| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 708, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.002824858757062147, | |
| "grad_norm": 0.6147084683934281, | |
| "learning_rate": 9.999950776495983e-06, | |
| "loss": 0.1718, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005649717514124294, | |
| "grad_norm": 0.6665089926445491, | |
| "learning_rate": 9.99980310695311e-06, | |
| "loss": 0.1906, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00847457627118644, | |
| "grad_norm": 0.5879608524504671, | |
| "learning_rate": 9.99955699427891e-06, | |
| "loss": 0.1905, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011299435028248588, | |
| "grad_norm": 0.5728624077474995, | |
| "learning_rate": 9.999212443319191e-06, | |
| "loss": 0.1806, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.014124293785310734, | |
| "grad_norm": 0.45407189292974287, | |
| "learning_rate": 9.998769460857955e-06, | |
| "loss": 0.1587, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01694915254237288, | |
| "grad_norm": 0.695598470337452, | |
| "learning_rate": 9.998228055617264e-06, | |
| "loss": 0.3073, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01977401129943503, | |
| "grad_norm": 0.4388502111835372, | |
| "learning_rate": 9.99758823825706e-06, | |
| "loss": 0.1658, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.022598870056497175, | |
| "grad_norm": 0.506295761059801, | |
| "learning_rate": 9.996850021374969e-06, | |
| "loss": 0.1822, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.025423728813559324, | |
| "grad_norm": 0.5054469922635372, | |
| "learning_rate": 9.996013419506035e-06, | |
| "loss": 0.1878, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.02824858757062147, | |
| "grad_norm": 0.5236855115593358, | |
| "learning_rate": 9.99507844912245e-06, | |
| "loss": 0.164, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.031073446327683617, | |
| "grad_norm": 0.6005034865395634, | |
| "learning_rate": 9.994045128633221e-06, | |
| "loss": 0.2148, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03389830508474576, | |
| "grad_norm": 0.39196864541665416, | |
| "learning_rate": 9.99291347838381e-06, | |
| "loss": 0.1439, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03672316384180791, | |
| "grad_norm": 0.3520974148515624, | |
| "learning_rate": 9.991683520655735e-06, | |
| "loss": 0.138, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03954802259887006, | |
| "grad_norm": 0.39891278345339304, | |
| "learning_rate": 9.990355279666124e-06, | |
| "loss": 0.1778, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0423728813559322, | |
| "grad_norm": 0.38823463524310226, | |
| "learning_rate": 9.988928781567251e-06, | |
| "loss": 0.1802, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04519774011299435, | |
| "grad_norm": 0.3456396477919156, | |
| "learning_rate": 9.987404054446009e-06, | |
| "loss": 0.1537, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0480225988700565, | |
| "grad_norm": 0.4433642134268957, | |
| "learning_rate": 9.98578112832336e-06, | |
| "loss": 0.2148, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05084745762711865, | |
| "grad_norm": 0.5978046488506871, | |
| "learning_rate": 9.984060035153752e-06, | |
| "loss": 0.1769, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05367231638418079, | |
| "grad_norm": 0.3537901451183181, | |
| "learning_rate": 9.982240808824477e-06, | |
| "loss": 0.1323, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05649717514124294, | |
| "grad_norm": 0.5417108653864326, | |
| "learning_rate": 9.980323485155013e-06, | |
| "loss": 0.1666, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.059322033898305086, | |
| "grad_norm": 0.35289882003600165, | |
| "learning_rate": 9.978308101896318e-06, | |
| "loss": 0.1205, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.062146892655367235, | |
| "grad_norm": 0.36004933523243315, | |
| "learning_rate": 9.97619469873008e-06, | |
| "loss": 0.1371, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06497175141242938, | |
| "grad_norm": 0.3854571301167785, | |
| "learning_rate": 9.973983317267944e-06, | |
| "loss": 0.1568, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06779661016949153, | |
| "grad_norm": 0.6487420228246645, | |
| "learning_rate": 9.971674001050687e-06, | |
| "loss": 0.1647, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07062146892655367, | |
| "grad_norm": 0.3854893336014017, | |
| "learning_rate": 9.969266795547364e-06, | |
| "loss": 0.1854, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07344632768361582, | |
| "grad_norm": 0.3832699272168604, | |
| "learning_rate": 9.96676174815441e-06, | |
| "loss": 0.1955, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07627118644067797, | |
| "grad_norm": 0.31760668668906533, | |
| "learning_rate": 9.964158908194708e-06, | |
| "loss": 0.1195, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07909604519774012, | |
| "grad_norm": 0.41114801745145746, | |
| "learning_rate": 9.961458326916624e-06, | |
| "loss": 0.1414, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08192090395480225, | |
| "grad_norm": 0.3240630544735708, | |
| "learning_rate": 9.958660057492982e-06, | |
| "loss": 0.1562, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0847457627118644, | |
| "grad_norm": 0.3317790911132787, | |
| "learning_rate": 9.955764155020037e-06, | |
| "loss": 0.1263, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08757062146892655, | |
| "grad_norm": 0.3730810477019189, | |
| "learning_rate": 9.952770676516372e-06, | |
| "loss": 0.1549, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0903954802259887, | |
| "grad_norm": 0.3594977608871381, | |
| "learning_rate": 9.94967968092179e-06, | |
| "loss": 0.1362, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09322033898305085, | |
| "grad_norm": 0.5395386567672124, | |
| "learning_rate": 9.946491229096143e-06, | |
| "loss": 0.1447, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.096045197740113, | |
| "grad_norm": 0.34069006187102624, | |
| "learning_rate": 9.943205383818142e-06, | |
| "loss": 0.1602, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09887005649717515, | |
| "grad_norm": 0.40507986945355756, | |
| "learning_rate": 9.93982220978411e-06, | |
| "loss": 0.1727, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1016949152542373, | |
| "grad_norm": 0.3101237427572441, | |
| "learning_rate": 9.936341773606723e-06, | |
| "loss": 0.1328, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.10451977401129943, | |
| "grad_norm": 0.32711902171143986, | |
| "learning_rate": 9.932764143813686e-06, | |
| "loss": 0.1377, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.10734463276836158, | |
| "grad_norm": 0.3698266114267764, | |
| "learning_rate": 9.929089390846389e-06, | |
| "loss": 0.1593, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11016949152542373, | |
| "grad_norm": 0.36372051261110894, | |
| "learning_rate": 9.925317587058516e-06, | |
| "loss": 0.1238, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11299435028248588, | |
| "grad_norm": 0.36933238520645995, | |
| "learning_rate": 9.92144880671463e-06, | |
| "loss": 0.1678, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11581920903954802, | |
| "grad_norm": 0.3460461190525915, | |
| "learning_rate": 9.9174831259887e-06, | |
| "loss": 0.1578, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11864406779661017, | |
| "grad_norm": 0.5499266672914876, | |
| "learning_rate": 9.913420622962606e-06, | |
| "loss": 0.1437, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12146892655367232, | |
| "grad_norm": 0.4049849988270768, | |
| "learning_rate": 9.909261377624601e-06, | |
| "loss": 0.187, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12429378531073447, | |
| "grad_norm": 0.47235530071518683, | |
| "learning_rate": 9.90500547186774e-06, | |
| "loss": 0.1449, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1271186440677966, | |
| "grad_norm": 0.36137299735697387, | |
| "learning_rate": 9.900652989488255e-06, | |
| "loss": 0.1505, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.12994350282485875, | |
| "grad_norm": 0.372928772406179, | |
| "learning_rate": 9.896204016183924e-06, | |
| "loss": 0.1984, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1327683615819209, | |
| "grad_norm": 0.31608553050802646, | |
| "learning_rate": 9.891658639552368e-06, | |
| "loss": 0.1233, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13559322033898305, | |
| "grad_norm": 0.34009496825621116, | |
| "learning_rate": 9.887016949089334e-06, | |
| "loss": 0.1533, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1384180790960452, | |
| "grad_norm": 0.3475244874058269, | |
| "learning_rate": 9.882279036186927e-06, | |
| "loss": 0.1348, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14124293785310735, | |
| "grad_norm": 0.3465923771954804, | |
| "learning_rate": 9.87744499413182e-06, | |
| "loss": 0.1211, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1440677966101695, | |
| "grad_norm": 0.4625874588181501, | |
| "learning_rate": 9.872514918103407e-06, | |
| "loss": 0.1316, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14689265536723164, | |
| "grad_norm": 0.4219123981937324, | |
| "learning_rate": 9.867488905171934e-06, | |
| "loss": 0.1499, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1497175141242938, | |
| "grad_norm": 0.28327838865809335, | |
| "learning_rate": 9.86236705429659e-06, | |
| "loss": 0.118, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.15254237288135594, | |
| "grad_norm": 0.42996698854118004, | |
| "learning_rate": 9.85714946632355e-06, | |
| "loss": 0.1445, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.1553672316384181, | |
| "grad_norm": 0.3158297884352036, | |
| "learning_rate": 9.851836243984005e-06, | |
| "loss": 0.1026, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15819209039548024, | |
| "grad_norm": 0.34425358123925115, | |
| "learning_rate": 9.846427491892117e-06, | |
| "loss": 0.1296, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16101694915254236, | |
| "grad_norm": 0.26454441870053136, | |
| "learning_rate": 9.840923316542984e-06, | |
| "loss": 0.0945, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1638418079096045, | |
| "grad_norm": 0.47330273293079517, | |
| "learning_rate": 9.835323826310522e-06, | |
| "loss": 0.1645, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.30609709369908644, | |
| "learning_rate": 9.829629131445342e-06, | |
| "loss": 0.122, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1694915254237288, | |
| "grad_norm": 0.4330091411356326, | |
| "learning_rate": 9.823839344072582e-06, | |
| "loss": 0.1494, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17231638418079095, | |
| "grad_norm": 0.42693180952210663, | |
| "learning_rate": 9.817954578189686e-06, | |
| "loss": 0.1235, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1751412429378531, | |
| "grad_norm": 0.3324864445850298, | |
| "learning_rate": 9.811974949664176e-06, | |
| "loss": 0.1327, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.17796610169491525, | |
| "grad_norm": 0.41054779677427783, | |
| "learning_rate": 9.805900576231358e-06, | |
| "loss": 0.1741, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.1807909604519774, | |
| "grad_norm": 0.33830483714924636, | |
| "learning_rate": 9.79973157749201e-06, | |
| "loss": 0.1265, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.18361581920903955, | |
| "grad_norm": 0.3152187519500513, | |
| "learning_rate": 9.793468074910028e-06, | |
| "loss": 0.1202, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.1864406779661017, | |
| "grad_norm": 0.368592840409881, | |
| "learning_rate": 9.787110191810027e-06, | |
| "loss": 0.1292, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18926553672316385, | |
| "grad_norm": 0.3471829002234472, | |
| "learning_rate": 9.780658053374923e-06, | |
| "loss": 0.1532, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.192090395480226, | |
| "grad_norm": 0.3461593726912688, | |
| "learning_rate": 9.77411178664346e-06, | |
| "loss": 0.1352, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.19491525423728814, | |
| "grad_norm": 0.647720230993773, | |
| "learning_rate": 9.767471520507713e-06, | |
| "loss": 0.1291, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.1977401129943503, | |
| "grad_norm": 0.31371308918917606, | |
| "learning_rate": 9.760737385710546e-06, | |
| "loss": 0.1363, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.20056497175141244, | |
| "grad_norm": 0.36616052339629535, | |
| "learning_rate": 9.753909514843047e-06, | |
| "loss": 0.163, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2033898305084746, | |
| "grad_norm": 0.3366775524198522, | |
| "learning_rate": 9.746988042341907e-06, | |
| "loss": 0.1211, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2062146892655367, | |
| "grad_norm": 0.32238872392227863, | |
| "learning_rate": 9.739973104486777e-06, | |
| "loss": 0.1244, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.20903954802259886, | |
| "grad_norm": 0.309699834626684, | |
| "learning_rate": 9.732864839397585e-06, | |
| "loss": 0.1123, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.211864406779661, | |
| "grad_norm": 0.40415497402853257, | |
| "learning_rate": 9.725663387031818e-06, | |
| "loss": 0.1185, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.21468926553672316, | |
| "grad_norm": 0.43571692379160615, | |
| "learning_rate": 9.718368889181763e-06, | |
| "loss": 0.1205, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2175141242937853, | |
| "grad_norm": 0.3220023585216405, | |
| "learning_rate": 9.710981489471721e-06, | |
| "loss": 0.1513, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.22033898305084745, | |
| "grad_norm": 0.41630953777961766, | |
| "learning_rate": 9.703501333355167e-06, | |
| "loss": 0.1249, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2231638418079096, | |
| "grad_norm": 0.5019640716602216, | |
| "learning_rate": 9.6959285681119e-06, | |
| "loss": 0.1914, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.22598870056497175, | |
| "grad_norm": 0.4023458378359013, | |
| "learning_rate": 9.68826334284514e-06, | |
| "loss": 0.1346, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2288135593220339, | |
| "grad_norm": 0.34337841648791245, | |
| "learning_rate": 9.680505808478583e-06, | |
| "loss": 0.1272, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.23163841807909605, | |
| "grad_norm": 0.32382210914063797, | |
| "learning_rate": 9.672656117753435e-06, | |
| "loss": 0.1155, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2344632768361582, | |
| "grad_norm": 0.4098493186152155, | |
| "learning_rate": 9.664714425225414e-06, | |
| "loss": 0.155, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.23728813559322035, | |
| "grad_norm": 0.30680580576204597, | |
| "learning_rate": 9.656680887261693e-06, | |
| "loss": 0.1268, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2401129943502825, | |
| "grad_norm": 0.34530888636266843, | |
| "learning_rate": 9.648555662037826e-06, | |
| "loss": 0.1255, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.24293785310734464, | |
| "grad_norm": 0.31982112281364816, | |
| "learning_rate": 9.640338909534636e-06, | |
| "loss": 0.1187, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.2457627118644068, | |
| "grad_norm": 0.3249369835609272, | |
| "learning_rate": 9.632030791535063e-06, | |
| "loss": 0.1094, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.24858757062146894, | |
| "grad_norm": 0.4202492399280893, | |
| "learning_rate": 9.62363147162098e-06, | |
| "loss": 0.1473, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2514124293785311, | |
| "grad_norm": 0.4206786168286477, | |
| "learning_rate": 9.615141115169968e-06, | |
| "loss": 0.1321, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2542372881355932, | |
| "grad_norm": 0.35455170569024247, | |
| "learning_rate": 9.606559889352065e-06, | |
| "loss": 0.1333, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2570621468926554, | |
| "grad_norm": 0.30654274314933255, | |
| "learning_rate": 9.597887963126476e-06, | |
| "loss": 0.1324, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2598870056497175, | |
| "grad_norm": 0.31426903973238157, | |
| "learning_rate": 9.589125507238234e-06, | |
| "loss": 0.0965, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2627118644067797, | |
| "grad_norm": 0.39888445390797206, | |
| "learning_rate": 9.580272694214855e-06, | |
| "loss": 0.1795, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2655367231638418, | |
| "grad_norm": 0.32688334776107986, | |
| "learning_rate": 9.571329698362931e-06, | |
| "loss": 0.0993, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.268361581920904, | |
| "grad_norm": 0.37334590333180434, | |
| "learning_rate": 9.562296695764695e-06, | |
| "loss": 0.1201, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2711864406779661, | |
| "grad_norm": 0.3189896682514011, | |
| "learning_rate": 9.553173864274567e-06, | |
| "loss": 0.1261, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2740112994350282, | |
| "grad_norm": 0.34974753780587814, | |
| "learning_rate": 9.543961383515638e-06, | |
| "loss": 0.1476, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.2768361581920904, | |
| "grad_norm": 0.3593944813062625, | |
| "learning_rate": 9.53465943487614e-06, | |
| "loss": 0.123, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2796610169491525, | |
| "grad_norm": 0.35812916144207474, | |
| "learning_rate": 9.52526820150588e-06, | |
| "loss": 0.1257, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2824858757062147, | |
| "grad_norm": 0.32235620707593265, | |
| "learning_rate": 9.51578786831262e-06, | |
| "loss": 0.1493, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2853107344632768, | |
| "grad_norm": 0.37142164119731536, | |
| "learning_rate": 9.506218621958448e-06, | |
| "loss": 0.1278, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.288135593220339, | |
| "grad_norm": 0.44435482682594635, | |
| "learning_rate": 9.496560650856097e-06, | |
| "loss": 0.1443, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.2909604519774011, | |
| "grad_norm": 0.3566707987782221, | |
| "learning_rate": 9.486814145165242e-06, | |
| "loss": 0.1122, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2937853107344633, | |
| "grad_norm": 0.33349931081864886, | |
| "learning_rate": 9.476979296788746e-06, | |
| "loss": 0.1185, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2966101694915254, | |
| "grad_norm": 0.3196942221194822, | |
| "learning_rate": 9.467056299368888e-06, | |
| "loss": 0.1405, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2994350282485876, | |
| "grad_norm": 0.300108694500991, | |
| "learning_rate": 9.457045348283552e-06, | |
| "loss": 0.1032, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3022598870056497, | |
| "grad_norm": 0.2628183218575623, | |
| "learning_rate": 9.446946640642372e-06, | |
| "loss": 0.0891, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3050847457627119, | |
| "grad_norm": 0.3173943921412448, | |
| "learning_rate": 9.436760375282858e-06, | |
| "loss": 0.1389, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.307909604519774, | |
| "grad_norm": 0.3374023308082592, | |
| "learning_rate": 9.426486752766481e-06, | |
| "loss": 0.1433, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3107344632768362, | |
| "grad_norm": 0.32296890768155634, | |
| "learning_rate": 9.416125975374722e-06, | |
| "loss": 0.1489, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3135593220338983, | |
| "grad_norm": 0.809125754824113, | |
| "learning_rate": 9.405678247105083e-06, | |
| "loss": 0.1197, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3163841807909605, | |
| "grad_norm": 0.2899410860114942, | |
| "learning_rate": 9.395143773667089e-06, | |
| "loss": 0.1101, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3192090395480226, | |
| "grad_norm": 0.35261776609510237, | |
| "learning_rate": 9.38452276247821e-06, | |
| "loss": 0.15, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3220338983050847, | |
| "grad_norm": 0.40136031754220597, | |
| "learning_rate": 9.373815422659806e-06, | |
| "loss": 0.1822, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3248587570621469, | |
| "grad_norm": 0.313017574757418, | |
| "learning_rate": 9.363021965032993e-06, | |
| "loss": 0.1188, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.327683615819209, | |
| "grad_norm": 0.373452387713479, | |
| "learning_rate": 9.352142602114487e-06, | |
| "loss": 0.137, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.3305084745762712, | |
| "grad_norm": 0.3420346614883704, | |
| "learning_rate": 9.341177548112437e-06, | |
| "loss": 0.1344, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.4036780338239374, | |
| "learning_rate": 9.330127018922195e-06, | |
| "loss": 0.133, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.3361581920903955, | |
| "grad_norm": 0.32756640437501205, | |
| "learning_rate": 9.318991232122065e-06, | |
| "loss": 0.1321, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.3389830508474576, | |
| "grad_norm": 0.345171287549607, | |
| "learning_rate": 9.307770406969032e-06, | |
| "loss": 0.1202, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3418079096045198, | |
| "grad_norm": 0.437595297190657, | |
| "learning_rate": 9.296464764394422e-06, | |
| "loss": 0.1824, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.3446327683615819, | |
| "grad_norm": 0.2838090630299304, | |
| "learning_rate": 9.285074526999577e-06, | |
| "loss": 0.114, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3474576271186441, | |
| "grad_norm": 0.33725871702683274, | |
| "learning_rate": 9.273599919051452e-06, | |
| "loss": 0.1254, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.3502824858757062, | |
| "grad_norm": 0.2980485886022378, | |
| "learning_rate": 9.262041166478215e-06, | |
| "loss": 0.1139, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3531073446327684, | |
| "grad_norm": 0.35961892985059185, | |
| "learning_rate": 9.250398496864782e-06, | |
| "loss": 0.114, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3559322033898305, | |
| "grad_norm": 0.44215669470603974, | |
| "learning_rate": 9.238672139448354e-06, | |
| "loss": 0.1133, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.3587570621468927, | |
| "grad_norm": 0.4377017180308674, | |
| "learning_rate": 9.226862325113894e-06, | |
| "loss": 0.1438, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.3615819209039548, | |
| "grad_norm": 0.3248130129102471, | |
| "learning_rate": 9.214969286389577e-06, | |
| "loss": 0.1079, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3644067796610169, | |
| "grad_norm": 0.3687271690510821, | |
| "learning_rate": 9.202993257442216e-06, | |
| "loss": 0.1576, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.3672316384180791, | |
| "grad_norm": 0.31854432765060964, | |
| "learning_rate": 9.190934474072658e-06, | |
| "loss": 0.1128, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3700564971751412, | |
| "grad_norm": 0.43632199490503926, | |
| "learning_rate": 9.178793173711133e-06, | |
| "loss": 0.16, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3728813559322034, | |
| "grad_norm": 0.3184995213048756, | |
| "learning_rate": 9.166569595412576e-06, | |
| "loss": 0.1087, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3757062146892655, | |
| "grad_norm": 0.2852410374038083, | |
| "learning_rate": 9.154263979851932e-06, | |
| "loss": 0.1017, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3785310734463277, | |
| "grad_norm": 0.4894599992720483, | |
| "learning_rate": 9.141876569319405e-06, | |
| "loss": 0.1338, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3813559322033898, | |
| "grad_norm": 0.3025442389978783, | |
| "learning_rate": 9.129407607715697e-06, | |
| "loss": 0.1029, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.384180790960452, | |
| "grad_norm": 0.3362331028333517, | |
| "learning_rate": 9.116857340547203e-06, | |
| "loss": 0.1171, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3870056497175141, | |
| "grad_norm": 0.3058727798799934, | |
| "learning_rate": 9.104226014921171e-06, | |
| "loss": 0.1242, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3898305084745763, | |
| "grad_norm": 0.32225074355114075, | |
| "learning_rate": 9.091513879540845e-06, | |
| "loss": 0.1199, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3926553672316384, | |
| "grad_norm": 0.4160727630117183, | |
| "learning_rate": 9.078721184700565e-06, | |
| "loss": 0.1703, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3954802259887006, | |
| "grad_norm": 0.3825137129544069, | |
| "learning_rate": 9.065848182280835e-06, | |
| "loss": 0.1417, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3983050847457627, | |
| "grad_norm": 0.3109793025220252, | |
| "learning_rate": 9.05289512574337e-06, | |
| "loss": 0.1079, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4011299435028249, | |
| "grad_norm": 0.32403338158384704, | |
| "learning_rate": 9.039862270126102e-06, | |
| "loss": 0.1304, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.403954802259887, | |
| "grad_norm": 0.333516403508159, | |
| "learning_rate": 9.026749872038161e-06, | |
| "loss": 0.1215, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4067796610169492, | |
| "grad_norm": 0.5512785845644815, | |
| "learning_rate": 9.013558189654819e-06, | |
| "loss": 0.119, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.4096045197740113, | |
| "grad_norm": 0.3342692670079559, | |
| "learning_rate": 9.000287482712407e-06, | |
| "loss": 0.1327, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4124293785310734, | |
| "grad_norm": 0.3844315934040279, | |
| "learning_rate": 8.986938012503203e-06, | |
| "loss": 0.1354, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.4152542372881356, | |
| "grad_norm": 0.3704715200009876, | |
| "learning_rate": 8.973510041870287e-06, | |
| "loss": 0.116, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4180790960451977, | |
| "grad_norm": 0.3731805906895666, | |
| "learning_rate": 8.960003835202369e-06, | |
| "loss": 0.1447, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.4209039548022599, | |
| "grad_norm": 0.3074985687330806, | |
| "learning_rate": 8.946419658428573e-06, | |
| "loss": 0.0944, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.423728813559322, | |
| "grad_norm": 0.43341288159333946, | |
| "learning_rate": 8.932757779013214e-06, | |
| "loss": 0.1266, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4265536723163842, | |
| "grad_norm": 0.336284725389875, | |
| "learning_rate": 8.919018465950517e-06, | |
| "loss": 0.1398, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4293785310734463, | |
| "grad_norm": 0.35240140744565995, | |
| "learning_rate": 8.90520198975934e-06, | |
| "loss": 0.161, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4322033898305085, | |
| "grad_norm": 0.33406617309219283, | |
| "learning_rate": 8.89130862247783e-06, | |
| "loss": 0.1452, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.4350282485875706, | |
| "grad_norm": 0.4037611933501478, | |
| "learning_rate": 8.877338637658074e-06, | |
| "loss": 0.1355, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.4378531073446328, | |
| "grad_norm": 0.3475789016676025, | |
| "learning_rate": 8.863292310360716e-06, | |
| "loss": 0.1298, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4406779661016949, | |
| "grad_norm": 0.4841263150458153, | |
| "learning_rate": 8.849169917149532e-06, | |
| "loss": 0.1207, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4435028248587571, | |
| "grad_norm": 0.30177409440633907, | |
| "learning_rate": 8.834971736085995e-06, | |
| "loss": 0.1092, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.4463276836158192, | |
| "grad_norm": 0.3405364099196524, | |
| "learning_rate": 8.820698046723796e-06, | |
| "loss": 0.1196, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.4491525423728814, | |
| "grad_norm": 0.3399165082210464, | |
| "learning_rate": 8.806349130103334e-06, | |
| "loss": 0.1215, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.4519774011299435, | |
| "grad_norm": 0.30309252729845493, | |
| "learning_rate": 8.791925268746193e-06, | |
| "loss": 0.133, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4548022598870056, | |
| "grad_norm": 0.4127573259391668, | |
| "learning_rate": 8.777426746649571e-06, | |
| "loss": 0.1131, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.4576271186440678, | |
| "grad_norm": 0.3210525339974507, | |
| "learning_rate": 8.762853849280692e-06, | |
| "loss": 0.1097, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.4604519774011299, | |
| "grad_norm": 0.38978996569997443, | |
| "learning_rate": 8.748206863571188e-06, | |
| "loss": 0.1259, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.4632768361581921, | |
| "grad_norm": 0.3200510621145131, | |
| "learning_rate": 8.73348607791144e-06, | |
| "loss": 0.1028, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4661016949152542, | |
| "grad_norm": 0.5781086641593365, | |
| "learning_rate": 8.718691782144908e-06, | |
| "loss": 0.1504, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4689265536723164, | |
| "grad_norm": 0.3161434045596249, | |
| "learning_rate": 8.703824267562424e-06, | |
| "loss": 0.1341, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.4717514124293785, | |
| "grad_norm": 0.2920424513956196, | |
| "learning_rate": 8.688883826896458e-06, | |
| "loss": 0.0985, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.4745762711864407, | |
| "grad_norm": 0.3862783048815386, | |
| "learning_rate": 8.673870754315336e-06, | |
| "loss": 0.1352, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.4774011299435028, | |
| "grad_norm": 0.4426996657474187, | |
| "learning_rate": 8.658785345417484e-06, | |
| "loss": 0.1414, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.480225988700565, | |
| "grad_norm": 0.461223196307408, | |
| "learning_rate": 8.64362789722557e-06, | |
| "loss": 0.1397, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4830508474576271, | |
| "grad_norm": 0.3236362093117533, | |
| "learning_rate": 8.62839870818068e-06, | |
| "loss": 0.1185, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.4858757062146893, | |
| "grad_norm": 0.5017393956700863, | |
| "learning_rate": 8.613098078136436e-06, | |
| "loss": 0.1301, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.4887005649717514, | |
| "grad_norm": 0.3476632540894821, | |
| "learning_rate": 8.597726308353085e-06, | |
| "loss": 0.1265, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.4915254237288136, | |
| "grad_norm": 0.38990017498501656, | |
| "learning_rate": 8.582283701491576e-06, | |
| "loss": 0.152, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.4943502824858757, | |
| "grad_norm": 0.3190826675735708, | |
| "learning_rate": 8.566770561607598e-06, | |
| "loss": 0.1281, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4971751412429379, | |
| "grad_norm": 0.3967141406957403, | |
| "learning_rate": 8.551187194145591e-06, | |
| "loss": 0.1546, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.3328836982483565, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 0.1314, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5028248587570622, | |
| "grad_norm": 0.374457685382645, | |
| "learning_rate": 8.519811005172916e-06, | |
| "loss": 0.1165, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5056497175141242, | |
| "grad_norm": 0.363385614117873, | |
| "learning_rate": 8.50401880144063e-06, | |
| "loss": 0.1226, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5084745762711864, | |
| "grad_norm": 0.32620237927339973, | |
| "learning_rate": 8.488157605674924e-06, | |
| "loss": 0.1203, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5112994350282486, | |
| "grad_norm": 0.2961670016082417, | |
| "learning_rate": 8.472227730173252e-06, | |
| "loss": 0.104, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5141242937853108, | |
| "grad_norm": 0.3350355060825277, | |
| "learning_rate": 8.456229488585328e-06, | |
| "loss": 0.1136, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5169491525423728, | |
| "grad_norm": 0.3540976396926151, | |
| "learning_rate": 8.440163195906959e-06, | |
| "loss": 0.0946, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.519774011299435, | |
| "grad_norm": 0.33467249667837795, | |
| "learning_rate": 8.424029168473829e-06, | |
| "loss": 0.1341, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5225988700564972, | |
| "grad_norm": 0.3582483372686221, | |
| "learning_rate": 8.407827723955287e-06, | |
| "loss": 0.101, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5254237288135594, | |
| "grad_norm": 0.3871399637156461, | |
| "learning_rate": 8.391559181348081e-06, | |
| "loss": 0.1368, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5282485875706214, | |
| "grad_norm": 0.44494561948427913, | |
| "learning_rate": 8.375223860970078e-06, | |
| "loss": 0.1542, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5310734463276836, | |
| "grad_norm": 0.342778980721638, | |
| "learning_rate": 8.358822084453964e-06, | |
| "loss": 0.1343, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5338983050847458, | |
| "grad_norm": 0.3190954383119015, | |
| "learning_rate": 8.342354174740904e-06, | |
| "loss": 0.1217, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.536723163841808, | |
| "grad_norm": 0.32918038764737034, | |
| "learning_rate": 8.325820456074181e-06, | |
| "loss": 0.1158, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.53954802259887, | |
| "grad_norm": 0.3879838615458632, | |
| "learning_rate": 8.309221253992825e-06, | |
| "loss": 0.1256, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5423728813559322, | |
| "grad_norm": 0.35618247459755487, | |
| "learning_rate": 8.292556895325195e-06, | |
| "loss": 0.1099, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5451977401129944, | |
| "grad_norm": 0.3234827304096444, | |
| "learning_rate": 8.275827708182536e-06, | |
| "loss": 0.1284, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.5480225988700564, | |
| "grad_norm": 0.34766152172703946, | |
| "learning_rate": 8.259034021952537e-06, | |
| "loss": 0.1356, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5508474576271186, | |
| "grad_norm": 0.36732776065701966, | |
| "learning_rate": 8.242176167292827e-06, | |
| "loss": 0.1151, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.5536723163841808, | |
| "grad_norm": 0.33889281731020826, | |
| "learning_rate": 8.225254476124479e-06, | |
| "loss": 0.1063, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.556497175141243, | |
| "grad_norm": 0.3193880715488108, | |
| "learning_rate": 8.208269281625466e-06, | |
| "loss": 0.1276, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.559322033898305, | |
| "grad_norm": 0.3204272104884364, | |
| "learning_rate": 8.191220918224102e-06, | |
| "loss": 0.1223, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5621468926553672, | |
| "grad_norm": 0.2837424255636162, | |
| "learning_rate": 8.174109721592463e-06, | |
| "loss": 0.1004, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.5649717514124294, | |
| "grad_norm": 0.37977501376826955, | |
| "learning_rate": 8.156936028639768e-06, | |
| "loss": 0.1343, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5677966101694916, | |
| "grad_norm": 0.3802933186191023, | |
| "learning_rate": 8.13970017750576e-06, | |
| "loss": 0.1229, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.5706214689265536, | |
| "grad_norm": 0.3336231226315229, | |
| "learning_rate": 8.12240250755403e-06, | |
| "loss": 0.1071, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.5734463276836158, | |
| "grad_norm": 0.31611617409721354, | |
| "learning_rate": 8.10504335936535e-06, | |
| "loss": 0.1315, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.576271186440678, | |
| "grad_norm": 0.37986304879280974, | |
| "learning_rate": 8.08762307473096e-06, | |
| "loss": 0.1488, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.5790960451977402, | |
| "grad_norm": 0.47382310798593613, | |
| "learning_rate": 8.07014199664584e-06, | |
| "loss": 0.1199, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5819209039548022, | |
| "grad_norm": 0.328423980785905, | |
| "learning_rate": 8.052600469301958e-06, | |
| "loss": 0.1094, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.5847457627118644, | |
| "grad_norm": 0.36866153232988097, | |
| "learning_rate": 8.03499883808149e-06, | |
| "loss": 0.1215, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.5875706214689266, | |
| "grad_norm": 0.4258404561855114, | |
| "learning_rate": 8.01733744955002e-06, | |
| "loss": 0.1728, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.5903954802259888, | |
| "grad_norm": 0.4379066754162319, | |
| "learning_rate": 7.999616651449722e-06, | |
| "loss": 0.1334, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.5932203389830508, | |
| "grad_norm": 0.3298246566080818, | |
| "learning_rate": 7.981836792692508e-06, | |
| "loss": 0.1321, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.596045197740113, | |
| "grad_norm": 0.3213552929669426, | |
| "learning_rate": 7.963998223353154e-06, | |
| "loss": 0.1475, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.5988700564971752, | |
| "grad_norm": 0.6800071694421395, | |
| "learning_rate": 7.946101294662418e-06, | |
| "loss": 0.1521, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.6016949152542372, | |
| "grad_norm": 0.31394438576238753, | |
| "learning_rate": 7.928146359000117e-06, | |
| "loss": 0.1269, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6045197740112994, | |
| "grad_norm": 0.34115929397266076, | |
| "learning_rate": 7.91013376988819e-06, | |
| "loss": 0.1079, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6073446327683616, | |
| "grad_norm": 0.3070888783690557, | |
| "learning_rate": 7.892063881983736e-06, | |
| "loss": 0.1037, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6101694915254238, | |
| "grad_norm": 0.35233334420808005, | |
| "learning_rate": 7.873937051072037e-06, | |
| "loss": 0.1291, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.6129943502824858, | |
| "grad_norm": 0.37926219175637094, | |
| "learning_rate": 7.855753634059543e-06, | |
| "loss": 0.1997, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.615819209039548, | |
| "grad_norm": 0.281725523224709, | |
| "learning_rate": 7.83751398896686e-06, | |
| "loss": 0.0946, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6186440677966102, | |
| "grad_norm": 0.33366851056170144, | |
| "learning_rate": 7.81921847492168e-06, | |
| "loss": 0.1355, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6214689265536724, | |
| "grad_norm": 0.4693528685955929, | |
| "learning_rate": 7.80086745215173e-06, | |
| "loss": 0.1235, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6242937853107344, | |
| "grad_norm": 0.348566291026797, | |
| "learning_rate": 7.782461281977668e-06, | |
| "loss": 0.1463, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6271186440677966, | |
| "grad_norm": 0.3432846241066091, | |
| "learning_rate": 7.764000326805967e-06, | |
| "loss": 0.1253, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6299435028248588, | |
| "grad_norm": 0.3907373096790576, | |
| "learning_rate": 7.74548495012179e-06, | |
| "loss": 0.1391, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.632768361581921, | |
| "grad_norm": 0.3728412944790509, | |
| "learning_rate": 7.726915516481824e-06, | |
| "loss": 0.122, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.635593220338983, | |
| "grad_norm": 0.5220133391248876, | |
| "learning_rate": 7.708292391507105e-06, | |
| "loss": 0.154, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6384180790960452, | |
| "grad_norm": 0.4146492597116334, | |
| "learning_rate": 7.68961594187582e-06, | |
| "loss": 0.1359, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6412429378531074, | |
| "grad_norm": 0.31397589485063837, | |
| "learning_rate": 7.670886535316086e-06, | |
| "loss": 0.1219, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6440677966101694, | |
| "grad_norm": 0.3176441428020906, | |
| "learning_rate": 7.652104540598712e-06, | |
| "loss": 0.1178, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6468926553672316, | |
| "grad_norm": 0.4044788848890745, | |
| "learning_rate": 7.633270327529936e-06, | |
| "loss": 0.0976, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6497175141242938, | |
| "grad_norm": 0.4182319443410884, | |
| "learning_rate": 7.614384266944139e-06, | |
| "loss": 0.1645, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.652542372881356, | |
| "grad_norm": 0.456563352937022, | |
| "learning_rate": 7.595446730696554e-06, | |
| "loss": 0.1382, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.655367231638418, | |
| "grad_norm": 0.32903524423468056, | |
| "learning_rate": 7.5764580916559405e-06, | |
| "loss": 0.1326, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.6581920903954802, | |
| "grad_norm": 0.28365921307195663, | |
| "learning_rate": 7.5574187236972344e-06, | |
| "loss": 0.0935, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.6610169491525424, | |
| "grad_norm": 0.33982089630336487, | |
| "learning_rate": 7.5383290016942e-06, | |
| "loss": 0.1454, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.6638418079096046, | |
| "grad_norm": 0.3772064426329603, | |
| "learning_rate": 7.519189301512042e-06, | |
| "loss": 0.1138, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.32618022313934375, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1334, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.6694915254237288, | |
| "grad_norm": 0.45333599138089803, | |
| "learning_rate": 7.480761474983943e-06, | |
| "loss": 0.1124, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.672316384180791, | |
| "grad_norm": 0.2925984327006227, | |
| "learning_rate": 7.461474105258911e-06, | |
| "loss": 0.1186, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.6751412429378532, | |
| "grad_norm": 0.29274180541284806, | |
| "learning_rate": 7.442138270581676e-06, | |
| "loss": 0.1152, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.6779661016949152, | |
| "grad_norm": 0.31809376762962216, | |
| "learning_rate": 7.422754351663252e-06, | |
| "loss": 0.1305, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6807909604519774, | |
| "grad_norm": 0.38676571872596516, | |
| "learning_rate": 7.403322730161402e-06, | |
| "loss": 0.1282, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.6836158192090396, | |
| "grad_norm": 0.4018879090622138, | |
| "learning_rate": 7.3838437886731264e-06, | |
| "loss": 0.1183, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.6864406779661016, | |
| "grad_norm": 0.3776316331740439, | |
| "learning_rate": 7.364317910727128e-06, | |
| "loss": 0.1222, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.6892655367231638, | |
| "grad_norm": 0.3824776982755315, | |
| "learning_rate": 7.3447454807762565e-06, | |
| "loss": 0.1428, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.692090395480226, | |
| "grad_norm": 0.3047645047687849, | |
| "learning_rate": 7.325126884189948e-06, | |
| "loss": 0.1385, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.6949152542372882, | |
| "grad_norm": 0.3314073663562561, | |
| "learning_rate": 7.30546250724663e-06, | |
| "loss": 0.1206, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.6977401129943502, | |
| "grad_norm": 0.4400078921981207, | |
| "learning_rate": 7.285752737126117e-06, | |
| "loss": 0.1327, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.7005649717514124, | |
| "grad_norm": 0.4934623443358996, | |
| "learning_rate": 7.265997961901987e-06, | |
| "loss": 0.1564, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.7033898305084746, | |
| "grad_norm": 0.30484309715193525, | |
| "learning_rate": 7.246198570533944e-06, | |
| "loss": 0.1242, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7062146892655368, | |
| "grad_norm": 0.29354477145041785, | |
| "learning_rate": 7.226354952860157e-06, | |
| "loss": 0.1149, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7090395480225988, | |
| "grad_norm": 0.3505364073788135, | |
| "learning_rate": 7.206467499589584e-06, | |
| "loss": 0.1087, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.711864406779661, | |
| "grad_norm": 0.29464691426886963, | |
| "learning_rate": 7.186536602294278e-06, | |
| "loss": 0.1142, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.7146892655367232, | |
| "grad_norm": 0.39810478648654263, | |
| "learning_rate": 7.166562653401681e-06, | |
| "loss": 0.1723, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7175141242937854, | |
| "grad_norm": 0.6996654555301565, | |
| "learning_rate": 7.146546046186893e-06, | |
| "loss": 0.1509, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.7203389830508474, | |
| "grad_norm": 0.43941978897146655, | |
| "learning_rate": 7.126487174764936e-06, | |
| "loss": 0.1214, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7231638418079096, | |
| "grad_norm": 0.8711552264444874, | |
| "learning_rate": 7.106386434082979e-06, | |
| "loss": 0.1814, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7259887005649718, | |
| "grad_norm": 0.44400959950712626, | |
| "learning_rate": 7.0862442199125836e-06, | |
| "loss": 0.1358, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7288135593220338, | |
| "grad_norm": 0.35970291529551507, | |
| "learning_rate": 7.066060928841891e-06, | |
| "loss": 0.168, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.731638418079096, | |
| "grad_norm": 0.3562466168617285, | |
| "learning_rate": 7.0458369582678276e-06, | |
| "loss": 0.1436, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.7344632768361582, | |
| "grad_norm": 0.3120380022750767, | |
| "learning_rate": 7.025572706388268e-06, | |
| "loss": 0.1146, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7372881355932204, | |
| "grad_norm": 0.3774265959308726, | |
| "learning_rate": 7.005268572194208e-06, | |
| "loss": 0.1034, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.7401129943502824, | |
| "grad_norm": 0.2979654585888925, | |
| "learning_rate": 6.984924955461901e-06, | |
| "loss": 0.1314, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.7429378531073446, | |
| "grad_norm": 0.581337456503481, | |
| "learning_rate": 6.964542256744986e-06, | |
| "loss": 0.1417, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7457627118644068, | |
| "grad_norm": 0.38453399260208176, | |
| "learning_rate": 6.944120877366605e-06, | |
| "loss": 0.1564, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.748587570621469, | |
| "grad_norm": 0.4514847039771076, | |
| "learning_rate": 6.923661219411494e-06, | |
| "loss": 0.1104, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.751412429378531, | |
| "grad_norm": 0.32941464728785497, | |
| "learning_rate": 6.9031636857180795e-06, | |
| "loss": 0.1232, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.7542372881355932, | |
| "grad_norm": 0.32342200359362855, | |
| "learning_rate": 6.8826286798705325e-06, | |
| "loss": 0.1298, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.7570621468926554, | |
| "grad_norm": 0.31600309169763335, | |
| "learning_rate": 6.86205660619083e-06, | |
| "loss": 0.1052, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.7598870056497176, | |
| "grad_norm": 0.3385392307218001, | |
| "learning_rate": 6.841447869730794e-06, | |
| "loss": 0.1078, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.7627118644067796, | |
| "grad_norm": 0.3999058208022931, | |
| "learning_rate": 6.820802876264112e-06, | |
| "loss": 0.1002, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7655367231638418, | |
| "grad_norm": 0.3086907716307657, | |
| "learning_rate": 6.800122032278351e-06, | |
| "loss": 0.1057, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.768361581920904, | |
| "grad_norm": 0.3019429125850021, | |
| "learning_rate": 6.7794057449669545e-06, | |
| "loss": 0.1224, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.7711864406779662, | |
| "grad_norm": 0.3649142358181941, | |
| "learning_rate": 6.758654422221225e-06, | |
| "loss": 0.1229, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.7740112994350282, | |
| "grad_norm": 0.3886884309856128, | |
| "learning_rate": 6.7378684726222875e-06, | |
| "loss": 0.1347, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.7768361581920904, | |
| "grad_norm": 0.39178721337354694, | |
| "learning_rate": 6.717048305433053e-06, | |
| "loss": 0.1395, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.7796610169491526, | |
| "grad_norm": 0.3560643954364327, | |
| "learning_rate": 6.6961943305901515e-06, | |
| "loss": 0.0996, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.7824858757062146, | |
| "grad_norm": 0.2987347215445713, | |
| "learning_rate": 6.675306958695874e-06, | |
| "loss": 0.0932, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.7853107344632768, | |
| "grad_norm": 0.391815080682088, | |
| "learning_rate": 6.65438660101007e-06, | |
| "loss": 0.1008, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.788135593220339, | |
| "grad_norm": 0.36385558646484645, | |
| "learning_rate": 6.633433669442066e-06, | |
| "loss": 0.1477, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.7909604519774012, | |
| "grad_norm": 0.3950859179251757, | |
| "learning_rate": 6.612448576542545e-06, | |
| "loss": 0.1546, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7937853107344632, | |
| "grad_norm": 0.3508433770997797, | |
| "learning_rate": 6.59143173549543e-06, | |
| "loss": 0.1226, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.7966101694915254, | |
| "grad_norm": 0.31172563185935787, | |
| "learning_rate": 6.570383560109745e-06, | |
| "loss": 0.1159, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.7994350282485876, | |
| "grad_norm": 0.3722434789366503, | |
| "learning_rate": 6.549304464811467e-06, | |
| "loss": 0.1718, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.8022598870056498, | |
| "grad_norm": 0.5468457727438586, | |
| "learning_rate": 6.52819486463537e-06, | |
| "loss": 0.129, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8050847457627118, | |
| "grad_norm": 0.37527355148100994, | |
| "learning_rate": 6.50705517521685e-06, | |
| "loss": 0.1245, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.807909604519774, | |
| "grad_norm": 0.3327457659403298, | |
| "learning_rate": 6.48588581278374e-06, | |
| "loss": 0.1058, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8107344632768362, | |
| "grad_norm": 0.32768437622701796, | |
| "learning_rate": 6.464687194148121e-06, | |
| "loss": 0.1215, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8135593220338984, | |
| "grad_norm": 0.2981605604482326, | |
| "learning_rate": 6.443459736698106e-06, | |
| "loss": 0.107, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.8163841807909604, | |
| "grad_norm": 0.39440686793408264, | |
| "learning_rate": 6.422203858389633e-06, | |
| "loss": 0.099, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.8192090395480226, | |
| "grad_norm": 0.3079389916506814, | |
| "learning_rate": 6.400919977738222e-06, | |
| "loss": 0.1261, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8220338983050848, | |
| "grad_norm": 0.39230281512992937, | |
| "learning_rate": 6.379608513810753e-06, | |
| "loss": 0.1388, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8248587570621468, | |
| "grad_norm": 0.34403030238315363, | |
| "learning_rate": 6.3582698862171945e-06, | |
| "loss": 0.1144, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.827683615819209, | |
| "grad_norm": 0.3554434392149389, | |
| "learning_rate": 6.336904515102355e-06, | |
| "loss": 0.1401, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8305084745762712, | |
| "grad_norm": 0.3903419801304912, | |
| "learning_rate": 6.315512821137606e-06, | |
| "loss": 0.1166, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.29926598642859675, | |
| "learning_rate": 6.294095225512604e-06, | |
| "loss": 0.1172, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.8361581920903954, | |
| "grad_norm": 0.3247406039107872, | |
| "learning_rate": 6.272652149926989e-06, | |
| "loss": 0.1206, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.8389830508474576, | |
| "grad_norm": 0.4155920338425029, | |
| "learning_rate": 6.251184016582088e-06, | |
| "loss": 0.1569, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.8418079096045198, | |
| "grad_norm": 0.49517100510425244, | |
| "learning_rate": 6.229691248172599e-06, | |
| "loss": 0.1269, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.844632768361582, | |
| "grad_norm": 0.3069243773291746, | |
| "learning_rate": 6.208174267878272e-06, | |
| "loss": 0.1039, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.847457627118644, | |
| "grad_norm": 0.30008373909078834, | |
| "learning_rate": 6.186633499355576e-06, | |
| "loss": 0.1011, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8502824858757062, | |
| "grad_norm": 0.39498835620426576, | |
| "learning_rate": 6.165069366729347e-06, | |
| "loss": 0.1262, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.8531073446327684, | |
| "grad_norm": 0.45259238504149724, | |
| "learning_rate": 6.143482294584459e-06, | |
| "loss": 0.1555, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.8559322033898306, | |
| "grad_norm": 0.3111562216471757, | |
| "learning_rate": 6.121872707957441e-06, | |
| "loss": 0.1037, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.8587570621468926, | |
| "grad_norm": 0.38394089498308964, | |
| "learning_rate": 6.100241032328125e-06, | |
| "loss": 0.1381, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.8615819209039548, | |
| "grad_norm": 0.4084090521249512, | |
| "learning_rate": 6.078587693611258e-06, | |
| "loss": 0.132, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.864406779661017, | |
| "grad_norm": 0.32206751376923765, | |
| "learning_rate": 6.056913118148122e-06, | |
| "loss": 0.115, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.867231638418079, | |
| "grad_norm": 0.39859711839150824, | |
| "learning_rate": 6.035217732698141e-06, | |
| "loss": 0.0989, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.8700564971751412, | |
| "grad_norm": 0.3223901842186738, | |
| "learning_rate": 6.013501964430468e-06, | |
| "loss": 0.1129, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.8728813559322034, | |
| "grad_norm": 0.37501991260972894, | |
| "learning_rate": 5.9917662409155896e-06, | |
| "loss": 0.1158, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.8757062146892656, | |
| "grad_norm": 0.3680155603064568, | |
| "learning_rate": 5.970010990116892e-06, | |
| "loss": 0.135, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.8785310734463276, | |
| "grad_norm": 0.4225399124226897, | |
| "learning_rate": 5.948236640382249e-06, | |
| "loss": 0.1597, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.8813559322033898, | |
| "grad_norm": 0.3062593607691136, | |
| "learning_rate": 5.926443620435572e-06, | |
| "loss": 0.1136, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.884180790960452, | |
| "grad_norm": 0.39959497875600314, | |
| "learning_rate": 5.904632359368388e-06, | |
| "loss": 0.1177, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.8870056497175142, | |
| "grad_norm": 1.0041774505409626, | |
| "learning_rate": 5.8828032866313725e-06, | |
| "loss": 0.1129, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.8898305084745762, | |
| "grad_norm": 0.33893621066441904, | |
| "learning_rate": 5.860956832025907e-06, | |
| "loss": 0.1375, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.8926553672316384, | |
| "grad_norm": 0.39383272399940145, | |
| "learning_rate": 5.839093425695609e-06, | |
| "loss": 0.1422, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.8954802259887006, | |
| "grad_norm": 0.35738567030204327, | |
| "learning_rate": 5.817213498117866e-06, | |
| "loss": 0.1529, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.8983050847457628, | |
| "grad_norm": 0.39608069155077835, | |
| "learning_rate": 5.795317480095361e-06, | |
| "loss": 0.1716, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.9011299435028248, | |
| "grad_norm": 0.3631332118845927, | |
| "learning_rate": 5.773405802747585e-06, | |
| "loss": 0.1555, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.903954802259887, | |
| "grad_norm": 0.3841861480992692, | |
| "learning_rate": 5.751478897502353e-06, | |
| "loss": 0.1894, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9067796610169492, | |
| "grad_norm": 0.29759516462695884, | |
| "learning_rate": 5.729537196087309e-06, | |
| "loss": 0.112, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9096045197740112, | |
| "grad_norm": 0.28809510654440856, | |
| "learning_rate": 5.707581130521424e-06, | |
| "loss": 0.1134, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9124293785310734, | |
| "grad_norm": 0.397848892460871, | |
| "learning_rate": 5.685611133106491e-06, | |
| "loss": 0.1297, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.9152542372881356, | |
| "grad_norm": 0.3160261105495359, | |
| "learning_rate": 5.663627636418611e-06, | |
| "loss": 0.1023, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.9180790960451978, | |
| "grad_norm": 0.31177416648537104, | |
| "learning_rate": 5.64163107329968e-06, | |
| "loss": 0.1143, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.9209039548022598, | |
| "grad_norm": 0.32018560545646063, | |
| "learning_rate": 5.619621876848864e-06, | |
| "loss": 0.0991, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.923728813559322, | |
| "grad_norm": 0.2963637468584666, | |
| "learning_rate": 5.597600480414069e-06, | |
| "loss": 0.1292, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.9265536723163842, | |
| "grad_norm": 0.44573857637587677, | |
| "learning_rate": 5.575567317583415e-06, | |
| "loss": 0.1217, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9293785310734464, | |
| "grad_norm": 0.39445386690118284, | |
| "learning_rate": 5.553522822176694e-06, | |
| "loss": 0.1684, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9322033898305084, | |
| "grad_norm": 0.5570629744547292, | |
| "learning_rate": 5.531467428236827e-06, | |
| "loss": 0.121, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9350282485875706, | |
| "grad_norm": 0.2927441446558258, | |
| "learning_rate": 5.5094015700213254e-06, | |
| "loss": 0.1199, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.9378531073446328, | |
| "grad_norm": 0.31083045505241014, | |
| "learning_rate": 5.4873256819937325e-06, | |
| "loss": 0.1299, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.940677966101695, | |
| "grad_norm": 0.3685953460491956, | |
| "learning_rate": 5.465240198815073e-06, | |
| "loss": 0.1432, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.943502824858757, | |
| "grad_norm": 0.358140929608912, | |
| "learning_rate": 5.443145555335296e-06, | |
| "loss": 0.1148, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.9463276836158192, | |
| "grad_norm": 0.4047657464869157, | |
| "learning_rate": 5.421042186584708e-06, | |
| "loss": 0.1339, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.9491525423728814, | |
| "grad_norm": 0.30030457933241933, | |
| "learning_rate": 5.398930527765416e-06, | |
| "loss": 0.1301, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.9519774011299436, | |
| "grad_norm": 0.31945028324973423, | |
| "learning_rate": 5.376811014242749e-06, | |
| "loss": 0.1147, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.9548022598870056, | |
| "grad_norm": 0.48059842793389146, | |
| "learning_rate": 5.354684081536693e-06, | |
| "loss": 0.1251, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.9576271186440678, | |
| "grad_norm": 0.3547595891598906, | |
| "learning_rate": 5.332550165313312e-06, | |
| "loss": 0.1256, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.96045197740113, | |
| "grad_norm": 0.4010586374877903, | |
| "learning_rate": 5.31040970137617e-06, | |
| "loss": 0.1521, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.963276836158192, | |
| "grad_norm": 0.30749195608479307, | |
| "learning_rate": 5.288263125657757e-06, | |
| "loss": 0.0898, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.9661016949152542, | |
| "grad_norm": 0.38150734834003086, | |
| "learning_rate": 5.266110874210893e-06, | |
| "loss": 0.1153, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.9689265536723164, | |
| "grad_norm": 0.4214873565398611, | |
| "learning_rate": 5.2439533832001565e-06, | |
| "loss": 0.1148, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.9717514124293786, | |
| "grad_norm": 0.6122493487962786, | |
| "learning_rate": 5.221791088893282e-06, | |
| "loss": 0.1104, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.9745762711864406, | |
| "grad_norm": 0.3363807115202252, | |
| "learning_rate": 5.199624427652589e-06, | |
| "loss": 0.1223, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.9774011299435028, | |
| "grad_norm": 0.4560391650148292, | |
| "learning_rate": 5.177453835926366e-06, | |
| "loss": 0.1279, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.980225988700565, | |
| "grad_norm": 0.34518676510036406, | |
| "learning_rate": 5.155279750240302e-06, | |
| "loss": 0.111, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.9830508474576272, | |
| "grad_norm": 0.37562743319165254, | |
| "learning_rate": 5.133102607188875e-06, | |
| "loss": 0.1223, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.9858757062146892, | |
| "grad_norm": 0.37060789770525393, | |
| "learning_rate": 5.1109228434267585e-06, | |
| "loss": 0.1205, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.9887005649717514, | |
| "grad_norm": 0.30512645778837, | |
| "learning_rate": 5.0887408956602316e-06, | |
| "loss": 0.1123, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.9915254237288136, | |
| "grad_norm": 0.31734580505159626, | |
| "learning_rate": 5.06655720063857e-06, | |
| "loss": 0.1393, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.9943502824858758, | |
| "grad_norm": 0.4252275291418614, | |
| "learning_rate": 5.044372195145455e-06, | |
| "loss": 0.1804, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.9971751412429378, | |
| "grad_norm": 0.445405112876313, | |
| "learning_rate": 5.022186315990371e-06, | |
| "loss": 0.1466, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.33905894838175626, | |
| "learning_rate": 5e-06, | |
| "loss": 0.123, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.002824858757062, | |
| "grad_norm": 0.2724848502115793, | |
| "learning_rate": 4.97781368400963e-06, | |
| "loss": 0.0967, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.0056497175141244, | |
| "grad_norm": 0.25595208945562964, | |
| "learning_rate": 4.9556278048545445e-06, | |
| "loss": 0.0704, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.0084745762711864, | |
| "grad_norm": 0.2425492096389186, | |
| "learning_rate": 4.933442799361432e-06, | |
| "loss": 0.0799, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.0112994350282485, | |
| "grad_norm": 0.2761823631954267, | |
| "learning_rate": 4.911259104339771e-06, | |
| "loss": 0.0936, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.0141242937853108, | |
| "grad_norm": 0.27454279772458723, | |
| "learning_rate": 4.889077156573242e-06, | |
| "loss": 0.1175, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.0169491525423728, | |
| "grad_norm": 0.31992512429885583, | |
| "learning_rate": 4.866897392811127e-06, | |
| "loss": 0.0968, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0197740112994351, | |
| "grad_norm": 0.2894306047919742, | |
| "learning_rate": 4.8447202497596975e-06, | |
| "loss": 0.1236, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.0225988700564972, | |
| "grad_norm": 0.24920669865954365, | |
| "learning_rate": 4.822546164073635e-06, | |
| "loss": 0.0852, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.0254237288135593, | |
| "grad_norm": 0.2952175126202192, | |
| "learning_rate": 4.800375572347414e-06, | |
| "loss": 0.0991, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.0282485875706215, | |
| "grad_norm": 0.23148883860994288, | |
| "learning_rate": 4.778208911106718e-06, | |
| "loss": 0.066, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.0310734463276836, | |
| "grad_norm": 0.2692480902893908, | |
| "learning_rate": 4.756046616799845e-06, | |
| "loss": 0.0973, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.0338983050847457, | |
| "grad_norm": 0.2760736585863661, | |
| "learning_rate": 4.7338891257891085e-06, | |
| "loss": 0.0912, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.036723163841808, | |
| "grad_norm": 0.24876270637682962, | |
| "learning_rate": 4.7117368743422435e-06, | |
| "loss": 0.0837, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.03954802259887, | |
| "grad_norm": 0.2494244392514283, | |
| "learning_rate": 4.689590298623831e-06, | |
| "loss": 0.0811, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.042372881355932, | |
| "grad_norm": 0.24104486268426012, | |
| "learning_rate": 4.667449834686689e-06, | |
| "loss": 0.076, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.0451977401129944, | |
| "grad_norm": 0.2789423674939665, | |
| "learning_rate": 4.645315918463308e-06, | |
| "loss": 0.086, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.0480225988700564, | |
| "grad_norm": 0.2770971973884279, | |
| "learning_rate": 4.623188985757252e-06, | |
| "loss": 0.089, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.0508474576271187, | |
| "grad_norm": 0.28375117683833695, | |
| "learning_rate": 4.601069472234584e-06, | |
| "loss": 0.1046, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.0536723163841808, | |
| "grad_norm": 0.23874038173398437, | |
| "learning_rate": 4.578957813415293e-06, | |
| "loss": 0.0657, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.0564971751412429, | |
| "grad_norm": 0.27927992062438906, | |
| "learning_rate": 4.556854444664706e-06, | |
| "loss": 0.0823, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.0593220338983051, | |
| "grad_norm": 0.25006740348604295, | |
| "learning_rate": 4.534759801184928e-06, | |
| "loss": 0.084, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.0621468926553672, | |
| "grad_norm": 0.26172192084359713, | |
| "learning_rate": 4.512674318006268e-06, | |
| "loss": 0.0688, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.0649717514124293, | |
| "grad_norm": 0.27984219569057084, | |
| "learning_rate": 4.490598429978676e-06, | |
| "loss": 0.1003, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.0677966101694916, | |
| "grad_norm": 0.31355314424830794, | |
| "learning_rate": 4.468532571763174e-06, | |
| "loss": 0.093, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.0706214689265536, | |
| "grad_norm": 0.2789313738912747, | |
| "learning_rate": 4.446477177823308e-06, | |
| "loss": 0.0891, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.073446327683616, | |
| "grad_norm": 0.2647947856217315, | |
| "learning_rate": 4.424432682416585e-06, | |
| "loss": 0.0657, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.076271186440678, | |
| "grad_norm": 0.309205913837631, | |
| "learning_rate": 4.402399519585932e-06, | |
| "loss": 0.091, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.07909604519774, | |
| "grad_norm": 0.27983746723387665, | |
| "learning_rate": 4.380378123151139e-06, | |
| "loss": 0.0758, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.0819209039548023, | |
| "grad_norm": 0.265317674784406, | |
| "learning_rate": 4.358368926700321e-06, | |
| "loss": 0.0744, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.0847457627118644, | |
| "grad_norm": 0.3844212564195145, | |
| "learning_rate": 4.336372363581391e-06, | |
| "loss": 0.1193, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.0875706214689265, | |
| "grad_norm": 0.301325518174874, | |
| "learning_rate": 4.314388866893512e-06, | |
| "loss": 0.0954, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.0903954802259888, | |
| "grad_norm": 0.2778786659311699, | |
| "learning_rate": 4.292418869478577e-06, | |
| "loss": 0.0791, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.0932203389830508, | |
| "grad_norm": 0.2662950349405817, | |
| "learning_rate": 4.270462803912692e-06, | |
| "loss": 0.076, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.0960451977401129, | |
| "grad_norm": 0.2796794355932831, | |
| "learning_rate": 4.248521102497649e-06, | |
| "loss": 0.0804, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.0988700564971752, | |
| "grad_norm": 0.2768998598394203, | |
| "learning_rate": 4.226594197252417e-06, | |
| "loss": 0.0834, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.1016949152542372, | |
| "grad_norm": 0.27374836470708835, | |
| "learning_rate": 4.204682519904641e-06, | |
| "loss": 0.0718, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1045197740112995, | |
| "grad_norm": 0.3445589105129125, | |
| "learning_rate": 4.182786501882135e-06, | |
| "loss": 0.1162, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.1073446327683616, | |
| "grad_norm": 0.34154494685980724, | |
| "learning_rate": 4.160906574304392e-06, | |
| "loss": 0.0821, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.1101694915254237, | |
| "grad_norm": 0.28827252450009566, | |
| "learning_rate": 4.139043167974096e-06, | |
| "loss": 0.0789, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.112994350282486, | |
| "grad_norm": 0.29707210161706604, | |
| "learning_rate": 4.117196713368629e-06, | |
| "loss": 0.0826, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.115819209039548, | |
| "grad_norm": 0.27591994189237984, | |
| "learning_rate": 4.095367640631614e-06, | |
| "loss": 0.0703, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.11864406779661, | |
| "grad_norm": 0.31721715937650236, | |
| "learning_rate": 4.073556379564429e-06, | |
| "loss": 0.0741, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.1214689265536724, | |
| "grad_norm": 0.2809918490407584, | |
| "learning_rate": 4.051763359617753e-06, | |
| "loss": 0.0768, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.1242937853107344, | |
| "grad_norm": 0.3012659075431546, | |
| "learning_rate": 4.0299890098831096e-06, | |
| "loss": 0.0899, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.1271186440677967, | |
| "grad_norm": 0.41884111681760783, | |
| "learning_rate": 4.00823375908441e-06, | |
| "loss": 0.1056, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.1299435028248588, | |
| "grad_norm": 0.3072172569806948, | |
| "learning_rate": 3.986498035569533e-06, | |
| "loss": 0.0946, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1327683615819208, | |
| "grad_norm": 0.30733421162176133, | |
| "learning_rate": 3.964782267301861e-06, | |
| "loss": 0.1148, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.1355932203389831, | |
| "grad_norm": 0.3062196925146934, | |
| "learning_rate": 3.9430868818518786e-06, | |
| "loss": 0.0939, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.1384180790960452, | |
| "grad_norm": 0.31120677637400174, | |
| "learning_rate": 3.921412306388744e-06, | |
| "loss": 0.0907, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.1412429378531073, | |
| "grad_norm": 0.29992317179141076, | |
| "learning_rate": 3.899758967671879e-06, | |
| "loss": 0.0936, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.1440677966101696, | |
| "grad_norm": 0.3192209928402148, | |
| "learning_rate": 3.8781272920425605e-06, | |
| "loss": 0.0926, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.1468926553672316, | |
| "grad_norm": 0.2758355027322586, | |
| "learning_rate": 3.856517705415543e-06, | |
| "loss": 0.0716, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.1497175141242937, | |
| "grad_norm": 0.30041138229139885, | |
| "learning_rate": 3.834930633270654e-06, | |
| "loss": 0.0915, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.152542372881356, | |
| "grad_norm": 0.2793345364877422, | |
| "learning_rate": 3.813366500644426e-06, | |
| "loss": 0.084, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.155367231638418, | |
| "grad_norm": 0.31892388449193476, | |
| "learning_rate": 3.791825732121729e-06, | |
| "loss": 0.0874, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.1581920903954803, | |
| "grad_norm": 0.31584149456312116, | |
| "learning_rate": 3.770308751827402e-06, | |
| "loss": 0.0973, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.1610169491525424, | |
| "grad_norm": 0.34820980091040554, | |
| "learning_rate": 3.748815983417914e-06, | |
| "loss": 0.1253, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.1638418079096045, | |
| "grad_norm": 0.28853430489950965, | |
| "learning_rate": 3.727347850073012e-06, | |
| "loss": 0.0759, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.1666666666666667, | |
| "grad_norm": 0.3062765715619059, | |
| "learning_rate": 3.705904774487396e-06, | |
| "loss": 0.0933, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.1694915254237288, | |
| "grad_norm": 0.32486365661487987, | |
| "learning_rate": 3.6844871788623946e-06, | |
| "loss": 0.0911, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.1723163841807909, | |
| "grad_norm": 0.3026538149404601, | |
| "learning_rate": 3.6630954848976472e-06, | |
| "loss": 0.0942, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.1751412429378532, | |
| "grad_norm": 0.3953970274281564, | |
| "learning_rate": 3.641730113782807e-06, | |
| "loss": 0.0779, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.1779661016949152, | |
| "grad_norm": 0.2978638146695922, | |
| "learning_rate": 3.6203914861892483e-06, | |
| "loss": 0.0907, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.1807909604519775, | |
| "grad_norm": 0.29056550406150716, | |
| "learning_rate": 3.5990800222617774e-06, | |
| "loss": 0.0754, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.1836158192090396, | |
| "grad_norm": 0.2519599426657091, | |
| "learning_rate": 3.577796141610369e-06, | |
| "loss": 0.0632, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.1864406779661016, | |
| "grad_norm": 0.2867955638349113, | |
| "learning_rate": 3.5565402633018963e-06, | |
| "loss": 0.0854, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.189265536723164, | |
| "grad_norm": 0.27040139234217375, | |
| "learning_rate": 3.535312805851881e-06, | |
| "loss": 0.0676, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.192090395480226, | |
| "grad_norm": 0.3332989661333647, | |
| "learning_rate": 3.5141141872162613e-06, | |
| "loss": 0.1127, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.194915254237288, | |
| "grad_norm": 0.309745904183909, | |
| "learning_rate": 3.4929448247831523e-06, | |
| "loss": 0.0917, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.1977401129943503, | |
| "grad_norm": 0.45096384475777856, | |
| "learning_rate": 3.4718051353646304e-06, | |
| "loss": 0.1173, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.2005649717514124, | |
| "grad_norm": 0.30878271329326906, | |
| "learning_rate": 3.4506955351885346e-06, | |
| "loss": 0.0919, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.2033898305084745, | |
| "grad_norm": 0.320682805049876, | |
| "learning_rate": 3.4296164398902576e-06, | |
| "loss": 0.0922, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.2062146892655368, | |
| "grad_norm": 0.30581016394054344, | |
| "learning_rate": 3.408568264504571e-06, | |
| "loss": 0.0809, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.2090395480225988, | |
| "grad_norm": 0.2747339191019564, | |
| "learning_rate": 3.387551423457456e-06, | |
| "loss": 0.0802, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.211864406779661, | |
| "grad_norm": 0.552467904262321, | |
| "learning_rate": 3.366566330557935e-06, | |
| "loss": 0.1036, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.2146892655367232, | |
| "grad_norm": 0.32041597951648537, | |
| "learning_rate": 3.345613398989932e-06, | |
| "loss": 0.0849, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.2175141242937852, | |
| "grad_norm": 0.32164554989021144, | |
| "learning_rate": 3.324693041304128e-06, | |
| "loss": 0.0901, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.2203389830508475, | |
| "grad_norm": 0.297941242679515, | |
| "learning_rate": 3.3038056694098485e-06, | |
| "loss": 0.0857, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.2231638418079096, | |
| "grad_norm": 0.29988786294219155, | |
| "learning_rate": 3.2829516945669493e-06, | |
| "loss": 0.0658, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.2259887005649717, | |
| "grad_norm": 0.28671491672159266, | |
| "learning_rate": 3.262131527377715e-06, | |
| "loss": 0.0825, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.228813559322034, | |
| "grad_norm": 0.29202073423769753, | |
| "learning_rate": 3.241345577778775e-06, | |
| "loss": 0.0793, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.231638418079096, | |
| "grad_norm": 0.26686163546586056, | |
| "learning_rate": 3.220594255033046e-06, | |
| "loss": 0.0621, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.2344632768361583, | |
| "grad_norm": 0.2713137149878859, | |
| "learning_rate": 3.1998779677216508e-06, | |
| "loss": 0.0731, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.2372881355932204, | |
| "grad_norm": 0.2775500944738776, | |
| "learning_rate": 3.1791971237358893e-06, | |
| "loss": 0.0734, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.2401129943502824, | |
| "grad_norm": 0.28779045523216457, | |
| "learning_rate": 3.1585521302692073e-06, | |
| "loss": 0.0924, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.2429378531073447, | |
| "grad_norm": 0.3012961352670417, | |
| "learning_rate": 3.1379433938091695e-06, | |
| "loss": 0.0977, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.2457627118644068, | |
| "grad_norm": 0.2790213382568145, | |
| "learning_rate": 3.117371320129469e-06, | |
| "loss": 0.0638, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.2485875706214689, | |
| "grad_norm": 0.289333568849024, | |
| "learning_rate": 3.0968363142819226e-06, | |
| "loss": 0.0835, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.2514124293785311, | |
| "grad_norm": 0.27842001426208673, | |
| "learning_rate": 3.076338780588507e-06, | |
| "loss": 0.0744, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.2542372881355932, | |
| "grad_norm": 0.29218223586498643, | |
| "learning_rate": 3.0558791226333974e-06, | |
| "loss": 0.084, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.2570621468926553, | |
| "grad_norm": 0.27223825874055874, | |
| "learning_rate": 3.035457743255016e-06, | |
| "loss": 0.0836, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.2598870056497176, | |
| "grad_norm": 0.27957551489646953, | |
| "learning_rate": 3.0150750445380995e-06, | |
| "loss": 0.0782, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.2627118644067796, | |
| "grad_norm": 0.3122940814886727, | |
| "learning_rate": 2.9947314278057927e-06, | |
| "loss": 0.1053, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.2655367231638417, | |
| "grad_norm": 0.3191436186366097, | |
| "learning_rate": 2.9744272936117323e-06, | |
| "loss": 0.1014, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.268361581920904, | |
| "grad_norm": 0.2842992416668331, | |
| "learning_rate": 2.954163041732174e-06, | |
| "loss": 0.0749, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.271186440677966, | |
| "grad_norm": 0.30400106395182974, | |
| "learning_rate": 2.9339390711581105e-06, | |
| "loss": 0.0887, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.274011299435028, | |
| "grad_norm": 0.29506183884480336, | |
| "learning_rate": 2.9137557800874177e-06, | |
| "loss": 0.091, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.2768361581920904, | |
| "grad_norm": 0.3144746461494332, | |
| "learning_rate": 2.8936135659170217e-06, | |
| "loss": 0.1059, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.2796610169491525, | |
| "grad_norm": 0.2996152337603787, | |
| "learning_rate": 2.8735128252350677e-06, | |
| "loss": 0.0786, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.2824858757062148, | |
| "grad_norm": 0.34649915198509984, | |
| "learning_rate": 2.853453953813108e-06, | |
| "loss": 0.0802, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.2853107344632768, | |
| "grad_norm": 0.3012239481267599, | |
| "learning_rate": 2.8334373465983216e-06, | |
| "loss": 0.0895, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.288135593220339, | |
| "grad_norm": 0.28862860268152546, | |
| "learning_rate": 2.8134633977057236e-06, | |
| "loss": 0.0839, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.2909604519774012, | |
| "grad_norm": 0.3084982778477155, | |
| "learning_rate": 2.7935325004104164e-06, | |
| "loss": 0.1009, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.2937853107344632, | |
| "grad_norm": 0.3282395905268862, | |
| "learning_rate": 2.7736450471398435e-06, | |
| "loss": 0.0652, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.2966101694915255, | |
| "grad_norm": 0.2793458687882274, | |
| "learning_rate": 2.7538014294660564e-06, | |
| "loss": 0.06, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.2994350282485876, | |
| "grad_norm": 0.29032428847483527, | |
| "learning_rate": 2.734002038098015e-06, | |
| "loss": 0.0674, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.3022598870056497, | |
| "grad_norm": 0.26031067828121474, | |
| "learning_rate": 2.7142472628738846e-06, | |
| "loss": 0.0628, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.305084745762712, | |
| "grad_norm": 0.3479546918961412, | |
| "learning_rate": 2.69453749275337e-06, | |
| "loss": 0.105, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.307909604519774, | |
| "grad_norm": 0.2777871642249738, | |
| "learning_rate": 2.6748731158100528e-06, | |
| "loss": 0.0733, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.310734463276836, | |
| "grad_norm": 0.29860877000280583, | |
| "learning_rate": 2.655254519223746e-06, | |
| "loss": 0.0956, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.3135593220338984, | |
| "grad_norm": 0.39509978738969753, | |
| "learning_rate": 2.6356820892728752e-06, | |
| "loss": 0.098, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.3163841807909604, | |
| "grad_norm": 0.26842617785571943, | |
| "learning_rate": 2.616156211326875e-06, | |
| "loss": 0.0683, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.3192090395480225, | |
| "grad_norm": 0.2638729229718093, | |
| "learning_rate": 2.5966772698386e-06, | |
| "loss": 0.0697, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.3220338983050848, | |
| "grad_norm": 0.2885132528975007, | |
| "learning_rate": 2.57724564833675e-06, | |
| "loss": 0.0853, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.3248587570621468, | |
| "grad_norm": 0.2909952803458812, | |
| "learning_rate": 2.557861729418326e-06, | |
| "loss": 0.0702, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.327683615819209, | |
| "grad_norm": 0.3508969831429909, | |
| "learning_rate": 2.5385258947410908e-06, | |
| "loss": 0.1163, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.3305084745762712, | |
| "grad_norm": 0.3137041512371342, | |
| "learning_rate": 2.5192385250160587e-06, | |
| "loss": 0.0791, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.34011082522993374, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.0863, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.3361581920903955, | |
| "grad_norm": 0.31378850435891975, | |
| "learning_rate": 2.4808106984879597e-06, | |
| "loss": 0.1031, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.3389830508474576, | |
| "grad_norm": 0.32916522472842985, | |
| "learning_rate": 2.461670998305802e-06, | |
| "loss": 0.1104, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.34180790960452, | |
| "grad_norm": 0.2858816116663427, | |
| "learning_rate": 2.4425812763027672e-06, | |
| "loss": 0.082, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.344632768361582, | |
| "grad_norm": 0.3311097135552006, | |
| "learning_rate": 2.4235419083440615e-06, | |
| "loss": 0.1001, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.347457627118644, | |
| "grad_norm": 0.3212815875251671, | |
| "learning_rate": 2.404553269303448e-06, | |
| "loss": 0.0706, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.3502824858757063, | |
| "grad_norm": 0.3059752435224393, | |
| "learning_rate": 2.3856157330558625e-06, | |
| "loss": 0.0858, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.3531073446327684, | |
| "grad_norm": 0.3389049114062409, | |
| "learning_rate": 2.366729672470065e-06, | |
| "loss": 0.0853, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.3559322033898304, | |
| "grad_norm": 0.32200853396437623, | |
| "learning_rate": 2.3478954594012884e-06, | |
| "loss": 0.11, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3587570621468927, | |
| "grad_norm": 0.30985593467605343, | |
| "learning_rate": 2.329113464683913e-06, | |
| "loss": 0.0925, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.3615819209039548, | |
| "grad_norm": 0.29010419081013045, | |
| "learning_rate": 2.310384058124181e-06, | |
| "loss": 0.079, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.3644067796610169, | |
| "grad_norm": 0.2868431818032766, | |
| "learning_rate": 2.2917076084928953e-06, | |
| "loss": 0.0691, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.3672316384180792, | |
| "grad_norm": 0.32921780136851597, | |
| "learning_rate": 2.273084483518176e-06, | |
| "loss": 0.1029, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.3700564971751412, | |
| "grad_norm": 0.3443609157960376, | |
| "learning_rate": 2.25451504987821e-06, | |
| "loss": 0.1094, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.3728813559322033, | |
| "grad_norm": 0.3442323075700346, | |
| "learning_rate": 2.2359996731940348e-06, | |
| "loss": 0.1148, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.3757062146892656, | |
| "grad_norm": 0.31019876061048274, | |
| "learning_rate": 2.2175387180223333e-06, | |
| "loss": 0.0846, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.3785310734463276, | |
| "grad_norm": 0.27898269961579986, | |
| "learning_rate": 2.1991325478482695e-06, | |
| "loss": 0.0858, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.3813559322033897, | |
| "grad_norm": 0.30200643340593814, | |
| "learning_rate": 2.1807815250783194e-06, | |
| "loss": 0.0901, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.384180790960452, | |
| "grad_norm": 0.28412971697416345, | |
| "learning_rate": 2.162486011033142e-06, | |
| "loss": 0.0649, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.387005649717514, | |
| "grad_norm": 0.28849690690010993, | |
| "learning_rate": 2.1442463659404587e-06, | |
| "loss": 0.0734, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.3898305084745763, | |
| "grad_norm": 0.2872214027286925, | |
| "learning_rate": 2.1260629489279662e-06, | |
| "loss": 0.0744, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.3926553672316384, | |
| "grad_norm": 0.2856113105572892, | |
| "learning_rate": 2.1079361180162657e-06, | |
| "loss": 0.0772, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.3954802259887007, | |
| "grad_norm": 0.2917591710852941, | |
| "learning_rate": 2.089866230111813e-06, | |
| "loss": 0.0872, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.3983050847457628, | |
| "grad_norm": 0.3156299163799424, | |
| "learning_rate": 2.0718536409998834e-06, | |
| "loss": 0.0755, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.4011299435028248, | |
| "grad_norm": 0.3420052483929326, | |
| "learning_rate": 2.053898705337583e-06, | |
| "loss": 0.0833, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.4039548022598871, | |
| "grad_norm": 0.3007655048264405, | |
| "learning_rate": 2.0360017766468466e-06, | |
| "loss": 0.0755, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.4067796610169492, | |
| "grad_norm": 0.3285257384928867, | |
| "learning_rate": 2.0181632073074925e-06, | |
| "loss": 0.0882, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.4096045197740112, | |
| "grad_norm": 0.27048721301742223, | |
| "learning_rate": 2.000383348550279e-06, | |
| "loss": 0.0739, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.4124293785310735, | |
| "grad_norm": 0.2984410410038522, | |
| "learning_rate": 1.9826625504499807e-06, | |
| "loss": 0.0954, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.4124293785310735, | |
| "eval_loss": 0.130197674036026, | |
| "eval_runtime": 1.5872, | |
| "eval_samples_per_second": 18.271, | |
| "eval_steps_per_second": 5.04, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.4152542372881356, | |
| "grad_norm": 0.2898926398978837, | |
| "learning_rate": 1.965001161918513e-06, | |
| "loss": 0.0789, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.4180790960451977, | |
| "grad_norm": 0.3465927210704245, | |
| "learning_rate": 1.947399530698043e-06, | |
| "loss": 0.0979, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.42090395480226, | |
| "grad_norm": 0.3475682598971018, | |
| "learning_rate": 1.92985800335416e-06, | |
| "loss": 0.0843, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.423728813559322, | |
| "grad_norm": 0.3294028190875718, | |
| "learning_rate": 1.912376925269041e-06, | |
| "loss": 0.1121, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.426553672316384, | |
| "grad_norm": 0.3498640725406821, | |
| "learning_rate": 1.894956640634652e-06, | |
| "loss": 0.0828, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.4293785310734464, | |
| "grad_norm": 0.30225327794900386, | |
| "learning_rate": 1.8775974924459716e-06, | |
| "loss": 0.085, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.4322033898305084, | |
| "grad_norm": 0.3931329736136659, | |
| "learning_rate": 1.860299822494241e-06, | |
| "loss": 0.0724, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.4350282485875705, | |
| "grad_norm": 0.28157973731776237, | |
| "learning_rate": 1.8430639713602317e-06, | |
| "loss": 0.0658, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.4378531073446328, | |
| "grad_norm": 0.28515233111654203, | |
| "learning_rate": 1.8258902784075394e-06, | |
| "loss": 0.0847, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.4406779661016949, | |
| "grad_norm": 0.2760261233151599, | |
| "learning_rate": 1.808779081775901e-06, | |
| "loss": 0.066, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.4435028248587571, | |
| "grad_norm": 0.2941924396596464, | |
| "learning_rate": 1.7917307183745353e-06, | |
| "loss": 0.0884, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.4463276836158192, | |
| "grad_norm": 0.2686743706277037, | |
| "learning_rate": 1.7747455238755223e-06, | |
| "loss": 0.0743, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.4491525423728815, | |
| "grad_norm": 0.3151180489919417, | |
| "learning_rate": 1.757823832707175e-06, | |
| "loss": 0.1007, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.4519774011299436, | |
| "grad_norm": 0.34222558948663734, | |
| "learning_rate": 1.7409659780474652e-06, | |
| "loss": 0.103, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.4548022598870056, | |
| "grad_norm": 0.3402549223734959, | |
| "learning_rate": 1.7241722918174642e-06, | |
| "loss": 0.1213, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.457627118644068, | |
| "grad_norm": 0.3145312403253552, | |
| "learning_rate": 1.7074431046748075e-06, | |
| "loss": 0.0969, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.46045197740113, | |
| "grad_norm": 0.2896418247469403, | |
| "learning_rate": 1.6907787460071756e-06, | |
| "loss": 0.0862, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.463276836158192, | |
| "grad_norm": 0.31049200696233425, | |
| "learning_rate": 1.6741795439258218e-06, | |
| "loss": 0.098, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.4661016949152543, | |
| "grad_norm": 0.36636752026678193, | |
| "learning_rate": 1.6576458252590988e-06, | |
| "loss": 0.1391, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.4689265536723164, | |
| "grad_norm": 0.27909992931423844, | |
| "learning_rate": 1.641177915546036e-06, | |
| "loss": 0.0744, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.4717514124293785, | |
| "grad_norm": 0.2864859483455, | |
| "learning_rate": 1.6247761390299221e-06, | |
| "loss": 0.0898, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.4745762711864407, | |
| "grad_norm": 0.4362808575365348, | |
| "learning_rate": 1.6084408186519195e-06, | |
| "loss": 0.0734, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.4774011299435028, | |
| "grad_norm": 0.28051391012639115, | |
| "learning_rate": 1.5921722760447144e-06, | |
| "loss": 0.0678, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.4802259887005649, | |
| "grad_norm": 0.30037283427959355, | |
| "learning_rate": 1.5759708315261724e-06, | |
| "loss": 0.0932, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.4830508474576272, | |
| "grad_norm": 0.2982894826882516, | |
| "learning_rate": 1.5598368040930427e-06, | |
| "loss": 0.0735, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.4858757062146892, | |
| "grad_norm": 0.3166739516240939, | |
| "learning_rate": 1.5437705114146735e-06, | |
| "loss": 0.1003, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.4887005649717513, | |
| "grad_norm": 0.3288809776102775, | |
| "learning_rate": 1.527772269826749e-06, | |
| "loss": 0.0984, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.4915254237288136, | |
| "grad_norm": 0.29168718949906514, | |
| "learning_rate": 1.511842394325077e-06, | |
| "loss": 0.0907, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.4943502824858756, | |
| "grad_norm": 0.2993454545746122, | |
| "learning_rate": 1.4959811985593707e-06, | |
| "loss": 0.0648, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.497175141242938, | |
| "grad_norm": 0.2901988801448637, | |
| "learning_rate": 1.4801889948270852e-06, | |
| "loss": 0.0843, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.276078183792322, | |
| "learning_rate": 1.4644660940672628e-06, | |
| "loss": 0.0646, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.5028248587570623, | |
| "grad_norm": 0.2957096527763229, | |
| "learning_rate": 1.44881280585441e-06, | |
| "loss": 0.0838, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.5056497175141241, | |
| "grad_norm": 0.3074950453422565, | |
| "learning_rate": 1.4332294383924034e-06, | |
| "loss": 0.0976, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.5084745762711864, | |
| "grad_norm": 0.28111249969205165, | |
| "learning_rate": 1.4177162985084242e-06, | |
| "loss": 0.07, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.5112994350282487, | |
| "grad_norm": 0.3040746365690992, | |
| "learning_rate": 1.4022736916469166e-06, | |
| "loss": 0.0675, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.5141242937853108, | |
| "grad_norm": 0.2970452295573905, | |
| "learning_rate": 1.3869019218635644e-06, | |
| "loss": 0.0937, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.5169491525423728, | |
| "grad_norm": 0.3091154520174239, | |
| "learning_rate": 1.3716012918193206e-06, | |
| "loss": 0.0761, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.5197740112994351, | |
| "grad_norm": 0.33856268143824403, | |
| "learning_rate": 1.3563721027744309e-06, | |
| "loss": 0.0941, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.5225988700564972, | |
| "grad_norm": 0.2960053858988549, | |
| "learning_rate": 1.3412146545825166e-06, | |
| "loss": 0.0731, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.5254237288135593, | |
| "grad_norm": 0.3000001098205527, | |
| "learning_rate": 1.3261292456846648e-06, | |
| "loss": 0.0777, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.5282485875706215, | |
| "grad_norm": 0.32275915585442194, | |
| "learning_rate": 1.3111161731035448e-06, | |
| "loss": 0.1028, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.5310734463276836, | |
| "grad_norm": 0.28047948839083625, | |
| "learning_rate": 1.2961757324375768e-06, | |
| "loss": 0.0773, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.5338983050847457, | |
| "grad_norm": 0.2897511789785588, | |
| "learning_rate": 1.2813082178550929e-06, | |
| "loss": 0.0761, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.536723163841808, | |
| "grad_norm": 0.3604669071306025, | |
| "learning_rate": 1.2665139220885615e-06, | |
| "loss": 0.0966, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.53954802259887, | |
| "grad_norm": 0.35066125768752815, | |
| "learning_rate": 1.2517931364288133e-06, | |
| "loss": 0.1189, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.542372881355932, | |
| "grad_norm": 0.36481440937249643, | |
| "learning_rate": 1.2371461507193077e-06, | |
| "loss": 0.0854, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.5451977401129944, | |
| "grad_norm": 0.2705446394892136, | |
| "learning_rate": 1.2225732533504309e-06, | |
| "loss": 0.0681, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.5480225988700564, | |
| "grad_norm": 0.2709042435292725, | |
| "learning_rate": 1.2080747312538082e-06, | |
| "loss": 0.0605, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.5508474576271185, | |
| "grad_norm": 0.2941674674541573, | |
| "learning_rate": 1.1936508698966664e-06, | |
| "loss": 0.0759, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.5536723163841808, | |
| "grad_norm": 0.3301045484204278, | |
| "learning_rate": 1.1793019532762057e-06, | |
| "loss": 0.09, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.556497175141243, | |
| "grad_norm": 0.5929461456725253, | |
| "learning_rate": 1.1650282639140066e-06, | |
| "loss": 0.115, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.559322033898305, | |
| "grad_norm": 0.29282230586457825, | |
| "learning_rate": 1.1508300828504682e-06, | |
| "loss": 0.068, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.5621468926553672, | |
| "grad_norm": 0.290916040961216, | |
| "learning_rate": 1.1367076896392853e-06, | |
| "loss": 0.0759, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.5649717514124295, | |
| "grad_norm": 0.3308510796313253, | |
| "learning_rate": 1.122661362341927e-06, | |
| "loss": 0.107, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.5677966101694916, | |
| "grad_norm": 0.2902882953470368, | |
| "learning_rate": 1.1086913775221709e-06, | |
| "loss": 0.0817, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.5706214689265536, | |
| "grad_norm": 0.25808920224703275, | |
| "learning_rate": 1.0947980102406597e-06, | |
| "loss": 0.063, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.573446327683616, | |
| "grad_norm": 0.3042205877096776, | |
| "learning_rate": 1.0809815340494822e-06, | |
| "loss": 0.0755, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.576271186440678, | |
| "grad_norm": 0.3051764415373886, | |
| "learning_rate": 1.0672422209867879e-06, | |
| "loss": 0.0652, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.57909604519774, | |
| "grad_norm": 0.3146319364999993, | |
| "learning_rate": 1.053580341571428e-06, | |
| "loss": 0.1059, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.5819209039548023, | |
| "grad_norm": 0.3027084741124625, | |
| "learning_rate": 1.0399961647976315e-06, | |
| "loss": 0.0812, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.5847457627118644, | |
| "grad_norm": 0.3278175149471125, | |
| "learning_rate": 1.0264899581297121e-06, | |
| "loss": 0.1192, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.5875706214689265, | |
| "grad_norm": 0.28592160027084834, | |
| "learning_rate": 1.0130619874967983e-06, | |
| "loss": 0.0752, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.5903954802259888, | |
| "grad_norm": 0.2930417798367745, | |
| "learning_rate": 9.997125172875943e-07, | |
| "loss": 0.0879, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.5932203389830508, | |
| "grad_norm": 0.2704337798545791, | |
| "learning_rate": 9.86441810345183e-07, | |
| "loss": 0.0624, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.5960451977401129, | |
| "grad_norm": 0.3166751927355104, | |
| "learning_rate": 9.732501279618388e-07, | |
| "loss": 0.0848, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.5988700564971752, | |
| "grad_norm": 0.29461324641929826, | |
| "learning_rate": 9.60137729873898e-07, | |
| "loss": 0.0789, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.6016949152542372, | |
| "grad_norm": 0.31189484709881815, | |
| "learning_rate": 9.471048742566313e-07, | |
| "loss": 0.0822, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.6045197740112993, | |
| "grad_norm": 0.3466231703998608, | |
| "learning_rate": 9.34151817719166e-07, | |
| "loss": 0.0767, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.6073446327683616, | |
| "grad_norm": 0.30675238542761885, | |
| "learning_rate": 9.212788152994367e-07, | |
| "loss": 0.1034, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.6101694915254239, | |
| "grad_norm": 0.522761335835565, | |
| "learning_rate": 9.08486120459155e-07, | |
| "loss": 0.1273, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.6129943502824857, | |
| "grad_norm": 0.2810136760249764, | |
| "learning_rate": 8.957739850788288e-07, | |
| "loss": 0.073, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.615819209039548, | |
| "grad_norm": 0.31293166014889473, | |
| "learning_rate": 8.831426594527976e-07, | |
| "loss": 0.0956, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.6186440677966103, | |
| "grad_norm": 0.3408526367512878, | |
| "learning_rate": 8.705923922843041e-07, | |
| "loss": 0.0891, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.6214689265536724, | |
| "grad_norm": 0.30421762095488025, | |
| "learning_rate": 8.581234306805969e-07, | |
| "loss": 0.0946, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.6242937853107344, | |
| "grad_norm": 0.2940599119195987, | |
| "learning_rate": 8.457360201480702e-07, | |
| "loss": 0.0692, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.6271186440677967, | |
| "grad_norm": 0.3002501406760772, | |
| "learning_rate": 8.334304045874248e-07, | |
| "loss": 0.0815, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.6299435028248588, | |
| "grad_norm": 0.268425275016888, | |
| "learning_rate": 8.212068262888684e-07, | |
| "loss": 0.0751, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.6327683615819208, | |
| "grad_norm": 0.2928986782866679, | |
| "learning_rate": 8.090655259273428e-07, | |
| "loss": 0.0918, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.6355932203389831, | |
| "grad_norm": 0.32485035024590025, | |
| "learning_rate": 7.970067425577849e-07, | |
| "loss": 0.0933, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.6384180790960452, | |
| "grad_norm": 0.3234267210299417, | |
| "learning_rate": 7.850307136104246e-07, | |
| "loss": 0.0904, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.6412429378531073, | |
| "grad_norm": 0.30188930742886005, | |
| "learning_rate": 7.731376748861069e-07, | |
| "loss": 0.0889, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.6440677966101696, | |
| "grad_norm": 0.335467078244967, | |
| "learning_rate": 7.613278605516455e-07, | |
| "loss": 0.1325, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.6468926553672316, | |
| "grad_norm": 0.3072516075801986, | |
| "learning_rate": 7.4960150313522e-07, | |
| "loss": 0.0783, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.6497175141242937, | |
| "grad_norm": 0.3131137120089587, | |
| "learning_rate": 7.379588335217875e-07, | |
| "loss": 0.0995, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.652542372881356, | |
| "grad_norm": 0.2914572623071712, | |
| "learning_rate": 7.264000809485483e-07, | |
| "loss": 0.0863, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.655367231638418, | |
| "grad_norm": 0.32502736910136926, | |
| "learning_rate": 7.149254730004246e-07, | |
| "loss": 0.1124, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.65819209039548, | |
| "grad_norm": 0.326075318059859, | |
| "learning_rate": 7.035352356055786e-07, | |
| "loss": 0.1201, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.6610169491525424, | |
| "grad_norm": 0.37300992352749307, | |
| "learning_rate": 6.922295930309691e-07, | |
| "loss": 0.1073, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.6638418079096047, | |
| "grad_norm": 0.28876296213713953, | |
| "learning_rate": 6.810087678779353e-07, | |
| "loss": 0.0743, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.30368795978426233, | |
| "learning_rate": 6.698729810778065e-07, | |
| "loss": 0.0798, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.6694915254237288, | |
| "grad_norm": 0.3072791432365542, | |
| "learning_rate": 6.588224518875647e-07, | |
| "loss": 0.0812, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.672316384180791, | |
| "grad_norm": 0.3056293727238639, | |
| "learning_rate": 6.478573978855146e-07, | |
| "loss": 0.0684, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.6751412429378532, | |
| "grad_norm": 0.30153905844016693, | |
| "learning_rate": 6.369780349670085e-07, | |
| "loss": 0.0779, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.6779661016949152, | |
| "grad_norm": 0.2858390899342033, | |
| "learning_rate": 6.261845773401936e-07, | |
| "loss": 0.0713, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.6807909604519775, | |
| "grad_norm": 0.30891966645412655, | |
| "learning_rate": 6.154772375217905e-07, | |
| "loss": 0.0837, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.6836158192090396, | |
| "grad_norm": 0.29088564996940475, | |
| "learning_rate": 6.048562263329139e-07, | |
| "loss": 0.0825, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.6864406779661016, | |
| "grad_norm": 0.29812823046797693, | |
| "learning_rate": 5.943217528949169e-07, | |
| "loss": 0.0927, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.689265536723164, | |
| "grad_norm": 0.30652461054045976, | |
| "learning_rate": 5.838740246252794e-07, | |
| "loss": 0.0766, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.692090395480226, | |
| "grad_norm": 0.3012630776892009, | |
| "learning_rate": 5.735132472335192e-07, | |
| "loss": 0.0893, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.694915254237288, | |
| "grad_norm": 0.30023304223451514, | |
| "learning_rate": 5.632396247171429e-07, | |
| "loss": 0.1049, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.6977401129943503, | |
| "grad_norm": 0.3280835461780043, | |
| "learning_rate": 5.530533593576292e-07, | |
| "loss": 0.116, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.7005649717514124, | |
| "grad_norm": 0.2762418019865388, | |
| "learning_rate": 5.429546517164486e-07, | |
| "loss": 0.067, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.7033898305084745, | |
| "grad_norm": 0.31675842207409677, | |
| "learning_rate": 5.329437006311122e-07, | |
| "loss": 0.0872, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.7062146892655368, | |
| "grad_norm": 0.3041486434567392, | |
| "learning_rate": 5.230207032112549e-07, | |
| "loss": 0.0752, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.7090395480225988, | |
| "grad_norm": 0.29993780041723445, | |
| "learning_rate": 5.131858548347596e-07, | |
| "loss": 0.0717, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.711864406779661, | |
| "grad_norm": 0.30787432406626875, | |
| "learning_rate": 5.034393491439044e-07, | |
| "loss": 0.0802, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.7146892655367232, | |
| "grad_norm": 0.28990243338730465, | |
| "learning_rate": 4.93781378041554e-07, | |
| "loss": 0.0871, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.7175141242937855, | |
| "grad_norm": 0.3024236161639379, | |
| "learning_rate": 4.842121316873821e-07, | |
| "loss": 0.0855, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.7203389830508473, | |
| "grad_norm": 0.3186961475088875, | |
| "learning_rate": 4.747317984941213e-07, | |
| "loss": 0.0875, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.7231638418079096, | |
| "grad_norm": 0.2907838927015749, | |
| "learning_rate": 4.653405651238607e-07, | |
| "loss": 0.0908, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.725988700564972, | |
| "grad_norm": 0.30763579782876077, | |
| "learning_rate": 4.560386164843639e-07, | |
| "loss": 0.0964, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.7288135593220337, | |
| "grad_norm": 0.30987458360594455, | |
| "learning_rate": 4.468261357254339e-07, | |
| "loss": 0.0947, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.731638418079096, | |
| "grad_norm": 0.29429416026094735, | |
| "learning_rate": 4.3770330423530626e-07, | |
| "loss": 0.0834, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.7344632768361583, | |
| "grad_norm": 0.3063918655948546, | |
| "learning_rate": 4.286703016370719e-07, | |
| "loss": 0.0925, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.7372881355932204, | |
| "grad_norm": 0.33329444172148553, | |
| "learning_rate": 4.197273057851464e-07, | |
| "loss": 0.0983, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.7401129943502824, | |
| "grad_norm": 0.29709804495802744, | |
| "learning_rate": 4.108744927617669e-07, | |
| "loss": 0.079, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.7429378531073447, | |
| "grad_norm": 0.2937681249333296, | |
| "learning_rate": 4.021120368735254e-07, | |
| "loss": 0.088, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.7457627118644068, | |
| "grad_norm": 0.33235581919645196, | |
| "learning_rate": 3.934401106479352e-07, | |
| "loss": 0.093, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.7485875706214689, | |
| "grad_norm": 0.3052937062176937, | |
| "learning_rate": 3.8485888483003384e-07, | |
| "loss": 0.0987, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.7514124293785311, | |
| "grad_norm": 0.2953169881958288, | |
| "learning_rate": 3.763685283790208e-07, | |
| "loss": 0.0861, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.7542372881355932, | |
| "grad_norm": 0.39089257775250585, | |
| "learning_rate": 3.679692084649372e-07, | |
| "loss": 0.1092, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.7570621468926553, | |
| "grad_norm": 0.298962166079189, | |
| "learning_rate": 3.596610904653652e-07, | |
| "loss": 0.0877, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.7598870056497176, | |
| "grad_norm": 0.33068479547784435, | |
| "learning_rate": 3.5144433796217515e-07, | |
| "loss": 0.0868, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.7627118644067796, | |
| "grad_norm": 0.29230226311663704, | |
| "learning_rate": 3.433191127383079e-07, | |
| "loss": 0.0786, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.7655367231638417, | |
| "grad_norm": 0.342045177865139, | |
| "learning_rate": 3.352855747745859e-07, | |
| "loss": 0.1034, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.768361581920904, | |
| "grad_norm": 0.3538547033051671, | |
| "learning_rate": 3.2734388224656575e-07, | |
| "loss": 0.0913, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.7711864406779663, | |
| "grad_norm": 0.3022791955228267, | |
| "learning_rate": 3.1949419152142e-07, | |
| "loss": 0.0912, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.774011299435028, | |
| "grad_norm": 0.3286478841800299, | |
| "learning_rate": 3.1173665715486076e-07, | |
| "loss": 0.1005, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.7768361581920904, | |
| "grad_norm": 0.2914491072414654, | |
| "learning_rate": 3.0407143188809885e-07, | |
| "loss": 0.087, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.7796610169491527, | |
| "grad_norm": 0.26788446393967724, | |
| "learning_rate": 2.9649866664483387e-07, | |
| "loss": 0.06, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.7824858757062145, | |
| "grad_norm": 0.29391848362521833, | |
| "learning_rate": 2.8901851052828e-07, | |
| "loss": 0.0789, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.7853107344632768, | |
| "grad_norm": 0.2778847008048511, | |
| "learning_rate": 2.816311108182368e-07, | |
| "loss": 0.0626, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.788135593220339, | |
| "grad_norm": 0.3154659365103027, | |
| "learning_rate": 2.743366129681824e-07, | |
| "loss": 0.101, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.7909604519774012, | |
| "grad_norm": 0.31293085126205966, | |
| "learning_rate": 2.671351606024153e-07, | |
| "loss": 0.0762, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.7937853107344632, | |
| "grad_norm": 0.47243232664184365, | |
| "learning_rate": 2.6002689551322403e-07, | |
| "loss": 0.1006, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.7966101694915255, | |
| "grad_norm": 0.2781939275244853, | |
| "learning_rate": 2.530119576580936e-07, | |
| "loss": 0.0638, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.7994350282485876, | |
| "grad_norm": 0.2796358570748197, | |
| "learning_rate": 2.460904851569534e-07, | |
| "loss": 0.0636, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.8022598870056497, | |
| "grad_norm": 0.31697163166683606, | |
| "learning_rate": 2.3926261428945386e-07, | |
| "loss": 0.0707, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.805084745762712, | |
| "grad_norm": 0.2764809135900223, | |
| "learning_rate": 2.325284794922883e-07, | |
| "loss": 0.0674, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.807909604519774, | |
| "grad_norm": 0.3238518566445213, | |
| "learning_rate": 2.2588821335654044e-07, | |
| "loss": 0.0824, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.810734463276836, | |
| "grad_norm": 0.34129160389189483, | |
| "learning_rate": 2.1934194662507736e-07, | |
| "loss": 0.0851, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.8135593220338984, | |
| "grad_norm": 0.2825014150604838, | |
| "learning_rate": 2.1288980818997272e-07, | |
| "loss": 0.077, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.8163841807909604, | |
| "grad_norm": 0.29713126389115424, | |
| "learning_rate": 2.0653192508997222e-07, | |
| "loss": 0.0762, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.8192090395480225, | |
| "grad_norm": 0.300338264829181, | |
| "learning_rate": 2.0026842250799038e-07, | |
| "loss": 0.0878, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.8220338983050848, | |
| "grad_norm": 1.4166868383475169, | |
| "learning_rate": 1.9409942376864333e-07, | |
| "loss": 0.0867, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.8248587570621468, | |
| "grad_norm": 0.25772375662960606, | |
| "learning_rate": 1.8802505033582608e-07, | |
| "loss": 0.0604, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.827683615819209, | |
| "grad_norm": 0.31296870168050195, | |
| "learning_rate": 1.8204542181031572e-07, | |
| "loss": 0.0909, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.8305084745762712, | |
| "grad_norm": 0.3064057159229424, | |
| "learning_rate": 1.7616065592742038e-07, | |
| "loss": 0.0881, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.8333333333333335, | |
| "grad_norm": 0.2998465592381362, | |
| "learning_rate": 1.7037086855465902e-07, | |
| "loss": 0.0629, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.8361581920903953, | |
| "grad_norm": 0.29682999878586, | |
| "learning_rate": 1.6467617368947918e-07, | |
| "loss": 0.0786, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.8389830508474576, | |
| "grad_norm": 0.2781558427742286, | |
| "learning_rate": 1.5907668345701732e-07, | |
| "loss": 0.0818, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.84180790960452, | |
| "grad_norm": 0.27914990667274464, | |
| "learning_rate": 1.5357250810788316e-07, | |
| "loss": 0.0739, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.844632768361582, | |
| "grad_norm": 0.3438724850417393, | |
| "learning_rate": 1.4816375601599653e-07, | |
| "loss": 0.0723, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.847457627118644, | |
| "grad_norm": 0.31828540808245065, | |
| "learning_rate": 1.4285053367645074e-07, | |
| "loss": 0.077, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.8502824858757063, | |
| "grad_norm": 0.27972916277751064, | |
| "learning_rate": 1.37632945703412e-07, | |
| "loss": 0.0705, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.8531073446327684, | |
| "grad_norm": 0.24716233730443476, | |
| "learning_rate": 1.3251109482806667e-07, | |
| "loss": 0.0489, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.8559322033898304, | |
| "grad_norm": 0.308468550353476, | |
| "learning_rate": 1.2748508189659447e-07, | |
| "loss": 0.0866, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.8587570621468927, | |
| "grad_norm": 0.3092098443706496, | |
| "learning_rate": 1.2255500586818015e-07, | |
| "loss": 0.1055, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.8615819209039548, | |
| "grad_norm": 0.28162343440142174, | |
| "learning_rate": 1.177209638130733e-07, | |
| "loss": 0.072, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.8644067796610169, | |
| "grad_norm": 0.3007912961091226, | |
| "learning_rate": 1.1298305091066664e-07, | |
| "loss": 0.0908, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.8672316384180792, | |
| "grad_norm": 0.3136447204914414, | |
| "learning_rate": 1.0834136044763188e-07, | |
| "loss": 0.0836, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.8700564971751412, | |
| "grad_norm": 0.31357010636279004, | |
| "learning_rate": 1.0379598381607681e-07, | |
| "loss": 0.0807, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.8728813559322033, | |
| "grad_norm": 0.3090649577202879, | |
| "learning_rate": 9.93470105117461e-08, | |
| "loss": 0.0949, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.8757062146892656, | |
| "grad_norm": 0.30061286344301524, | |
| "learning_rate": 9.499452813226284e-08, | |
| "loss": 0.0832, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.8785310734463276, | |
| "grad_norm": 0.3207780749688742, | |
| "learning_rate": 9.073862237539977e-08, | |
| "loss": 0.0922, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.8813559322033897, | |
| "grad_norm": 0.35510559900468425, | |
| "learning_rate": 8.657937703739516e-08, | |
| "loss": 0.0989, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.884180790960452, | |
| "grad_norm": 0.29122472340879935, | |
| "learning_rate": 8.251687401130137e-08, | |
| "loss": 0.0806, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.8870056497175143, | |
| "grad_norm": 0.3387325114735546, | |
| "learning_rate": 7.855119328537109e-08, | |
| "loss": 0.1179, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.8898305084745761, | |
| "grad_norm": 0.3838105969146293, | |
| "learning_rate": 7.468241294148471e-08, | |
| "loss": 0.1056, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.8926553672316384, | |
| "grad_norm": 0.34521981231042703, | |
| "learning_rate": 7.09106091536127e-08, | |
| "loss": 0.0708, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.8954802259887007, | |
| "grad_norm": 0.2860135017170368, | |
| "learning_rate": 6.723585618631456e-08, | |
| "loss": 0.0807, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.8983050847457628, | |
| "grad_norm": 0.25713094736882913, | |
| "learning_rate": 6.365822639327724e-08, | |
| "loss": 0.0596, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.9011299435028248, | |
| "grad_norm": 0.32493231522498606, | |
| "learning_rate": 6.017779021589065e-08, | |
| "loss": 0.0783, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.9039548022598871, | |
| "grad_norm": 0.3195456319023967, | |
| "learning_rate": 5.679461618185944e-08, | |
| "loss": 0.0989, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.9067796610169492, | |
| "grad_norm": 0.28832414052354244, | |
| "learning_rate": 5.350877090385731e-08, | |
| "loss": 0.0842, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.9096045197740112, | |
| "grad_norm": 0.29862804180466584, | |
| "learning_rate": 5.032031907821089e-08, | |
| "loss": 0.0799, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.9124293785310735, | |
| "grad_norm": 0.39261040413818743, | |
| "learning_rate": 4.722932348362852e-08, | |
| "loss": 0.0763, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.9152542372881356, | |
| "grad_norm": 0.28345109997512263, | |
| "learning_rate": 4.423584497996458e-08, | |
| "loss": 0.0623, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.9180790960451977, | |
| "grad_norm": 0.3730843254054893, | |
| "learning_rate": 4.1339942507018225e-08, | |
| "loss": 0.1051, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.92090395480226, | |
| "grad_norm": 0.408781043992154, | |
| "learning_rate": 3.8541673083377086e-08, | |
| "loss": 0.0972, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.923728813559322, | |
| "grad_norm": 0.27235202225360894, | |
| "learning_rate": 3.584109180529205e-08, | |
| "loss": 0.078, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.926553672316384, | |
| "grad_norm": 0.3249573151531456, | |
| "learning_rate": 3.323825184559204e-08, | |
| "loss": 0.0665, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.9293785310734464, | |
| "grad_norm": 0.3917754153996301, | |
| "learning_rate": 3.073320445263817e-08, | |
| "loss": 0.0948, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.9322033898305084, | |
| "grad_norm": 0.3436451083397938, | |
| "learning_rate": 2.8325998949314536e-08, | |
| "loss": 0.1001, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.9350282485875705, | |
| "grad_norm": 0.2962089436476239, | |
| "learning_rate": 2.6016682732057375e-08, | |
| "loss": 0.092, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.9378531073446328, | |
| "grad_norm": 0.31332687230877443, | |
| "learning_rate": 2.3805301269920754e-08, | |
| "loss": 0.0719, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.940677966101695, | |
| "grad_norm": 0.29206089331438057, | |
| "learning_rate": 2.1691898103682885e-08, | |
| "loss": 0.0803, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.943502824858757, | |
| "grad_norm": 0.29588225232531956, | |
| "learning_rate": 1.9676514844987338e-08, | |
| "loss": 0.0746, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.9463276836158192, | |
| "grad_norm": 0.3052178637723924, | |
| "learning_rate": 1.775919117552427e-08, | |
| "loss": 0.0683, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.9491525423728815, | |
| "grad_norm": 0.33628210770097194, | |
| "learning_rate": 1.593996484624938e-08, | |
| "loss": 0.0876, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.9519774011299436, | |
| "grad_norm": 0.41716535023044066, | |
| "learning_rate": 1.42188716766406e-08, | |
| "loss": 0.0797, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.9548022598870056, | |
| "grad_norm": 0.2721830233815887, | |
| "learning_rate": 1.2595945553992572e-08, | |
| "loss": 0.0746, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.957627118644068, | |
| "grad_norm": 0.3093867803026701, | |
| "learning_rate": 1.1071218432749942e-08, | |
| "loss": 0.0927, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.96045197740113, | |
| "grad_norm": 0.28442945395766916, | |
| "learning_rate": 9.6447203338762e-09, | |
| "loss": 0.0787, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.963276836158192, | |
| "grad_norm": 0.30806352104616475, | |
| "learning_rate": 8.316479344266382e-09, | |
| "loss": 0.0888, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.9661016949152543, | |
| "grad_norm": 0.366990265292286, | |
| "learning_rate": 7.0865216161902785e-09, | |
| "loss": 0.1067, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.9689265536723164, | |
| "grad_norm": 0.37187519570368377, | |
| "learning_rate": 5.954871366779525e-09, | |
| "loss": 0.0814, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.9717514124293785, | |
| "grad_norm": 0.3167633384396195, | |
| "learning_rate": 4.921550877550752e-09, | |
| "loss": 0.0687, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.9745762711864407, | |
| "grad_norm": 0.301910108787057, | |
| "learning_rate": 3.9865804939659414e-09, | |
| "loss": 0.0748, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.9774011299435028, | |
| "grad_norm": 0.32697329312245565, | |
| "learning_rate": 3.1499786250321904e-09, | |
| "loss": 0.1215, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.9802259887005649, | |
| "grad_norm": 0.31440661932949104, | |
| "learning_rate": 2.411761742939778e-09, | |
| "loss": 0.0798, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.9830508474576272, | |
| "grad_norm": 0.32934297243481625, | |
| "learning_rate": 1.7719443827368677e-09, | |
| "loss": 0.0759, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.9858757062146892, | |
| "grad_norm": 0.3018134458199549, | |
| "learning_rate": 1.2305391420458502e-09, | |
| "loss": 0.0935, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.9887005649717513, | |
| "grad_norm": 0.2717967961520749, | |
| "learning_rate": 7.875566808107638e-10, | |
| "loss": 0.0621, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.9915254237288136, | |
| "grad_norm": 0.30491685297395554, | |
| "learning_rate": 4.4300572109134965e-10, | |
| "loss": 0.0913, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.9943502824858759, | |
| "grad_norm": 0.2812227615626389, | |
| "learning_rate": 1.9689304688985667e-10, | |
| "loss": 0.0699, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.9971751412429377, | |
| "grad_norm": 0.27084127166875765, | |
| "learning_rate": 4.922350401781461e-11, | |
| "loss": 0.0745, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.29812406633307104, | |
| "learning_rate": 0.0, | |
| "loss": 0.0894, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 708, | |
| "total_flos": 20169300639744.0, | |
| "train_loss": 0.10943256686362675, | |
| "train_runtime": 738.5688, | |
| "train_samples_per_second": 7.666, | |
| "train_steps_per_second": 0.959 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 708, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50000000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 20169300639744.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |