |
| "learning_rate": 0.00014032786885245904, | |
| "loss": 0.1037, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 45.003050640634534, | |
| "grad_norm": 2.5847785472869873, | |
| "learning_rate": 0.0001401967213114754, | |
| "loss": 0.1025, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 45.100671140939596, | |
| "grad_norm": 3.0032763481140137, | |
| "learning_rate": 0.00014006557377049182, | |
| "loss": 0.0794, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 45.19829164124466, | |
| "grad_norm": 3.35851788520813, | |
| "learning_rate": 0.0001399344262295082, | |
| "loss": 0.091, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 45.29591214154973, | |
| "grad_norm": 2.5405397415161133, | |
| "learning_rate": 0.0001398032786885246, | |
| "loss": 0.0839, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 45.39353264185479, | |
| "grad_norm": 3.7618603706359863, | |
| "learning_rate": 0.000139672131147541, | |
| "loss": 0.0909, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 45.49115314215985, | |
| "grad_norm": 2.1094727516174316, | |
| "learning_rate": 0.00013954098360655738, | |
| "loss": 0.0884, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 45.588773642464915, | |
| "grad_norm": 2.060983657836914, | |
| "learning_rate": 0.0001394098360655738, | |
| "loss": 0.0918, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 45.686394142769984, | |
| "grad_norm": 2.426604747772217, | |
| "learning_rate": 0.00013927868852459016, | |
| "loss": 0.0943, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 45.78401464307505, | |
| "grad_norm": 2.716404438018799, | |
| "learning_rate": 0.00013914754098360657, | |
| "loss": 0.0985, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 45.88163514338011, | |
| "grad_norm": 2.1184298992156982, | |
| "learning_rate": 0.00013901639344262296, | |
| "loss": 0.0941, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 45.97925564368517, | |
| "grad_norm": 2.4884073734283447, | |
| "learning_rate": 0.00013888524590163935, | |
| "loss": 0.1097, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 46.07687614399024, | |
| "grad_norm": 2.272751808166504, | |
| "learning_rate": 0.00013875409836065574, | |
| "loss": 0.088, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 46.1744966442953, | |
| "grad_norm": 2.296656847000122, | |
| "learning_rate": 0.00013862295081967213, | |
| "loss": 0.0813, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 46.272117144600365, | |
| "grad_norm": 2.6703407764434814, | |
| "learning_rate": 0.00013849180327868855, | |
| "loss": 0.0837, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 46.36973764490543, | |
| "grad_norm": 3.1443161964416504, | |
| "learning_rate": 0.0001383606557377049, | |
| "loss": 0.0876, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 46.4673581452105, | |
| "grad_norm": 2.6503586769104004, | |
| "learning_rate": 0.00013822950819672133, | |
| "loss": 0.0835, | |
| "step": 9520 | |
| }, | |
| { | |
| "epoch": 46.56497864551556, | |
| "grad_norm": 2.3241240978240967, | |
| "learning_rate": 0.00013809836065573771, | |
| "loss": 0.0996, | |
| "step": 9540 | |
| }, | |
| { | |
| "epoch": 46.66259914582062, | |
| "grad_norm": 2.8960094451904297, | |
| "learning_rate": 0.0001379672131147541, | |
| "loss": 0.0904, | |
| "step": 9560 | |
| }, | |
| { | |
| "epoch": 46.760219646125684, | |
| "grad_norm": 3.4559171199798584, | |
| "learning_rate": 0.0001378360655737705, | |
| "loss": 0.0906, | |
| "step": 9580 | |
| }, | |
| { | |
| "epoch": 46.85784014643075, | |
| "grad_norm": 2.325892925262451, | |
| "learning_rate": 0.00013770491803278688, | |
| "loss": 0.1012, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 46.955460646735816, | |
| "grad_norm": 2.833704948425293, | |
| "learning_rate": 0.0001375737704918033, | |
| "loss": 0.0947, | |
| "step": 9620 | |
| }, | |
| { | |
| "epoch": 47.05308114704088, | |
| "grad_norm": 2.060310125350952, | |
| "learning_rate": 0.0001374426229508197, | |
| "loss": 0.0861, | |
| "step": 9640 | |
| }, | |
| { | |
| "epoch": 47.15070164734594, | |
| "grad_norm": 2.1415886878967285, | |
| "learning_rate": 0.00013731147540983608, | |
| "loss": 0.0775, | |
| "step": 9660 | |
| }, | |
| { | |
| "epoch": 47.24832214765101, | |
| "grad_norm": 1.9478198289871216, | |
| "learning_rate": 0.00013718032786885247, | |
| "loss": 0.082, | |
| "step": 9680 | |
| }, | |
| { | |
| "epoch": 47.34594264795607, | |
| "grad_norm": 4.712246417999268, | |
| "learning_rate": 0.00013704918032786886, | |
| "loss": 0.0753, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 47.443563148261134, | |
| "grad_norm": 2.2505266666412354, | |
| "learning_rate": 0.00013691803278688525, | |
| "loss": 0.0809, | |
| "step": 9720 | |
| }, | |
| { | |
| "epoch": 47.5411836485662, | |
| "grad_norm": 2.4680440425872803, | |
| "learning_rate": 0.00013678688524590164, | |
| "loss": 0.0923, | |
| "step": 9740 | |
| }, | |
| { | |
| "epoch": 47.638804148871266, | |
| "grad_norm": 2.5806798934936523, | |
| "learning_rate": 0.00013665573770491805, | |
| "loss": 0.0938, | |
| "step": 9760 | |
| }, | |
| { | |
| "epoch": 47.73642464917633, | |
| "grad_norm": 2.5440475940704346, | |
| "learning_rate": 0.00013652459016393444, | |
| "loss": 0.0911, | |
| "step": 9780 | |
| }, | |
| { | |
| "epoch": 47.83404514948139, | |
| "grad_norm": 2.316375732421875, | |
| "learning_rate": 0.00013639344262295083, | |
| "loss": 0.0958, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 47.93166564978645, | |
| "grad_norm": 2.1451222896575928, | |
| "learning_rate": 0.00013626229508196722, | |
| "loss": 0.0971, | |
| "step": 9820 | |
| }, | |
| { | |
| "epoch": 48.02928615009152, | |
| "grad_norm": 2.381439685821533, | |
| "learning_rate": 0.0001361311475409836, | |
| "loss": 0.0944, | |
| "step": 9840 | |
| }, | |
| { | |
| "epoch": 48.126906650396585, | |
| "grad_norm": 2.9612882137298584, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 0.0676, | |
| "step": 9860 | |
| }, | |
| { | |
| "epoch": 48.22452715070165, | |
| "grad_norm": 1.7020941972732544, | |
| "learning_rate": 0.0001358688524590164, | |
| "loss": 0.0835, | |
| "step": 9880 | |
| }, | |
| { | |
| "epoch": 48.32214765100671, | |
| "grad_norm": 1.820823311805725, | |
| "learning_rate": 0.0001357377049180328, | |
| "loss": 0.0773, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 48.41976815131178, | |
| "grad_norm": 1.7456953525543213, | |
| "learning_rate": 0.0001356065573770492, | |
| "loss": 0.0847, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 48.51738865161684, | |
| "grad_norm": 2.277359962463379, | |
| "learning_rate": 0.00013547540983606556, | |
| "loss": 0.0879, | |
| "step": 9940 | |
| }, | |
| { | |
| "epoch": 48.6150091519219, | |
| "grad_norm": 1.9201923608779907, | |
| "learning_rate": 0.00013534426229508197, | |
| "loss": 0.0886, | |
| "step": 9960 | |
| }, | |
| { | |
| "epoch": 48.712629652226966, | |
| "grad_norm": 2.5674726963043213, | |
| "learning_rate": 0.00013521311475409836, | |
| "loss": 0.0828, | |
| "step": 9980 | |
| }, | |
| { | |
| "epoch": 48.810250152532035, | |
| "grad_norm": 2.4883131980895996, | |
| "learning_rate": 0.00013508196721311478, | |
| "loss": 0.0949, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 48.9078706528371, | |
| "grad_norm": 2.472350835800171, | |
| "learning_rate": 0.00013495081967213114, | |
| "loss": 0.0868, | |
| "step": 10020 | |
| }, | |
| { | |
| "epoch": 49.00549115314216, | |
| "grad_norm": 2.2226836681365967, | |
| "learning_rate": 0.00013481967213114756, | |
| "loss": 0.0984, | |
| "step": 10040 | |
| }, | |
| { | |
| "epoch": 49.10311165344722, | |
| "grad_norm": 1.5109550952911377, | |
| "learning_rate": 0.00013468852459016395, | |
| "loss": 0.0608, | |
| "step": 10060 | |
| }, | |
| { | |
| "epoch": 49.20073215375229, | |
| "grad_norm": 2.532590627670288, | |
| "learning_rate": 0.00013455737704918034, | |
| "loss": 0.0695, | |
| "step": 10080 | |
| }, | |
| { | |
| "epoch": 49.298352654057354, | |
| "grad_norm": 2.6609067916870117, | |
| "learning_rate": 0.00013442622950819673, | |
| "loss": 0.082, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 49.395973154362416, | |
| "grad_norm": 2.5846164226531982, | |
| "learning_rate": 0.00013429508196721311, | |
| "loss": 0.08, | |
| "step": 10120 | |
| }, | |
| { | |
| "epoch": 49.49359365466748, | |
| "grad_norm": 2.4516260623931885, | |
| "learning_rate": 0.00013416393442622953, | |
| "loss": 0.0796, | |
| "step": 10140 | |
| }, | |
| { | |
| "epoch": 49.59121415497255, | |
| "grad_norm": 2.357771158218384, | |
| "learning_rate": 0.0001340327868852459, | |
| "loss": 0.0823, | |
| "step": 10160 | |
| }, | |
| { | |
| "epoch": 49.68883465527761, | |
| "grad_norm": 2.575505018234253, | |
| "learning_rate": 0.0001339016393442623, | |
| "loss": 0.0899, | |
| "step": 10180 | |
| }, | |
| { | |
| "epoch": 49.78645515558267, | |
| "grad_norm": 2.2076785564422607, | |
| "learning_rate": 0.0001337704918032787, | |
| "loss": 0.0822, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 49.884075655887735, | |
| "grad_norm": 2.361694097518921, | |
| "learning_rate": 0.0001336393442622951, | |
| "loss": 0.0913, | |
| "step": 10220 | |
| }, | |
| { | |
| "epoch": 49.981696156192804, | |
| "grad_norm": 2.3063478469848633, | |
| "learning_rate": 0.00013350819672131148, | |
| "loss": 0.0919, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 50.079316656497866, | |
| "grad_norm": 1.9585156440734863, | |
| "learning_rate": 0.00013337704918032787, | |
| "loss": 0.0767, | |
| "step": 10260 | |
| }, | |
| { | |
| "epoch": 50.17693715680293, | |
| "grad_norm": 3.1279983520507812, | |
| "learning_rate": 0.00013324590163934428, | |
| "loss": 0.0755, | |
| "step": 10280 | |
| }, | |
| { | |
| "epoch": 50.27455765710799, | |
| "grad_norm": 2.0095512866973877, | |
| "learning_rate": 0.00013311475409836067, | |
| "loss": 0.0735, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 50.37217815741306, | |
| "grad_norm": 3.011828899383545, | |
| "learning_rate": 0.00013298360655737706, | |
| "loss": 0.0766, | |
| "step": 10320 | |
| }, | |
| { | |
| "epoch": 50.46979865771812, | |
| "grad_norm": 2.0386946201324463, | |
| "learning_rate": 0.00013285245901639345, | |
| "loss": 0.0738, | |
| "step": 10340 | |
| }, | |
| { | |
| "epoch": 50.567419158023185, | |
| "grad_norm": 2.0549607276916504, | |
| "learning_rate": 0.00013272131147540984, | |
| "loss": 0.089, | |
| "step": 10360 | |
| }, | |
| { | |
| "epoch": 50.66503965832825, | |
| "grad_norm": 2.816814422607422, | |
| "learning_rate": 0.00013259016393442623, | |
| "loss": 0.0849, | |
| "step": 10380 | |
| }, | |
| { | |
| "epoch": 50.76266015863331, | |
| "grad_norm": 2.2058887481689453, | |
| "learning_rate": 0.00013245901639344262, | |
| "loss": 0.0842, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 50.86028065893838, | |
| "grad_norm": 2.2805745601654053, | |
| "learning_rate": 0.00013232786885245904, | |
| "loss": 0.0888, | |
| "step": 10420 | |
| }, | |
| { | |
| "epoch": 50.95790115924344, | |
| "grad_norm": 2.602515697479248, | |
| "learning_rate": 0.00013219672131147543, | |
| "loss": 0.0867, | |
| "step": 10440 | |
| }, | |
| { | |
| "epoch": 51.0555216595485, | |
| "grad_norm": 1.9022042751312256, | |
| "learning_rate": 0.0001320655737704918, | |
| "loss": 0.0748, | |
| "step": 10460 | |
| }, | |
| { | |
| "epoch": 51.153142159853566, | |
| "grad_norm": 2.4302797317504883, | |
| "learning_rate": 0.0001319344262295082, | |
| "loss": 0.0693, | |
| "step": 10480 | |
| }, | |
| { | |
| "epoch": 51.250762660158635, | |
| "grad_norm": 1.5306838750839233, | |
| "learning_rate": 0.0001318032786885246, | |
| "loss": 0.0748, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 51.3483831604637, | |
| "grad_norm": 2.035921812057495, | |
| "learning_rate": 0.000131672131147541, | |
| "loss": 0.0728, | |
| "step": 10520 | |
| }, | |
| { | |
| "epoch": 51.44600366076876, | |
| "grad_norm": 1.8543293476104736, | |
| "learning_rate": 0.00013154098360655737, | |
| "loss": 0.0752, | |
| "step": 10540 | |
| }, | |
| { | |
| "epoch": 51.54362416107382, | |
| "grad_norm": 2.405411720275879, | |
| "learning_rate": 0.0001314098360655738, | |
| "loss": 0.0775, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 51.64124466137889, | |
| "grad_norm": 2.767051935195923, | |
| "learning_rate": 0.00013127868852459018, | |
| "loss": 0.075, | |
| "step": 10580 | |
| }, | |
| { | |
| "epoch": 51.738865161683954, | |
| "grad_norm": 2.200438976287842, | |
| "learning_rate": 0.00013114754098360654, | |
| "loss": 0.084, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 51.836485661989016, | |
| "grad_norm": 2.7286131381988525, | |
| "learning_rate": 0.00013101639344262296, | |
| "loss": 0.0863, | |
| "step": 10620 | |
| }, | |
| { | |
| "epoch": 51.93410616229408, | |
| "grad_norm": 1.557892084121704, | |
| "learning_rate": 0.00013088524590163935, | |
| "loss": 0.0832, | |
| "step": 10640 | |
| }, | |
| { | |
| "epoch": 52.03172666259915, | |
| "grad_norm": 1.6213957071304321, | |
| "learning_rate": 0.00013075409836065576, | |
| "loss": 0.0835, | |
| "step": 10660 | |
| }, | |
| { | |
| "epoch": 52.12934716290421, | |
| "grad_norm": 1.9538192749023438, | |
| "learning_rate": 0.00013062295081967213, | |
| "loss": 0.0632, | |
| "step": 10680 | |
| }, | |
| { | |
| "epoch": 52.22696766320927, | |
| "grad_norm": 1.5224443674087524, | |
| "learning_rate": 0.00013049180327868854, | |
| "loss": 0.0691, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 52.324588163514335, | |
| "grad_norm": 2.0320827960968018, | |
| "learning_rate": 0.00013036065573770493, | |
| "loss": 0.0777, | |
| "step": 10720 | |
| }, | |
| { | |
| "epoch": 52.422208663819404, | |
| "grad_norm": 2.9247395992279053, | |
| "learning_rate": 0.00013022950819672132, | |
| "loss": 0.0738, | |
| "step": 10740 | |
| }, | |
| { | |
| "epoch": 52.51982916412447, | |
| "grad_norm": 2.0458271503448486, | |
| "learning_rate": 0.0001300983606557377, | |
| "loss": 0.0716, | |
| "step": 10760 | |
| }, | |
| { | |
| "epoch": 52.61744966442953, | |
| "grad_norm": 2.1657023429870605, | |
| "learning_rate": 0.0001299672131147541, | |
| "loss": 0.0798, | |
| "step": 10780 | |
| }, | |
| { | |
| "epoch": 52.71507016473459, | |
| "grad_norm": 2.819549798965454, | |
| "learning_rate": 0.00012983606557377052, | |
| "loss": 0.0869, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 52.81269066503966, | |
| "grad_norm": 2.455775737762451, | |
| "learning_rate": 0.00012970491803278688, | |
| "loss": 0.0784, | |
| "step": 10820 | |
| }, | |
| { | |
| "epoch": 52.91031116534472, | |
| "grad_norm": 2.803501844406128, | |
| "learning_rate": 0.0001295737704918033, | |
| "loss": 0.0802, | |
| "step": 10840 | |
| }, | |
| { | |
| "epoch": 53.007931665649785, | |
| "grad_norm": 2.2946598529815674, | |
| "learning_rate": 0.00012944262295081968, | |
| "loss": 0.0767, | |
| "step": 10860 | |
| }, | |
| { | |
| "epoch": 53.10555216595485, | |
| "grad_norm": 1.9233254194259644, | |
| "learning_rate": 0.00012931147540983607, | |
| "loss": 0.0652, | |
| "step": 10880 | |
| }, | |
| { | |
| "epoch": 53.20317266625992, | |
| "grad_norm": 1.8651479482650757, | |
| "learning_rate": 0.00012918032786885246, | |
| "loss": 0.0675, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 53.30079316656498, | |
| "grad_norm": 2.216233253479004, | |
| "learning_rate": 0.00012904918032786885, | |
| "loss": 0.0716, | |
| "step": 10920 | |
| }, | |
| { | |
| "epoch": 53.39841366687004, | |
| "grad_norm": 2.591519355773926, | |
| "learning_rate": 0.00012891803278688527, | |
| "loss": 0.073, | |
| "step": 10940 | |
| }, | |
| { | |
| "epoch": 53.496034167175104, | |
| "grad_norm": 2.490187406539917, | |
| "learning_rate": 0.00012878688524590166, | |
| "loss": 0.0715, | |
| "step": 10960 | |
| }, | |
| { | |
| "epoch": 53.59365466748017, | |
| "grad_norm": 2.2473397254943848, | |
| "learning_rate": 0.00012865573770491802, | |
| "loss": 0.0715, | |
| "step": 10980 | |
| }, | |
| { | |
| "epoch": 53.691275167785236, | |
| "grad_norm": 1.9002201557159424, | |
| "learning_rate": 0.00012852459016393444, | |
| "loss": 0.0762, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 53.7888956680903, | |
| "grad_norm": 2.4991517066955566, | |
| "learning_rate": 0.00012839344262295083, | |
| "loss": 0.0869, | |
| "step": 11020 | |
| }, | |
| { | |
| "epoch": 53.88651616839536, | |
| "grad_norm": 2.5355913639068604, | |
| "learning_rate": 0.00012826229508196722, | |
| "loss": 0.0801, | |
| "step": 11040 | |
| }, | |
| { | |
| "epoch": 53.98413666870043, | |
| "grad_norm": 1.895386815071106, | |
| "learning_rate": 0.0001281311475409836, | |
| "loss": 0.0747, | |
| "step": 11060 | |
| }, | |
| { | |
| "epoch": 54.08175716900549, | |
| "grad_norm": 2.150261163711548, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 0.0646, | |
| "step": 11080 | |
| }, | |
| { | |
| "epoch": 54.179377669310554, | |
| "grad_norm": 4.12515926361084, | |
| "learning_rate": 0.0001278688524590164, | |
| "loss": 0.0693, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 54.27699816961562, | |
| "grad_norm": 1.3180917501449585, | |
| "learning_rate": 0.00012773770491803277, | |
| "loss": 0.069, | |
| "step": 11120 | |
| }, | |
| { | |
| "epoch": 54.374618669920686, | |
| "grad_norm": 1.714634895324707, | |
| "learning_rate": 0.0001276065573770492, | |
| "loss": 0.0701, | |
| "step": 11140 | |
| }, | |
| { | |
| "epoch": 54.47223917022575, | |
| "grad_norm": 1.93068265914917, | |
| "learning_rate": 0.00012747540983606558, | |
| "loss": 0.0682, | |
| "step": 11160 | |
| }, | |
| { | |
| "epoch": 54.56985967053081, | |
| "grad_norm": 1.9187488555908203, | |
| "learning_rate": 0.000127344262295082, | |
| "loss": 0.0778, | |
| "step": 11180 | |
| }, | |
| { | |
| "epoch": 54.66748017083587, | |
| "grad_norm": 2.3723368644714355, | |
| "learning_rate": 0.00012721311475409836, | |
| "loss": 0.0769, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 54.76510067114094, | |
| "grad_norm": 1.7806050777435303, | |
| "learning_rate": 0.00012708196721311477, | |
| "loss": 0.0729, | |
| "step": 11220 | |
| }, | |
| { | |
| "epoch": 54.862721171446005, | |
| "grad_norm": 1.839896321296692, | |
| "learning_rate": 0.00012695081967213116, | |
| "loss": 0.0711, | |
| "step": 11240 | |
| }, | |
| { | |
| "epoch": 54.96034167175107, | |
| "grad_norm": 3.055255889892578, | |
| "learning_rate": 0.00012681967213114753, | |
| "loss": 0.077, | |
| "step": 11260 | |
| }, | |
| { | |
| "epoch": 55.05796217205613, | |
| "grad_norm": 1.504418134689331, | |
| "learning_rate": 0.00012668852459016394, | |
| "loss": 0.0711, | |
| "step": 11280 | |
| }, | |
| { | |
| "epoch": 55.1555826723612, | |
| "grad_norm": 3.407508373260498, | |
| "learning_rate": 0.00012655737704918033, | |
| "loss": 0.0762, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 55.25320317266626, | |
| "grad_norm": 1.7444705963134766, | |
| "learning_rate": 0.00012642622950819675, | |
| "loss": 0.0676, | |
| "step": 11320 | |
| }, | |
| { | |
| "epoch": 55.35082367297132, | |
| "grad_norm": 1.7877843379974365, | |
| "learning_rate": 0.0001262950819672131, | |
| "loss": 0.0611, | |
| "step": 11340 | |
| }, | |
| { | |
| "epoch": 55.448444173276386, | |
| "grad_norm": 2.1501216888427734, | |
| "learning_rate": 0.0001261639344262295, | |
| "loss": 0.0702, | |
| "step": 11360 | |
| }, | |
| { | |
| "epoch": 55.546064673581455, | |
| "grad_norm": 1.4569560289382935, | |
| "learning_rate": 0.00012603278688524592, | |
| "loss": 0.0651, | |
| "step": 11380 | |
| }, | |
| { | |
| "epoch": 55.64368517388652, | |
| "grad_norm": 1.5365070104599, | |
| "learning_rate": 0.0001259016393442623, | |
| "loss": 0.0708, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 55.74130567419158, | |
| "grad_norm": 2.2428839206695557, | |
| "learning_rate": 0.0001257704918032787, | |
| "loss": 0.0694, | |
| "step": 11420 | |
| }, | |
| { | |
| "epoch": 55.83892617449664, | |
| "grad_norm": 2.054964542388916, | |
| "learning_rate": 0.00012563934426229508, | |
| "loss": 0.0754, | |
| "step": 11440 | |
| }, | |
| { | |
| "epoch": 55.93654667480171, | |
| "grad_norm": 3.42903733253479, | |
| "learning_rate": 0.0001255081967213115, | |
| "loss": 0.0752, | |
| "step": 11460 | |
| }, | |
| { | |
| "epoch": 56.034167175106774, | |
| "grad_norm": 1.6482346057891846, | |
| "learning_rate": 0.00012537704918032786, | |
| "loss": 0.068, | |
| "step": 11480 | |
| }, | |
| { | |
| "epoch": 56.131787675411836, | |
| "grad_norm": 2.0840466022491455, | |
| "learning_rate": 0.00012524590163934425, | |
| "loss": 0.0624, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 56.2294081757169, | |
| "grad_norm": 2.7554879188537598, | |
| "learning_rate": 0.00012511475409836067, | |
| "loss": 0.0592, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 56.32702867602197, | |
| "grad_norm": 2.8247487545013428, | |
| "learning_rate": 0.00012498360655737706, | |
| "loss": 0.0661, | |
| "step": 11540 | |
| }, | |
| { | |
| "epoch": 56.42464917632703, | |
| "grad_norm": 2.2485480308532715, | |
| "learning_rate": 0.00012485245901639345, | |
| "loss": 0.0712, | |
| "step": 11560 | |
| }, | |
| { | |
| "epoch": 56.52226967663209, | |
| "grad_norm": 2.1642680168151855, | |
| "learning_rate": 0.00012472131147540984, | |
| "loss": 0.0752, | |
| "step": 11580 | |
| }, | |
| { | |
| "epoch": 56.619890176937155, | |
| "grad_norm": 1.827771544456482, | |
| "learning_rate": 0.00012459016393442625, | |
| "loss": 0.0707, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 56.717510677242224, | |
| "grad_norm": 1.9134409427642822, | |
| "learning_rate": 0.00012445901639344262, | |
| "loss": 0.0696, | |
| "step": 11620 | |
| }, | |
| { | |
| "epoch": 56.815131177547286, | |
| "grad_norm": 2.3444812297821045, | |
| "learning_rate": 0.000124327868852459, | |
| "loss": 0.065, | |
| "step": 11640 | |
| }, | |
| { | |
| "epoch": 56.91275167785235, | |
| "grad_norm": 2.0354135036468506, | |
| "learning_rate": 0.00012419672131147542, | |
| "loss": 0.0746, | |
| "step": 11660 | |
| }, | |
| { | |
| "epoch": 57.01037217815741, | |
| "grad_norm": 1.5275322198867798, | |
| "learning_rate": 0.0001240655737704918, | |
| "loss": 0.0756, | |
| "step": 11680 | |
| }, | |
| { | |
| "epoch": 57.10799267846248, | |
| "grad_norm": 2.310797929763794, | |
| "learning_rate": 0.0001239344262295082, | |
| "loss": 0.057, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 57.20561317876754, | |
| "grad_norm": 1.903204083442688, | |
| "learning_rate": 0.0001238032786885246, | |
| "loss": 0.0657, | |
| "step": 11720 | |
| }, | |
| { | |
| "epoch": 57.303233679072605, | |
| "grad_norm": 2.115591049194336, | |
| "learning_rate": 0.000123672131147541, | |
| "loss": 0.0624, | |
| "step": 11740 | |
| }, | |
| { | |
| "epoch": 57.40085417937767, | |
| "grad_norm": 2.1732656955718994, | |
| "learning_rate": 0.0001235409836065574, | |
| "loss": 0.0636, | |
| "step": 11760 | |
| }, | |
| { | |
| "epoch": 57.49847467968274, | |
| "grad_norm": 2.2554845809936523, | |
| "learning_rate": 0.00012340983606557376, | |
| "loss": 0.0691, | |
| "step": 11780 | |
| }, | |
| { | |
| "epoch": 57.5960951799878, | |
| "grad_norm": 2.994563341140747, | |
| "learning_rate": 0.00012327868852459017, | |
| "loss": 0.0724, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 57.69371568029286, | |
| "grad_norm": 2.824934482574463, | |
| "learning_rate": 0.00012314754098360656, | |
| "loss": 0.07, | |
| "step": 11820 | |
| }, | |
| { | |
| "epoch": 57.79133618059792, | |
| "grad_norm": 2.435310125350952, | |
| "learning_rate": 0.00012301639344262295, | |
| "loss": 0.0753, | |
| "step": 11840 | |
| }, | |
| { | |
| "epoch": 57.88895668090299, | |
| "grad_norm": 2.140200138092041, | |
| "learning_rate": 0.00012288524590163934, | |
| "loss": 0.0762, | |
| "step": 11860 | |
| }, | |
| { | |
| "epoch": 57.986577181208055, | |
| "grad_norm": 2.013983726501465, | |
| "learning_rate": 0.00012275409836065573, | |
| "loss": 0.067, | |
| "step": 11880 | |
| }, | |
| { | |
| "epoch": 58.08419768151312, | |
| "grad_norm": 1.8855304718017578, | |
| "learning_rate": 0.00012262295081967215, | |
| "loss": 0.0606, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 58.18181818181818, | |
| "grad_norm": 2.3649916648864746, | |
| "learning_rate": 0.0001224918032786885, | |
| "loss": 0.0626, | |
| "step": 11920 | |
| }, | |
| { | |
| "epoch": 58.27943868212325, | |
| "grad_norm": 2.566420078277588, | |
| "learning_rate": 0.00012236065573770493, | |
| "loss": 0.059, | |
| "step": 11940 | |
| }, | |
| { | |
| "epoch": 58.37705918242831, | |
| "grad_norm": 2.3648924827575684, | |
| "learning_rate": 0.00012222950819672132, | |
| "loss": 0.0644, | |
| "step": 11960 | |
| }, | |
| { | |
| "epoch": 58.474679682733374, | |
| "grad_norm": 1.7778581380844116, | |
| "learning_rate": 0.00012209836065573773, | |
| "loss": 0.0691, | |
| "step": 11980 | |
| }, | |
| { | |
| "epoch": 58.572300183038436, | |
| "grad_norm": 1.8090124130249023, | |
| "learning_rate": 0.0001219672131147541, | |
| "loss": 0.0685, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 58.669920683343506, | |
| "grad_norm": 2.2762486934661865, | |
| "learning_rate": 0.0001218360655737705, | |
| "loss": 0.0634, | |
| "step": 12020 | |
| }, | |
| { | |
| "epoch": 58.76754118364857, | |
| "grad_norm": 2.279273509979248, | |
| "learning_rate": 0.0001217049180327869, | |
| "loss": 0.0681, | |
| "step": 12040 | |
| }, | |
| { | |
| "epoch": 58.86516168395363, | |
| "grad_norm": 1.5724430084228516, | |
| "learning_rate": 0.00012157377049180328, | |
| "loss": 0.0712, | |
| "step": 12060 | |
| }, | |
| { | |
| "epoch": 58.96278218425869, | |
| "grad_norm": 1.7655867338180542, | |
| "learning_rate": 0.00012144262295081968, | |
| "loss": 0.0711, | |
| "step": 12080 | |
| }, | |
| { | |
| "epoch": 59.060402684563755, | |
| "grad_norm": 1.0987945795059204, | |
| "learning_rate": 0.00012131147540983607, | |
| "loss": 0.0614, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 59.158023184868824, | |
| "grad_norm": 1.433910608291626, | |
| "learning_rate": 0.00012118032786885247, | |
| "loss": 0.0522, | |
| "step": 12120 | |
| }, | |
| { | |
| "epoch": 59.25564368517389, | |
| "grad_norm": 1.4189090728759766, | |
| "learning_rate": 0.00012104918032786885, | |
| "loss": 0.0603, | |
| "step": 12140 | |
| }, | |
| { | |
| "epoch": 59.35326418547895, | |
| "grad_norm": 1.915209174156189, | |
| "learning_rate": 0.00012091803278688525, | |
| "loss": 0.0631, | |
| "step": 12160 | |
| }, | |
| { | |
| "epoch": 59.45088468578401, | |
| "grad_norm": 2.0415852069854736, | |
| "learning_rate": 0.00012078688524590165, | |
| "loss": 0.065, | |
| "step": 12180 | |
| }, | |
| { | |
| "epoch": 59.54850518608908, | |
| "grad_norm": 1.8674949407577515, | |
| "learning_rate": 0.00012065573770491804, | |
| "loss": 0.0636, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 59.64612568639414, | |
| "grad_norm": 3.768343687057495, | |
| "learning_rate": 0.00012052459016393443, | |
| "loss": 0.0666, | |
| "step": 12220 | |
| }, | |
| { | |
| "epoch": 59.743746186699205, | |
| "grad_norm": 1.8944804668426514, | |
| "learning_rate": 0.00012039344262295082, | |
| "loss": 0.0698, | |
| "step": 12240 | |
| }, | |
| { | |
| "epoch": 59.84136668700427, | |
| "grad_norm": 2.296515464782715, | |
| "learning_rate": 0.00012026229508196722, | |
| "loss": 0.0715, | |
| "step": 12260 | |
| }, | |
| { | |
| "epoch": 59.93898718730934, | |
| "grad_norm": 1.957593560218811, | |
| "learning_rate": 0.0001201311475409836, | |
| "loss": 0.0681, | |
| "step": 12280 | |
| }, | |
| { | |
| "epoch": 60.0366076876144, | |
| "grad_norm": 1.8776521682739258, | |
| "learning_rate": 0.00012, | |
| "loss": 0.0676, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 60.13422818791946, | |
| "grad_norm": 1.3199673891067505, | |
| "learning_rate": 0.0001198688524590164, | |
| "loss": 0.0515, | |
| "step": 12320 | |
| }, | |
| { | |
| "epoch": 60.231848688224524, | |
| "grad_norm": 1.8994742631912231, | |
| "learning_rate": 0.0001197377049180328, | |
| "loss": 0.0566, | |
| "step": 12340 | |
| }, | |
| { | |
| "epoch": 60.32946918852959, | |
| "grad_norm": 1.8091648817062378, | |
| "learning_rate": 0.00011960655737704917, | |
| "loss": 0.0628, | |
| "step": 12360 | |
| }, | |
| { | |
| "epoch": 60.427089688834656, | |
| "grad_norm": 1.7513519525527954, | |
| "learning_rate": 0.00011947540983606557, | |
| "loss": 0.0652, | |
| "step": 12380 | |
| }, | |
| { | |
| "epoch": 60.52471018913972, | |
| "grad_norm": 1.8451545238494873, | |
| "learning_rate": 0.00011934426229508198, | |
| "loss": 0.063, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 60.62233068944478, | |
| "grad_norm": 1.6873937845230103, | |
| "learning_rate": 0.00011921311475409838, | |
| "loss": 0.0666, | |
| "step": 12420 | |
| }, | |
| { | |
| "epoch": 60.71995118974985, | |
| "grad_norm": 2.2382214069366455, | |
| "learning_rate": 0.00011908196721311476, | |
| "loss": 0.0593, | |
| "step": 12440 | |
| }, | |
| { | |
| "epoch": 60.81757169005491, | |
| "grad_norm": 1.4879833459854126, | |
| "learning_rate": 0.00011895081967213116, | |
| "loss": 0.0718, | |
| "step": 12460 | |
| }, | |
| { | |
| "epoch": 60.915192190359974, | |
| "grad_norm": 1.8051949739456177, | |
| "learning_rate": 0.00011881967213114755, | |
| "loss": 0.0659, | |
| "step": 12480 | |
| }, | |
| { | |
| "epoch": 61.01281269066504, | |
| "grad_norm": 2.070537567138672, | |
| "learning_rate": 0.00011868852459016392, | |
| "loss": 0.0647, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 61.110433190970106, | |
| "grad_norm": 2.195227861404419, | |
| "learning_rate": 0.00011855737704918033, | |
| "loss": 0.0516, | |
| "step": 12520 | |
| }, | |
| { | |
| "epoch": 61.20805369127517, | |
| "grad_norm": 2.3762006759643555, | |
| "learning_rate": 0.00011842622950819673, | |
| "loss": 0.056, | |
| "step": 12540 | |
| }, | |
| { | |
| "epoch": 61.30567419158023, | |
| "grad_norm": 1.836585521697998, | |
| "learning_rate": 0.00011829508196721313, | |
| "loss": 0.0624, | |
| "step": 12560 | |
| }, | |
| { | |
| "epoch": 61.40329469188529, | |
| "grad_norm": 1.6185029745101929, | |
| "learning_rate": 0.00011816393442622951, | |
| "loss": 0.0591, | |
| "step": 12580 | |
| }, | |
| { | |
| "epoch": 61.50091519219036, | |
| "grad_norm": 2.116314649581909, | |
| "learning_rate": 0.00011803278688524591, | |
| "loss": 0.0615, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 61.598535692495425, | |
| "grad_norm": 2.873701333999634, | |
| "learning_rate": 0.0001179016393442623, | |
| "loss": 0.0624, | |
| "step": 12620 | |
| }, | |
| { | |
| "epoch": 61.69615619280049, | |
| "grad_norm": 1.8966227769851685, | |
| "learning_rate": 0.0001177704918032787, | |
| "loss": 0.0643, | |
| "step": 12640 | |
| }, | |
| { | |
| "epoch": 61.79377669310555, | |
| "grad_norm": 2.343330144882202, | |
| "learning_rate": 0.00011763934426229508, | |
| "loss": 0.0637, | |
| "step": 12660 | |
| }, | |
| { | |
| "epoch": 61.89139719341062, | |
| "grad_norm": 2.882538318634033, | |
| "learning_rate": 0.00011750819672131148, | |
| "loss": 0.0678, | |
| "step": 12680 | |
| }, | |
| { | |
| "epoch": 61.98901769371568, | |
| "grad_norm": 3.441783905029297, | |
| "learning_rate": 0.00011737704918032789, | |
| "loss": 0.0667, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 62.08663819402074, | |
| "grad_norm": 1.9193288087844849, | |
| "learning_rate": 0.00011724590163934426, | |
| "loss": 0.0512, | |
| "step": 12720 | |
| }, | |
| { | |
| "epoch": 62.184258694325806, | |
| "grad_norm": 1.326977252960205, | |
| "learning_rate": 0.00011711475409836066, | |
| "loss": 0.0602, | |
| "step": 12740 | |
| }, | |
| { | |
| "epoch": 62.281879194630875, | |
| "grad_norm": 1.6590323448181152, | |
| "learning_rate": 0.00011698360655737705, | |
| "loss": 0.0564, | |
| "step": 12760 | |
| }, | |
| { | |
| "epoch": 62.37949969493594, | |
| "grad_norm": 1.5968765020370483, | |
| "learning_rate": 0.00011685245901639346, | |
| "loss": 0.0591, | |
| "step": 12780 | |
| }, | |
| { | |
| "epoch": 62.477120195241, | |
| "grad_norm": 1.674349069595337, | |
| "learning_rate": 0.00011672131147540983, | |
| "loss": 0.0603, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 62.57474069554606, | |
| "grad_norm": 2.2285475730895996, | |
| "learning_rate": 0.00011659016393442623, | |
| "loss": 0.0578, | |
| "step": 12820 | |
| }, | |
| { | |
| "epoch": 62.67236119585113, | |
| "grad_norm": 1.4100092649459839, | |
| "learning_rate": 0.00011645901639344264, | |
| "loss": 0.0652, | |
| "step": 12840 | |
| }, | |
| { | |
| "epoch": 62.769981696156194, | |
| "grad_norm": 2.5096209049224854, | |
| "learning_rate": 0.00011632786885245903, | |
| "loss": 0.0551, | |
| "step": 12860 | |
| }, | |
| { | |
| "epoch": 62.867602196461256, | |
| "grad_norm": 3.1331403255462646, | |
| "learning_rate": 0.0001161967213114754, | |
| "loss": 0.0662, | |
| "step": 12880 | |
| }, | |
| { | |
| "epoch": 62.96522269676632, | |
| "grad_norm": 1.8345986604690552, | |
| "learning_rate": 0.0001160655737704918, | |
| "loss": 0.0671, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 63.06284319707139, | |
| "grad_norm": 1.7236154079437256, | |
| "learning_rate": 0.00011593442622950821, | |
| "loss": 0.0553, | |
| "step": 12920 | |
| }, | |
| { | |
| "epoch": 63.16046369737645, | |
| "grad_norm": 1.664116621017456, | |
| "learning_rate": 0.00011580327868852458, | |
| "loss": 0.0554, | |
| "step": 12940 | |
| }, | |
| { | |
| "epoch": 63.25808419768151, | |
| "grad_norm": 1.4789968729019165, | |
| "learning_rate": 0.00011567213114754099, | |
| "loss": 0.0562, | |
| "step": 12960 | |
| }, | |
| { | |
| "epoch": 63.355704697986575, | |
| "grad_norm": 2.4718666076660156, | |
| "learning_rate": 0.00011554098360655739, | |
| "loss": 0.0625, | |
| "step": 12980 | |
| }, | |
| { | |
| "epoch": 63.453325198291644, | |
| "grad_norm": 2.611966133117676, | |
| "learning_rate": 0.00011540983606557378, | |
| "loss": 0.0564, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 63.550945698596706, | |
| "grad_norm": 2.4028069972991943, | |
| "learning_rate": 0.00011527868852459016, | |
| "loss": 0.0629, | |
| "step": 13020 | |
| }, | |
| { | |
| "epoch": 63.64856619890177, | |
| "grad_norm": 1.251297950744629, | |
| "learning_rate": 0.00011514754098360656, | |
| "loss": 0.0608, | |
| "step": 13040 | |
| }, | |
| { | |
| "epoch": 63.74618669920683, | |
| "grad_norm": 2.8238635063171387, | |
| "learning_rate": 0.00011501639344262296, | |
| "loss": 0.0619, | |
| "step": 13060 | |
| }, | |
| { | |
| "epoch": 63.8438071995119, | |
| "grad_norm": 2.705125570297241, | |
| "learning_rate": 0.00011488524590163936, | |
| "loss": 0.0631, | |
| "step": 13080 | |
| }, | |
| { | |
| "epoch": 63.94142769981696, | |
| "grad_norm": 1.7994593381881714, | |
| "learning_rate": 0.00011475409836065574, | |
| "loss": 0.0664, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 64.03904820012202, | |
| "grad_norm": 3.055631399154663, | |
| "learning_rate": 0.00011462295081967214, | |
| "loss": 0.0561, | |
| "step": 13120 | |
| }, | |
| { | |
| "epoch": 64.1366687004271, | |
| "grad_norm": 1.7646046876907349, | |
| "learning_rate": 0.00011449180327868853, | |
| "loss": 0.0542, | |
| "step": 13140 | |
| }, | |
| { | |
| "epoch": 64.23428920073215, | |
| "grad_norm": 1.33475923538208, | |
| "learning_rate": 0.00011436065573770491, | |
| "loss": 0.0535, | |
| "step": 13160 | |
| }, | |
| { | |
| "epoch": 64.33190970103722, | |
| "grad_norm": 1.4737622737884521, | |
| "learning_rate": 0.00011422950819672131, | |
| "loss": 0.0527, | |
| "step": 13180 | |
| }, | |
| { | |
| "epoch": 64.42953020134229, | |
| "grad_norm": 2.2712624073028564, | |
| "learning_rate": 0.00011409836065573771, | |
| "loss": 0.0563, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 64.52715070164734, | |
| "grad_norm": 1.3113006353378296, | |
| "learning_rate": 0.00011396721311475412, | |
| "loss": 0.0529, | |
| "step": 13220 | |
| }, | |
| { | |
| "epoch": 64.62477120195241, | |
| "grad_norm": 1.557981014251709, | |
| "learning_rate": 0.00011383606557377049, | |
| "loss": 0.0625, | |
| "step": 13240 | |
| }, | |
| { | |
| "epoch": 64.72239170225747, | |
| "grad_norm": 1.8564097881317139, | |
| "learning_rate": 0.00011370491803278688, | |
| "loss": 0.0638, | |
| "step": 13260 | |
| }, | |
| { | |
| "epoch": 64.82001220256254, | |
| "grad_norm": 1.741255521774292, | |
| "learning_rate": 0.00011357377049180329, | |
| "loss": 0.0591, | |
| "step": 13280 | |
| }, | |
| { | |
| "epoch": 64.9176327028676, | |
| "grad_norm": 1.85543692111969, | |
| "learning_rate": 0.00011344262295081969, | |
| "loss": 0.0628, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 65.01525320317266, | |
| "grad_norm": 1.9299827814102173, | |
| "learning_rate": 0.00011331147540983606, | |
| "loss": 0.0553, | |
| "step": 13320 | |
| }, | |
| { | |
| "epoch": 65.11287370347773, | |
| "grad_norm": 1.5189353227615356, | |
| "learning_rate": 0.00011318032786885247, | |
| "loss": 0.0502, | |
| "step": 13340 | |
| }, | |
| { | |
| "epoch": 65.2104942037828, | |
| "grad_norm": 1.8111690282821655, | |
| "learning_rate": 0.00011304918032786887, | |
| "loss": 0.0514, | |
| "step": 13360 | |
| }, | |
| { | |
| "epoch": 65.30811470408786, | |
| "grad_norm": 2.5628137588500977, | |
| "learning_rate": 0.00011291803278688525, | |
| "loss": 0.0479, | |
| "step": 13380 | |
| }, | |
| { | |
| "epoch": 65.40573520439293, | |
| "grad_norm": 1.4838755130767822, | |
| "learning_rate": 0.00011278688524590164, | |
| "loss": 0.0591, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 65.50335570469798, | |
| "grad_norm": 1.512497067451477, | |
| "learning_rate": 0.00011265573770491804, | |
| "loss": 0.0554, | |
| "step": 13420 | |
| }, | |
| { | |
| "epoch": 65.60097620500305, | |
| "grad_norm": 2.546452283859253, | |
| "learning_rate": 0.00011252459016393444, | |
| "loss": 0.0567, | |
| "step": 13440 | |
| }, | |
| { | |
| "epoch": 65.69859670530812, | |
| "grad_norm": 3.046600580215454, | |
| "learning_rate": 0.00011239344262295082, | |
| "loss": 0.0614, | |
| "step": 13460 | |
| }, | |
| { | |
| "epoch": 65.79621720561317, | |
| "grad_norm": 2.1981520652770996, | |
| "learning_rate": 0.00011226229508196722, | |
| "loss": 0.0631, | |
| "step": 13480 | |
| }, | |
| { | |
| "epoch": 65.89383770591824, | |
| "grad_norm": 2.0264618396759033, | |
| "learning_rate": 0.00011213114754098362, | |
| "loss": 0.0613, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 65.99145820622331, | |
| "grad_norm": 5.12288236618042, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 0.06, | |
| "step": 13520 | |
| }, | |
| { | |
| "epoch": 66.08907870652837, | |
| "grad_norm": 1.6284067630767822, | |
| "learning_rate": 0.00011186885245901639, | |
| "loss": 0.0492, | |
| "step": 13540 | |
| }, | |
| { | |
| "epoch": 66.18669920683344, | |
| "grad_norm": 1.5839916467666626, | |
| "learning_rate": 0.00011173770491803279, | |
| "loss": 0.0522, | |
| "step": 13560 | |
| }, | |
| { | |
| "epoch": 66.2843197071385, | |
| "grad_norm": 2.381410598754883, | |
| "learning_rate": 0.0001116065573770492, | |
| "loss": 0.054, | |
| "step": 13580 | |
| }, | |
| { | |
| "epoch": 66.38194020744356, | |
| "grad_norm": 2.1883935928344727, | |
| "learning_rate": 0.00011147540983606557, | |
| "loss": 0.0539, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 66.47956070774863, | |
| "grad_norm": 1.3095104694366455, | |
| "learning_rate": 0.00011134426229508197, | |
| "loss": 0.0548, | |
| "step": 13620 | |
| }, | |
| { | |
| "epoch": 66.57718120805369, | |
| "grad_norm": 3.0383543968200684, | |
| "learning_rate": 0.00011121311475409838, | |
| "loss": 0.0559, | |
| "step": 13640 | |
| }, | |
| { | |
| "epoch": 66.67480170835876, | |
| "grad_norm": 1.848026990890503, | |
| "learning_rate": 0.00011108196721311476, | |
| "loss": 0.0582, | |
| "step": 13660 | |
| }, | |
| { | |
| "epoch": 66.77242220866383, | |
| "grad_norm": 3.3856444358825684, | |
| "learning_rate": 0.00011095081967213114, | |
| "loss": 0.0583, | |
| "step": 13680 | |
| }, | |
| { | |
| "epoch": 66.87004270896888, | |
| "grad_norm": 2.1376798152923584, | |
| "learning_rate": 0.00011081967213114754, | |
| "loss": 0.0542, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 66.96766320927395, | |
| "grad_norm": 1.6410188674926758, | |
| "learning_rate": 0.00011068852459016395, | |
| "loss": 0.0648, | |
| "step": 13720 | |
| }, | |
| { | |
| "epoch": 67.065283709579, | |
| "grad_norm": 2.1899428367614746, | |
| "learning_rate": 0.00011055737704918035, | |
| "loss": 0.0518, | |
| "step": 13740 | |
| }, | |
| { | |
| "epoch": 67.16290420988408, | |
| "grad_norm": 1.662888526916504, | |
| "learning_rate": 0.00011042622950819672, | |
| "loss": 0.0481, | |
| "step": 13760 | |
| }, | |
| { | |
| "epoch": 67.26052471018915, | |
| "grad_norm": 1.8363763093948364, | |
| "learning_rate": 0.00011029508196721311, | |
| "loss": 0.0521, | |
| "step": 13780 | |
| }, | |
| { | |
| "epoch": 67.3581452104942, | |
| "grad_norm": 1.5296106338500977, | |
| "learning_rate": 0.00011016393442622952, | |
| "loss": 0.0569, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 67.45576571079927, | |
| "grad_norm": 3.030101776123047, | |
| "learning_rate": 0.00011003278688524589, | |
| "loss": 0.0542, | |
| "step": 13820 | |
| }, | |
| { | |
| "epoch": 67.55338621110434, | |
| "grad_norm": 1.4792683124542236, | |
| "learning_rate": 0.0001099016393442623, | |
| "loss": 0.0571, | |
| "step": 13840 | |
| }, | |
| { | |
| "epoch": 67.6510067114094, | |
| "grad_norm": 1.8390675783157349, | |
| "learning_rate": 0.0001097704918032787, | |
| "loss": 0.0562, | |
| "step": 13860 | |
| }, | |
| { | |
| "epoch": 67.74862721171446, | |
| "grad_norm": 1.6094483137130737, | |
| "learning_rate": 0.0001096393442622951, | |
| "loss": 0.0529, | |
| "step": 13880 | |
| }, | |
| { | |
| "epoch": 67.84624771201952, | |
| "grad_norm": 2.030127763748169, | |
| "learning_rate": 0.00010950819672131148, | |
| "loss": 0.0572, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 67.94386821232459, | |
| "grad_norm": 1.7216037511825562, | |
| "learning_rate": 0.00010937704918032787, | |
| "loss": 0.0618, | |
| "step": 13920 | |
| }, | |
| { | |
| "epoch": 68.04148871262966, | |
| "grad_norm": 2.081637382507324, | |
| "learning_rate": 0.00010924590163934427, | |
| "loss": 0.0524, | |
| "step": 13940 | |
| }, | |
| { | |
| "epoch": 68.13910921293471, | |
| "grad_norm": 1.6611350774765015, | |
| "learning_rate": 0.00010911475409836067, | |
| "loss": 0.0463, | |
| "step": 13960 | |
| }, | |
| { | |
| "epoch": 68.23672971323978, | |
| "grad_norm": 1.8240830898284912, | |
| "learning_rate": 0.00010898360655737705, | |
| "loss": 0.0528, | |
| "step": 13980 | |
| }, | |
| { | |
| "epoch": 68.33435021354484, | |
| "grad_norm": 2.299060344696045, | |
| "learning_rate": 0.00010885245901639345, | |
| "loss": 0.0525, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 68.4319707138499, | |
| "grad_norm": 2.1000702381134033, | |
| "learning_rate": 0.00010872131147540985, | |
| "loss": 0.0534, | |
| "step": 14020 | |
| }, | |
| { | |
| "epoch": 68.52959121415498, | |
| "grad_norm": 1.4629288911819458, | |
| "learning_rate": 0.00010859016393442623, | |
| "loss": 0.05, | |
| "step": 14040 | |
| }, | |
| { | |
| "epoch": 68.62721171446003, | |
| "grad_norm": 1.571041464805603, | |
| "learning_rate": 0.00010845901639344262, | |
| "loss": 0.0535, | |
| "step": 14060 | |
| }, | |
| { | |
| "epoch": 68.7248322147651, | |
| "grad_norm": 1.6722811460494995, | |
| "learning_rate": 0.00010832786885245902, | |
| "loss": 0.0498, | |
| "step": 14080 | |
| }, | |
| { | |
| "epoch": 68.82245271507017, | |
| "grad_norm": 1.8763340711593628, | |
| "learning_rate": 0.00010819672131147543, | |
| "loss": 0.0547, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 68.92007321537523, | |
| "grad_norm": 2.936068534851074, | |
| "learning_rate": 0.0001080655737704918, | |
| "loss": 0.055, | |
| "step": 14120 | |
| }, | |
| { | |
| "epoch": 69.0176937156803, | |
| "grad_norm": 2.24238920211792, | |
| "learning_rate": 0.0001079344262295082, | |
| "loss": 0.059, | |
| "step": 14140 | |
| }, | |
| { | |
| "epoch": 69.11531421598535, | |
| "grad_norm": 1.4389880895614624, | |
| "learning_rate": 0.00010780327868852461, | |
| "loss": 0.0453, | |
| "step": 14160 | |
| }, | |
| { | |
| "epoch": 69.21293471629042, | |
| "grad_norm": 1.5990976095199585, | |
| "learning_rate": 0.00010767213114754098, | |
| "loss": 0.0483, | |
| "step": 14180 | |
| }, | |
| { | |
| "epoch": 69.31055521659549, | |
| "grad_norm": 1.6288248300552368, | |
| "learning_rate": 0.00010754098360655737, | |
| "loss": 0.0492, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 69.40817571690054, | |
| "grad_norm": 1.6476655006408691, | |
| "learning_rate": 0.00010740983606557378, | |
| "loss": 0.0485, | |
| "step": 14220 | |
| }, | |
| { | |
| "epoch": 69.50579621720561, | |
| "grad_norm": 1.6670396327972412, | |
| "learning_rate": 0.00010727868852459018, | |
| "loss": 0.0573, | |
| "step": 14240 | |
| }, | |
| { | |
| "epoch": 69.60341671751068, | |
| "grad_norm": 1.6438359022140503, | |
| "learning_rate": 0.00010714754098360655, | |
| "loss": 0.0511, | |
| "step": 14260 | |
| }, | |
| { | |
| "epoch": 69.70103721781574, | |
| "grad_norm": 2.0703489780426025, | |
| "learning_rate": 0.00010701639344262296, | |
| "loss": 0.056, | |
| "step": 14280 | |
| }, | |
| { | |
| "epoch": 69.79865771812081, | |
| "grad_norm": 1.2886505126953125, | |
| "learning_rate": 0.00010688524590163935, | |
| "loss": 0.0518, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 69.89627821842586, | |
| "grad_norm": 1.869315505027771, | |
| "learning_rate": 0.00010675409836065575, | |
| "loss": 0.0613, | |
| "step": 14320 | |
| }, | |
| { | |
| "epoch": 69.99389871873093, | |
| "grad_norm": 1.5631581544876099, | |
| "learning_rate": 0.00010662295081967212, | |
| "loss": 0.0501, | |
| "step": 14340 | |
| }, | |
| { | |
| "epoch": 70.091519219036, | |
| "grad_norm": 2.5238254070281982, | |
| "learning_rate": 0.00010649180327868853, | |
| "loss": 0.0429, | |
| "step": 14360 | |
| }, | |
| { | |
| "epoch": 70.18913971934106, | |
| "grad_norm": 1.3227728605270386, | |
| "learning_rate": 0.00010636065573770493, | |
| "loss": 0.0474, | |
| "step": 14380 | |
| }, | |
| { | |
| "epoch": 70.28676021964613, | |
| "grad_norm": 2.284006118774414, | |
| "learning_rate": 0.0001062295081967213, | |
| "loss": 0.0513, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 70.3843807199512, | |
| "grad_norm": 1.3161708116531372, | |
| "learning_rate": 0.00010609836065573771, | |
| "loss": 0.0517, | |
| "step": 14420 | |
| }, | |
| { | |
| "epoch": 70.48200122025625, | |
| "grad_norm": 1.7371370792388916, | |
| "learning_rate": 0.0001059672131147541, | |
| "loss": 0.0494, | |
| "step": 14440 | |
| }, | |
| { | |
| "epoch": 70.57962172056132, | |
| "grad_norm": 3.2471296787261963, | |
| "learning_rate": 0.0001058360655737705, | |
| "loss": 0.0501, | |
| "step": 14460 | |
| }, | |
| { | |
| "epoch": 70.67724222086638, | |
| "grad_norm": 1.9515056610107422, | |
| "learning_rate": 0.00010570491803278688, | |
| "loss": 0.0562, | |
| "step": 14480 | |
| }, | |
| { | |
| "epoch": 70.77486272117144, | |
| "grad_norm": 1.9817485809326172, | |
| "learning_rate": 0.00010557377049180328, | |
| "loss": 0.0555, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 70.87248322147651, | |
| "grad_norm": 1.4870388507843018, | |
| "learning_rate": 0.00010544262295081968, | |
| "loss": 0.0529, | |
| "step": 14520 | |
| }, | |
| { | |
| "epoch": 70.97010372178157, | |
| "grad_norm": 1.6690961122512817, | |
| "learning_rate": 0.00010531147540983609, | |
| "loss": 0.0525, | |
| "step": 14540 | |
| }, | |
| { | |
| "epoch": 71.06772422208664, | |
| "grad_norm": 1.4471197128295898, | |
| "learning_rate": 0.00010518032786885246, | |
| "loss": 0.0473, | |
| "step": 14560 | |
| }, | |
| { | |
| "epoch": 71.16534472239171, | |
| "grad_norm": 2.1219794750213623, | |
| "learning_rate": 0.00010504918032786885, | |
| "loss": 0.044, | |
| "step": 14580 | |
| }, | |
| { | |
| "epoch": 71.26296522269676, | |
| "grad_norm": 2.1904103755950928, | |
| "learning_rate": 0.00010491803278688525, | |
| "loss": 0.0455, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 71.36058572300183, | |
| "grad_norm": 2.576287269592285, | |
| "learning_rate": 0.00010478688524590163, | |
| "loss": 0.047, | |
| "step": 14620 | |
| }, | |
| { | |
| "epoch": 71.45820622330689, | |
| "grad_norm": 2.6635916233062744, | |
| "learning_rate": 0.00010465573770491803, | |
| "loss": 0.0461, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 71.55582672361196, | |
| "grad_norm": 1.234979510307312, | |
| "learning_rate": 0.00010452459016393444, | |
| "loss": 0.0464, | |
| "step": 14660 | |
| }, | |
| { | |
| "epoch": 71.65344722391703, | |
| "grad_norm": 1.6037148237228394, | |
| "learning_rate": 0.00010439344262295083, | |
| "loss": 0.0522, | |
| "step": 14680 | |
| }, | |
| { | |
| "epoch": 71.75106772422208, | |
| "grad_norm": 1.990586280822754, | |
| "learning_rate": 0.00010426229508196721, | |
| "loss": 0.0577, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 71.84868822452715, | |
| "grad_norm": 1.5060235261917114, | |
| "learning_rate": 0.0001041311475409836, | |
| "loss": 0.0582, | |
| "step": 14720 | |
| }, | |
| { | |
| "epoch": 71.94630872483222, | |
| "grad_norm": 2.0390243530273438, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 0.0559, | |
| "step": 14740 | |
| }, | |
| { | |
| "epoch": 72.04392922513728, | |
| "grad_norm": 1.5087133646011353, | |
| "learning_rate": 0.00010386885245901641, | |
| "loss": 0.0538, | |
| "step": 14760 | |
| }, | |
| { | |
| "epoch": 72.14154972544235, | |
| "grad_norm": 2.0309317111968994, | |
| "learning_rate": 0.00010373770491803279, | |
| "loss": 0.0455, | |
| "step": 14780 | |
| }, | |
| { | |
| "epoch": 72.2391702257474, | |
| "grad_norm": 1.643815279006958, | |
| "learning_rate": 0.00010360655737704919, | |
| "loss": 0.0449, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 72.33679072605247, | |
| "grad_norm": 1.881631851196289, | |
| "learning_rate": 0.00010347540983606558, | |
| "loss": 0.0473, | |
| "step": 14820 | |
| }, | |
| { | |
| "epoch": 72.43441122635754, | |
| "grad_norm": 1.6459357738494873, | |
| "learning_rate": 0.00010334426229508197, | |
| "loss": 0.048, | |
| "step": 14840 | |
| }, | |
| { | |
| "epoch": 72.5320317266626, | |
| "grad_norm": 2.058401346206665, | |
| "learning_rate": 0.00010321311475409836, | |
| "loss": 0.0435, | |
| "step": 14860 | |
| }, | |
| { | |
| "epoch": 72.62965222696766, | |
| "grad_norm": 1.428796410560608, | |
| "learning_rate": 0.00010308196721311476, | |
| "loss": 0.0501, | |
| "step": 14880 | |
| }, | |
| { | |
| "epoch": 72.72727272727273, | |
| "grad_norm": 1.7262030839920044, | |
| "learning_rate": 0.00010295081967213116, | |
| "loss": 0.0512, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 72.82489322757779, | |
| "grad_norm": 1.5179914236068726, | |
| "learning_rate": 0.00010281967213114754, | |
| "loss": 0.0524, | |
| "step": 14920 | |
| }, | |
| { | |
| "epoch": 72.92251372788286, | |
| "grad_norm": 1.5609859228134155, | |
| "learning_rate": 0.00010268852459016394, | |
| "loss": 0.0547, | |
| "step": 14940 | |
| }, | |
| { | |
| "epoch": 73.02013422818791, | |
| "grad_norm": 1.1576975584030151, | |
| "learning_rate": 0.00010255737704918033, | |
| "loss": 0.0506, | |
| "step": 14960 | |
| }, | |
| { | |
| "epoch": 73.11775472849298, | |
| "grad_norm": 1.4885412454605103, | |
| "learning_rate": 0.00010242622950819673, | |
| "loss": 0.0424, | |
| "step": 14980 | |
| }, | |
| { | |
| "epoch": 73.21537522879805, | |
| "grad_norm": 1.0229533910751343, | |
| "learning_rate": 0.00010229508196721311, | |
| "loss": 0.0448, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 73.31299572910311, | |
| "grad_norm": 1.9551867246627808, | |
| "learning_rate": 0.00010216393442622951, | |
| "loss": 0.0468, | |
| "step": 15020 | |
| }, | |
| { | |
| "epoch": 73.41061622940818, | |
| "grad_norm": 2.8374905586242676, | |
| "learning_rate": 0.00010203278688524592, | |
| "loss": 0.0473, | |
| "step": 15040 | |
| }, | |
| { | |
| "epoch": 73.50823672971325, | |
| "grad_norm": 1.2752137184143066, | |
| "learning_rate": 0.00010190163934426229, | |
| "loss": 0.0487, | |
| "step": 15060 | |
| }, | |
| { | |
| "epoch": 73.6058572300183, | |
| "grad_norm": 1.7868965864181519, | |
| "learning_rate": 0.0001017704918032787, | |
| "loss": 0.0479, | |
| "step": 15080 | |
| }, | |
| { | |
| "epoch": 73.70347773032337, | |
| "grad_norm": 1.7490606307983398, | |
| "learning_rate": 0.00010163934426229508, | |
| "loss": 0.0466, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 73.80109823062843, | |
| "grad_norm": 2.4176886081695557, | |
| "learning_rate": 0.00010150819672131149, | |
| "loss": 0.0539, | |
| "step": 15120 | |
| }, | |
| { | |
| "epoch": 73.8987187309335, | |
| "grad_norm": 1.9388647079467773, | |
| "learning_rate": 0.00010137704918032786, | |
| "loss": 0.0527, | |
| "step": 15140 | |
| }, | |
| { | |
| "epoch": 73.99633923123857, | |
| "grad_norm": 1.9763060808181763, | |
| "learning_rate": 0.00010124590163934427, | |
| "loss": 0.0512, | |
| "step": 15160 | |
| }, | |
| { | |
| "epoch": 74.09395973154362, | |
| "grad_norm": 1.1881762742996216, | |
| "learning_rate": 0.00010111475409836067, | |
| "loss": 0.0406, | |
| "step": 15180 | |
| }, | |
| { | |
| "epoch": 74.19158023184869, | |
| "grad_norm": 1.5949184894561768, | |
| "learning_rate": 0.00010098360655737706, | |
| "loss": 0.0446, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 74.28920073215376, | |
| "grad_norm": 1.3817694187164307, | |
| "learning_rate": 0.00010085245901639345, | |
| "loss": 0.0468, | |
| "step": 15220 | |
| }, | |
| { | |
| "epoch": 74.38682123245881, | |
| "grad_norm": 1.4940481185913086, | |
| "learning_rate": 0.00010072131147540984, | |
| "loss": 0.0528, | |
| "step": 15240 | |
| }, | |
| { | |
| "epoch": 74.48444173276388, | |
| "grad_norm": 1.5146404504776, | |
| "learning_rate": 0.00010059016393442624, | |
| "loss": 0.0503, | |
| "step": 15260 | |
| }, | |
| { | |
| "epoch": 74.58206223306894, | |
| "grad_norm": 1.4908167123794556, | |
| "learning_rate": 0.00010045901639344261, | |
| "loss": 0.0474, | |
| "step": 15280 | |
| }, | |
| { | |
| "epoch": 74.67968273337401, | |
| "grad_norm": 1.9127382040023804, | |
| "learning_rate": 0.00010032786885245902, | |
| "loss": 0.0475, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 74.77730323367908, | |
| "grad_norm": 1.9283052682876587, | |
| "learning_rate": 0.00010019672131147542, | |
| "loss": 0.0482, | |
| "step": 15320 | |
| }, | |
| { | |
| "epoch": 74.87492373398413, | |
| "grad_norm": 1.434121012687683, | |
| "learning_rate": 0.00010006557377049181, | |
| "loss": 0.0484, | |
| "step": 15340 | |
| }, | |
| { | |
| "epoch": 74.9725442342892, | |
| "grad_norm": 1.2398452758789062, | |
| "learning_rate": 9.99344262295082e-05, | |
| "loss": 0.0474, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 75.07016473459427, | |
| "grad_norm": 1.463592529296875, | |
| "learning_rate": 9.980327868852459e-05, | |
| "loss": 0.0435, | |
| "step": 15380 | |
| }, | |
| { | |
| "epoch": 75.16778523489933, | |
| "grad_norm": 1.5821679830551147, | |
| "learning_rate": 9.967213114754099e-05, | |
| "loss": 0.0447, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 75.2654057352044, | |
| "grad_norm": 1.525728702545166, | |
| "learning_rate": 9.954098360655738e-05, | |
| "loss": 0.0442, | |
| "step": 15420 | |
| }, | |
| { | |
| "epoch": 75.36302623550945, | |
| "grad_norm": 1.2761436700820923, | |
| "learning_rate": 9.940983606557378e-05, | |
| "loss": 0.0464, | |
| "step": 15440 | |
| }, | |
| { | |
| "epoch": 75.46064673581452, | |
| "grad_norm": 1.4206050634384155, | |
| "learning_rate": 9.927868852459017e-05, | |
| "loss": 0.0436, | |
| "step": 15460 | |
| }, | |
| { | |
| "epoch": 75.55826723611959, | |
| "grad_norm": 2.7865281105041504, | |
| "learning_rate": 9.914754098360656e-05, | |
| "loss": 0.0505, | |
| "step": 15480 | |
| }, | |
| { | |
| "epoch": 75.65588773642465, | |
| "grad_norm": 1.249638557434082, | |
| "learning_rate": 9.901639344262295e-05, | |
| "loss": 0.0461, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 75.75350823672972, | |
| "grad_norm": 1.9104782342910767, | |
| "learning_rate": 9.888524590163934e-05, | |
| "loss": 0.0465, | |
| "step": 15520 | |
| }, | |
| { | |
| "epoch": 75.85112873703477, | |
| "grad_norm": 1.3011826276779175, | |
| "learning_rate": 9.875409836065574e-05, | |
| "loss": 0.0488, | |
| "step": 15540 | |
| }, | |
| { | |
| "epoch": 75.94874923733984, | |
| "grad_norm": 2.3834664821624756, | |
| "learning_rate": 9.862295081967213e-05, | |
| "loss": 0.0519, | |
| "step": 15560 | |
| }, | |
| { | |
| "epoch": 76.04636973764491, | |
| "grad_norm": 1.3640445470809937, | |
| "learning_rate": 9.849180327868854e-05, | |
| "loss": 0.0449, | |
| "step": 15580 | |
| }, | |
| { | |
| "epoch": 76.14399023794996, | |
| "grad_norm": 1.3013081550598145, | |
| "learning_rate": 9.836065573770493e-05, | |
| "loss": 0.0399, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 76.24161073825503, | |
| "grad_norm": 2.726545810699463, | |
| "learning_rate": 9.822950819672132e-05, | |
| "loss": 0.0427, | |
| "step": 15620 | |
| }, | |
| { | |
| "epoch": 76.3392312385601, | |
| "grad_norm": 1.2742412090301514, | |
| "learning_rate": 9.80983606557377e-05, | |
| "loss": 0.0439, | |
| "step": 15640 | |
| }, | |
| { | |
| "epoch": 76.43685173886516, | |
| "grad_norm": 1.67985999584198, | |
| "learning_rate": 9.796721311475411e-05, | |
| "loss": 0.0483, | |
| "step": 15660 | |
| }, | |
| { | |
| "epoch": 76.53447223917023, | |
| "grad_norm": 1.9676116704940796, | |
| "learning_rate": 9.78360655737705e-05, | |
| "loss": 0.0442, | |
| "step": 15680 | |
| }, | |
| { | |
| "epoch": 76.63209273947528, | |
| "grad_norm": 2.992687463760376, | |
| "learning_rate": 9.770491803278689e-05, | |
| "loss": 0.0429, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 76.72971323978035, | |
| "grad_norm": 1.0530943870544434, | |
| "learning_rate": 9.757377049180329e-05, | |
| "loss": 0.0488, | |
| "step": 15720 | |
| }, | |
| { | |
| "epoch": 76.82733374008542, | |
| "grad_norm": 1.5308395624160767, | |
| "learning_rate": 9.744262295081968e-05, | |
| "loss": 0.0442, | |
| "step": 15740 | |
| }, | |
| { | |
| "epoch": 76.92495424039048, | |
| "grad_norm": 1.6086573600769043, | |
| "learning_rate": 9.731147540983607e-05, | |
| "loss": 0.0504, | |
| "step": 15760 | |
| }, | |
| { | |
| "epoch": 77.02257474069555, | |
| "grad_norm": 0.7401487827301025, | |
| "learning_rate": 9.718032786885246e-05, | |
| "loss": 0.0469, | |
| "step": 15780 | |
| }, | |
| { | |
| "epoch": 77.12019524100062, | |
| "grad_norm": 1.6960521936416626, | |
| "learning_rate": 9.704918032786886e-05, | |
| "loss": 0.041, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 77.21781574130567, | |
| "grad_norm": 1.638244390487671, | |
| "learning_rate": 9.691803278688525e-05, | |
| "loss": 0.0381, | |
| "step": 15820 | |
| }, | |
| { | |
| "epoch": 77.31543624161074, | |
| "grad_norm": 1.749157190322876, | |
| "learning_rate": 9.678688524590165e-05, | |
| "loss": 0.0425, | |
| "step": 15840 | |
| }, | |
| { | |
| "epoch": 77.4130567419158, | |
| "grad_norm": 1.9708433151245117, | |
| "learning_rate": 9.665573770491804e-05, | |
| "loss": 0.046, | |
| "step": 15860 | |
| }, | |
| { | |
| "epoch": 77.51067724222086, | |
| "grad_norm": 1.3981733322143555, | |
| "learning_rate": 9.652459016393443e-05, | |
| "loss": 0.0489, | |
| "step": 15880 | |
| }, | |
| { | |
| "epoch": 77.60829774252593, | |
| "grad_norm": 2.0385096073150635, | |
| "learning_rate": 9.639344262295082e-05, | |
| "loss": 0.0442, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 77.70591824283099, | |
| "grad_norm": 1.5670779943466187, | |
| "learning_rate": 9.626229508196721e-05, | |
| "loss": 0.0412, | |
| "step": 15920 | |
| }, | |
| { | |
| "epoch": 77.80353874313606, | |
| "grad_norm": 1.3687881231307983, | |
| "learning_rate": 9.613114754098361e-05, | |
| "loss": 0.0469, | |
| "step": 15940 | |
| }, | |
| { | |
| "epoch": 77.90115924344113, | |
| "grad_norm": 1.388791799545288, | |
| "learning_rate": 9.6e-05, | |
| "loss": 0.0456, | |
| "step": 15960 | |
| }, | |
| { | |
| "epoch": 77.99877974374618, | |
| "grad_norm": 1.5988264083862305, | |
| "learning_rate": 9.58688524590164e-05, | |
| "loss": 0.048, | |
| "step": 15980 | |
| }, | |
| { | |
| "epoch": 78.09640024405125, | |
| "grad_norm": 2.0128464698791504, | |
| "learning_rate": 9.57377049180328e-05, | |
| "loss": 0.0406, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 78.19402074435631, | |
| "grad_norm": 1.9334806203842163, | |
| "learning_rate": 9.560655737704918e-05, | |
| "loss": 0.0402, | |
| "step": 16020 | |
| }, | |
| { | |
| "epoch": 78.29164124466138, | |
| "grad_norm": 2.799999952316284, | |
| "learning_rate": 9.547540983606557e-05, | |
| "loss": 0.0415, | |
| "step": 16040 | |
| }, | |
| { | |
| "epoch": 78.38926174496645, | |
| "grad_norm": 1.2322758436203003, | |
| "learning_rate": 9.534426229508198e-05, | |
| "loss": 0.0396, | |
| "step": 16060 | |
| }, | |
| { | |
| "epoch": 78.4868822452715, | |
| "grad_norm": 2.0646209716796875, | |
| "learning_rate": 9.521311475409837e-05, | |
| "loss": 0.0449, | |
| "step": 16080 | |
| }, | |
| { | |
| "epoch": 78.58450274557657, | |
| "grad_norm": 1.3239318132400513, | |
| "learning_rate": 9.508196721311476e-05, | |
| "loss": 0.0473, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 78.68212324588164, | |
| "grad_norm": 1.0266462564468384, | |
| "learning_rate": 9.495081967213116e-05, | |
| "loss": 0.047, | |
| "step": 16120 | |
| }, | |
| { | |
| "epoch": 78.7797437461867, | |
| "grad_norm": 1.3568265438079834, | |
| "learning_rate": 9.481967213114755e-05, | |
| "loss": 0.0472, | |
| "step": 16140 | |
| }, | |
| { | |
| "epoch": 78.87736424649177, | |
| "grad_norm": 1.8407416343688965, | |
| "learning_rate": 9.468852459016394e-05, | |
| "loss": 0.0489, | |
| "step": 16160 | |
| }, | |
| { | |
| "epoch": 78.97498474679682, | |
| "grad_norm": 3.1566367149353027, | |
| "learning_rate": 9.455737704918033e-05, | |
| "loss": 0.0447, | |
| "step": 16180 | |
| }, | |
| { | |
| "epoch": 79.07260524710189, | |
| "grad_norm": 1.4140985012054443, | |
| "learning_rate": 9.442622950819673e-05, | |
| "loss": 0.0397, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 79.17022574740696, | |
| "grad_norm": 1.774674654006958, | |
| "learning_rate": 9.429508196721312e-05, | |
| "loss": 0.0394, | |
| "step": 16220 | |
| }, | |
| { | |
| "epoch": 79.26784624771201, | |
| "grad_norm": 1.8382582664489746, | |
| "learning_rate": 9.416393442622952e-05, | |
| "loss": 0.0391, | |
| "step": 16240 | |
| }, | |
| { | |
| "epoch": 79.36546674801708, | |
| "grad_norm": 1.465022087097168, | |
| "learning_rate": 9.403278688524591e-05, | |
| "loss": 0.0427, | |
| "step": 16260 | |
| }, | |
| { | |
| "epoch": 79.46308724832215, | |
| "grad_norm": 1.5447912216186523, | |
| "learning_rate": 9.39016393442623e-05, | |
| "loss": 0.0435, | |
| "step": 16280 | |
| }, | |
| { | |
| "epoch": 79.56070774862721, | |
| "grad_norm": 1.5630700588226318, | |
| "learning_rate": 9.377049180327869e-05, | |
| "loss": 0.0426, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 79.65832824893228, | |
| "grad_norm": 2.1536877155303955, | |
| "learning_rate": 9.363934426229508e-05, | |
| "loss": 0.0436, | |
| "step": 16320 | |
| }, | |
| { | |
| "epoch": 79.75594874923733, | |
| "grad_norm": 1.5898195505142212, | |
| "learning_rate": 9.350819672131148e-05, | |
| "loss": 0.0488, | |
| "step": 16340 | |
| }, | |
| { | |
| "epoch": 79.8535692495424, | |
| "grad_norm": 1.5002379417419434, | |
| "learning_rate": 9.337704918032787e-05, | |
| "loss": 0.0453, | |
| "step": 16360 | |
| }, | |
| { | |
| "epoch": 79.95118974984747, | |
| "grad_norm": 1.7117916345596313, | |
| "learning_rate": 9.324590163934427e-05, | |
| "loss": 0.0483, | |
| "step": 16380 | |
| }, | |
| { | |
| "epoch": 80.04881025015253, | |
| "grad_norm": 1.672136664390564, | |
| "learning_rate": 9.311475409836066e-05, | |
| "loss": 0.0444, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 80.1464307504576, | |
| "grad_norm": 1.2889288663864136, | |
| "learning_rate": 9.298360655737705e-05, | |
| "loss": 0.0377, | |
| "step": 16420 | |
| }, | |
| { | |
| "epoch": 80.24405125076267, | |
| "grad_norm": 1.602657675743103, | |
| "learning_rate": 9.285245901639344e-05, | |
| "loss": 0.0431, | |
| "step": 16440 | |
| }, | |
| { | |
| "epoch": 80.34167175106772, | |
| "grad_norm": 1.673161506652832, | |
| "learning_rate": 9.272131147540985e-05, | |
| "loss": 0.0392, | |
| "step": 16460 | |
| }, | |
| { | |
| "epoch": 80.43929225137279, | |
| "grad_norm": 1.5590755939483643, | |
| "learning_rate": 9.259016393442623e-05, | |
| "loss": 0.0401, | |
| "step": 16480 | |
| }, | |
| { | |
| "epoch": 80.53691275167785, | |
| "grad_norm": 1.3792154788970947, | |
| "learning_rate": 9.245901639344264e-05, | |
| "loss": 0.0434, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 80.63453325198292, | |
| "grad_norm": 1.0110834836959839, | |
| "learning_rate": 9.232786885245903e-05, | |
| "loss": 0.0455, | |
| "step": 16520 | |
| }, | |
| { | |
| "epoch": 80.73215375228799, | |
| "grad_norm": 1.3876726627349854, | |
| "learning_rate": 9.21967213114754e-05, | |
| "loss": 0.0457, | |
| "step": 16540 | |
| }, | |
| { | |
| "epoch": 80.82977425259304, | |
| "grad_norm": 2.110999822616577, | |
| "learning_rate": 9.20655737704918e-05, | |
| "loss": 0.0467, | |
| "step": 16560 | |
| }, | |
| { | |
| "epoch": 80.92739475289811, | |
| "grad_norm": 1.9339549541473389, | |
| "learning_rate": 9.19344262295082e-05, | |
| "loss": 0.0432, | |
| "step": 16580 | |
| }, | |
| { | |
| "epoch": 81.02501525320318, | |
| "grad_norm": 1.1022757291793823, | |
| "learning_rate": 9.18032786885246e-05, | |
| "loss": 0.0454, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 81.12263575350823, | |
| "grad_norm": 1.7495752573013306, | |
| "learning_rate": 9.167213114754099e-05, | |
| "loss": 0.0375, | |
| "step": 16620 | |
| }, | |
| { | |
| "epoch": 81.2202562538133, | |
| "grad_norm": 1.6110937595367432, | |
| "learning_rate": 9.154098360655739e-05, | |
| "loss": 0.0399, | |
| "step": 16640 | |
| }, | |
| { | |
| "epoch": 81.31787675411836, | |
| "grad_norm": 1.7473942041397095, | |
| "learning_rate": 9.140983606557378e-05, | |
| "loss": 0.0399, | |
| "step": 16660 | |
| }, | |
| { | |
| "epoch": 81.41549725442343, | |
| "grad_norm": 1.6311129331588745, | |
| "learning_rate": 9.127868852459017e-05, | |
| "loss": 0.0397, | |
| "step": 16680 | |
| }, | |
| { | |
| "epoch": 81.5131177547285, | |
| "grad_norm": 2.0772838592529297, | |
| "learning_rate": 9.114754098360656e-05, | |
| "loss": 0.042, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 81.61073825503355, | |
| "grad_norm": 1.5818514823913574, | |
| "learning_rate": 9.101639344262296e-05, | |
| "loss": 0.0419, | |
| "step": 16720 | |
| }, | |
| { | |
| "epoch": 81.70835875533862, | |
| "grad_norm": 1.0751603841781616, | |
| "learning_rate": 9.088524590163935e-05, | |
| "loss": 0.0447, | |
| "step": 16740 | |
| }, | |
| { | |
| "epoch": 81.80597925564369, | |
| "grad_norm": 1.2915217876434326, | |
| "learning_rate": 9.075409836065574e-05, | |
| "loss": 0.0425, | |
| "step": 16760 | |
| }, | |
| { | |
| "epoch": 81.90359975594875, | |
| "grad_norm": 1.4814674854278564, | |
| "learning_rate": 9.062295081967214e-05, | |
| "loss": 0.0455, | |
| "step": 16780 | |
| }, | |
| { | |
| "epoch": 82.00122025625382, | |
| "grad_norm": 1.1471761465072632, | |
| "learning_rate": 9.049180327868852e-05, | |
| "loss": 0.0418, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 82.09884075655887, | |
| "grad_norm": 1.663212537765503, | |
| "learning_rate": 9.036065573770492e-05, | |
| "loss": 0.0348, | |
| "step": 16820 | |
| }, | |
| { | |
| "epoch": 82.19646125686394, | |
| "grad_norm": 2.423266887664795, | |
| "learning_rate": 9.022950819672131e-05, | |
| "loss": 0.0367, | |
| "step": 16840 | |
| }, | |
| { | |
| "epoch": 82.29408175716901, | |
| "grad_norm": 1.5083626508712769, | |
| "learning_rate": 9.009836065573771e-05, | |
| "loss": 0.038, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 82.39170225747407, | |
| "grad_norm": 1.2111772298812866, | |
| "learning_rate": 8.99672131147541e-05, | |
| "loss": 0.0404, | |
| "step": 16880 | |
| }, | |
| { | |
| "epoch": 82.48932275777914, | |
| "grad_norm": 2.1727235317230225, | |
| "learning_rate": 8.98360655737705e-05, | |
| "loss": 0.0434, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 82.5869432580842, | |
| "grad_norm": 1.3636887073516846, | |
| "learning_rate": 8.97049180327869e-05, | |
| "loss": 0.0404, | |
| "step": 16920 | |
| }, | |
| { | |
| "epoch": 82.68456375838926, | |
| "grad_norm": 2.5128073692321777, | |
| "learning_rate": 8.957377049180328e-05, | |
| "loss": 0.0442, | |
| "step": 16940 | |
| }, | |
| { | |
| "epoch": 82.78218425869433, | |
| "grad_norm": 1.3435509204864502, | |
| "learning_rate": 8.944262295081967e-05, | |
| "loss": 0.044, | |
| "step": 16960 | |
| }, | |
| { | |
| "epoch": 82.87980475899938, | |
| "grad_norm": 2.397655487060547, | |
| "learning_rate": 8.931147540983606e-05, | |
| "loss": 0.0447, | |
| "step": 16980 | |
| }, | |
| { | |
| "epoch": 82.97742525930445, | |
| "grad_norm": 1.5072938203811646, | |
| "learning_rate": 8.918032786885247e-05, | |
| "loss": 0.0411, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 83.07504575960952, | |
| "grad_norm": 1.9065088033676147, | |
| "learning_rate": 8.904918032786886e-05, | |
| "loss": 0.036, | |
| "step": 17020 | |
| }, | |
| { | |
| "epoch": 83.17266625991458, | |
| "grad_norm": 1.4267394542694092, | |
| "learning_rate": 8.891803278688526e-05, | |
| "loss": 0.035, | |
| "step": 17040 | |
| }, | |
| { | |
| "epoch": 83.27028676021965, | |
| "grad_norm": 1.677395224571228, | |
| "learning_rate": 8.878688524590163e-05, | |
| "loss": 0.0425, | |
| "step": 17060 | |
| }, | |
| { | |
| "epoch": 83.36790726052472, | |
| "grad_norm": 1.5407129526138306, | |
| "learning_rate": 8.865573770491804e-05, | |
| "loss": 0.0362, | |
| "step": 17080 | |
| }, | |
| { | |
| "epoch": 83.46552776082977, | |
| "grad_norm": 1.1154581308364868, | |
| "learning_rate": 8.852459016393443e-05, | |
| "loss": 0.0403, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 83.56314826113484, | |
| "grad_norm": 1.6043624877929688, | |
| "learning_rate": 8.839344262295083e-05, | |
| "loss": 0.0413, | |
| "step": 17120 | |
| }, | |
| { | |
| "epoch": 83.6607687614399, | |
| "grad_norm": 1.4952675104141235, | |
| "learning_rate": 8.826229508196722e-05, | |
| "loss": 0.0397, | |
| "step": 17140 | |
| }, | |
| { | |
| "epoch": 83.75838926174497, | |
| "grad_norm": 1.4162040948867798, | |
| "learning_rate": 8.813114754098362e-05, | |
| "loss": 0.0435, | |
| "step": 17160 | |
| }, | |
| { | |
| "epoch": 83.85600976205004, | |
| "grad_norm": 1.616909384727478, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.0455, | |
| "step": 17180 | |
| }, | |
| { | |
| "epoch": 83.95363026235509, | |
| "grad_norm": 1.3712050914764404, | |
| "learning_rate": 8.786885245901639e-05, | |
| "loss": 0.0442, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 84.05125076266016, | |
| "grad_norm": 1.0349308252334595, | |
| "learning_rate": 8.773770491803279e-05, | |
| "loss": 0.0368, | |
| "step": 17220 | |
| }, | |
| { | |
| "epoch": 84.14887126296523, | |
| "grad_norm": 1.427959680557251, | |
| "learning_rate": 8.760655737704918e-05, | |
| "loss": 0.0318, | |
| "step": 17240 | |
| }, | |
| { | |
| "epoch": 84.24649176327028, | |
| "grad_norm": 1.1814582347869873, | |
| "learning_rate": 8.747540983606558e-05, | |
| "loss": 0.0386, | |
| "step": 17260 | |
| }, | |
| { | |
| "epoch": 84.34411226357535, | |
| "grad_norm": 1.0893490314483643, | |
| "learning_rate": 8.734426229508197e-05, | |
| "loss": 0.0398, | |
| "step": 17280 | |
| }, | |
| { | |
| "epoch": 84.44173276388041, | |
| "grad_norm": 1.2510510683059692, | |
| "learning_rate": 8.721311475409837e-05, | |
| "loss": 0.0425, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 84.53935326418548, | |
| "grad_norm": 0.9642720818519592, | |
| "learning_rate": 8.708196721311475e-05, | |
| "loss": 0.0384, | |
| "step": 17320 | |
| }, | |
| { | |
| "epoch": 84.63697376449055, | |
| "grad_norm": 1.2420659065246582, | |
| "learning_rate": 8.695081967213115e-05, | |
| "loss": 0.038, | |
| "step": 17340 | |
| }, | |
| { | |
| "epoch": 84.7345942647956, | |
| "grad_norm": 1.0976778268814087, | |
| "learning_rate": 8.681967213114754e-05, | |
| "loss": 0.0428, | |
| "step": 17360 | |
| }, | |
| { | |
| "epoch": 84.83221476510067, | |
| "grad_norm": 1.282233476638794, | |
| "learning_rate": 8.668852459016393e-05, | |
| "loss": 0.0418, | |
| "step": 17380 | |
| }, | |
| { | |
| "epoch": 84.92983526540573, | |
| "grad_norm": 2.0834672451019287, | |
| "learning_rate": 8.655737704918033e-05, | |
| "loss": 0.0441, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 85.0274557657108, | |
| "grad_norm": 1.0747671127319336, | |
| "learning_rate": 8.642622950819672e-05, | |
| "loss": 0.0415, | |
| "step": 17420 | |
| }, | |
| { | |
| "epoch": 85.12507626601587, | |
| "grad_norm": 1.6715941429138184, | |
| "learning_rate": 8.629508196721313e-05, | |
| "loss": 0.0366, | |
| "step": 17440 | |
| }, | |
| { | |
| "epoch": 85.22269676632092, | |
| "grad_norm": 1.115537166595459, | |
| "learning_rate": 8.61639344262295e-05, | |
| "loss": 0.0326, | |
| "step": 17460 | |
| }, | |
| { | |
| "epoch": 85.32031726662599, | |
| "grad_norm": 1.118455410003662, | |
| "learning_rate": 8.60327868852459e-05, | |
| "loss": 0.0384, | |
| "step": 17480 | |
| }, | |
| { | |
| "epoch": 85.41793776693106, | |
| "grad_norm": 1.6129100322723389, | |
| "learning_rate": 8.59016393442623e-05, | |
| "loss": 0.0376, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 85.51555826723612, | |
| "grad_norm": 2.3377175331115723, | |
| "learning_rate": 8.57704918032787e-05, | |
| "loss": 0.0388, | |
| "step": 17520 | |
| }, | |
| { | |
| "epoch": 85.61317876754119, | |
| "grad_norm": 1.2189383506774902, | |
| "learning_rate": 8.563934426229509e-05, | |
| "loss": 0.0405, | |
| "step": 17540 | |
| }, | |
| { | |
| "epoch": 85.71079926784624, | |
| "grad_norm": 1.24251389503479, | |
| "learning_rate": 8.550819672131149e-05, | |
| "loss": 0.0426, | |
| "step": 17560 | |
| }, | |
| { | |
| "epoch": 85.80841976815131, | |
| "grad_norm": 1.1355634927749634, | |
| "learning_rate": 8.537704918032787e-05, | |
| "loss": 0.0407, | |
| "step": 17580 | |
| }, | |
| { | |
| "epoch": 85.90604026845638, | |
| "grad_norm": 1.5692590475082397, | |
| "learning_rate": 8.524590163934426e-05, | |
| "loss": 0.0438, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 86.00366076876143, | |
| "grad_norm": 0.9955014586448669, | |
| "learning_rate": 8.511475409836066e-05, | |
| "loss": 0.0414, | |
| "step": 17620 | |
| }, | |
| { | |
| "epoch": 86.1012812690665, | |
| "grad_norm": 1.3448798656463623, | |
| "learning_rate": 8.498360655737705e-05, | |
| "loss": 0.0307, | |
| "step": 17640 | |
| }, | |
| { | |
| "epoch": 86.19890176937157, | |
| "grad_norm": 2.263423442840576, | |
| "learning_rate": 8.485245901639345e-05, | |
| "loss": 0.0354, | |
| "step": 17660 | |
| }, | |
| { | |
| "epoch": 86.29652226967663, | |
| "grad_norm": 2.385218620300293, | |
| "learning_rate": 8.472131147540984e-05, | |
| "loss": 0.0373, | |
| "step": 17680 | |
| }, | |
| { | |
| "epoch": 86.3941427699817, | |
| "grad_norm": 1.5091354846954346, | |
| "learning_rate": 8.459016393442624e-05, | |
| "loss": 0.0366, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 86.49176327028675, | |
| "grad_norm": 1.300026297569275, | |
| "learning_rate": 8.445901639344262e-05, | |
| "loss": 0.0423, | |
| "step": 17720 | |
| }, | |
| { | |
| "epoch": 86.58938377059182, | |
| "grad_norm": 1.4994693994522095, | |
| "learning_rate": 8.432786885245902e-05, | |
| "loss": 0.0406, | |
| "step": 17740 | |
| }, | |
| { | |
| "epoch": 86.68700427089689, | |
| "grad_norm": 1.3377670049667358, | |
| "learning_rate": 8.419672131147541e-05, | |
| "loss": 0.0414, | |
| "step": 17760 | |
| }, | |
| { | |
| "epoch": 86.78462477120195, | |
| "grad_norm": 0.9744328260421753, | |
| "learning_rate": 8.406557377049181e-05, | |
| "loss": 0.0391, | |
| "step": 17780 | |
| }, | |
| { | |
| "epoch": 86.88224527150702, | |
| "grad_norm": 1.1345022916793823, | |
| "learning_rate": 8.39344262295082e-05, | |
| "loss": 0.0404, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 86.97986577181209, | |
| "grad_norm": 1.7516177892684937, | |
| "learning_rate": 8.380327868852459e-05, | |
| "loss": 0.0439, | |
| "step": 17820 | |
| }, | |
| { | |
| "epoch": 87.07748627211714, | |
| "grad_norm": 1.2439887523651123, | |
| "learning_rate": 8.367213114754098e-05, | |
| "loss": 0.0335, | |
| "step": 17840 | |
| }, | |
| { | |
| "epoch": 87.17510677242221, | |
| "grad_norm": 1.586146354675293, | |
| "learning_rate": 8.354098360655737e-05, | |
| "loss": 0.031, | |
| "step": 17860 | |
| }, | |
| { | |
| "epoch": 87.27272727272727, | |
| "grad_norm": 1.3947432041168213, | |
| "learning_rate": 8.340983606557377e-05, | |
| "loss": 0.0385, | |
| "step": 17880 | |
| }, | |
| { | |
| "epoch": 87.37034777303234, | |
| "grad_norm": 0.9418930411338806, | |
| "learning_rate": 8.327868852459016e-05, | |
| "loss": 0.0364, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 87.4679682733374, | |
| "grad_norm": 0.8570541739463806, | |
| "learning_rate": 8.314754098360657e-05, | |
| "loss": 0.0378, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 87.56558877364246, | |
| "grad_norm": 1.2564805746078491, | |
| "learning_rate": 8.301639344262296e-05, | |
| "loss": 0.0373, | |
| "step": 17940 | |
| }, | |
| { | |
| "epoch": 87.66320927394753, | |
| "grad_norm": 1.062628149986267, | |
| "learning_rate": 8.288524590163935e-05, | |
| "loss": 0.0372, | |
| "step": 17960 | |
| }, | |
| { | |
| "epoch": 87.7608297742526, | |
| "grad_norm": 1.4381693601608276, | |
| "learning_rate": 8.275409836065573e-05, | |
| "loss": 0.0417, | |
| "step": 17980 | |
| }, | |
| { | |
| "epoch": 87.85845027455765, | |
| "grad_norm": 1.6939609050750732, | |
| "learning_rate": 8.262295081967214e-05, | |
| "loss": 0.0414, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 87.95607077486272, | |
| "grad_norm": 1.5280972719192505, | |
| "learning_rate": 8.249180327868853e-05, | |
| "loss": 0.0411, | |
| "step": 18020 | |
| }, | |
| { | |
| "epoch": 88.05369127516778, | |
| "grad_norm": 1.0062308311462402, | |
| "learning_rate": 8.236065573770492e-05, | |
| "loss": 0.0354, | |
| "step": 18040 | |
| }, | |
| { | |
| "epoch": 88.15131177547285, | |
| "grad_norm": 2.354576826095581, | |
| "learning_rate": 8.222950819672132e-05, | |
| "loss": 0.0318, | |
| "step": 18060 | |
| }, | |
| { | |
| "epoch": 88.24893227577792, | |
| "grad_norm": 1.3578128814697266, | |
| "learning_rate": 8.209836065573771e-05, | |
| "loss": 0.0355, | |
| "step": 18080 | |
| }, | |
| { | |
| "epoch": 88.34655277608297, | |
| "grad_norm": 1.0361335277557373, | |
| "learning_rate": 8.19672131147541e-05, | |
| "loss": 0.0332, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 88.44417327638804, | |
| "grad_norm": 1.3577088117599487, | |
| "learning_rate": 8.183606557377049e-05, | |
| "loss": 0.0372, | |
| "step": 18120 | |
| }, | |
| { | |
| "epoch": 88.54179377669311, | |
| "grad_norm": 1.5548292398452759, | |
| "learning_rate": 8.170491803278689e-05, | |
| "loss": 0.0384, | |
| "step": 18140 | |
| }, | |
| { | |
| "epoch": 88.63941427699817, | |
| "grad_norm": 0.9935953617095947, | |
| "learning_rate": 8.157377049180328e-05, | |
| "loss": 0.0364, | |
| "step": 18160 | |
| }, | |
| { | |
| "epoch": 88.73703477730324, | |
| "grad_norm": 1.093231201171875, | |
| "learning_rate": 8.144262295081968e-05, | |
| "loss": 0.0369, | |
| "step": 18180 | |
| }, | |
| { | |
| "epoch": 88.83465527760829, | |
| "grad_norm": 1.0061986446380615, | |
| "learning_rate": 8.131147540983607e-05, | |
| "loss": 0.0415, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 88.93227577791336, | |
| "grad_norm": 1.2928169965744019, | |
| "learning_rate": 8.118032786885246e-05, | |
| "loss": 0.0405, | |
| "step": 18220 | |
| }, | |
| { | |
| "epoch": 89.02989627821843, | |
| "grad_norm": 1.4209178686141968, | |
| "learning_rate": 8.104918032786885e-05, | |
| "loss": 0.0382, | |
| "step": 18240 | |
| }, | |
| { | |
| "epoch": 89.12751677852349, | |
| "grad_norm": 1.7799944877624512, | |
| "learning_rate": 8.091803278688524e-05, | |
| "loss": 0.0338, | |
| "step": 18260 | |
| }, | |
| { | |
| "epoch": 89.22513727882856, | |
| "grad_norm": 0.9708789587020874, | |
| "learning_rate": 8.078688524590164e-05, | |
| "loss": 0.0362, | |
| "step": 18280 | |
| }, | |
| { | |
| "epoch": 89.32275777913362, | |
| "grad_norm": 1.9509353637695312, | |
| "learning_rate": 8.065573770491803e-05, | |
| "loss": 0.0339, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 89.42037827943868, | |
| "grad_norm": 1.2407267093658447, | |
| "learning_rate": 8.052459016393444e-05, | |
| "loss": 0.0368, | |
| "step": 18320 | |
| }, | |
| { | |
| "epoch": 89.51799877974375, | |
| "grad_norm": 1.3398857116699219, | |
| "learning_rate": 8.039344262295082e-05, | |
| "loss": 0.0368, | |
| "step": 18340 | |
| }, | |
| { | |
| "epoch": 89.6156192800488, | |
| "grad_norm": 1.7856793403625488, | |
| "learning_rate": 8.026229508196721e-05, | |
| "loss": 0.0421, | |
| "step": 18360 | |
| }, | |
| { | |
| "epoch": 89.71323978035387, | |
| "grad_norm": 0.8198001980781555, | |
| "learning_rate": 8.01311475409836e-05, | |
| "loss": 0.0343, | |
| "step": 18380 | |
| }, | |
| { | |
| "epoch": 89.81086028065894, | |
| "grad_norm": 1.2887758016586304, | |
| "learning_rate": 8e-05, | |
| "loss": 0.037, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 89.908480780964, | |
| "grad_norm": 1.6205062866210938, | |
| "learning_rate": 7.98688524590164e-05, | |
| "loss": 0.0395, | |
| "step": 18420 | |
| }, | |
| { | |
| "epoch": 90.00610128126907, | |
| "grad_norm": 1.4648716449737549, | |
| "learning_rate": 7.97377049180328e-05, | |
| "loss": 0.0387, | |
| "step": 18440 | |
| }, | |
| { | |
| "epoch": 90.10372178157414, | |
| "grad_norm": 0.8236134648323059, | |
| "learning_rate": 7.960655737704919e-05, | |
| "loss": 0.0317, | |
| "step": 18460 | |
| }, | |
| { | |
| "epoch": 90.20134228187919, | |
| "grad_norm": 1.4608750343322754, | |
| "learning_rate": 7.947540983606558e-05, | |
| "loss": 0.0347, | |
| "step": 18480 | |
| }, | |
| { | |
| "epoch": 90.29896278218426, | |
| "grad_norm": 0.9114894866943359, | |
| "learning_rate": 7.934426229508197e-05, | |
| "loss": 0.0347, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 90.39658328248932, | |
| "grad_norm": 0.7654961347579956, | |
| "learning_rate": 7.921311475409836e-05, | |
| "loss": 0.0324, | |
| "step": 18520 | |
| }, | |
| { | |
| "epoch": 90.49420378279439, | |
| "grad_norm": 1.1844931840896606, | |
| "learning_rate": 7.908196721311476e-05, | |
| "loss": 0.0352, | |
| "step": 18540 | |
| }, | |
| { | |
| "epoch": 90.59182428309946, | |
| "grad_norm": 1.6406841278076172, | |
| "learning_rate": 7.895081967213115e-05, | |
| "loss": 0.0352, | |
| "step": 18560 | |
| }, | |
| { | |
| "epoch": 90.68944478340451, | |
| "grad_norm": 1.5059739351272583, | |
| "learning_rate": 7.881967213114755e-05, | |
| "loss": 0.0386, | |
| "step": 18580 | |
| }, | |
| { | |
| "epoch": 90.78706528370958, | |
| "grad_norm": 0.7421912550926208, | |
| "learning_rate": 7.868852459016394e-05, | |
| "loss": 0.0381, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 90.88468578401465, | |
| "grad_norm": 1.2397936582565308, | |
| "learning_rate": 7.855737704918033e-05, | |
| "loss": 0.0365, | |
| "step": 18620 | |
| }, | |
| { | |
| "epoch": 90.9823062843197, | |
| "grad_norm": 1.0268443822860718, | |
| "learning_rate": 7.842622950819672e-05, | |
| "loss": 0.0402, | |
| "step": 18640 | |
| }, | |
| { | |
| "epoch": 91.07992678462477, | |
| "grad_norm": 0.87629234790802, | |
| "learning_rate": 7.829508196721311e-05, | |
| "loss": 0.0308, | |
| "step": 18660 | |
| }, | |
| { | |
| "epoch": 91.17754728492983, | |
| "grad_norm": 2.2132532596588135, | |
| "learning_rate": 7.816393442622951e-05, | |
| "loss": 0.031, | |
| "step": 18680 | |
| }, | |
| { | |
| "epoch": 91.2751677852349, | |
| "grad_norm": 1.2565170526504517, | |
| "learning_rate": 7.80327868852459e-05, | |
| "loss": 0.034, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 91.37278828553997, | |
| "grad_norm": 1.191254734992981, | |
| "learning_rate": 7.79016393442623e-05, | |
| "loss": 0.0348, | |
| "step": 18720 | |
| }, | |
| { | |
| "epoch": 91.47040878584502, | |
| "grad_norm": 0.711301326751709, | |
| "learning_rate": 7.77704918032787e-05, | |
| "loss": 0.0342, | |
| "step": 18740 | |
| }, | |
| { | |
| "epoch": 91.5680292861501, | |
| "grad_norm": 0.8898797035217285, | |
| "learning_rate": 7.763934426229508e-05, | |
| "loss": 0.0348, | |
| "step": 18760 | |
| }, | |
| { | |
| "epoch": 91.66564978645516, | |
| "grad_norm": 0.8842754364013672, | |
| "learning_rate": 7.750819672131147e-05, | |
| "loss": 0.0343, | |
| "step": 18780 | |
| }, | |
| { | |
| "epoch": 91.76327028676022, | |
| "grad_norm": 1.7899420261383057, | |
| "learning_rate": 7.737704918032788e-05, | |
| "loss": 0.0395, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 91.86089078706529, | |
| "grad_norm": 1.710175633430481, | |
| "learning_rate": 7.724590163934426e-05, | |
| "loss": 0.0387, | |
| "step": 18820 | |
| }, | |
| { | |
| "epoch": 91.95851128737034, | |
| "grad_norm": 1.8949339389801025, | |
| "learning_rate": 7.711475409836067e-05, | |
| "loss": 0.0409, | |
| "step": 18840 | |
| }, | |
| { | |
| "epoch": 92.05613178767541, | |
| "grad_norm": 1.6512871980667114, | |
| "learning_rate": 7.698360655737706e-05, | |
| "loss": 0.0353, | |
| "step": 18860 | |
| }, | |
| { | |
| "epoch": 92.15375228798048, | |
| "grad_norm": 1.1428453922271729, | |
| "learning_rate": 7.685245901639345e-05, | |
| "loss": 0.031, | |
| "step": 18880 | |
| }, | |
| { | |
| "epoch": 92.25137278828554, | |
| "grad_norm": 1.9489351511001587, | |
| "learning_rate": 7.672131147540984e-05, | |
| "loss": 0.0328, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 92.3489932885906, | |
| "grad_norm": 1.148255467414856, | |
| "learning_rate": 7.659016393442622e-05, | |
| "loss": 0.0334, | |
| "step": 18920 | |
| }, | |
| { | |
| "epoch": 92.44661378889568, | |
| "grad_norm": 1.2118985652923584, | |
| "learning_rate": 7.645901639344263e-05, | |
| "loss": 0.0319, | |
| "step": 18940 | |
| }, | |
| { | |
| "epoch": 92.54423428920073, | |
| "grad_norm": 0.9434294700622559, | |
| "learning_rate": 7.632786885245902e-05, | |
| "loss": 0.0377, | |
| "step": 18960 | |
| }, | |
| { | |
| "epoch": 92.6418547895058, | |
| "grad_norm": 1.136614441871643, | |
| "learning_rate": 7.619672131147542e-05, | |
| "loss": 0.0358, | |
| "step": 18980 | |
| }, | |
| { | |
| "epoch": 92.73947528981085, | |
| "grad_norm": 1.6465675830841064, | |
| "learning_rate": 7.606557377049181e-05, | |
| "loss": 0.0335, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 92.83709579011592, | |
| "grad_norm": 1.623631477355957, | |
| "learning_rate": 7.59344262295082e-05, | |
| "loss": 0.0377, | |
| "step": 19020 | |
| }, | |
| { | |
| "epoch": 92.934716290421, | |
| "grad_norm": 1.2394886016845703, | |
| "learning_rate": 7.580327868852459e-05, | |
| "loss": 0.0378, | |
| "step": 19040 | |
| }, | |
| { | |
| "epoch": 93.03233679072605, | |
| "grad_norm": 1.0836046934127808, | |
| "learning_rate": 7.567213114754099e-05, | |
| "loss": 0.0374, | |
| "step": 19060 | |
| }, | |
| { | |
| "epoch": 93.12995729103112, | |
| "grad_norm": 1.2980749607086182, | |
| "learning_rate": 7.554098360655738e-05, | |
| "loss": 0.0299, | |
| "step": 19080 | |
| }, | |
| { | |
| "epoch": 93.22757779133617, | |
| "grad_norm": 0.7896084785461426, | |
| "learning_rate": 7.540983606557377e-05, | |
| "loss": 0.0338, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 93.32519829164124, | |
| "grad_norm": 1.4562427997589111, | |
| "learning_rate": 7.527868852459017e-05, | |
| "loss": 0.0325, | |
| "step": 19120 | |
| }, | |
| { | |
| "epoch": 93.42281879194631, | |
| "grad_norm": 1.0484211444854736, | |
| "learning_rate": 7.514754098360656e-05, | |
| "loss": 0.0319, | |
| "step": 19140 | |
| }, | |
| { | |
| "epoch": 93.52043929225137, | |
| "grad_norm": 1.2180018424987793, | |
| "learning_rate": 7.501639344262295e-05, | |
| "loss": 0.0358, | |
| "step": 19160 | |
| }, | |
| { | |
| "epoch": 93.61805979255644, | |
| "grad_norm": 1.5725558996200562, | |
| "learning_rate": 7.488524590163934e-05, | |
| "loss": 0.0371, | |
| "step": 19180 | |
| }, | |
| { | |
| "epoch": 93.7156802928615, | |
| "grad_norm": 0.962448000907898, | |
| "learning_rate": 7.475409836065574e-05, | |
| "loss": 0.0329, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 93.81330079316656, | |
| "grad_norm": 1.2971309423446655, | |
| "learning_rate": 7.462295081967213e-05, | |
| "loss": 0.0388, | |
| "step": 19220 | |
| }, | |
| { | |
| "epoch": 93.91092129347163, | |
| "grad_norm": 1.621389627456665, | |
| "learning_rate": 7.449180327868854e-05, | |
| "loss": 0.035, | |
| "step": 19240 | |
| }, | |
| { | |
| "epoch": 94.00854179377669, | |
| "grad_norm": 1.1524312496185303, | |
| "learning_rate": 7.436065573770493e-05, | |
| "loss": 0.036, | |
| "step": 19260 | |
| }, | |
| { | |
| "epoch": 94.10616229408176, | |
| "grad_norm": 1.0361751317977905, | |
| "learning_rate": 7.422950819672131e-05, | |
| "loss": 0.0311, | |
| "step": 19280 | |
| }, | |
| { | |
| "epoch": 94.20378279438683, | |
| "grad_norm": 1.2280365228652954, | |
| "learning_rate": 7.40983606557377e-05, | |
| "loss": 0.032, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 94.30140329469188, | |
| "grad_norm": 1.39139723777771, | |
| "learning_rate": 7.39672131147541e-05, | |
| "loss": 0.0312, | |
| "step": 19320 | |
| }, | |
| { | |
| "epoch": 94.39902379499695, | |
| "grad_norm": 2.101762533187866, | |
| "learning_rate": 7.38360655737705e-05, | |
| "loss": 0.0304, | |
| "step": 19340 | |
| }, | |
| { | |
| "epoch": 94.49664429530202, | |
| "grad_norm": 1.5655161142349243, | |
| "learning_rate": 7.370491803278689e-05, | |
| "loss": 0.0332, | |
| "step": 19360 | |
| }, | |
| { | |
| "epoch": 94.59426479560707, | |
| "grad_norm": 1.1902984380722046, | |
| "learning_rate": 7.357377049180329e-05, | |
| "loss": 0.0308, | |
| "step": 19380 | |
| }, | |
| { | |
| "epoch": 94.69188529591214, | |
| "grad_norm": 1.3503782749176025, | |
| "learning_rate": 7.344262295081968e-05, | |
| "loss": 0.0342, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 94.7895057962172, | |
| "grad_norm": 1.6430237293243408, | |
| "learning_rate": 7.331147540983607e-05, | |
| "loss": 0.0375, | |
| "step": 19420 | |
| }, | |
| { | |
| "epoch": 94.88712629652227, | |
| "grad_norm": 1.4753270149230957, | |
| "learning_rate": 7.318032786885246e-05, | |
| "loss": 0.0361, | |
| "step": 19440 | |
| }, | |
| { | |
| "epoch": 94.98474679682734, | |
| "grad_norm": 1.6849883794784546, | |
| "learning_rate": 7.304918032786886e-05, | |
| "loss": 0.0377, | |
| "step": 19460 | |
| }, | |
| { | |
| "epoch": 95.0823672971324, | |
| "grad_norm": 0.9980133175849915, | |
| "learning_rate": 7.291803278688525e-05, | |
| "loss": 0.0286, | |
| "step": 19480 | |
| }, | |
| { | |
| "epoch": 95.17998779743746, | |
| "grad_norm": 0.911799430847168, | |
| "learning_rate": 7.278688524590165e-05, | |
| "loss": 0.0304, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 95.27760829774253, | |
| "grad_norm": 1.0748980045318604, | |
| "learning_rate": 7.265573770491804e-05, | |
| "loss": 0.0312, | |
| "step": 19520 | |
| }, | |
| { | |
| "epoch": 95.37522879804759, | |
| "grad_norm": 1.6856895685195923, | |
| "learning_rate": 7.252459016393443e-05, | |
| "loss": 0.0313, | |
| "step": 19540 | |
| }, | |
| { | |
| "epoch": 95.47284929835266, | |
| "grad_norm": 1.281265139579773, | |
| "learning_rate": 7.239344262295082e-05, | |
| "loss": 0.0325, | |
| "step": 19560 | |
| }, | |
| { | |
| "epoch": 95.57046979865771, | |
| "grad_norm": 1.3441650867462158, | |
| "learning_rate": 7.226229508196721e-05, | |
| "loss": 0.0369, | |
| "step": 19580 | |
| }, | |
| { | |
| "epoch": 95.66809029896278, | |
| "grad_norm": 0.9338245391845703, | |
| "learning_rate": 7.213114754098361e-05, | |
| "loss": 0.0328, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 95.76571079926785, | |
| "grad_norm": 1.0416558980941772, | |
| "learning_rate": 7.2e-05, | |
| "loss": 0.0313, | |
| "step": 19620 | |
| }, | |
| { | |
| "epoch": 95.8633312995729, | |
| "grad_norm": 1.187015175819397, | |
| "learning_rate": 7.18688524590164e-05, | |
| "loss": 0.0389, | |
| "step": 19640 | |
| }, | |
| { | |
| "epoch": 95.96095179987798, | |
| "grad_norm": 1.7248045206069946, | |
| "learning_rate": 7.17377049180328e-05, | |
| "loss": 0.0389, | |
| "step": 19660 | |
| }, | |
| { | |
| "epoch": 96.05857230018304, | |
| "grad_norm": 1.244112253189087, | |
| "learning_rate": 7.160655737704918e-05, | |
| "loss": 0.0282, | |
| "step": 19680 | |
| }, | |
| { | |
| "epoch": 96.1561928004881, | |
| "grad_norm": 0.868563711643219, | |
| "learning_rate": 7.147540983606557e-05, | |
| "loss": 0.03, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 96.25381330079317, | |
| "grad_norm": 0.6851422190666199, | |
| "learning_rate": 7.134426229508198e-05, | |
| "loss": 0.0293, | |
| "step": 19720 | |
| }, | |
| { | |
| "epoch": 96.35143380109822, | |
| "grad_norm": 1.0482094287872314, | |
| "learning_rate": 7.121311475409837e-05, | |
| "loss": 0.0348, | |
| "step": 19740 | |
| }, | |
| { | |
| "epoch": 96.4490543014033, | |
| "grad_norm": 0.9821768403053284, | |
| "learning_rate": 7.108196721311475e-05, | |
| "loss": 0.0346, | |
| "step": 19760 | |
| }, | |
| { | |
| "epoch": 96.54667480170836, | |
| "grad_norm": 1.2944159507751465, | |
| "learning_rate": 7.095081967213116e-05, | |
| "loss": 0.033, | |
| "step": 19780 | |
| }, | |
| { | |
| "epoch": 96.64429530201342, | |
| "grad_norm": 1.797796368598938, | |
| "learning_rate": 7.081967213114755e-05, | |
| "loss": 0.0313, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 96.74191580231849, | |
| "grad_norm": 2.1053693294525146, | |
| "learning_rate": 7.068852459016394e-05, | |
| "loss": 0.0321, | |
| "step": 19820 | |
| }, | |
| { | |
| "epoch": 96.83953630262356, | |
| "grad_norm": 0.6129661202430725, | |
| "learning_rate": 7.055737704918033e-05, | |
| "loss": 0.0343, | |
| "step": 19840 | |
| }, | |
| { | |
| "epoch": 96.93715680292861, | |
| "grad_norm": 1.184906005859375, | |
| "learning_rate": 7.042622950819673e-05, | |
| "loss": 0.0335, | |
| "step": 19860 | |
| }, | |
| { | |
| "epoch": 97.03477730323368, | |
| "grad_norm": 1.0629398822784424, | |
| "learning_rate": 7.029508196721312e-05, | |
| "loss": 0.0336, | |
| "step": 19880 | |
| }, | |
| { | |
| "epoch": 97.13239780353874, | |
| "grad_norm": 1.0311895608901978, | |
| "learning_rate": 7.016393442622952e-05, | |
| "loss": 0.028, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 97.2300183038438, | |
| "grad_norm": 1.3785455226898193, | |
| "learning_rate": 7.003278688524591e-05, | |
| "loss": 0.0304, | |
| "step": 19920 | |
| }, | |
| { | |
| "epoch": 97.32763880414888, | |
| "grad_norm": 1.515749216079712, | |
| "learning_rate": 6.99016393442623e-05, | |
| "loss": 0.0314, | |
| "step": 19940 | |
| }, | |
| { | |
| "epoch": 97.42525930445393, | |
| "grad_norm": 1.5662028789520264, | |
| "learning_rate": 6.977049180327869e-05, | |
| "loss": 0.0302, | |
| "step": 19960 | |
| }, | |
| { | |
| "epoch": 97.522879804759, | |
| "grad_norm": 1.1467646360397339, | |
| "learning_rate": 6.963934426229508e-05, | |
| "loss": 0.0317, | |
| "step": 19980 | |
| }, | |
| { | |
| "epoch": 97.62050030506407, | |
| "grad_norm": 0.9327864646911621, | |
| "learning_rate": 6.950819672131148e-05, | |
| "loss": 0.0323, | |
| "step": 20000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 30600, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 150, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.1916893437732045e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |