| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.999628666914222, | |
| "eval_steps": 337, | |
| "global_step": 1346, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007426661715558856, | |
| "grad_norm": 1.4470734596252441, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 1.3568, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0007426661715558856, | |
| "eval_loss": 0.4649902582168579, | |
| "eval_runtime": 189.2701, | |
| "eval_samples_per_second": 5.991, | |
| "eval_steps_per_second": 2.996, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0014853323431117712, | |
| "grad_norm": 1.335021734237671, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 1.7148, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0022279985146676567, | |
| "grad_norm": 2.0106494426727295, | |
| "learning_rate": 6e-06, | |
| "loss": 1.9387, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0029706646862235424, | |
| "grad_norm": 2.0463790893554688, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.1194, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003713330857779428, | |
| "grad_norm": 2.6946020126342773, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2371, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004455997029335313, | |
| "grad_norm": 2.0213823318481445, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.4407, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0051986632008912, | |
| "grad_norm": 2.055884599685669, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 2.5741, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005941329372447085, | |
| "grad_norm": 2.562527656555176, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.7273, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006683995544002971, | |
| "grad_norm": 2.70515775680542, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.9015, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007426661715558856, | |
| "grad_norm": 2.6423349380493164, | |
| "learning_rate": 2e-05, | |
| "loss": 2.9069, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008169327887114742, | |
| "grad_norm": 2.7331981658935547, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 2.8755, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.008911994058670627, | |
| "grad_norm": 4.324583053588867, | |
| "learning_rate": 2.4e-05, | |
| "loss": 3.0139, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009654660230226514, | |
| "grad_norm": 3.278507947921753, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 3.1708, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0103973264017824, | |
| "grad_norm": 3.9470252990722656, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 3.2871, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.011139992573338284, | |
| "grad_norm": 4.76015567779541, | |
| "learning_rate": 3e-05, | |
| "loss": 2.9986, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01188265874489417, | |
| "grad_norm": 6.300633907318115, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 3.4458, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.012625324916450055, | |
| "grad_norm": 4.9842095375061035, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 3.2977, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.013367991088005942, | |
| "grad_norm": 4.979720592498779, | |
| "learning_rate": 3.6e-05, | |
| "loss": 3.285, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.014110657259561827, | |
| "grad_norm": 5.599627494812012, | |
| "learning_rate": 3.8e-05, | |
| "loss": 3.0219, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.014853323431117713, | |
| "grad_norm": 5.2525153160095215, | |
| "learning_rate": 4e-05, | |
| "loss": 2.9258, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015595989602673598, | |
| "grad_norm": 5.435408115386963, | |
| "learning_rate": 4.2e-05, | |
| "loss": 2.7997, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.016338655774229483, | |
| "grad_norm": 7.129375457763672, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 2.8325, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01708132194578537, | |
| "grad_norm": 6.04606819152832, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 2.4109, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.017823988117341254, | |
| "grad_norm": 7.30807638168335, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.5344, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01856665428889714, | |
| "grad_norm": 6.1737775802612305, | |
| "learning_rate": 5e-05, | |
| "loss": 2.2762, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.019309320460453028, | |
| "grad_norm": 1.2856624126434326, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 0.4641, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.02005198663200891, | |
| "grad_norm": 1.5805637836456299, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 0.4788, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0207946528035648, | |
| "grad_norm": 1.7802543640136719, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 0.4178, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.021537318975120682, | |
| "grad_norm": 1.6685826778411865, | |
| "learning_rate": 5.8e-05, | |
| "loss": 0.3613, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.02227998514667657, | |
| "grad_norm": 1.2964603900909424, | |
| "learning_rate": 6e-05, | |
| "loss": 0.2742, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.023022651318232456, | |
| "grad_norm": 1.162868618965149, | |
| "learning_rate": 6.2e-05, | |
| "loss": 0.1826, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02376531748978834, | |
| "grad_norm": 1.0783743858337402, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 0.1383, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.024507983661344226, | |
| "grad_norm": 0.7907594442367554, | |
| "learning_rate": 6.6e-05, | |
| "loss": 0.0578, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02525064983290011, | |
| "grad_norm": 0.45891255140304565, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 0.0416, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.025993316004455997, | |
| "grad_norm": 0.2138753980398178, | |
| "learning_rate": 7e-05, | |
| "loss": 0.0117, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.026735982176011884, | |
| "grad_norm": 0.09297079592943192, | |
| "learning_rate": 7.2e-05, | |
| "loss": 0.0046, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.027478648347567768, | |
| "grad_norm": 0.27470022439956665, | |
| "learning_rate": 7.4e-05, | |
| "loss": 0.0102, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.028221314519123655, | |
| "grad_norm": 0.05123934894800186, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.0022, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.028963980690679538, | |
| "grad_norm": 0.22415509819984436, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 0.0058, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.029706646862235425, | |
| "grad_norm": 1.0602823495864868, | |
| "learning_rate": 8e-05, | |
| "loss": 0.0308, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.030449313033791312, | |
| "grad_norm": 0.019605087116360664, | |
| "learning_rate": 8.2e-05, | |
| "loss": 0.0008, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.031191979205347196, | |
| "grad_norm": 0.020324615761637688, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.001, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03193464537690308, | |
| "grad_norm": 0.0388132706284523, | |
| "learning_rate": 8.6e-05, | |
| "loss": 0.0009, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.032677311548458966, | |
| "grad_norm": 0.11963042616844177, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.0016, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03341997772001486, | |
| "grad_norm": 0.04200240597128868, | |
| "learning_rate": 9e-05, | |
| "loss": 0.0007, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03416264389157074, | |
| "grad_norm": 1.10586416721344, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 0.0202, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.034905310063126624, | |
| "grad_norm": 0.007396018132567406, | |
| "learning_rate": 9.4e-05, | |
| "loss": 0.0004, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03564797623468251, | |
| "grad_norm": 0.012978550978004932, | |
| "learning_rate": 9.6e-05, | |
| "loss": 0.0005, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0363906424062384, | |
| "grad_norm": 0.006396067328751087, | |
| "learning_rate": 9.8e-05, | |
| "loss": 0.0002, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03713330857779428, | |
| "grad_norm": 0.024195007979869843, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0007, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.037875974749350165, | |
| "grad_norm": 1.803991436958313, | |
| "learning_rate": 9.999985309738107e-05, | |
| "loss": 0.0337, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.038618640920906055, | |
| "grad_norm": 3.9776086807250977, | |
| "learning_rate": 9.999941239038748e-05, | |
| "loss": 0.0281, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03936130709246194, | |
| "grad_norm": 0.039939988404512405, | |
| "learning_rate": 9.999867788160888e-05, | |
| "loss": 0.0005, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.04010397326401782, | |
| "grad_norm": 2.4416584968566895, | |
| "learning_rate": 9.999764957536131e-05, | |
| "loss": 0.0216, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04084663943557371, | |
| "grad_norm": 0.005858874414116144, | |
| "learning_rate": 9.999632747768722e-05, | |
| "loss": 0.0003, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0415893056071296, | |
| "grad_norm": 0.004351081792265177, | |
| "learning_rate": 9.999471159635539e-05, | |
| "loss": 0.0003, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.04233197177868548, | |
| "grad_norm": 0.37215837836265564, | |
| "learning_rate": 9.999280194086089e-05, | |
| "loss": 0.0037, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.043074637950241364, | |
| "grad_norm": 0.005829016678035259, | |
| "learning_rate": 9.999059852242507e-05, | |
| "loss": 0.0004, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.043817304121797254, | |
| "grad_norm": 0.11440680176019669, | |
| "learning_rate": 9.998810135399546e-05, | |
| "loss": 0.0011, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.04455997029335314, | |
| "grad_norm": 0.023830680176615715, | |
| "learning_rate": 9.998531045024566e-05, | |
| "loss": 0.0006, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04530263646490902, | |
| "grad_norm": 5.839324474334717, | |
| "learning_rate": 9.998222582757533e-05, | |
| "loss": 0.0621, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.04604530263646491, | |
| "grad_norm": 0.17139233648777008, | |
| "learning_rate": 9.997884750411005e-05, | |
| "loss": 0.0021, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.046787968808020795, | |
| "grad_norm": 0.030187880620360374, | |
| "learning_rate": 9.997517549970115e-05, | |
| "loss": 0.0007, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.04753063497957668, | |
| "grad_norm": 0.015891535207629204, | |
| "learning_rate": 9.997120983592574e-05, | |
| "loss": 0.0004, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04827330115113257, | |
| "grad_norm": 0.001739410450682044, | |
| "learning_rate": 9.996695053608651e-05, | |
| "loss": 0.0002, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04901596732268845, | |
| "grad_norm": 0.010511704720556736, | |
| "learning_rate": 9.996239762521151e-05, | |
| "loss": 0.0003, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.049758633494244336, | |
| "grad_norm": 0.0017766759265214205, | |
| "learning_rate": 9.995755113005414e-05, | |
| "loss": 0.0002, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.05050129966580022, | |
| "grad_norm": 0.004469791427254677, | |
| "learning_rate": 9.99524110790929e-05, | |
| "loss": 0.0003, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05124396583735611, | |
| "grad_norm": 0.0036557905841618776, | |
| "learning_rate": 9.994697750253127e-05, | |
| "loss": 0.0004, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.051986632008911994, | |
| "grad_norm": 0.0024416742380708456, | |
| "learning_rate": 9.994125043229752e-05, | |
| "loss": 0.0003, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.05272929818046788, | |
| "grad_norm": 0.015205912292003632, | |
| "learning_rate": 9.993522990204453e-05, | |
| "loss": 0.0004, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.05347196435202377, | |
| "grad_norm": 0.014475185424089432, | |
| "learning_rate": 9.992891594714954e-05, | |
| "loss": 0.0005, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.05421463052357965, | |
| "grad_norm": 0.007128004450351, | |
| "learning_rate": 9.992230860471402e-05, | |
| "loss": 0.0004, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.054957296695135535, | |
| "grad_norm": 0.014512602239847183, | |
| "learning_rate": 9.991540791356342e-05, | |
| "loss": 0.0004, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.055699962866691426, | |
| "grad_norm": 0.006873926613479853, | |
| "learning_rate": 9.990821391424689e-05, | |
| "loss": 0.0003, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05644262903824731, | |
| "grad_norm": 0.4753952920436859, | |
| "learning_rate": 9.990072664903717e-05, | |
| "loss": 0.0152, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05718529520980319, | |
| "grad_norm": 0.006181332748383284, | |
| "learning_rate": 9.989294616193017e-05, | |
| "loss": 0.0004, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.057927961381359076, | |
| "grad_norm": 0.005821248050779104, | |
| "learning_rate": 9.988487249864489e-05, | |
| "loss": 0.0004, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05867062755291497, | |
| "grad_norm": 0.005324830766767263, | |
| "learning_rate": 9.9876505706623e-05, | |
| "loss": 0.0003, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.05941329372447085, | |
| "grad_norm": 0.0035147082526236773, | |
| "learning_rate": 9.986784583502862e-05, | |
| "loss": 0.0003, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.060155959896026734, | |
| "grad_norm": 0.011637026444077492, | |
| "learning_rate": 9.98588929347481e-05, | |
| "loss": 0.0006, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.060898626067582624, | |
| "grad_norm": 0.006236548535525799, | |
| "learning_rate": 9.98496470583896e-05, | |
| "loss": 0.0006, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.06164129223913851, | |
| "grad_norm": 0.006345892325043678, | |
| "learning_rate": 9.984010826028288e-05, | |
| "loss": 0.0003, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.06238395841069439, | |
| "grad_norm": 0.016584008932113647, | |
| "learning_rate": 9.98302765964789e-05, | |
| "loss": 0.0006, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06312662458225028, | |
| "grad_norm": 0.0018910560756921768, | |
| "learning_rate": 9.982015212474955e-05, | |
| "loss": 0.0002, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06386929075380617, | |
| "grad_norm": 0.0022837521973997355, | |
| "learning_rate": 9.980973490458728e-05, | |
| "loss": 0.0002, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06461195692536205, | |
| "grad_norm": 1.3271163702011108, | |
| "learning_rate": 9.979902499720477e-05, | |
| "loss": 0.0071, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06535462309691793, | |
| "grad_norm": 0.011062121950089931, | |
| "learning_rate": 9.978802246553459e-05, | |
| "loss": 0.0003, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06609728926847382, | |
| "grad_norm": 0.007717825006693602, | |
| "learning_rate": 9.97767273742287e-05, | |
| "loss": 0.0004, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.06683995544002971, | |
| "grad_norm": 0.010918798856437206, | |
| "learning_rate": 9.976513978965829e-05, | |
| "loss": 0.0006, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0675826216115856, | |
| "grad_norm": 0.006601768545806408, | |
| "learning_rate": 9.975325977991322e-05, | |
| "loss": 0.0003, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.06832528778314148, | |
| "grad_norm": 0.009449873119592667, | |
| "learning_rate": 9.974108741480165e-05, | |
| "loss": 0.0004, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06906795395469736, | |
| "grad_norm": 0.015310431830585003, | |
| "learning_rate": 9.97286227658497e-05, | |
| "loss": 0.0004, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.06981062012625325, | |
| "grad_norm": 0.09934096038341522, | |
| "learning_rate": 9.971586590630093e-05, | |
| "loss": 0.0019, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.07055328629780913, | |
| "grad_norm": 0.1965462863445282, | |
| "learning_rate": 9.970281691111598e-05, | |
| "loss": 0.0012, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07129595246936501, | |
| "grad_norm": 0.05416925996541977, | |
| "learning_rate": 9.968947585697214e-05, | |
| "loss": 0.0007, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07203861864092091, | |
| "grad_norm": 0.012609624303877354, | |
| "learning_rate": 9.967584282226281e-05, | |
| "loss": 0.0005, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0727812848124768, | |
| "grad_norm": 0.4876333177089691, | |
| "learning_rate": 9.966191788709716e-05, | |
| "loss": 0.0031, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.07352395098403268, | |
| "grad_norm": 0.01990601420402527, | |
| "learning_rate": 9.964770113329953e-05, | |
| "loss": 0.0009, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07426661715558856, | |
| "grad_norm": 0.00828948151320219, | |
| "learning_rate": 9.96331926444091e-05, | |
| "loss": 0.0004, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07500928332714445, | |
| "grad_norm": 0.005279685370624065, | |
| "learning_rate": 9.961839250567924e-05, | |
| "loss": 0.0004, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07575194949870033, | |
| "grad_norm": 1.349811315536499, | |
| "learning_rate": 9.960330080407711e-05, | |
| "loss": 0.1017, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07649461567025621, | |
| "grad_norm": 0.006720814388245344, | |
| "learning_rate": 9.958791762828317e-05, | |
| "loss": 0.0005, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07723728184181211, | |
| "grad_norm": 0.026319369673728943, | |
| "learning_rate": 9.957224306869053e-05, | |
| "loss": 0.0007, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.077979948013368, | |
| "grad_norm": 0.013718683272600174, | |
| "learning_rate": 9.955627721740454e-05, | |
| "loss": 0.0007, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07872261418492388, | |
| "grad_norm": 0.030150871723890305, | |
| "learning_rate": 9.954002016824227e-05, | |
| "loss": 0.0008, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07946528035647976, | |
| "grad_norm": 0.03413335233926773, | |
| "learning_rate": 9.95234720167318e-05, | |
| "loss": 0.0009, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.08020794652803565, | |
| "grad_norm": 0.03414030373096466, | |
| "learning_rate": 9.950663286011179e-05, | |
| "loss": 0.0009, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.08095061269959153, | |
| "grad_norm": 0.7618962526321411, | |
| "learning_rate": 9.948950279733093e-05, | |
| "loss": 0.0031, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.08169327887114743, | |
| "grad_norm": 0.01711968518793583, | |
| "learning_rate": 9.947208192904722e-05, | |
| "loss": 0.0005, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08243594504270331, | |
| "grad_norm": 0.014231199398636818, | |
| "learning_rate": 9.945437035762754e-05, | |
| "loss": 0.0005, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0831786112142592, | |
| "grad_norm": 0.08236388117074966, | |
| "learning_rate": 9.943636818714695e-05, | |
| "loss": 0.0021, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08392127738581508, | |
| "grad_norm": 0.008401082828640938, | |
| "learning_rate": 9.941807552338804e-05, | |
| "loss": 0.0004, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.08466394355737096, | |
| "grad_norm": 0.00980079360306263, | |
| "learning_rate": 9.939949247384046e-05, | |
| "loss": 0.0005, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.08540660972892684, | |
| "grad_norm": 0.01246787328273058, | |
| "learning_rate": 9.938061914770012e-05, | |
| "loss": 0.0006, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.08614927590048273, | |
| "grad_norm": 0.011044768616557121, | |
| "learning_rate": 9.936145565586871e-05, | |
| "loss": 0.0005, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08689194207203862, | |
| "grad_norm": 0.0040915291756391525, | |
| "learning_rate": 9.934200211095288e-05, | |
| "loss": 0.0003, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.08763460824359451, | |
| "grad_norm": 0.004765264689922333, | |
| "learning_rate": 9.93222586272637e-05, | |
| "loss": 0.0004, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08837727441515039, | |
| "grad_norm": 0.0034806388430297375, | |
| "learning_rate": 9.930222532081597e-05, | |
| "loss": 0.0003, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08911994058670628, | |
| "grad_norm": 0.005507381167262793, | |
| "learning_rate": 9.928190230932746e-05, | |
| "loss": 0.0003, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08986260675826216, | |
| "grad_norm": 0.42908811569213867, | |
| "learning_rate": 9.926128971221835e-05, | |
| "loss": 0.0026, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.09060527292981804, | |
| "grad_norm": 0.00801977701485157, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 0.0003, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.09134793910137393, | |
| "grad_norm": 0.007514494471251965, | |
| "learning_rate": 9.921919624732635e-05, | |
| "loss": 0.0004, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.09209060527292982, | |
| "grad_norm": 0.006028357893228531, | |
| "learning_rate": 9.919771562688904e-05, | |
| "loss": 0.0002, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.09283327144448571, | |
| "grad_norm": 0.009683290496468544, | |
| "learning_rate": 9.917594591552089e-05, | |
| "loss": 0.0005, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09357593761604159, | |
| "grad_norm": 0.01857328973710537, | |
| "learning_rate": 9.915388724114301e-05, | |
| "loss": 0.0011, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.09431860378759747, | |
| "grad_norm": 0.0045487345196306705, | |
| "learning_rate": 9.913153973337446e-05, | |
| "loss": 0.0004, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.09506126995915336, | |
| "grad_norm": 0.019089125096797943, | |
| "learning_rate": 9.910890352353153e-05, | |
| "loss": 0.0009, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09580393613070924, | |
| "grad_norm": 0.01162141002714634, | |
| "learning_rate": 9.908597874462699e-05, | |
| "loss": 0.0004, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.09654660230226514, | |
| "grad_norm": 0.0069759683683514595, | |
| "learning_rate": 9.906276553136923e-05, | |
| "loss": 0.0005, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09728926847382102, | |
| "grad_norm": 1.388071894645691, | |
| "learning_rate": 9.903926402016153e-05, | |
| "loss": 0.0127, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0980319346453769, | |
| "grad_norm": 0.006189362611621618, | |
| "learning_rate": 9.901547434910122e-05, | |
| "loss": 0.0004, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.09877460081693279, | |
| "grad_norm": 0.0009308147127740085, | |
| "learning_rate": 9.899139665797887e-05, | |
| "loss": 0.0001, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.09951726698848867, | |
| "grad_norm": 0.003263165010139346, | |
| "learning_rate": 9.896703108827759e-05, | |
| "loss": 0.0002, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.10025993316004456, | |
| "grad_norm": 0.0018793240888044238, | |
| "learning_rate": 9.894237778317195e-05, | |
| "loss": 0.0002, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.10100259933160044, | |
| "grad_norm": 0.2546728551387787, | |
| "learning_rate": 9.891743688752738e-05, | |
| "loss": 0.0021, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.10174526550315634, | |
| "grad_norm": 0.01736506260931492, | |
| "learning_rate": 9.88922085478992e-05, | |
| "loss": 0.0004, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.10248793167471222, | |
| "grad_norm": 0.003645472228527069, | |
| "learning_rate": 9.88666929125318e-05, | |
| "loss": 0.0003, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.1032305978462681, | |
| "grad_norm": 0.002554230624809861, | |
| "learning_rate": 9.884089013135766e-05, | |
| "loss": 0.0002, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.10397326401782399, | |
| "grad_norm": 0.003567540319636464, | |
| "learning_rate": 9.881480035599667e-05, | |
| "loss": 0.0002, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10471593018937987, | |
| "grad_norm": 0.0011336462339386344, | |
| "learning_rate": 9.87884237397551e-05, | |
| "loss": 0.0002, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.10545859636093576, | |
| "grad_norm": 0.0014857546193525195, | |
| "learning_rate": 9.876176043762467e-05, | |
| "loss": 0.0001, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.10620126253249164, | |
| "grad_norm": 0.022715391591191292, | |
| "learning_rate": 9.873481060628174e-05, | |
| "loss": 0.0006, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.10694392870404754, | |
| "grad_norm": 1.0306154489517212, | |
| "learning_rate": 9.870757440408638e-05, | |
| "loss": 0.0073, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.10768659487560342, | |
| "grad_norm": 0.002907106187194586, | |
| "learning_rate": 9.868005199108133e-05, | |
| "loss": 0.0002, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.1084292610471593, | |
| "grad_norm": 0.0016110404394567013, | |
| "learning_rate": 9.865224352899119e-05, | |
| "loss": 0.0002, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.10917192721871519, | |
| "grad_norm": 0.0025813078973442316, | |
| "learning_rate": 9.862414918122141e-05, | |
| "loss": 0.0002, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.10991459339027107, | |
| "grad_norm": 0.0018323366530239582, | |
| "learning_rate": 9.859576911285728e-05, | |
| "loss": 0.0002, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.11065725956182695, | |
| "grad_norm": 0.7255179286003113, | |
| "learning_rate": 9.856710349066307e-05, | |
| "loss": 0.0081, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.11139992573338285, | |
| "grad_norm": 3.8413281440734863, | |
| "learning_rate": 9.853815248308101e-05, | |
| "loss": 0.2887, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11214259190493873, | |
| "grad_norm": 0.01481606438755989, | |
| "learning_rate": 9.850891626023022e-05, | |
| "loss": 0.0009, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.11288525807649462, | |
| "grad_norm": 0.005497956182807684, | |
| "learning_rate": 9.84793949939058e-05, | |
| "loss": 0.0004, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.1136279242480505, | |
| "grad_norm": 0.05774915963411331, | |
| "learning_rate": 9.844958885757784e-05, | |
| "loss": 0.0008, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.11437059041960639, | |
| "grad_norm": 0.22279635071754456, | |
| "learning_rate": 9.84194980263903e-05, | |
| "loss": 0.0023, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.11511325659116227, | |
| "grad_norm": 0.0037247207947075367, | |
| "learning_rate": 9.838912267716005e-05, | |
| "loss": 0.0003, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.11585592276271815, | |
| "grad_norm": 0.0031422695610672235, | |
| "learning_rate": 9.835846298837584e-05, | |
| "loss": 0.0003, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.11659858893427405, | |
| "grad_norm": 0.004694859962910414, | |
| "learning_rate": 9.83275191401972e-05, | |
| "loss": 0.0003, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.11734125510582993, | |
| "grad_norm": 0.009304534643888474, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 0.0004, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.11808392127738582, | |
| "grad_norm": 0.0023734685964882374, | |
| "learning_rate": 9.826477969464249e-05, | |
| "loss": 0.0003, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.1188265874489417, | |
| "grad_norm": 0.004084714688360691, | |
| "learning_rate": 9.823298446592998e-05, | |
| "loss": 0.0003, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11956925362049758, | |
| "grad_norm": 0.005245341453701258, | |
| "learning_rate": 9.820090581514797e-05, | |
| "loss": 0.0003, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.12031191979205347, | |
| "grad_norm": 0.00829974003136158, | |
| "learning_rate": 9.816854393079403e-05, | |
| "loss": 0.0004, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.12105458596360935, | |
| "grad_norm": 0.07217549532651901, | |
| "learning_rate": 9.81358990030299e-05, | |
| "loss": 0.001, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.12179725213516525, | |
| "grad_norm": 0.0019713877700269222, | |
| "learning_rate": 9.810297122368067e-05, | |
| "loss": 0.0002, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.12253991830672113, | |
| "grad_norm": 0.008508375845849514, | |
| "learning_rate": 9.806976078623337e-05, | |
| "loss": 0.0005, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12328258447827702, | |
| "grad_norm": 0.0026160285342484713, | |
| "learning_rate": 9.803626788583603e-05, | |
| "loss": 0.0002, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.1240252506498329, | |
| "grad_norm": 0.004115263931453228, | |
| "learning_rate": 9.800249271929645e-05, | |
| "loss": 0.0003, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.12476791682138878, | |
| "grad_norm": 0.009765752591192722, | |
| "learning_rate": 9.796843548508101e-05, | |
| "loss": 0.0004, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.12551058299294468, | |
| "grad_norm": 0.032703742384910583, | |
| "learning_rate": 9.793409638331363e-05, | |
| "loss": 0.0009, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.12625324916450056, | |
| "grad_norm": 0.005391250364482403, | |
| "learning_rate": 9.789947561577445e-05, | |
| "loss": 0.0003, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.12699591533605645, | |
| "grad_norm": 0.2329409420490265, | |
| "learning_rate": 9.786457338589872e-05, | |
| "loss": 0.0022, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.12773858150761233, | |
| "grad_norm": 0.006565611343830824, | |
| "learning_rate": 9.782938989877562e-05, | |
| "loss": 0.0003, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.12848124767916821, | |
| "grad_norm": 0.00805259495973587, | |
| "learning_rate": 9.779392536114698e-05, | |
| "loss": 0.0005, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.1292239138507241, | |
| "grad_norm": 0.05728701129555702, | |
| "learning_rate": 9.775817998140616e-05, | |
| "loss": 0.0008, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.12996658002227998, | |
| "grad_norm": 0.030414534732699394, | |
| "learning_rate": 9.772215396959674e-05, | |
| "loss": 0.0006, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13070924619383587, | |
| "grad_norm": 0.003174105193465948, | |
| "learning_rate": 9.768584753741134e-05, | |
| "loss": 0.0003, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.13145191236539175, | |
| "grad_norm": 0.005474573001265526, | |
| "learning_rate": 9.764926089819038e-05, | |
| "loss": 0.0005, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.13219457853694763, | |
| "grad_norm": 0.01835116557776928, | |
| "learning_rate": 9.761239426692077e-05, | |
| "loss": 0.0006, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.13293724470850352, | |
| "grad_norm": 0.07570798695087433, | |
| "learning_rate": 9.757524786023468e-05, | |
| "loss": 0.0025, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.13367991088005943, | |
| "grad_norm": 0.0024187075905501842, | |
| "learning_rate": 9.753782189640834e-05, | |
| "loss": 0.0002, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1344225770516153, | |
| "grad_norm": 0.014368316158652306, | |
| "learning_rate": 9.750011659536058e-05, | |
| "loss": 0.0006, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.1351652432231712, | |
| "grad_norm": 0.013278312981128693, | |
| "learning_rate": 9.74621321786517e-05, | |
| "loss": 0.0005, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.13590790939472708, | |
| "grad_norm": 0.01532573252916336, | |
| "learning_rate": 9.742386886948213e-05, | |
| "loss": 0.0003, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.13665057556628296, | |
| "grad_norm": 0.0070452457293868065, | |
| "learning_rate": 9.738532689269112e-05, | |
| "loss": 0.0003, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.13739324173783884, | |
| "grad_norm": 0.5086126923561096, | |
| "learning_rate": 9.73465064747553e-05, | |
| "loss": 0.0046, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.13813590790939473, | |
| "grad_norm": 0.00677888048812747, | |
| "learning_rate": 9.730740784378753e-05, | |
| "loss": 0.0004, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1388785740809506, | |
| "grad_norm": 0.014153935015201569, | |
| "learning_rate": 9.726803122953547e-05, | |
| "loss": 0.0005, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1396212402525065, | |
| "grad_norm": 0.005835146643221378, | |
| "learning_rate": 9.722837686338025e-05, | |
| "loss": 0.0002, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.14036390642406238, | |
| "grad_norm": 0.004742791876196861, | |
| "learning_rate": 9.718844497833504e-05, | |
| "loss": 0.0003, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.14110657259561826, | |
| "grad_norm": 0.002392592839896679, | |
| "learning_rate": 9.71482358090438e-05, | |
| "loss": 0.0002, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.14184923876717415, | |
| "grad_norm": 0.009971565566956997, | |
| "learning_rate": 9.710774959177983e-05, | |
| "loss": 0.0004, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.14259190493873003, | |
| "grad_norm": 0.000985423568636179, | |
| "learning_rate": 9.706698656444438e-05, | |
| "loss": 0.0001, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1433345711102859, | |
| "grad_norm": 0.002976580522954464, | |
| "learning_rate": 9.702594696656524e-05, | |
| "loss": 0.0003, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.14407723728184182, | |
| "grad_norm": 0.004122724290937185, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 0.0003, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1448199034533977, | |
| "grad_norm": 0.004222078714519739, | |
| "learning_rate": 9.694303902541163e-05, | |
| "loss": 0.0002, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1455625696249536, | |
| "grad_norm": 0.002287927782163024, | |
| "learning_rate": 9.69011711693129e-05, | |
| "loss": 0.0002, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.14630523579650948, | |
| "grad_norm": 0.001492491108365357, | |
| "learning_rate": 9.685902771701913e-05, | |
| "loss": 0.0002, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.14704790196806536, | |
| "grad_norm": 0.00485377898439765, | |
| "learning_rate": 9.681660891616966e-05, | |
| "loss": 0.0003, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.14779056813962124, | |
| "grad_norm": 0.014361650682985783, | |
| "learning_rate": 9.677391501602182e-05, | |
| "loss": 0.0005, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.14853323431117713, | |
| "grad_norm": 1.621951699256897, | |
| "learning_rate": 9.673094626744942e-05, | |
| "loss": 0.0662, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.149275900482733, | |
| "grad_norm": 0.0026435288600623608, | |
| "learning_rate": 9.668770292294136e-05, | |
| "loss": 0.0002, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.1500185666542889, | |
| "grad_norm": 0.0024962294846773148, | |
| "learning_rate": 9.664418523660004e-05, | |
| "loss": 0.0002, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.15076123282584478, | |
| "grad_norm": 0.002211767714470625, | |
| "learning_rate": 9.660039346413994e-05, | |
| "loss": 0.0002, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.15150389899740066, | |
| "grad_norm": 0.002533277263864875, | |
| "learning_rate": 9.65563278628861e-05, | |
| "loss": 0.0003, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.15224656516895654, | |
| "grad_norm": 0.006425884552299976, | |
| "learning_rate": 9.651198869177263e-05, | |
| "loss": 0.0004, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.15298923134051243, | |
| "grad_norm": 0.0048479656688869, | |
| "learning_rate": 9.646737621134112e-05, | |
| "loss": 0.0003, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.15373189751206834, | |
| "grad_norm": 0.004739306401461363, | |
| "learning_rate": 9.642249068373921e-05, | |
| "loss": 0.0004, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.15447456368362422, | |
| "grad_norm": 0.0035690851509571075, | |
| "learning_rate": 9.637733237271894e-05, | |
| "loss": 0.0002, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.1552172298551801, | |
| "grad_norm": 0.001971122808754444, | |
| "learning_rate": 9.633190154363527e-05, | |
| "loss": 0.0002, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.155959896026736, | |
| "grad_norm": 0.0017060886602848768, | |
| "learning_rate": 9.628619846344454e-05, | |
| "loss": 0.0002, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15670256219829187, | |
| "grad_norm": 0.011329671368002892, | |
| "learning_rate": 9.624022340070279e-05, | |
| "loss": 0.0003, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.15744522836984776, | |
| "grad_norm": 0.002570141339674592, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 0.0002, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.15818789454140364, | |
| "grad_norm": 0.0023831925354897976, | |
| "learning_rate": 9.614745840978008e-05, | |
| "loss": 0.0002, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.15893056071295952, | |
| "grad_norm": 0.016878001391887665, | |
| "learning_rate": 9.610066902669592e-05, | |
| "loss": 0.0007, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.1596732268845154, | |
| "grad_norm": 0.008881067857146263, | |
| "learning_rate": 9.605360875125117e-05, | |
| "loss": 0.0002, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.1604158930560713, | |
| "grad_norm": 0.0010246317833662033, | |
| "learning_rate": 9.600627785997696e-05, | |
| "loss": 0.0001, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.16115855922762717, | |
| "grad_norm": 0.0019209448946639895, | |
| "learning_rate": 9.595867663099453e-05, | |
| "loss": 0.0002, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.16190122539918306, | |
| "grad_norm": 0.0022903543431311846, | |
| "learning_rate": 9.591080534401371e-05, | |
| "loss": 0.0002, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.16264389157073894, | |
| "grad_norm": 0.009357557632029057, | |
| "learning_rate": 9.586266428033119e-05, | |
| "loss": 0.0003, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.16338655774229485, | |
| "grad_norm": 0.003212034935131669, | |
| "learning_rate": 9.581425372282891e-05, | |
| "loss": 0.0002, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16412922391385074, | |
| "grad_norm": 0.006715564522892237, | |
| "learning_rate": 9.576557395597236e-05, | |
| "loss": 0.0005, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.16487189008540662, | |
| "grad_norm": 0.009416691958904266, | |
| "learning_rate": 9.571662526580898e-05, | |
| "loss": 0.0007, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.1656145562569625, | |
| "grad_norm": 0.005933799315243959, | |
| "learning_rate": 9.566740793996637e-05, | |
| "loss": 0.0003, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.1663572224285184, | |
| "grad_norm": 0.0062180450186133385, | |
| "learning_rate": 9.561792226765072e-05, | |
| "loss": 0.0004, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.16709988860007427, | |
| "grad_norm": 0.007876376621425152, | |
| "learning_rate": 9.5568168539645e-05, | |
| "loss": 0.0004, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.16784255477163015, | |
| "grad_norm": 0.06271418929100037, | |
| "learning_rate": 9.551814704830734e-05, | |
| "loss": 0.0017, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.16858522094318604, | |
| "grad_norm": 0.002005163347348571, | |
| "learning_rate": 9.546785808756926e-05, | |
| "loss": 0.0002, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.16932788711474192, | |
| "grad_norm": 0.022918762639164925, | |
| "learning_rate": 9.541730195293397e-05, | |
| "loss": 0.0004, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.1700705532862978, | |
| "grad_norm": 0.018031733110547066, | |
| "learning_rate": 9.53664789414746e-05, | |
| "loss": 0.0004, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.1708132194578537, | |
| "grad_norm": 0.0034406071063131094, | |
| "learning_rate": 9.53153893518325e-05, | |
| "loss": 0.0002, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.17155588562940957, | |
| "grad_norm": 0.03680207580327988, | |
| "learning_rate": 9.526403348421544e-05, | |
| "loss": 0.0004, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.17229855180096545, | |
| "grad_norm": 0.001635130844078958, | |
| "learning_rate": 9.521241164039589e-05, | |
| "loss": 0.0002, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.17304121797252134, | |
| "grad_norm": 0.012099578976631165, | |
| "learning_rate": 9.516052412370921e-05, | |
| "loss": 0.0004, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.17378388414407725, | |
| "grad_norm": 0.001356737338937819, | |
| "learning_rate": 9.51083712390519e-05, | |
| "loss": 0.0001, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.17452655031563313, | |
| "grad_norm": 0.003389824880287051, | |
| "learning_rate": 9.505595329287972e-05, | |
| "loss": 0.0001, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.17526921648718902, | |
| "grad_norm": 0.004536811728030443, | |
| "learning_rate": 9.500327059320606e-05, | |
| "loss": 0.0003, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.1760118826587449, | |
| "grad_norm": 0.2518163323402405, | |
| "learning_rate": 9.495032344959998e-05, | |
| "loss": 0.0022, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.17675454883030078, | |
| "grad_norm": 0.0022397220600396395, | |
| "learning_rate": 9.48971121731844e-05, | |
| "loss": 0.0002, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.17749721500185667, | |
| "grad_norm": 0.0010416822042316198, | |
| "learning_rate": 9.484363707663442e-05, | |
| "loss": 0.0001, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.17823988117341255, | |
| "grad_norm": 0.0013076276518404484, | |
| "learning_rate": 9.478989847417526e-05, | |
| "loss": 0.0001, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17898254734496843, | |
| "grad_norm": 0.0012382504064589739, | |
| "learning_rate": 9.473589668158061e-05, | |
| "loss": 0.0002, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.17972521351652432, | |
| "grad_norm": 0.0214553065598011, | |
| "learning_rate": 9.468163201617062e-05, | |
| "loss": 0.0006, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.1804678796880802, | |
| "grad_norm": 0.002527498174458742, | |
| "learning_rate": 9.462710479681019e-05, | |
| "loss": 0.0002, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.18121054585963609, | |
| "grad_norm": 0.011280801147222519, | |
| "learning_rate": 9.457231534390694e-05, | |
| "loss": 0.0003, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.18195321203119197, | |
| "grad_norm": 0.0018357493681833148, | |
| "learning_rate": 9.451726397940945e-05, | |
| "loss": 0.0002, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.18269587820274785, | |
| "grad_norm": 0.08694746345281601, | |
| "learning_rate": 9.446195102680531e-05, | |
| "loss": 0.0009, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.18343854437430376, | |
| "grad_norm": 0.04296145588159561, | |
| "learning_rate": 9.440637681111922e-05, | |
| "loss": 0.0006, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.18418121054585965, | |
| "grad_norm": 0.006709881592541933, | |
| "learning_rate": 9.435054165891109e-05, | |
| "loss": 0.0004, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.18492387671741553, | |
| "grad_norm": 0.003305165795609355, | |
| "learning_rate": 9.429444589827412e-05, | |
| "loss": 0.0002, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.18566654288897141, | |
| "grad_norm": 0.0106744933873415, | |
| "learning_rate": 9.423808985883289e-05, | |
| "loss": 0.0004, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1864092090605273, | |
| "grad_norm": 0.2720184326171875, | |
| "learning_rate": 9.418147387174139e-05, | |
| "loss": 0.0074, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.18715187523208318, | |
| "grad_norm": 0.005303042940795422, | |
| "learning_rate": 9.412459826968108e-05, | |
| "loss": 0.0003, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.18789454140363906, | |
| "grad_norm": 0.0023732127156108618, | |
| "learning_rate": 9.406746338685895e-05, | |
| "loss": 0.0001, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.18863720757519495, | |
| "grad_norm": 0.006592648569494486, | |
| "learning_rate": 9.401006955900556e-05, | |
| "loss": 0.0002, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.18937987374675083, | |
| "grad_norm": 0.0008024009293876588, | |
| "learning_rate": 9.395241712337307e-05, | |
| "loss": 0.0001, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.19012253991830672, | |
| "grad_norm": 0.002137925708666444, | |
| "learning_rate": 9.389450641873323e-05, | |
| "loss": 0.0002, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.1908652060898626, | |
| "grad_norm": 0.005445448216050863, | |
| "learning_rate": 9.38363377853754e-05, | |
| "loss": 0.0002, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.19160787226141848, | |
| "grad_norm": 0.0010594564955681562, | |
| "learning_rate": 9.377791156510455e-05, | |
| "loss": 0.0001, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.19235053843297437, | |
| "grad_norm": 0.0036199286114424467, | |
| "learning_rate": 9.371922810123929e-05, | |
| "loss": 0.0002, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.19309320460453028, | |
| "grad_norm": 0.0007854366558603942, | |
| "learning_rate": 9.36602877386098e-05, | |
| "loss": 0.0001, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.19383587077608616, | |
| "grad_norm": 0.0012193581787869334, | |
| "learning_rate": 9.360109082355582e-05, | |
| "loss": 0.0001, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.19457853694764204, | |
| "grad_norm": 0.009622081182897091, | |
| "learning_rate": 9.354163770392461e-05, | |
| "loss": 0.0005, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.19532120311919793, | |
| "grad_norm": 0.0011145096505060792, | |
| "learning_rate": 9.348192872906896e-05, | |
| "loss": 0.0001, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.1960638692907538, | |
| "grad_norm": 0.004288391210138798, | |
| "learning_rate": 9.342196424984504e-05, | |
| "loss": 0.0001, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.1968065354623097, | |
| "grad_norm": 0.0054808189161121845, | |
| "learning_rate": 9.33617446186104e-05, | |
| "loss": 0.0003, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.19754920163386558, | |
| "grad_norm": 0.0028313531074672937, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 0.0002, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.19829186780542146, | |
| "grad_norm": 0.008807774633169174, | |
| "learning_rate": 9.324054131703371e-05, | |
| "loss": 0.0004, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.19903453397697735, | |
| "grad_norm": 0.008937092497944832, | |
| "learning_rate": 9.317955835889494e-05, | |
| "loss": 0.0002, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.19977720014853323, | |
| "grad_norm": 0.001885921461507678, | |
| "learning_rate": 9.311832167314787e-05, | |
| "loss": 0.0002, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.2005198663200891, | |
| "grad_norm": 0.0014201418962329626, | |
| "learning_rate": 9.305683161962569e-05, | |
| "loss": 0.0001, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.201262532491645, | |
| "grad_norm": 0.013191280886530876, | |
| "learning_rate": 9.299508855965039e-05, | |
| "loss": 0.0003, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.20200519866320088, | |
| "grad_norm": 1.5947636365890503, | |
| "learning_rate": 9.293309285603067e-05, | |
| "loss": 0.153, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.20274786483475676, | |
| "grad_norm": 0.005475195590406656, | |
| "learning_rate": 9.287084487305975e-05, | |
| "loss": 0.0002, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.20349053100631267, | |
| "grad_norm": 0.0020285584032535553, | |
| "learning_rate": 9.280834497651334e-05, | |
| "loss": 0.0002, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.20423319717786856, | |
| "grad_norm": 0.0018257065676152706, | |
| "learning_rate": 9.274559353364734e-05, | |
| "loss": 0.0001, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.20497586334942444, | |
| "grad_norm": 0.01402269210666418, | |
| "learning_rate": 9.268259091319582e-05, | |
| "loss": 0.0006, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.20571852952098033, | |
| "grad_norm": 0.05442607030272484, | |
| "learning_rate": 9.261933748536878e-05, | |
| "loss": 0.0006, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.2064611956925362, | |
| "grad_norm": 0.002209064783528447, | |
| "learning_rate": 9.255583362184999e-05, | |
| "loss": 0.0002, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.2072038618640921, | |
| "grad_norm": 0.019024129956960678, | |
| "learning_rate": 9.24920796957948e-05, | |
| "loss": 0.0003, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.20794652803564798, | |
| "grad_norm": 0.008843375369906425, | |
| "learning_rate": 9.242807608182795e-05, | |
| "loss": 0.0004, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.20868919420720386, | |
| "grad_norm": 0.11764897406101227, | |
| "learning_rate": 9.23638231560414e-05, | |
| "loss": 0.0015, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.20943186037875974, | |
| "grad_norm": 0.004572493955492973, | |
| "learning_rate": 9.229932129599205e-05, | |
| "loss": 0.0002, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.21017452655031563, | |
| "grad_norm": 0.006193062756210566, | |
| "learning_rate": 9.223457088069962e-05, | |
| "loss": 0.0003, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.2109171927218715, | |
| "grad_norm": 0.028236044570803642, | |
| "learning_rate": 9.21695722906443e-05, | |
| "loss": 0.0006, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.2116598588934274, | |
| "grad_norm": 0.012708684429526329, | |
| "learning_rate": 9.210432590776461e-05, | |
| "loss": 0.0008, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.21240252506498328, | |
| "grad_norm": 0.02218322828412056, | |
| "learning_rate": 9.203883211545517e-05, | |
| "loss": 0.0008, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.2131451912365392, | |
| "grad_norm": 0.015315636061131954, | |
| "learning_rate": 9.197309129856433e-05, | |
| "loss": 0.0005, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.21388785740809507, | |
| "grad_norm": 0.011535655707120895, | |
| "learning_rate": 9.190710384339203e-05, | |
| "loss": 0.0005, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.21463052357965096, | |
| "grad_norm": 0.013051263056695461, | |
| "learning_rate": 9.184087013768745e-05, | |
| "loss": 0.0003, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.21537318975120684, | |
| "grad_norm": 0.003687142627313733, | |
| "learning_rate": 9.177439057064683e-05, | |
| "loss": 0.0003, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.21611585592276272, | |
| "grad_norm": 0.007237662561237812, | |
| "learning_rate": 9.170766553291103e-05, | |
| "loss": 0.0003, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.2168585220943186, | |
| "grad_norm": 0.007928196340799332, | |
| "learning_rate": 9.164069541656337e-05, | |
| "loss": 0.0004, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.2176011882658745, | |
| "grad_norm": 0.04276131093502045, | |
| "learning_rate": 9.157348061512727e-05, | |
| "loss": 0.0005, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.21834385443743037, | |
| "grad_norm": 0.029039736837148666, | |
| "learning_rate": 9.150602152356395e-05, | |
| "loss": 0.0006, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.21908652060898626, | |
| "grad_norm": 0.08352446556091309, | |
| "learning_rate": 9.143831853827009e-05, | |
| "loss": 0.0015, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.21982918678054214, | |
| "grad_norm": 0.12498286366462708, | |
| "learning_rate": 9.137037205707552e-05, | |
| "loss": 0.0008, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.22057185295209802, | |
| "grad_norm": 0.012141775339841843, | |
| "learning_rate": 9.130218247924092e-05, | |
| "loss": 0.0005, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.2213145191236539, | |
| "grad_norm": 0.007531680166721344, | |
| "learning_rate": 9.123375020545535e-05, | |
| "loss": 0.0003, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.2220571852952098, | |
| "grad_norm": 0.009885065257549286, | |
| "learning_rate": 9.116507563783403e-05, | |
| "loss": 0.0004, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.2227998514667657, | |
| "grad_norm": 0.26482439041137695, | |
| "learning_rate": 9.109615917991591e-05, | |
| "loss": 0.0014, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2235425176383216, | |
| "grad_norm": 0.00270358519628644, | |
| "learning_rate": 9.102700123666132e-05, | |
| "loss": 0.0002, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.22428518380987747, | |
| "grad_norm": 1.3726491928100586, | |
| "learning_rate": 9.09576022144496e-05, | |
| "loss": 0.0146, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.22502784998143335, | |
| "grad_norm": 0.2558269500732422, | |
| "learning_rate": 9.088796252107665e-05, | |
| "loss": 0.0018, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.22577051615298924, | |
| "grad_norm": 0.7685271501541138, | |
| "learning_rate": 9.08180825657526e-05, | |
| "loss": 0.0064, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.22651318232454512, | |
| "grad_norm": 0.010196542367339134, | |
| "learning_rate": 9.07479627590994e-05, | |
| "loss": 0.0004, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.227255848496101, | |
| "grad_norm": 0.0029280134476721287, | |
| "learning_rate": 9.067760351314838e-05, | |
| "loss": 0.0002, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.2279985146676569, | |
| "grad_norm": 0.0070107802748680115, | |
| "learning_rate": 9.060700524133785e-05, | |
| "loss": 0.0003, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.22874118083921277, | |
| "grad_norm": 0.014984749257564545, | |
| "learning_rate": 9.053616835851062e-05, | |
| "loss": 0.0003, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.22948384701076865, | |
| "grad_norm": 0.0027229287661612034, | |
| "learning_rate": 9.046509328091166e-05, | |
| "loss": 0.0002, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.23022651318232454, | |
| "grad_norm": 0.006428726948797703, | |
| "learning_rate": 9.039378042618556e-05, | |
| "loss": 0.0003, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.23096917935388042, | |
| "grad_norm": 0.007419643457978964, | |
| "learning_rate": 9.032223021337414e-05, | |
| "loss": 0.0004, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.2317118455254363, | |
| "grad_norm": 0.0074731167405843735, | |
| "learning_rate": 9.025044306291392e-05, | |
| "loss": 0.0002, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.2324545116969922, | |
| "grad_norm": 0.0038742341566830873, | |
| "learning_rate": 9.017841939663374e-05, | |
| "loss": 0.0002, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.2331971778685481, | |
| "grad_norm": 0.010770438238978386, | |
| "learning_rate": 9.01061596377522e-05, | |
| "loss": 0.0004, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.23393984404010398, | |
| "grad_norm": 0.009036507457494736, | |
| "learning_rate": 9.003366421087521e-05, | |
| "loss": 0.0003, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.23468251021165987, | |
| "grad_norm": 0.008864130824804306, | |
| "learning_rate": 8.996093354199349e-05, | |
| "loss": 0.0003, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.23542517638321575, | |
| "grad_norm": 0.00652578379958868, | |
| "learning_rate": 8.988796805848007e-05, | |
| "loss": 0.0003, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.23616784255477163, | |
| "grad_norm": 0.006959845311939716, | |
| "learning_rate": 8.981476818908778e-05, | |
| "loss": 0.0002, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.23691050872632752, | |
| "grad_norm": 0.0059072221629321575, | |
| "learning_rate": 8.974133436394673e-05, | |
| "loss": 0.0003, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.2376531748978834, | |
| "grad_norm": 0.005236865486949682, | |
| "learning_rate": 8.966766701456177e-05, | |
| "loss": 0.0003, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.23839584106943928, | |
| "grad_norm": 0.01116922963410616, | |
| "learning_rate": 8.959376657380993e-05, | |
| "loss": 0.0003, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.23913850724099517, | |
| "grad_norm": 0.02754572220146656, | |
| "learning_rate": 8.951963347593797e-05, | |
| "loss": 0.0011, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.23988117341255105, | |
| "grad_norm": 0.05620993301272392, | |
| "learning_rate": 8.944526815655974e-05, | |
| "loss": 0.0008, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.24062383958410694, | |
| "grad_norm": 0.013477266766130924, | |
| "learning_rate": 8.937067105265362e-05, | |
| "loss": 0.0005, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.24136650575566282, | |
| "grad_norm": 0.008718527853488922, | |
| "learning_rate": 8.929584260256004e-05, | |
| "loss": 0.0002, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2421091719272187, | |
| "grad_norm": 0.0029952102340757847, | |
| "learning_rate": 8.922078324597879e-05, | |
| "loss": 0.0002, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.24285183809877461, | |
| "grad_norm": 0.19847828149795532, | |
| "learning_rate": 8.914549342396652e-05, | |
| "loss": 0.0012, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.2435945042703305, | |
| "grad_norm": 0.0039433506317436695, | |
| "learning_rate": 8.906997357893412e-05, | |
| "loss": 0.0002, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.24433717044188638, | |
| "grad_norm": 0.0013001116458326578, | |
| "learning_rate": 8.899422415464409e-05, | |
| "loss": 0.0001, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.24507983661344226, | |
| "grad_norm": 0.029752757400274277, | |
| "learning_rate": 8.891824559620801e-05, | |
| "loss": 0.0008, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.24582250278499815, | |
| "grad_norm": 0.009903517551720142, | |
| "learning_rate": 8.884203835008382e-05, | |
| "loss": 0.0003, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.24656516895655403, | |
| "grad_norm": 0.0017488920129835606, | |
| "learning_rate": 8.87656028640733e-05, | |
| "loss": 0.0002, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.24730783512810992, | |
| "grad_norm": 0.001140685984864831, | |
| "learning_rate": 8.868893958731937e-05, | |
| "loss": 0.0001, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.2480505012996658, | |
| "grad_norm": 0.001769506954587996, | |
| "learning_rate": 8.861204897030346e-05, | |
| "loss": 0.0002, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.24879316747122168, | |
| "grad_norm": 0.034780845046043396, | |
| "learning_rate": 8.853493146484291e-05, | |
| "loss": 0.0002, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.24953583364277757, | |
| "grad_norm": 0.0031120367348194122, | |
| "learning_rate": 8.845758752408826e-05, | |
| "loss": 0.0002, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.2502784998143335, | |
| "grad_norm": 0.0008751750574447215, | |
| "learning_rate": 8.838001760252059e-05, | |
| "loss": 0.0001, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.2502784998143335, | |
| "eval_loss": 8.811052975943312e-05, | |
| "eval_runtime": 190.5265, | |
| "eval_samples_per_second": 5.952, | |
| "eval_steps_per_second": 2.976, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.25102116598588936, | |
| "grad_norm": 0.001558976829983294, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 0.0001, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.25176383215744524, | |
| "grad_norm": 0.0019584286492317915, | |
| "learning_rate": 8.822420164150739e-05, | |
| "loss": 0.0002, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.25250649832900113, | |
| "grad_norm": 0.0017735590226948261, | |
| "learning_rate": 8.814595651765277e-05, | |
| "loss": 0.0002, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.253249164500557, | |
| "grad_norm": 0.0010270827915519476, | |
| "learning_rate": 8.806748724416156e-05, | |
| "loss": 0.0001, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.2539918306721129, | |
| "grad_norm": 0.0006894596735946834, | |
| "learning_rate": 8.798879428212747e-05, | |
| "loss": 0.0001, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.2547344968436688, | |
| "grad_norm": 0.8450746536254883, | |
| "learning_rate": 8.790987809395856e-05, | |
| "loss": 0.0079, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.25547716301522466, | |
| "grad_norm": 0.0008732525166124105, | |
| "learning_rate": 8.783073914337466e-05, | |
| "loss": 0.0001, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.25621982918678055, | |
| "grad_norm": 0.001687792013399303, | |
| "learning_rate": 8.775137789540446e-05, | |
| "loss": 0.0002, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.25696249535833643, | |
| "grad_norm": 0.0013438455061987042, | |
| "learning_rate": 8.767179481638303e-05, | |
| "loss": 0.0002, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.2577051615298923, | |
| "grad_norm": 0.0013538196217268705, | |
| "learning_rate": 8.759199037394887e-05, | |
| "loss": 0.0001, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.2584478277014482, | |
| "grad_norm": 0.0015409106854349375, | |
| "learning_rate": 8.751196503704123e-05, | |
| "loss": 0.0001, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.2591904938730041, | |
| "grad_norm": 0.006490611936897039, | |
| "learning_rate": 8.743171927589737e-05, | |
| "loss": 0.0002, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.25993316004455996, | |
| "grad_norm": 0.0011190741788595915, | |
| "learning_rate": 8.73512535620498e-05, | |
| "loss": 0.0001, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.26067582621611585, | |
| "grad_norm": 0.7256274819374084, | |
| "learning_rate": 8.727056836832348e-05, | |
| "loss": 0.0483, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.26141849238767173, | |
| "grad_norm": 0.0036242681089788675, | |
| "learning_rate": 8.718966416883306e-05, | |
| "loss": 0.0002, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.2621611585592276, | |
| "grad_norm": 0.002452226122841239, | |
| "learning_rate": 8.710854143898008e-05, | |
| "loss": 0.0002, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.2629038247307835, | |
| "grad_norm": 0.024639679118990898, | |
| "learning_rate": 8.702720065545024e-05, | |
| "loss": 0.0002, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2636464909023394, | |
| "grad_norm": 0.0007104252581484616, | |
| "learning_rate": 8.694564229621046e-05, | |
| "loss": 0.0001, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.26438915707389526, | |
| "grad_norm": 0.0009827159810811281, | |
| "learning_rate": 8.68638668405062e-05, | |
| "loss": 0.0002, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.26513182324545115, | |
| "grad_norm": 0.014522520825266838, | |
| "learning_rate": 8.678187476885864e-05, | |
| "loss": 0.0003, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.26587448941700703, | |
| "grad_norm": 0.004931020084768534, | |
| "learning_rate": 8.669966656306176e-05, | |
| "loss": 0.0002, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.2666171555885629, | |
| "grad_norm": 0.005294305272400379, | |
| "learning_rate": 8.661724270617962e-05, | |
| "loss": 0.0003, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.26735982176011885, | |
| "grad_norm": 0.0027490004431456327, | |
| "learning_rate": 8.653460368254338e-05, | |
| "loss": 0.0002, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.26810248793167474, | |
| "grad_norm": 0.0015370560577139258, | |
| "learning_rate": 8.645174997774864e-05, | |
| "loss": 0.0002, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2688451541032306, | |
| "grad_norm": 0.001747027155943215, | |
| "learning_rate": 8.636868207865244e-05, | |
| "loss": 0.0002, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2695878202747865, | |
| "grad_norm": 0.07692822813987732, | |
| "learning_rate": 8.628540047337045e-05, | |
| "loss": 0.0025, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.2703304864463424, | |
| "grad_norm": 0.003272181609645486, | |
| "learning_rate": 8.620190565127413e-05, | |
| "loss": 0.0002, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.27107315261789827, | |
| "grad_norm": 0.002899078419432044, | |
| "learning_rate": 8.611819810298778e-05, | |
| "loss": 0.0002, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.27181581878945416, | |
| "grad_norm": 0.0023670706432312727, | |
| "learning_rate": 8.603427832038574e-05, | |
| "loss": 0.0002, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.27255848496101004, | |
| "grad_norm": 0.006397952791303396, | |
| "learning_rate": 8.595014679658941e-05, | |
| "loss": 0.0004, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.2733011511325659, | |
| "grad_norm": 0.0371289923787117, | |
| "learning_rate": 8.586580402596446e-05, | |
| "loss": 0.0009, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.2740438173041218, | |
| "grad_norm": 0.0018306565470993519, | |
| "learning_rate": 8.578125050411787e-05, | |
| "loss": 0.0002, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.2747864834756777, | |
| "grad_norm": 0.003529072506353259, | |
| "learning_rate": 8.569648672789497e-05, | |
| "loss": 0.0002, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2755291496472336, | |
| "grad_norm": 0.008671244606375694, | |
| "learning_rate": 8.561151319537655e-05, | |
| "loss": 0.0003, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.27627181581878946, | |
| "grad_norm": 0.0039878771640360355, | |
| "learning_rate": 8.552633040587606e-05, | |
| "loss": 0.0002, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.27701448199034534, | |
| "grad_norm": 0.025948379188776016, | |
| "learning_rate": 8.544093885993643e-05, | |
| "loss": 0.0004, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.2777571481619012, | |
| "grad_norm": 0.00219323905184865, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 0.0002, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.2784998143334571, | |
| "grad_norm": 0.015180499292910099, | |
| "learning_rate": 8.526953150704229e-05, | |
| "loss": 0.0005, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.279242480505013, | |
| "grad_norm": 0.012962475419044495, | |
| "learning_rate": 8.518351670729529e-05, | |
| "loss": 0.0004, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2799851466765689, | |
| "grad_norm": 0.007179844658821821, | |
| "learning_rate": 8.509729516551841e-05, | |
| "loss": 0.0002, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.28072781284812476, | |
| "grad_norm": 0.06565556675195694, | |
| "learning_rate": 8.501086738835843e-05, | |
| "loss": 0.0012, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.28147047901968064, | |
| "grad_norm": 0.009149481542408466, | |
| "learning_rate": 8.492423388367403e-05, | |
| "loss": 0.0004, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.2822131451912365, | |
| "grad_norm": 0.007264537271112204, | |
| "learning_rate": 8.483739516053276e-05, | |
| "loss": 0.0003, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.2829558113627924, | |
| "grad_norm": 0.004194353707134724, | |
| "learning_rate": 8.475035172920804e-05, | |
| "loss": 0.0002, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2836984775343483, | |
| "grad_norm": 0.0032033442985266447, | |
| "learning_rate": 8.466310410117622e-05, | |
| "loss": 0.0003, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.2844411437059042, | |
| "grad_norm": 0.004893294535577297, | |
| "learning_rate": 8.457565278911348e-05, | |
| "loss": 0.0005, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.28518380987746006, | |
| "grad_norm": 0.13098792731761932, | |
| "learning_rate": 8.448799830689289e-05, | |
| "loss": 0.0042, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.28592647604901594, | |
| "grad_norm": 0.0038283192552626133, | |
| "learning_rate": 8.440014116958139e-05, | |
| "loss": 0.0002, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.2866691422205718, | |
| "grad_norm": 0.0029847463592886925, | |
| "learning_rate": 8.43120818934367e-05, | |
| "loss": 0.0002, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.28741180839212777, | |
| "grad_norm": 0.0012196919415146112, | |
| "learning_rate": 8.422382099590434e-05, | |
| "loss": 0.0001, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.28815447456368365, | |
| "grad_norm": 0.0022556744515895844, | |
| "learning_rate": 8.413535899561463e-05, | |
| "loss": 0.0002, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.28889714073523953, | |
| "grad_norm": 0.0016208338784053922, | |
| "learning_rate": 8.404669641237952e-05, | |
| "loss": 0.0002, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.2896398069067954, | |
| "grad_norm": 0.0015344212297350168, | |
| "learning_rate": 8.395783376718966e-05, | |
| "loss": 0.0002, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2903824730783513, | |
| "grad_norm": 0.0014779121847823262, | |
| "learning_rate": 8.386877158221125e-05, | |
| "loss": 0.0002, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.2911251392499072, | |
| "grad_norm": 0.0007657014648430049, | |
| "learning_rate": 8.377951038078302e-05, | |
| "loss": 0.0001, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.29186780542146307, | |
| "grad_norm": 0.6582887768745422, | |
| "learning_rate": 8.369005068741314e-05, | |
| "loss": 0.0099, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.29261047159301895, | |
| "grad_norm": 0.0017920684767886996, | |
| "learning_rate": 8.360039302777612e-05, | |
| "loss": 0.0002, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.29335313776457483, | |
| "grad_norm": 0.005679702386260033, | |
| "learning_rate": 8.35105379287098e-05, | |
| "loss": 0.0002, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.2940958039361307, | |
| "grad_norm": 0.0127026392146945, | |
| "learning_rate": 8.342048591821212e-05, | |
| "loss": 0.0003, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.2948384701076866, | |
| "grad_norm": 0.0027185885701328516, | |
| "learning_rate": 8.333023752543816e-05, | |
| "loss": 0.0002, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.2955811362792425, | |
| "grad_norm": 0.02255573309957981, | |
| "learning_rate": 8.323979328069689e-05, | |
| "loss": 0.0004, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.29632380245079837, | |
| "grad_norm": 0.005555687937885523, | |
| "learning_rate": 8.314915371544822e-05, | |
| "loss": 0.0002, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.29706646862235425, | |
| "grad_norm": 0.001257200026884675, | |
| "learning_rate": 8.305831936229966e-05, | |
| "loss": 0.0001, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.29780913479391014, | |
| "grad_norm": 0.07175086438655853, | |
| "learning_rate": 8.296729075500344e-05, | |
| "loss": 0.0013, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.298551800965466, | |
| "grad_norm": 0.0019333354430273175, | |
| "learning_rate": 8.28760684284532e-05, | |
| "loss": 0.0002, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.2992944671370219, | |
| "grad_norm": 0.17573364078998566, | |
| "learning_rate": 8.278465291868083e-05, | |
| "loss": 0.0023, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.3000371333085778, | |
| "grad_norm": 0.05853782966732979, | |
| "learning_rate": 8.269304476285349e-05, | |
| "loss": 0.0015, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.30077979948013367, | |
| "grad_norm": 0.09041046351194382, | |
| "learning_rate": 8.260124449927028e-05, | |
| "loss": 0.0022, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.30152246565168955, | |
| "grad_norm": 0.1666203737258911, | |
| "learning_rate": 8.250925266735918e-05, | |
| "loss": 0.002, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.30226513182324544, | |
| "grad_norm": 0.04917696863412857, | |
| "learning_rate": 8.241706980767381e-05, | |
| "loss": 0.0007, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.3030077979948013, | |
| "grad_norm": 0.0021273689344525337, | |
| "learning_rate": 8.232469646189032e-05, | |
| "loss": 0.0001, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.3037504641663572, | |
| "grad_norm": 0.0016986045520752668, | |
| "learning_rate": 8.223213317280419e-05, | |
| "loss": 0.0001, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.3044931303379131, | |
| "grad_norm": 0.03786956146359444, | |
| "learning_rate": 8.213938048432697e-05, | |
| "loss": 0.0006, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.30523579650946897, | |
| "grad_norm": 0.0048659974709153175, | |
| "learning_rate": 8.204643894148318e-05, | |
| "loss": 0.0002, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.30597846268102485, | |
| "grad_norm": 0.007152218371629715, | |
| "learning_rate": 8.195330909040708e-05, | |
| "loss": 0.0002, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.30672112885258074, | |
| "grad_norm": 0.0123271644115448, | |
| "learning_rate": 8.185999147833943e-05, | |
| "loss": 0.0004, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.3074637950241367, | |
| "grad_norm": 0.001973194070160389, | |
| "learning_rate": 8.176648665362425e-05, | |
| "loss": 0.0001, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.30820646119569256, | |
| "grad_norm": 0.007642180658876896, | |
| "learning_rate": 8.167279516570575e-05, | |
| "loss": 0.0005, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.30894912736724844, | |
| "grad_norm": 0.02631974406540394, | |
| "learning_rate": 8.157891756512488e-05, | |
| "loss": 0.0004, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.3096917935388043, | |
| "grad_norm": 0.0035103110130876303, | |
| "learning_rate": 8.148485440351629e-05, | |
| "loss": 0.0002, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.3104344597103602, | |
| "grad_norm": 0.004595728125423193, | |
| "learning_rate": 8.139060623360493e-05, | |
| "loss": 0.0002, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.3111771258819161, | |
| "grad_norm": 0.002736350754275918, | |
| "learning_rate": 8.129617360920296e-05, | |
| "loss": 0.0002, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.311919792053472, | |
| "grad_norm": 0.0011626658961176872, | |
| "learning_rate": 8.120155708520636e-05, | |
| "loss": 0.0001, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.31266245822502786, | |
| "grad_norm": 0.004247245844453573, | |
| "learning_rate": 8.110675721759171e-05, | |
| "loss": 0.0002, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.31340512439658375, | |
| "grad_norm": 0.001799743971787393, | |
| "learning_rate": 8.1011774563413e-05, | |
| "loss": 0.0001, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.31414779056813963, | |
| "grad_norm": 0.011404055170714855, | |
| "learning_rate": 8.091660968079826e-05, | |
| "loss": 0.0004, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.3148904567396955, | |
| "grad_norm": 0.00542003707960248, | |
| "learning_rate": 8.082126312894626e-05, | |
| "loss": 0.0003, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3156331229112514, | |
| "grad_norm": 0.1897122859954834, | |
| "learning_rate": 8.072573546812338e-05, | |
| "loss": 0.0017, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3163757890828073, | |
| "grad_norm": 0.004132653120905161, | |
| "learning_rate": 8.063002725966015e-05, | |
| "loss": 0.0002, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.31711845525436316, | |
| "grad_norm": 0.0019162169191986322, | |
| "learning_rate": 8.0534139065948e-05, | |
| "loss": 0.0002, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.31786112142591905, | |
| "grad_norm": 0.004962231498211622, | |
| "learning_rate": 8.043807145043604e-05, | |
| "loss": 0.0002, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.31860378759747493, | |
| "grad_norm": 0.0016735494136810303, | |
| "learning_rate": 8.034182497762762e-05, | |
| "loss": 0.0001, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3193464537690308, | |
| "grad_norm": 0.0013713808730244637, | |
| "learning_rate": 8.024540021307708e-05, | |
| "loss": 0.0001, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3200891199405867, | |
| "grad_norm": 0.00254115485586226, | |
| "learning_rate": 8.014879772338649e-05, | |
| "loss": 0.0002, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3208317861121426, | |
| "grad_norm": 0.0037052470725029707, | |
| "learning_rate": 8.005201807620215e-05, | |
| "loss": 0.0002, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.32157445228369846, | |
| "grad_norm": 0.0029492946341633797, | |
| "learning_rate": 7.995506184021142e-05, | |
| "loss": 0.0002, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.32231711845525435, | |
| "grad_norm": 0.001130992779508233, | |
| "learning_rate": 7.985792958513931e-05, | |
| "loss": 0.0001, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.32305978462681023, | |
| "grad_norm": 0.001874015899375081, | |
| "learning_rate": 7.976062188174512e-05, | |
| "loss": 0.0001, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.3238024507983661, | |
| "grad_norm": 0.007868077605962753, | |
| "learning_rate": 7.966313930181912e-05, | |
| "loss": 0.0002, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.324545116969922, | |
| "grad_norm": 0.0044991993345320225, | |
| "learning_rate": 7.956548241817912e-05, | |
| "loss": 0.0002, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.3252877831414779, | |
| "grad_norm": 0.0006172333378344774, | |
| "learning_rate": 7.946765180466724e-05, | |
| "loss": 0.0001, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.32603044931303377, | |
| "grad_norm": 0.0030133500695228577, | |
| "learning_rate": 7.936964803614641e-05, | |
| "loss": 0.0001, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3267731154845897, | |
| "grad_norm": 0.006758226081728935, | |
| "learning_rate": 7.927147168849704e-05, | |
| "loss": 0.0002, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3275157816561456, | |
| "grad_norm": 0.004247500095516443, | |
| "learning_rate": 7.91731233386136e-05, | |
| "loss": 0.0003, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.32825844782770147, | |
| "grad_norm": 0.005194083787500858, | |
| "learning_rate": 7.907460356440133e-05, | |
| "loss": 0.0002, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.32900111399925736, | |
| "grad_norm": 0.002930448157712817, | |
| "learning_rate": 7.897591294477275e-05, | |
| "loss": 0.0001, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.32974378017081324, | |
| "grad_norm": 0.0016131598968058825, | |
| "learning_rate": 7.887705205964426e-05, | |
| "loss": 0.0001, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.3304864463423691, | |
| "grad_norm": 0.002172644715756178, | |
| "learning_rate": 7.877802148993277e-05, | |
| "loss": 0.0001, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.331229112513925, | |
| "grad_norm": 0.027609726414084435, | |
| "learning_rate": 7.86788218175523e-05, | |
| "loss": 0.0006, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3319717786854809, | |
| "grad_norm": 0.015412206761538982, | |
| "learning_rate": 7.857945362541053e-05, | |
| "loss": 0.0005, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.3327144448570368, | |
| "grad_norm": 0.002325060311704874, | |
| "learning_rate": 7.847991749740533e-05, | |
| "loss": 0.0001, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.33345711102859266, | |
| "grad_norm": 0.0031837387941777706, | |
| "learning_rate": 7.838021401842144e-05, | |
| "loss": 0.0001, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.33419977720014854, | |
| "grad_norm": 0.003575611859560013, | |
| "learning_rate": 7.828034377432693e-05, | |
| "loss": 0.0001, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3349424433717044, | |
| "grad_norm": 0.0006315786740742624, | |
| "learning_rate": 7.818030735196984e-05, | |
| "loss": 0.0001, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.3356851095432603, | |
| "grad_norm": 0.005519942846149206, | |
| "learning_rate": 7.808010533917465e-05, | |
| "loss": 0.0002, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.3364277757148162, | |
| "grad_norm": 0.0023312256671488285, | |
| "learning_rate": 7.797973832473889e-05, | |
| "loss": 0.0002, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.3371704418863721, | |
| "grad_norm": 0.0030489324126392603, | |
| "learning_rate": 7.787920689842964e-05, | |
| "loss": 0.0001, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.33791310805792796, | |
| "grad_norm": 0.0010298596462234855, | |
| "learning_rate": 7.777851165098012e-05, | |
| "loss": 0.0001, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.33865577422948384, | |
| "grad_norm": 0.0006818081019446254, | |
| "learning_rate": 7.767765317408613e-05, | |
| "loss": 0.0001, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3393984404010397, | |
| "grad_norm": 0.0019495798042044044, | |
| "learning_rate": 7.757663206040264e-05, | |
| "loss": 0.0001, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.3401411065725956, | |
| "grad_norm": 0.012790280394256115, | |
| "learning_rate": 7.74754489035403e-05, | |
| "loss": 0.0002, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.3408837727441515, | |
| "grad_norm": 0.0026799726765602827, | |
| "learning_rate": 7.737410429806196e-05, | |
| "loss": 0.0001, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.3416264389157074, | |
| "grad_norm": 0.014524313621222973, | |
| "learning_rate": 7.727259883947913e-05, | |
| "loss": 0.0004, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.34236910508726326, | |
| "grad_norm": 0.0005140351131558418, | |
| "learning_rate": 7.71709331242485e-05, | |
| "loss": 0.0001, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.34311177125881914, | |
| "grad_norm": 0.003770928829908371, | |
| "learning_rate": 7.706910774976849e-05, | |
| "loss": 0.0001, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.343854437430375, | |
| "grad_norm": 0.0012231201399117708, | |
| "learning_rate": 7.696712331437565e-05, | |
| "loss": 0.0001, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.3445971036019309, | |
| "grad_norm": 0.005815563257783651, | |
| "learning_rate": 7.68649804173412e-05, | |
| "loss": 0.0001, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.3453397697734868, | |
| "grad_norm": 0.0009362814598716795, | |
| "learning_rate": 7.676267965886752e-05, | |
| "loss": 0.0001, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3460824359450427, | |
| "grad_norm": 0.005367044825106859, | |
| "learning_rate": 7.666022164008457e-05, | |
| "loss": 0.0001, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.3468251021165986, | |
| "grad_norm": 0.0023662326857447624, | |
| "learning_rate": 7.655760696304641e-05, | |
| "loss": 0.0001, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.3475677682881545, | |
| "grad_norm": 0.0014945007860660553, | |
| "learning_rate": 7.645483623072763e-05, | |
| "loss": 0.0001, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.3483104344597104, | |
| "grad_norm": 0.0008832917083054781, | |
| "learning_rate": 7.635191004701981e-05, | |
| "loss": 0.0001, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.34905310063126627, | |
| "grad_norm": 0.001919624744914472, | |
| "learning_rate": 7.6248829016728e-05, | |
| "loss": 0.0001, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.34979576680282215, | |
| "grad_norm": 0.003197997808456421, | |
| "learning_rate": 7.614559374556715e-05, | |
| "loss": 0.0003, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.35053843297437803, | |
| "grad_norm": 0.004050462041050196, | |
| "learning_rate": 7.60422048401585e-05, | |
| "loss": 0.0001, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3512810991459339, | |
| "grad_norm": 0.12640826404094696, | |
| "learning_rate": 7.593866290802608e-05, | |
| "loss": 0.0009, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.3520237653174898, | |
| "grad_norm": 0.003920804709196091, | |
| "learning_rate": 7.583496855759316e-05, | |
| "loss": 0.0002, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.3527664314890457, | |
| "grad_norm": 0.0023814719170331955, | |
| "learning_rate": 7.573112239817857e-05, | |
| "loss": 0.0001, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.35350909766060157, | |
| "grad_norm": 0.011592867784202099, | |
| "learning_rate": 7.562712503999327e-05, | |
| "loss": 0.0002, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.35425176383215745, | |
| "grad_norm": 0.0014887212309986353, | |
| "learning_rate": 7.552297709413658e-05, | |
| "loss": 0.0002, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.35499443000371333, | |
| "grad_norm": 0.0006836429820396006, | |
| "learning_rate": 7.541867917259277e-05, | |
| "loss": 0.0001, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.3557370961752692, | |
| "grad_norm": 0.025796182453632355, | |
| "learning_rate": 7.531423188822737e-05, | |
| "loss": 0.0005, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3564797623468251, | |
| "grad_norm": 0.0009914631955325603, | |
| "learning_rate": 7.520963585478353e-05, | |
| "loss": 0.0001, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.357222428518381, | |
| "grad_norm": 0.0003263329854235053, | |
| "learning_rate": 7.510489168687851e-05, | |
| "loss": 0.0001, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.35796509468993687, | |
| "grad_norm": 0.0006598143372684717, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.0001, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.35870776086149275, | |
| "grad_norm": 0.0005369288846850395, | |
| "learning_rate": 7.489496141050258e-05, | |
| "loss": 0.0001, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.35945042703304864, | |
| "grad_norm": 0.00033336677006445825, | |
| "learning_rate": 7.478977653560397e-05, | |
| "loss": 0.0001, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.3601930932046045, | |
| "grad_norm": 0.0009709247970022261, | |
| "learning_rate": 7.468444599338151e-05, | |
| "loss": 0.0001, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3609357593761604, | |
| "grad_norm": 0.0009371594642288983, | |
| "learning_rate": 7.457897040276853e-05, | |
| "loss": 0.0001, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.3616784255477163, | |
| "grad_norm": 0.0009692166349850595, | |
| "learning_rate": 7.447335038355063e-05, | |
| "loss": 0.0001, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.36242109171927217, | |
| "grad_norm": 0.0006186399841681123, | |
| "learning_rate": 7.436758655636212e-05, | |
| "loss": 0.0001, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.36316375789082805, | |
| "grad_norm": 0.0017053054179996252, | |
| "learning_rate": 7.426167954268231e-05, | |
| "loss": 0.0001, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.36390642406238394, | |
| "grad_norm": 0.0015286095440387726, | |
| "learning_rate": 7.415562996483192e-05, | |
| "loss": 0.0001, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3646490902339398, | |
| "grad_norm": 0.004435093142092228, | |
| "learning_rate": 7.404943844596939e-05, | |
| "loss": 0.0003, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.3653917564054957, | |
| "grad_norm": 0.00381348910741508, | |
| "learning_rate": 7.394310561008717e-05, | |
| "loss": 0.0002, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.3661344225770516, | |
| "grad_norm": 0.0006017435807734728, | |
| "learning_rate": 7.38366320820082e-05, | |
| "loss": 0.0001, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.3668770887486075, | |
| "grad_norm": 0.00891191978007555, | |
| "learning_rate": 7.373001848738202e-05, | |
| "loss": 0.0003, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.3676197549201634, | |
| "grad_norm": 0.0004915704485028982, | |
| "learning_rate": 7.362326545268133e-05, | |
| "loss": 0.0001, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3683624210917193, | |
| "grad_norm": 0.0015488892095163465, | |
| "learning_rate": 7.351637360519813e-05, | |
| "loss": 0.0001, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.3691050872632752, | |
| "grad_norm": 0.002463052747771144, | |
| "learning_rate": 7.34093435730401e-05, | |
| "loss": 0.0002, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.36984775343483106, | |
| "grad_norm": 0.0014587301993742585, | |
| "learning_rate": 7.330217598512695e-05, | |
| "loss": 0.0001, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.37059041960638694, | |
| "grad_norm": 0.001909661223180592, | |
| "learning_rate": 7.319487147118663e-05, | |
| "loss": 0.0001, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.37133308577794283, | |
| "grad_norm": 0.0004529351135715842, | |
| "learning_rate": 7.308743066175172e-05, | |
| "loss": 0.0001, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3720757519494987, | |
| "grad_norm": 0.03147125244140625, | |
| "learning_rate": 7.297985418815563e-05, | |
| "loss": 0.0006, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.3728184181210546, | |
| "grad_norm": 0.0007625695434398949, | |
| "learning_rate": 7.287214268252904e-05, | |
| "loss": 0.0001, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.3735610842926105, | |
| "grad_norm": 0.003061138093471527, | |
| "learning_rate": 7.276429677779602e-05, | |
| "loss": 0.0001, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.37430375046416636, | |
| "grad_norm": 0.000722390366718173, | |
| "learning_rate": 7.265631710767041e-05, | |
| "loss": 0.0001, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.37504641663572225, | |
| "grad_norm": 0.0014981662388890982, | |
| "learning_rate": 7.254820430665206e-05, | |
| "loss": 0.0001, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.37578908280727813, | |
| "grad_norm": 0.0009839913109317422, | |
| "learning_rate": 7.243995901002312e-05, | |
| "loss": 0.0001, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.376531748978834, | |
| "grad_norm": 0.0012856576358899474, | |
| "learning_rate": 7.233158185384426e-05, | |
| "loss": 0.0001, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.3772744151503899, | |
| "grad_norm": 0.0018630792619660497, | |
| "learning_rate": 7.222307347495105e-05, | |
| "loss": 0.0002, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.3780170813219458, | |
| "grad_norm": 0.0028617579955607653, | |
| "learning_rate": 7.211443451095007e-05, | |
| "loss": 0.0002, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.37875974749350166, | |
| "grad_norm": 0.0007013090653344989, | |
| "learning_rate": 7.200566560021524e-05, | |
| "loss": 0.0001, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.37950241366505755, | |
| "grad_norm": 0.00415767403319478, | |
| "learning_rate": 7.18967673818841e-05, | |
| "loss": 0.0002, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.38024507983661343, | |
| "grad_norm": 0.006695980206131935, | |
| "learning_rate": 7.178774049585397e-05, | |
| "loss": 0.0002, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3809877460081693, | |
| "grad_norm": 0.00166626728605479, | |
| "learning_rate": 7.167858558277827e-05, | |
| "loss": 0.0001, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.3817304121797252, | |
| "grad_norm": 0.0006517748697660863, | |
| "learning_rate": 7.156930328406268e-05, | |
| "loss": 0.0001, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.3824730783512811, | |
| "grad_norm": 0.0009984513744711876, | |
| "learning_rate": 7.145989424186146e-05, | |
| "loss": 0.0001, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.38321574452283697, | |
| "grad_norm": 0.0003130779368802905, | |
| "learning_rate": 7.135035909907358e-05, | |
| "loss": 0.0001, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.38395841069439285, | |
| "grad_norm": 0.006763228215277195, | |
| "learning_rate": 7.124069849933903e-05, | |
| "loss": 0.0003, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.38470107686594873, | |
| "grad_norm": 0.0008593749371357262, | |
| "learning_rate": 7.113091308703498e-05, | |
| "loss": 0.0001, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.3854437430375046, | |
| "grad_norm": 0.0006246920092962682, | |
| "learning_rate": 7.102100350727201e-05, | |
| "loss": 0.0001, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.38618640920906055, | |
| "grad_norm": 0.0012244340032339096, | |
| "learning_rate": 7.091097040589032e-05, | |
| "loss": 0.0001, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.38692907538061644, | |
| "grad_norm": 0.0009561783517710865, | |
| "learning_rate": 7.080081442945596e-05, | |
| "loss": 0.0001, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.3876717415521723, | |
| "grad_norm": 0.0005518002435564995, | |
| "learning_rate": 7.069053622525696e-05, | |
| "loss": 0.0001, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.3884144077237282, | |
| "grad_norm": 0.006016881670802832, | |
| "learning_rate": 7.058013644129962e-05, | |
| "loss": 0.0002, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.3891570738952841, | |
| "grad_norm": 0.002434986876323819, | |
| "learning_rate": 7.046961572630462e-05, | |
| "loss": 0.0002, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.38989974006684, | |
| "grad_norm": 0.001255060895346105, | |
| "learning_rate": 7.035897472970329e-05, | |
| "loss": 0.0001, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.39064240623839586, | |
| "grad_norm": 0.007994726300239563, | |
| "learning_rate": 7.024821410163368e-05, | |
| "loss": 0.0002, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.39138507240995174, | |
| "grad_norm": 0.002365407533943653, | |
| "learning_rate": 7.013733449293687e-05, | |
| "loss": 0.0002, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.3921277385815076, | |
| "grad_norm": 0.000903936626855284, | |
| "learning_rate": 7.002633655515303e-05, | |
| "loss": 0.0001, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.3928704047530635, | |
| "grad_norm": 0.001310791471041739, | |
| "learning_rate": 6.99152209405177e-05, | |
| "loss": 0.0001, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.3936130709246194, | |
| "grad_norm": 0.0015525859780609608, | |
| "learning_rate": 6.980398830195785e-05, | |
| "loss": 0.0001, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3943557370961753, | |
| "grad_norm": 0.0004448418621905148, | |
| "learning_rate": 6.969263929308812e-05, | |
| "loss": 0.0001, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.39509840326773116, | |
| "grad_norm": 0.0003616507747210562, | |
| "learning_rate": 6.958117456820696e-05, | |
| "loss": 0.0001, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.39584106943928704, | |
| "grad_norm": 0.0032966274302452803, | |
| "learning_rate": 6.946959478229276e-05, | |
| "loss": 0.0001, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.3965837356108429, | |
| "grad_norm": 0.0006540945032611489, | |
| "learning_rate": 6.935790059100003e-05, | |
| "loss": 0.0001, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.3973264017823988, | |
| "grad_norm": 0.0009542697225697339, | |
| "learning_rate": 6.924609265065556e-05, | |
| "loss": 0.0001, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.3980690679539547, | |
| "grad_norm": 0.004549324978142977, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 0.0003, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.3988117341255106, | |
| "grad_norm": 0.0008191264350898564, | |
| "learning_rate": 6.902213815145655e-05, | |
| "loss": 0.0001, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.39955440029706646, | |
| "grad_norm": 0.0007376139401458204, | |
| "learning_rate": 6.890999290858214e-05, | |
| "loss": 0.0001, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.40029706646862234, | |
| "grad_norm": 0.0011172041995450854, | |
| "learning_rate": 6.87977365486084e-05, | |
| "loss": 0.0001, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.4010397326401782, | |
| "grad_norm": 0.0010884919902309775, | |
| "learning_rate": 6.868536973116552e-05, | |
| "loss": 0.0001, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4017823988117341, | |
| "grad_norm": 0.0008193852263502777, | |
| "learning_rate": 6.857289311653268e-05, | |
| "loss": 0.0001, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.40252506498329, | |
| "grad_norm": 0.0013226550072431564, | |
| "learning_rate": 6.846030736563422e-05, | |
| "loss": 0.0001, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.4032677311548459, | |
| "grad_norm": 0.0019562705419957638, | |
| "learning_rate": 6.834761314003584e-05, | |
| "loss": 0.0001, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.40401039732640176, | |
| "grad_norm": 0.0033758783247321844, | |
| "learning_rate": 6.82348111019406e-05, | |
| "loss": 0.0001, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.40475306349795764, | |
| "grad_norm": 0.0009793438948690891, | |
| "learning_rate": 6.812190191418508e-05, | |
| "loss": 0.0001, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.4054957296695135, | |
| "grad_norm": 0.001416007406078279, | |
| "learning_rate": 6.800888624023553e-05, | |
| "loss": 0.0001, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.40623839584106947, | |
| "grad_norm": 0.0026139297988265753, | |
| "learning_rate": 6.789576474418386e-05, | |
| "loss": 0.0002, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.40698106201262535, | |
| "grad_norm": 1.4912738800048828, | |
| "learning_rate": 6.778253809074385e-05, | |
| "loss": 0.1274, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.40772372818418123, | |
| "grad_norm": 0.0013667610473930836, | |
| "learning_rate": 6.766920694524714e-05, | |
| "loss": 0.0001, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.4084663943557371, | |
| "grad_norm": 0.633527934551239, | |
| "learning_rate": 6.755577197363944e-05, | |
| "loss": 0.0168, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.409209060527293, | |
| "grad_norm": 0.0013449821854010224, | |
| "learning_rate": 6.744223384247655e-05, | |
| "loss": 0.0001, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.4099517266988489, | |
| "grad_norm": 0.0064977309666574, | |
| "learning_rate": 6.732859321892037e-05, | |
| "loss": 0.0003, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.41069439287040477, | |
| "grad_norm": 0.0006492669344879687, | |
| "learning_rate": 6.721485077073518e-05, | |
| "loss": 0.0001, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.41143705904196065, | |
| "grad_norm": 0.4942440986633301, | |
| "learning_rate": 6.710100716628344e-05, | |
| "loss": 0.0183, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.41217972521351653, | |
| "grad_norm": 0.0004884397494606674, | |
| "learning_rate": 6.698706307452216e-05, | |
| "loss": 0.0001, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.4129223913850724, | |
| "grad_norm": 0.0014100876869633794, | |
| "learning_rate": 6.687301916499871e-05, | |
| "loss": 0.0002, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.4136650575566283, | |
| "grad_norm": 0.0025068807881325483, | |
| "learning_rate": 6.675887610784708e-05, | |
| "loss": 0.0001, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.4144077237281842, | |
| "grad_norm": 0.0007504708482883871, | |
| "learning_rate": 6.664463457378383e-05, | |
| "loss": 0.0001, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.41515038989974007, | |
| "grad_norm": 0.0008488527382723987, | |
| "learning_rate": 6.653029523410417e-05, | |
| "loss": 0.0001, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.41589305607129595, | |
| "grad_norm": 3.241758108139038, | |
| "learning_rate": 6.641585876067807e-05, | |
| "loss": 0.0741, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.41663572224285184, | |
| "grad_norm": 0.0009746011346578598, | |
| "learning_rate": 6.630132582594617e-05, | |
| "loss": 0.0001, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.4173783884144077, | |
| "grad_norm": 0.004342979751527309, | |
| "learning_rate": 6.618669710291606e-05, | |
| "loss": 0.0002, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.4181210545859636, | |
| "grad_norm": 0.03014206513762474, | |
| "learning_rate": 6.607197326515808e-05, | |
| "loss": 0.001, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.4188637207575195, | |
| "grad_norm": 0.03773445263504982, | |
| "learning_rate": 6.595715498680156e-05, | |
| "loss": 0.0005, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.41960638692907537, | |
| "grad_norm": 0.23458726704120636, | |
| "learning_rate": 6.584224294253069e-05, | |
| "loss": 0.0019, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.42034905310063125, | |
| "grad_norm": 0.01963810622692108, | |
| "learning_rate": 6.572723780758069e-05, | |
| "loss": 0.0003, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.42109171927218714, | |
| "grad_norm": 0.002442006254568696, | |
| "learning_rate": 6.56121402577338e-05, | |
| "loss": 0.0002, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.421834385443743, | |
| "grad_norm": 0.026274830102920532, | |
| "learning_rate": 6.549695096931527e-05, | |
| "loss": 0.0002, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.4225770516152989, | |
| "grad_norm": 0.0038544712588191032, | |
| "learning_rate": 6.538167061918941e-05, | |
| "loss": 0.0002, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.4233197177868548, | |
| "grad_norm": 0.019080875441432, | |
| "learning_rate": 6.526629988475567e-05, | |
| "loss": 0.0007, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.42406238395841067, | |
| "grad_norm": 0.01698448695242405, | |
| "learning_rate": 6.515083944394453e-05, | |
| "loss": 0.0003, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.42480505012996655, | |
| "grad_norm": 0.006882249377667904, | |
| "learning_rate": 6.503528997521366e-05, | |
| "loss": 0.0004, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.42554771630152244, | |
| "grad_norm": 0.003293162677437067, | |
| "learning_rate": 6.491965215754384e-05, | |
| "loss": 0.0002, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.4262903824730784, | |
| "grad_norm": 0.004137948155403137, | |
| "learning_rate": 6.4803926670435e-05, | |
| "loss": 0.0002, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.42703304864463426, | |
| "grad_norm": 0.035909973084926605, | |
| "learning_rate": 6.468811419390222e-05, | |
| "loss": 0.0006, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.42777571481619014, | |
| "grad_norm": 0.004789507016539574, | |
| "learning_rate": 6.457221540847176e-05, | |
| "loss": 0.0003, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.42851838098774603, | |
| "grad_norm": 0.0077894036658108234, | |
| "learning_rate": 6.4456230995177e-05, | |
| "loss": 0.0002, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.4292610471593019, | |
| "grad_norm": 0.018438173457980156, | |
| "learning_rate": 6.434016163555452e-05, | |
| "loss": 0.0006, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.4300037133308578, | |
| "grad_norm": 0.004602341912686825, | |
| "learning_rate": 6.422400801164003e-05, | |
| "loss": 0.0002, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.4307463795024137, | |
| "grad_norm": 0.005459954962134361, | |
| "learning_rate": 6.41077708059644e-05, | |
| "loss": 0.0003, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.43148904567396956, | |
| "grad_norm": 0.2004091739654541, | |
| "learning_rate": 6.399145070154961e-05, | |
| "loss": 0.0013, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.43223171184552545, | |
| "grad_norm": 0.001691634999588132, | |
| "learning_rate": 6.387504838190479e-05, | |
| "loss": 0.0002, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.43297437801708133, | |
| "grad_norm": 0.002549877157434821, | |
| "learning_rate": 6.375856453102217e-05, | |
| "loss": 0.0002, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.4337170441886372, | |
| "grad_norm": 0.014543402940034866, | |
| "learning_rate": 6.364199983337306e-05, | |
| "loss": 0.0006, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.4344597103601931, | |
| "grad_norm": 0.025866270065307617, | |
| "learning_rate": 6.352535497390381e-05, | |
| "loss": 0.0006, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.435202376531749, | |
| "grad_norm": 0.001332697574980557, | |
| "learning_rate": 6.340863063803188e-05, | |
| "loss": 0.0001, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.43594504270330486, | |
| "grad_norm": 0.006850456353276968, | |
| "learning_rate": 6.329182751164165e-05, | |
| "loss": 0.0004, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.43668770887486075, | |
| "grad_norm": 0.002208834746852517, | |
| "learning_rate": 6.317494628108054e-05, | |
| "loss": 0.0002, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.43743037504641663, | |
| "grad_norm": 0.0022124634124338627, | |
| "learning_rate": 6.305798763315491e-05, | |
| "loss": 0.0002, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.4381730412179725, | |
| "grad_norm": 0.0013898340985178947, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 0.0001, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4389157073895284, | |
| "grad_norm": 0.0031000098679214716, | |
| "learning_rate": 6.282384083470605e-05, | |
| "loss": 0.0002, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.4396583735610843, | |
| "grad_norm": 0.004475159104913473, | |
| "learning_rate": 6.270665406005393e-05, | |
| "loss": 0.0002, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.44040103973264016, | |
| "grad_norm": 0.002459387294948101, | |
| "learning_rate": 6.258939261977143e-05, | |
| "loss": 0.0002, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.44114370590419605, | |
| "grad_norm": 0.0015267659910023212, | |
| "learning_rate": 6.247205720289907e-05, | |
| "loss": 0.0001, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.44188637207575193, | |
| "grad_norm": 0.0016903842333704233, | |
| "learning_rate": 6.235464849891205e-05, | |
| "loss": 0.0002, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.4426290382473078, | |
| "grad_norm": 0.008800814859569073, | |
| "learning_rate": 6.22371671977162e-05, | |
| "loss": 0.0005, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.4433717044188637, | |
| "grad_norm": 0.006040891632437706, | |
| "learning_rate": 6.211961398964396e-05, | |
| "loss": 0.0004, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.4441143705904196, | |
| "grad_norm": 0.014548071660101414, | |
| "learning_rate": 6.20019895654503e-05, | |
| "loss": 0.0004, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.44485703676197547, | |
| "grad_norm": 0.0014387888368219137, | |
| "learning_rate": 6.188429461630866e-05, | |
| "loss": 0.0001, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.4455997029335314, | |
| "grad_norm": 0.0014056372456252575, | |
| "learning_rate": 6.176652983380689e-05, | |
| "loss": 0.0001, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4463423691050873, | |
| "grad_norm": 0.0028386348858475685, | |
| "learning_rate": 6.164869590994317e-05, | |
| "loss": 0.0003, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.4470850352766432, | |
| "grad_norm": 0.0020914392080157995, | |
| "learning_rate": 6.153079353712201e-05, | |
| "loss": 0.0002, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.44782770144819906, | |
| "grad_norm": 0.0010804440826177597, | |
| "learning_rate": 6.14128234081501e-05, | |
| "loss": 0.0002, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.44857036761975494, | |
| "grad_norm": 0.11864292621612549, | |
| "learning_rate": 6.129478621623224e-05, | |
| "loss": 0.0011, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.4493130337913108, | |
| "grad_norm": 0.005421667359769344, | |
| "learning_rate": 6.117668265496737e-05, | |
| "loss": 0.0003, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.4500556999628667, | |
| "grad_norm": 0.002962655620649457, | |
| "learning_rate": 6.105851341834439e-05, | |
| "loss": 0.0002, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4507983661344226, | |
| "grad_norm": 0.0069792428985238075, | |
| "learning_rate": 6.094027920073811e-05, | |
| "loss": 0.0003, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.4515410323059785, | |
| "grad_norm": 0.004223910626024008, | |
| "learning_rate": 6.0821980696905146e-05, | |
| "loss": 0.0003, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.45228369847753436, | |
| "grad_norm": 0.004594327881932259, | |
| "learning_rate": 6.070361860197994e-05, | |
| "loss": 0.0003, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.45302636464909024, | |
| "grad_norm": 0.0028815660625696182, | |
| "learning_rate": 6.058519361147055e-05, | |
| "loss": 0.0002, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4537690308206461, | |
| "grad_norm": 0.05003628879785538, | |
| "learning_rate": 6.04667064212546e-05, | |
| "loss": 0.0015, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.454511696992202, | |
| "grad_norm": 0.004502360709011555, | |
| "learning_rate": 6.034815772757528e-05, | |
| "loss": 0.0003, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.4552543631637579, | |
| "grad_norm": 0.000712364970240742, | |
| "learning_rate": 6.022954822703709e-05, | |
| "loss": 0.0001, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.4559970293353138, | |
| "grad_norm": 0.010336031205952168, | |
| "learning_rate": 6.0110878616601904e-05, | |
| "loss": 0.0004, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.45673969550686966, | |
| "grad_norm": 0.0018599943723529577, | |
| "learning_rate": 5.999214959358477e-05, | |
| "loss": 0.0002, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.45748236167842554, | |
| "grad_norm": 0.0009950903477147222, | |
| "learning_rate": 5.9873361855649876e-05, | |
| "loss": 0.0001, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.4582250278499814, | |
| "grad_norm": 0.014687101356685162, | |
| "learning_rate": 5.9754516100806423e-05, | |
| "loss": 0.0002, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.4589676940215373, | |
| "grad_norm": 0.003395050298422575, | |
| "learning_rate": 5.963561302740449e-05, | |
| "loss": 0.0001, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.4597103601930932, | |
| "grad_norm": 0.0098502766340971, | |
| "learning_rate": 5.9516653334131015e-05, | |
| "loss": 0.0002, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.4604530263646491, | |
| "grad_norm": 0.011299900710582733, | |
| "learning_rate": 5.9397637720005595e-05, | |
| "loss": 0.0003, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.46119569253620496, | |
| "grad_norm": 0.001197346136905253, | |
| "learning_rate": 5.9278566884376474e-05, | |
| "loss": 0.0002, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.46193835870776084, | |
| "grad_norm": 0.0022983471862971783, | |
| "learning_rate": 5.915944152691634e-05, | |
| "loss": 0.0002, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.4626810248793167, | |
| "grad_norm": 0.001483081839978695, | |
| "learning_rate": 5.904026234761827e-05, | |
| "loss": 0.0001, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.4634236910508726, | |
| "grad_norm": 0.0636589452624321, | |
| "learning_rate": 5.8921030046791614e-05, | |
| "loss": 0.0008, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.4641663572224285, | |
| "grad_norm": 0.0022622577380388975, | |
| "learning_rate": 5.880174532505786e-05, | |
| "loss": 0.0002, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4649090233939844, | |
| "grad_norm": 0.006052135024219751, | |
| "learning_rate": 5.868240888334653e-05, | |
| "loss": 0.0004, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.4656516895655403, | |
| "grad_norm": 0.0013821636093780398, | |
| "learning_rate": 5.856302142289105e-05, | |
| "loss": 0.0001, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.4663943557370962, | |
| "grad_norm": 0.0037645832635462284, | |
| "learning_rate": 5.8443583645224655e-05, | |
| "loss": 0.0002, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4671370219086521, | |
| "grad_norm": 0.0014773118309676647, | |
| "learning_rate": 5.832409625217623e-05, | |
| "loss": 0.0002, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.46787968808020797, | |
| "grad_norm": 0.0014741544146090746, | |
| "learning_rate": 5.820455994586621e-05, | |
| "loss": 0.0001, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.46862235425176385, | |
| "grad_norm": 0.0006976615404710174, | |
| "learning_rate": 5.808497542870246e-05, | |
| "loss": 0.0001, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.46936502042331973, | |
| "grad_norm": 1.5406405925750732, | |
| "learning_rate": 5.796534340337614e-05, | |
| "loss": 0.0059, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4701076865948756, | |
| "grad_norm": 0.0043447320349514484, | |
| "learning_rate": 5.784566457285754e-05, | |
| "loss": 0.0002, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.4708503527664315, | |
| "grad_norm": 0.007482700981199741, | |
| "learning_rate": 5.772593964039203e-05, | |
| "loss": 0.0003, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.4715930189379874, | |
| "grad_norm": 0.0010534449247643352, | |
| "learning_rate": 5.7606169309495836e-05, | |
| "loss": 0.0002, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.47233568510954327, | |
| "grad_norm": 0.0039465283043682575, | |
| "learning_rate": 5.748635428395199e-05, | |
| "loss": 0.0002, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.47307835128109915, | |
| "grad_norm": 0.0014873096952214837, | |
| "learning_rate": 5.736649526780611e-05, | |
| "loss": 0.0001, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.47382101745265504, | |
| "grad_norm": 0.0004832752456422895, | |
| "learning_rate": 5.724659296536233e-05, | |
| "loss": 0.0001, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.4745636836242109, | |
| "grad_norm": 0.00103551906067878, | |
| "learning_rate": 5.712664808117918e-05, | |
| "loss": 0.0001, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.4753063497957668, | |
| "grad_norm": 0.000814305676613003, | |
| "learning_rate": 5.7006661320065315e-05, | |
| "loss": 0.0001, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4760490159673227, | |
| "grad_norm": 0.0019791568629443645, | |
| "learning_rate": 5.688663338707554e-05, | |
| "loss": 0.0003, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.47679168213887857, | |
| "grad_norm": 0.005826095584779978, | |
| "learning_rate": 5.6766564987506566e-05, | |
| "loss": 0.0002, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.47753434831043445, | |
| "grad_norm": 0.0011684205383062363, | |
| "learning_rate": 5.664645682689287e-05, | |
| "loss": 0.0001, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.47827701448199034, | |
| "grad_norm": 0.0011006612330675125, | |
| "learning_rate": 5.6526309611002594e-05, | |
| "loss": 0.0001, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.4790196806535462, | |
| "grad_norm": 0.0013071649009361863, | |
| "learning_rate": 5.640612404583337e-05, | |
| "loss": 0.0001, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.4797623468251021, | |
| "grad_norm": 0.0031639167573302984, | |
| "learning_rate": 5.628590083760814e-05, | |
| "loss": 0.0002, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.480505012996658, | |
| "grad_norm": 0.008663011714816093, | |
| "learning_rate": 5.61656406927711e-05, | |
| "loss": 0.0005, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.48124767916821387, | |
| "grad_norm": 0.004144098609685898, | |
| "learning_rate": 5.604534431798347e-05, | |
| "loss": 0.0002, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.48199034533976975, | |
| "grad_norm": 0.1093854233622551, | |
| "learning_rate": 5.5925012420119326e-05, | |
| "loss": 0.0024, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.48273301151132564, | |
| "grad_norm": 0.0017107095336541533, | |
| "learning_rate": 5.5804645706261514e-05, | |
| "loss": 0.0002, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4834756776828815, | |
| "grad_norm": 1.8285502195358276, | |
| "learning_rate": 5.568424488369746e-05, | |
| "loss": 0.0136, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.4842183438544374, | |
| "grad_norm": 0.00895615667104721, | |
| "learning_rate": 5.556381065991499e-05, | |
| "loss": 0.0003, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.4849610100259933, | |
| "grad_norm": 0.02741318568587303, | |
| "learning_rate": 5.544334374259823e-05, | |
| "loss": 0.0003, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.48570367619754923, | |
| "grad_norm": 0.007290950510650873, | |
| "learning_rate": 5.532284483962341e-05, | |
| "loss": 0.0003, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.4864463423691051, | |
| "grad_norm": 0.008065281435847282, | |
| "learning_rate": 5.520231465905471e-05, | |
| "loss": 0.0002, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.487189008540661, | |
| "grad_norm": 0.1876053661108017, | |
| "learning_rate": 5.5081753909140096e-05, | |
| "loss": 0.001, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.4879316747122169, | |
| "grad_norm": 0.00261944648809731, | |
| "learning_rate": 5.496116329830716e-05, | |
| "loss": 0.0002, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.48867434088377276, | |
| "grad_norm": 0.00967632420361042, | |
| "learning_rate": 5.484054353515896e-05, | |
| "loss": 0.0002, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.48941700705532865, | |
| "grad_norm": 0.002796781715005636, | |
| "learning_rate": 5.4719895328469874e-05, | |
| "loss": 0.0002, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.49015967322688453, | |
| "grad_norm": 0.5190077424049377, | |
| "learning_rate": 5.459921938718138e-05, | |
| "loss": 0.0018, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.4909023393984404, | |
| "grad_norm": 0.00399174727499485, | |
| "learning_rate": 5.447851642039798e-05, | |
| "loss": 0.0002, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.4916450055699963, | |
| "grad_norm": 0.0023514707572758198, | |
| "learning_rate": 5.435778713738292e-05, | |
| "loss": 0.0001, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.4923876717415522, | |
| "grad_norm": 0.006780529860407114, | |
| "learning_rate": 5.423703224755413e-05, | |
| "loss": 0.0002, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.49313033791310806, | |
| "grad_norm": 2.0502431392669678, | |
| "learning_rate": 5.411625246048e-05, | |
| "loss": 0.0082, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.49387300408466395, | |
| "grad_norm": 0.0009503086330369115, | |
| "learning_rate": 5.39954484858752e-05, | |
| "loss": 0.0001, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.49461567025621983, | |
| "grad_norm": 0.00197896221652627, | |
| "learning_rate": 5.387462103359655e-05, | |
| "loss": 0.0001, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.4953583364277757, | |
| "grad_norm": 0.0009312194306403399, | |
| "learning_rate": 5.3753770813638814e-05, | |
| "loss": 0.0001, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.4961010025993316, | |
| "grad_norm": 0.0009839091217145324, | |
| "learning_rate": 5.363289853613054e-05, | |
| "loss": 0.0001, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.4968436687708875, | |
| "grad_norm": 0.0011252929689362645, | |
| "learning_rate": 5.351200491132988e-05, | |
| "loss": 0.0001, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.49758633494244336, | |
| "grad_norm": 0.0006150620174594223, | |
| "learning_rate": 5.339109064962047e-05, | |
| "loss": 0.0001, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.49832900111399925, | |
| "grad_norm": 0.0013650684850290418, | |
| "learning_rate": 5.327015646150716e-05, | |
| "loss": 0.0001, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.49907166728555513, | |
| "grad_norm": 0.0012382761342450976, | |
| "learning_rate": 5.3149203057611906e-05, | |
| "loss": 0.0001, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.499814333457111, | |
| "grad_norm": 0.0007622497505508363, | |
| "learning_rate": 5.302823114866958e-05, | |
| "loss": 0.0001, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.500556999628667, | |
| "grad_norm": 0.0008362553780898452, | |
| "learning_rate": 5.290724144552379e-05, | |
| "loss": 0.0001, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.500556999628667, | |
| "eval_loss": 0.00016636255895718932, | |
| "eval_runtime": 190.6115, | |
| "eval_samples_per_second": 5.949, | |
| "eval_steps_per_second": 2.975, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5012996658002228, | |
| "grad_norm": 0.0007262330618686974, | |
| "learning_rate": 5.2786234659122726e-05, | |
| "loss": 0.0001, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5020423319717787, | |
| "grad_norm": 0.007732240483164787, | |
| "learning_rate": 5.266521150051492e-05, | |
| "loss": 0.0003, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.5027849981433345, | |
| "grad_norm": 0.0026611273642629385, | |
| "learning_rate": 5.254417268084514e-05, | |
| "loss": 0.0001, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.5035276643148905, | |
| "grad_norm": 0.00148585916031152, | |
| "learning_rate": 5.242311891135016e-05, | |
| "loss": 0.0002, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.5042703304864463, | |
| "grad_norm": 0.0006069787195883691, | |
| "learning_rate": 5.2302050903354606e-05, | |
| "loss": 0.0001, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.5050129966580023, | |
| "grad_norm": 0.002375812502577901, | |
| "learning_rate": 5.218096936826681e-05, | |
| "loss": 0.0002, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5057556628295581, | |
| "grad_norm": 0.003138911910355091, | |
| "learning_rate": 5.205987501757452e-05, | |
| "loss": 0.0002, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.506498329001114, | |
| "grad_norm": 0.001015645801089704, | |
| "learning_rate": 5.193876856284085e-05, | |
| "loss": 0.0001, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.5072409951726699, | |
| "grad_norm": 0.005205164197832346, | |
| "learning_rate": 5.1817650715699994e-05, | |
| "loss": 0.0002, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5079836613442258, | |
| "grad_norm": 0.0015316219069063663, | |
| "learning_rate": 5.1696522187853126e-05, | |
| "loss": 0.0002, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5087263275157816, | |
| "grad_norm": 0.0026633795350790024, | |
| "learning_rate": 5.157538369106414e-05, | |
| "loss": 0.0002, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5094689936873376, | |
| "grad_norm": 0.0016374588012695312, | |
| "learning_rate": 5.145423593715557e-05, | |
| "loss": 0.0001, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.5102116598588934, | |
| "grad_norm": 0.002312192926183343, | |
| "learning_rate": 5.133307963800429e-05, | |
| "loss": 0.0002, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5109543260304493, | |
| "grad_norm": 0.004305702168494463, | |
| "learning_rate": 5.121191550553741e-05, | |
| "loss": 0.0002, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5116969922020052, | |
| "grad_norm": 0.005227786023169756, | |
| "learning_rate": 5.1090744251728064e-05, | |
| "loss": 0.0002, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5124396583735611, | |
| "grad_norm": 0.0004952704184688628, | |
| "learning_rate": 5.096956658859122e-05, | |
| "loss": 0.0001, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5131823245451169, | |
| "grad_norm": 0.0006610776763409376, | |
| "learning_rate": 5.084838322817954e-05, | |
| "loss": 0.0001, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.5139249907166729, | |
| "grad_norm": 0.002141596982255578, | |
| "learning_rate": 5.072719488257914e-05, | |
| "loss": 0.0002, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.5146676568882287, | |
| "grad_norm": 0.0008098665275610983, | |
| "learning_rate": 5.060600226390543e-05, | |
| "loss": 0.0001, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.5154103230597846, | |
| "grad_norm": 0.0018381529953330755, | |
| "learning_rate": 5.048480608429893e-05, | |
| "loss": 0.0001, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.5161529892313406, | |
| "grad_norm": 0.0011757278116419911, | |
| "learning_rate": 5.0363607055921094e-05, | |
| "loss": 0.0002, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.5168956554028964, | |
| "grad_norm": 0.00043161376379430294, | |
| "learning_rate": 5.0242405890950093e-05, | |
| "loss": 0.0001, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5176383215744523, | |
| "grad_norm": 0.0008273344719782472, | |
| "learning_rate": 5.01212033015767e-05, | |
| "loss": 0.0001, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.5183809877460082, | |
| "grad_norm": 0.0035381957422941923, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0002, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5191236539175641, | |
| "grad_norm": 0.0015232819132506847, | |
| "learning_rate": 4.9878796698423324e-05, | |
| "loss": 0.0001, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.5198663200891199, | |
| "grad_norm": 0.0007589098531752825, | |
| "learning_rate": 4.975759410904992e-05, | |
| "loss": 0.0001, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5206089862606759, | |
| "grad_norm": 0.0017526590963825583, | |
| "learning_rate": 4.963639294407893e-05, | |
| "loss": 0.0001, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5213516524322317, | |
| "grad_norm": 0.0024000522680580616, | |
| "learning_rate": 4.951519391570108e-05, | |
| "loss": 0.0001, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.5220943186037876, | |
| "grad_norm": 0.027467429637908936, | |
| "learning_rate": 4.9393997736094596e-05, | |
| "loss": 0.0002, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5228369847753435, | |
| "grad_norm": 0.000491057347971946, | |
| "learning_rate": 4.9272805117420865e-05, | |
| "loss": 0.0001, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5235796509468994, | |
| "grad_norm": 0.0012586608063429594, | |
| "learning_rate": 4.915161677182048e-05, | |
| "loss": 0.0001, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5243223171184552, | |
| "grad_norm": 0.0025400435552001, | |
| "learning_rate": 4.903043341140879e-05, | |
| "loss": 0.0001, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.5250649832900112, | |
| "grad_norm": 0.0027627914678305387, | |
| "learning_rate": 4.890925574827195e-05, | |
| "loss": 0.0002, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.525807649461567, | |
| "grad_norm": 0.0010417604353278875, | |
| "learning_rate": 4.878808449446259e-05, | |
| "loss": 0.0001, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5265503156331229, | |
| "grad_norm": 0.0006638254853896797, | |
| "learning_rate": 4.866692036199572e-05, | |
| "loss": 0.0001, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5272929818046788, | |
| "grad_norm": 0.0013286214089021087, | |
| "learning_rate": 4.854576406284443e-05, | |
| "loss": 0.0001, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5280356479762347, | |
| "grad_norm": 0.004056660924106836, | |
| "learning_rate": 4.8424616308935875e-05, | |
| "loss": 0.0001, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5287783141477905, | |
| "grad_norm": 0.0007322711171582341, | |
| "learning_rate": 4.830347781214689e-05, | |
| "loss": 0.0001, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5295209803193465, | |
| "grad_norm": 0.004066810943186283, | |
| "learning_rate": 4.8182349284300024e-05, | |
| "loss": 0.0002, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.5302636464909023, | |
| "grad_norm": 0.0007455990416929126, | |
| "learning_rate": 4.806123143715916e-05, | |
| "loss": 0.0001, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5310063126624582, | |
| "grad_norm": 0.0005849769804626703, | |
| "learning_rate": 4.7940124982425497e-05, | |
| "loss": 0.0001, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5317489788340141, | |
| "grad_norm": 0.0005887853913009167, | |
| "learning_rate": 4.781903063173321e-05, | |
| "loss": 0.0001, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.53249164500557, | |
| "grad_norm": 0.0007402475457638502, | |
| "learning_rate": 4.76979490966454e-05, | |
| "loss": 0.0001, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5332343111771258, | |
| "grad_norm": 0.0029941131360828876, | |
| "learning_rate": 4.757688108864986e-05, | |
| "loss": 0.0002, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5339769773486818, | |
| "grad_norm": 0.0025099741760641336, | |
| "learning_rate": 4.7455827319154873e-05, | |
| "loss": 0.0002, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5347196435202377, | |
| "grad_norm": 0.0005227324436418712, | |
| "learning_rate": 4.73347884994851e-05, | |
| "loss": 0.0001, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5354623096917935, | |
| "grad_norm": 0.0016509697306901217, | |
| "learning_rate": 4.7213765340877286e-05, | |
| "loss": 0.0001, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5362049758633495, | |
| "grad_norm": 0.0011602011509239674, | |
| "learning_rate": 4.709275855447621e-05, | |
| "loss": 0.0001, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.5369476420349053, | |
| "grad_norm": 0.0008615905535407364, | |
| "learning_rate": 4.697176885133042e-05, | |
| "loss": 0.0001, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.5376903082064612, | |
| "grad_norm": 0.005168153438717127, | |
| "learning_rate": 4.6850796942388106e-05, | |
| "loss": 0.0002, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.5384329743780171, | |
| "grad_norm": 0.0014478538651019335, | |
| "learning_rate": 4.6729843538492847e-05, | |
| "loss": 0.0001, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.539175640549573, | |
| "grad_norm": 0.8397618532180786, | |
| "learning_rate": 4.660890935037954e-05, | |
| "loss": 0.013, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.5399183067211288, | |
| "grad_norm": 0.0026152590289711952, | |
| "learning_rate": 4.648799508867012e-05, | |
| "loss": 0.0002, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.5406609728926848, | |
| "grad_norm": 0.007058785296976566, | |
| "learning_rate": 4.636710146386947e-05, | |
| "loss": 0.0002, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5414036390642406, | |
| "grad_norm": 0.00043141836067661643, | |
| "learning_rate": 4.6246229186361184e-05, | |
| "loss": 0.0001, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5421463052357965, | |
| "grad_norm": 0.0015156837180256844, | |
| "learning_rate": 4.612537896640346e-05, | |
| "loss": 0.0002, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5428889714073524, | |
| "grad_norm": 0.0005109444609843194, | |
| "learning_rate": 4.600455151412482e-05, | |
| "loss": 0.0001, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.5436316375789083, | |
| "grad_norm": 0.0052424585446715355, | |
| "learning_rate": 4.588374753952001e-05, | |
| "loss": 0.0003, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.5443743037504641, | |
| "grad_norm": 0.0008050160249695182, | |
| "learning_rate": 4.576296775244588e-05, | |
| "loss": 0.0001, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.5451169699220201, | |
| "grad_norm": 0.0020469501614570618, | |
| "learning_rate": 4.564221286261709e-05, | |
| "loss": 0.0002, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.5458596360935759, | |
| "grad_norm": 0.000345387845300138, | |
| "learning_rate": 4.552148357960205e-05, | |
| "loss": 0.0001, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5466023022651318, | |
| "grad_norm": 0.002280315151438117, | |
| "learning_rate": 4.5400780612818626e-05, | |
| "loss": 0.0002, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5473449684366877, | |
| "grad_norm": 0.0018161894986405969, | |
| "learning_rate": 4.528010467153016e-05, | |
| "loss": 0.0002, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.5480876346082436, | |
| "grad_norm": 0.0008220197050832212, | |
| "learning_rate": 4.515945646484105e-05, | |
| "loss": 0.0001, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5488303007797994, | |
| "grad_norm": 0.000890806782990694, | |
| "learning_rate": 4.503883670169285e-05, | |
| "loss": 0.0001, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5495729669513554, | |
| "grad_norm": 0.0015704066026955843, | |
| "learning_rate": 4.491824609085991e-05, | |
| "loss": 0.0002, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5503156331229112, | |
| "grad_norm": 0.003599775955080986, | |
| "learning_rate": 4.4797685340945295e-05, | |
| "loss": 0.0002, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.5510582992944671, | |
| "grad_norm": 0.0044060661457479, | |
| "learning_rate": 4.467715516037659e-05, | |
| "loss": 0.0002, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.551800965466023, | |
| "grad_norm": 0.0007429459365084767, | |
| "learning_rate": 4.4556656257401786e-05, | |
| "loss": 0.0001, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.5525436316375789, | |
| "grad_norm": 0.0007204132853075862, | |
| "learning_rate": 4.443618934008502e-05, | |
| "loss": 0.0001, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5532862978091347, | |
| "grad_norm": 0.05012976750731468, | |
| "learning_rate": 4.4315755116302554e-05, | |
| "loss": 0.0007, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.5540289639806907, | |
| "grad_norm": 0.0010998403886333108, | |
| "learning_rate": 4.4195354293738484e-05, | |
| "loss": 0.0001, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5547716301522466, | |
| "grad_norm": 0.0022817838471382856, | |
| "learning_rate": 4.4074987579880685e-05, | |
| "loss": 0.0002, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.5555142963238024, | |
| "grad_norm": 0.009643745608627796, | |
| "learning_rate": 4.395465568201655e-05, | |
| "loss": 0.0005, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5562569624953584, | |
| "grad_norm": 0.0010438463650643826, | |
| "learning_rate": 4.38343593072289e-05, | |
| "loss": 0.0001, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5569996286669142, | |
| "grad_norm": 0.0021892802324146032, | |
| "learning_rate": 4.3714099162391875e-05, | |
| "loss": 0.0001, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5577422948384702, | |
| "grad_norm": 0.04245134815573692, | |
| "learning_rate": 4.359387595416665e-05, | |
| "loss": 0.001, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.558484961010026, | |
| "grad_norm": 0.001306456862948835, | |
| "learning_rate": 4.347369038899744e-05, | |
| "loss": 0.0001, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5592276271815819, | |
| "grad_norm": 0.0017448252765461802, | |
| "learning_rate": 4.335354317310715e-05, | |
| "loss": 0.0001, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5599702933531377, | |
| "grad_norm": 0.014149321243166924, | |
| "learning_rate": 4.323343501249346e-05, | |
| "loss": 0.0002, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5607129595246937, | |
| "grad_norm": 0.0010777488350868225, | |
| "learning_rate": 4.3113366612924465e-05, | |
| "loss": 0.0001, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5614556256962495, | |
| "grad_norm": 0.0005878026131540537, | |
| "learning_rate": 4.299333867993469e-05, | |
| "loss": 0.0001, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5621982918678055, | |
| "grad_norm": 0.004690335597842932, | |
| "learning_rate": 4.287335191882083e-05, | |
| "loss": 0.0003, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5629409580393613, | |
| "grad_norm": 0.002779589965939522, | |
| "learning_rate": 4.275340703463767e-05, | |
| "loss": 0.0002, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5636836242109172, | |
| "grad_norm": 0.01008251216262579, | |
| "learning_rate": 4.2633504732193896e-05, | |
| "loss": 0.0003, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.564426290382473, | |
| "grad_norm": 0.0012684455141425133, | |
| "learning_rate": 4.2513645716048025e-05, | |
| "loss": 0.0002, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.565168956554029, | |
| "grad_norm": 0.0006100367172621191, | |
| "learning_rate": 4.239383069050417e-05, | |
| "loss": 0.0001, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.5659116227255848, | |
| "grad_norm": 0.0015486053889617324, | |
| "learning_rate": 4.227406035960798e-05, | |
| "loss": 0.0001, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.5666542888971408, | |
| "grad_norm": 0.0052903578616678715, | |
| "learning_rate": 4.215433542714248e-05, | |
| "loss": 0.0003, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.5673969550686966, | |
| "grad_norm": 0.005356297362595797, | |
| "learning_rate": 4.203465659662388e-05, | |
| "loss": 0.0002, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5681396212402525, | |
| "grad_norm": 0.003509636502712965, | |
| "learning_rate": 4.191502457129756e-05, | |
| "loss": 0.0001, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.5688822874118084, | |
| "grad_norm": 0.013686946593225002, | |
| "learning_rate": 4.17954400541338e-05, | |
| "loss": 0.0004, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5696249535833643, | |
| "grad_norm": 0.0035359973553568125, | |
| "learning_rate": 4.16759037478238e-05, | |
| "loss": 0.0002, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.5703676197549201, | |
| "grad_norm": 0.0006897081038914621, | |
| "learning_rate": 4.1556416354775364e-05, | |
| "loss": 0.0001, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5711102859264761, | |
| "grad_norm": 0.000840733468066901, | |
| "learning_rate": 4.1436978577108975e-05, | |
| "loss": 0.0001, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.5718529520980319, | |
| "grad_norm": 0.001890829997137189, | |
| "learning_rate": 4.131759111665349e-05, | |
| "loss": 0.0001, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5725956182695878, | |
| "grad_norm": 0.006861279718577862, | |
| "learning_rate": 4.1198254674942154e-05, | |
| "loss": 0.0003, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.5733382844411437, | |
| "grad_norm": 0.020367255434393883, | |
| "learning_rate": 4.107896995320839e-05, | |
| "loss": 0.0001, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.5740809506126996, | |
| "grad_norm": 0.0010303305462002754, | |
| "learning_rate": 4.095973765238175e-05, | |
| "loss": 0.0001, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.5748236167842555, | |
| "grad_norm": 0.0006230950239114463, | |
| "learning_rate": 4.0840558473083664e-05, | |
| "loss": 0.0001, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.5755662829558114, | |
| "grad_norm": 0.06197100505232811, | |
| "learning_rate": 4.0721433115623544e-05, | |
| "loss": 0.0016, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5763089491273673, | |
| "grad_norm": 0.015006426721811295, | |
| "learning_rate": 4.060236227999441e-05, | |
| "loss": 0.0003, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.5770516152989231, | |
| "grad_norm": 0.0008427058928646147, | |
| "learning_rate": 4.0483346665869e-05, | |
| "loss": 0.0001, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.5777942814704791, | |
| "grad_norm": 0.0018404822330921888, | |
| "learning_rate": 4.036438697259551e-05, | |
| "loss": 0.0002, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.5785369476420349, | |
| "grad_norm": 0.0025395648553967476, | |
| "learning_rate": 4.0245483899193595e-05, | |
| "loss": 0.0001, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.5792796138135908, | |
| "grad_norm": 0.0011295550502836704, | |
| "learning_rate": 4.012663814435014e-05, | |
| "loss": 0.0001, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5800222799851467, | |
| "grad_norm": 0.0014332979917526245, | |
| "learning_rate": 4.0007850406415234e-05, | |
| "loss": 0.0002, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.5807649461567026, | |
| "grad_norm": 0.0006570751429535449, | |
| "learning_rate": 3.9889121383398115e-05, | |
| "loss": 0.0001, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.5815076123282584, | |
| "grad_norm": 0.0014288886450231075, | |
| "learning_rate": 3.977045177296291e-05, | |
| "loss": 0.0002, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.5822502784998144, | |
| "grad_norm": 0.000917198951356113, | |
| "learning_rate": 3.965184227242474e-05, | |
| "loss": 0.0001, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5829929446713702, | |
| "grad_norm": 0.002546856412664056, | |
| "learning_rate": 3.95332935787454e-05, | |
| "loss": 0.0001, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5837356108429261, | |
| "grad_norm": 0.0009835307719185948, | |
| "learning_rate": 3.941480638852948e-05, | |
| "loss": 0.0001, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.584478277014482, | |
| "grad_norm": 0.10560861229896545, | |
| "learning_rate": 3.929638139802007e-05, | |
| "loss": 0.0006, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.5852209431860379, | |
| "grad_norm": 0.0006187596009112895, | |
| "learning_rate": 3.917801930309486e-05, | |
| "loss": 0.0001, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.5859636093575937, | |
| "grad_norm": 0.027732783928513527, | |
| "learning_rate": 3.90597207992619e-05, | |
| "loss": 0.0005, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.5867062755291497, | |
| "grad_norm": 0.003759243292734027, | |
| "learning_rate": 3.8941486581655615e-05, | |
| "loss": 0.0002, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5874489417007055, | |
| "grad_norm": 0.002819777699187398, | |
| "learning_rate": 3.882331734503263e-05, | |
| "loss": 0.0001, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.5881916078722614, | |
| "grad_norm": 0.0007966597331687808, | |
| "learning_rate": 3.870521378376777e-05, | |
| "loss": 0.0001, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.5889342740438173, | |
| "grad_norm": 0.0005825799889862537, | |
| "learning_rate": 3.8587176591849916e-05, | |
| "loss": 0.0001, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.5896769402153732, | |
| "grad_norm": 0.002244536532089114, | |
| "learning_rate": 3.846920646287799e-05, | |
| "loss": 0.0001, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.590419606386929, | |
| "grad_norm": 0.002780564595013857, | |
| "learning_rate": 3.8351304090056825e-05, | |
| "loss": 0.0002, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.591162272558485, | |
| "grad_norm": 0.0009447059710510075, | |
| "learning_rate": 3.8233470166193126e-05, | |
| "loss": 0.0001, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.5919049387300408, | |
| "grad_norm": 0.001110229641199112, | |
| "learning_rate": 3.8115705383691355e-05, | |
| "loss": 0.0001, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.5926476049015967, | |
| "grad_norm": 0.0013889704132452607, | |
| "learning_rate": 3.799801043454971e-05, | |
| "loss": 0.0001, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.5933902710731526, | |
| "grad_norm": 0.0008447144064120948, | |
| "learning_rate": 3.7880386010356054e-05, | |
| "loss": 0.0001, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.5941329372447085, | |
| "grad_norm": 0.003610602580010891, | |
| "learning_rate": 3.776283280228381e-05, | |
| "loss": 0.0001, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5948756034162644, | |
| "grad_norm": 0.005582943558692932, | |
| "learning_rate": 3.764535150108798e-05, | |
| "loss": 0.0003, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.5956182695878203, | |
| "grad_norm": 0.003268187865614891, | |
| "learning_rate": 3.752794279710094e-05, | |
| "loss": 0.0002, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.5963609357593762, | |
| "grad_norm": 0.0018831202760338783, | |
| "learning_rate": 3.741060738022858e-05, | |
| "loss": 0.0002, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.597103601930932, | |
| "grad_norm": 0.0007239828119054437, | |
| "learning_rate": 3.7293345939946065e-05, | |
| "loss": 0.0001, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.597846268102488, | |
| "grad_norm": 0.21732911467552185, | |
| "learning_rate": 3.717615916529395e-05, | |
| "loss": 0.0017, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.5985889342740438, | |
| "grad_norm": 0.00127478688955307, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 0.0002, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.5993316004455997, | |
| "grad_norm": 0.0045968894846737385, | |
| "learning_rate": 3.6942012366845096e-05, | |
| "loss": 0.0001, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.6000742666171556, | |
| "grad_norm": 0.001871331944130361, | |
| "learning_rate": 3.6825053718919464e-05, | |
| "loss": 0.0002, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.6008169327887115, | |
| "grad_norm": 0.0010791551321744919, | |
| "learning_rate": 3.670817248835836e-05, | |
| "loss": 0.0001, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.6015595989602673, | |
| "grad_norm": 0.0006027501658536494, | |
| "learning_rate": 3.6591369361968124e-05, | |
| "loss": 0.0001, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6023022651318233, | |
| "grad_norm": 0.0032688211649656296, | |
| "learning_rate": 3.6474645026096185e-05, | |
| "loss": 0.0002, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.6030449313033791, | |
| "grad_norm": 0.00113378232344985, | |
| "learning_rate": 3.6358000166626966e-05, | |
| "loss": 0.0001, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.603787597474935, | |
| "grad_norm": 0.0009587003150954843, | |
| "learning_rate": 3.624143546897784e-05, | |
| "loss": 0.0001, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.6045302636464909, | |
| "grad_norm": 0.0017409041756764054, | |
| "learning_rate": 3.612495161809522e-05, | |
| "loss": 0.0001, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.6052729298180468, | |
| "grad_norm": 0.0004892799770459533, | |
| "learning_rate": 3.60085492984504e-05, | |
| "loss": 0.0001, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.6060155959896026, | |
| "grad_norm": 0.0016567412531003356, | |
| "learning_rate": 3.589222919403562e-05, | |
| "loss": 0.0002, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.6067582621611586, | |
| "grad_norm": 0.0016312827356159687, | |
| "learning_rate": 3.577599198835998e-05, | |
| "loss": 0.0001, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.6075009283327144, | |
| "grad_norm": 0.0007810606039129198, | |
| "learning_rate": 3.5659838364445505e-05, | |
| "loss": 0.0001, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.6082435945042703, | |
| "grad_norm": 0.0013547971611842513, | |
| "learning_rate": 3.554376900482301e-05, | |
| "loss": 0.0001, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.6089862606758262, | |
| "grad_norm": 0.0006009297212585807, | |
| "learning_rate": 3.542778459152826e-05, | |
| "loss": 0.0001, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6097289268473821, | |
| "grad_norm": 0.0010081218788400292, | |
| "learning_rate": 3.531188580609778e-05, | |
| "loss": 0.0001, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.6104715930189379, | |
| "grad_norm": 0.00042164497426711023, | |
| "learning_rate": 3.5196073329565015e-05, | |
| "loss": 0.0001, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.6112142591904939, | |
| "grad_norm": 0.0007461085333488882, | |
| "learning_rate": 3.508034784245616e-05, | |
| "loss": 0.0001, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.6119569253620497, | |
| "grad_norm": 1.3910738229751587, | |
| "learning_rate": 3.4964710024786354e-05, | |
| "loss": 0.0984, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.6126995915336056, | |
| "grad_norm": 0.0007660177652724087, | |
| "learning_rate": 3.484916055605548e-05, | |
| "loss": 0.0001, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6134422577051615, | |
| "grad_norm": 0.07438488304615021, | |
| "learning_rate": 3.473370011524435e-05, | |
| "loss": 0.0012, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.6141849238767174, | |
| "grad_norm": 0.0015772825572639704, | |
| "learning_rate": 3.461832938081059e-05, | |
| "loss": 0.0002, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.6149275900482734, | |
| "grad_norm": 0.0016182976542040706, | |
| "learning_rate": 3.450304903068475e-05, | |
| "loss": 0.0001, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.6156702562198292, | |
| "grad_norm": 0.0015479021240025759, | |
| "learning_rate": 3.4387859742266225e-05, | |
| "loss": 0.0001, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.6164129223913851, | |
| "grad_norm": 0.00273375422693789, | |
| "learning_rate": 3.427276219241933e-05, | |
| "loss": 0.0002, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.617155588562941, | |
| "grad_norm": 0.001004122314043343, | |
| "learning_rate": 3.415775705746934e-05, | |
| "loss": 0.0001, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6178982547344969, | |
| "grad_norm": 0.008125795051455498, | |
| "learning_rate": 3.4042845013198456e-05, | |
| "loss": 0.0002, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6186409209060527, | |
| "grad_norm": 0.13623641431331635, | |
| "learning_rate": 3.392802673484193e-05, | |
| "loss": 0.0018, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6193835870776087, | |
| "grad_norm": 0.000562251138035208, | |
| "learning_rate": 3.381330289708396e-05, | |
| "loss": 0.0001, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.6201262532491645, | |
| "grad_norm": 0.0013340302975848317, | |
| "learning_rate": 3.3698674174053844e-05, | |
| "loss": 0.0001, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.6208689194207204, | |
| "grad_norm": 0.0081217335537076, | |
| "learning_rate": 3.358414123932195e-05, | |
| "loss": 0.0005, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6216115855922763, | |
| "grad_norm": 0.059895727783441544, | |
| "learning_rate": 3.346970476589583e-05, | |
| "loss": 0.0021, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.6223542517638322, | |
| "grad_norm": 0.0007285134051926434, | |
| "learning_rate": 3.3355365426216164e-05, | |
| "loss": 0.0001, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.623096917935388, | |
| "grad_norm": 0.0223313607275486, | |
| "learning_rate": 3.324112389215293e-05, | |
| "loss": 0.0005, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.623839584106944, | |
| "grad_norm": 0.0015575829893350601, | |
| "learning_rate": 3.31269808350013e-05, | |
| "loss": 0.0001, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6245822502784998, | |
| "grad_norm": 0.001839678268879652, | |
| "learning_rate": 3.3012936925477857e-05, | |
| "loss": 0.0001, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.6253249164500557, | |
| "grad_norm": 0.0008310491102747619, | |
| "learning_rate": 3.289899283371657e-05, | |
| "loss": 0.0001, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.6260675826216116, | |
| "grad_norm": 0.0006663409876637161, | |
| "learning_rate": 3.278514922926484e-05, | |
| "loss": 0.0001, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6268102487931675, | |
| "grad_norm": 0.0015589938266202807, | |
| "learning_rate": 3.267140678107964e-05, | |
| "loss": 0.0001, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.6275529149647233, | |
| "grad_norm": 0.003392728278413415, | |
| "learning_rate": 3.2557766157523465e-05, | |
| "loss": 0.0001, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.6282955811362793, | |
| "grad_norm": 0.0010547033743932843, | |
| "learning_rate": 3.244422802636057e-05, | |
| "loss": 0.0001, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.6290382473078351, | |
| "grad_norm": 0.0024194575380533934, | |
| "learning_rate": 3.2330793054752873e-05, | |
| "loss": 0.0002, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.629780913479391, | |
| "grad_norm": 0.0006333817727863789, | |
| "learning_rate": 3.2217461909256185e-05, | |
| "loss": 0.0001, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.6305235796509469, | |
| "grad_norm": 0.0008936111116781831, | |
| "learning_rate": 3.210423525581615e-05, | |
| "loss": 0.0001, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6312662458225028, | |
| "grad_norm": 0.0010497324401512742, | |
| "learning_rate": 3.199111375976449e-05, | |
| "loss": 0.0001, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6320089119940586, | |
| "grad_norm": 0.010097804479300976, | |
| "learning_rate": 3.1878098085814924e-05, | |
| "loss": 0.0005, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.6327515781656146, | |
| "grad_norm": 0.003449138719588518, | |
| "learning_rate": 3.176518889805941e-05, | |
| "loss": 0.0001, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6334942443371704, | |
| "grad_norm": 0.020226208493113518, | |
| "learning_rate": 3.1652386859964156e-05, | |
| "loss": 0.0004, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6342369105087263, | |
| "grad_norm": 0.0008942090207710862, | |
| "learning_rate": 3.153969263436579e-05, | |
| "loss": 0.0001, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.6349795766802823, | |
| "grad_norm": 0.005887329112738371, | |
| "learning_rate": 3.142710688346733e-05, | |
| "loss": 0.0002, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.6357222428518381, | |
| "grad_norm": 0.0013140842784196138, | |
| "learning_rate": 3.131463026883449e-05, | |
| "loss": 0.0001, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.636464909023394, | |
| "grad_norm": 0.001480376347899437, | |
| "learning_rate": 3.12022634513916e-05, | |
| "loss": 0.0002, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.6372075751949499, | |
| "grad_norm": 0.0014736526645720005, | |
| "learning_rate": 3.109000709141788e-05, | |
| "loss": 0.0001, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6379502413665058, | |
| "grad_norm": 0.0005073303473182023, | |
| "learning_rate": 3.0977861848543444e-05, | |
| "loss": 0.0001, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6386929075380616, | |
| "grad_norm": 0.0009759328095242381, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 0.0001, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6394355737096176, | |
| "grad_norm": 0.0012429956113919616, | |
| "learning_rate": 3.0753907349344464e-05, | |
| "loss": 0.0001, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.6401782398811734, | |
| "grad_norm": 0.003437102073803544, | |
| "learning_rate": 3.064209940899998e-05, | |
| "loss": 0.0002, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.6409209060527293, | |
| "grad_norm": 0.0044972775503993034, | |
| "learning_rate": 3.0530405217707256e-05, | |
| "loss": 0.0003, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.6416635722242852, | |
| "grad_norm": 0.0003766078152693808, | |
| "learning_rate": 3.0418825431793052e-05, | |
| "loss": 0.0001, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6424062383958411, | |
| "grad_norm": 0.000993401394225657, | |
| "learning_rate": 3.0307360706911896e-05, | |
| "loss": 0.0001, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6431489045673969, | |
| "grad_norm": 0.0007390630780719221, | |
| "learning_rate": 3.019601169804216e-05, | |
| "loss": 0.0001, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6438915707389529, | |
| "grad_norm": 0.0011397154303267598, | |
| "learning_rate": 3.0084779059482322e-05, | |
| "loss": 0.0001, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.6446342369105087, | |
| "grad_norm": 0.004833963233977556, | |
| "learning_rate": 2.997366344484698e-05, | |
| "loss": 0.0002, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.6453769030820646, | |
| "grad_norm": 0.007242708466947079, | |
| "learning_rate": 2.9862665507063147e-05, | |
| "loss": 0.0002, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.6461195692536205, | |
| "grad_norm": 0.0021677876356989145, | |
| "learning_rate": 2.975178589836632e-05, | |
| "loss": 0.0001, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6468622354251764, | |
| "grad_norm": 0.0013036590535193682, | |
| "learning_rate": 2.9641025270296722e-05, | |
| "loss": 0.0001, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.6476049015967322, | |
| "grad_norm": 0.0008346426184289157, | |
| "learning_rate": 2.953038427369537e-05, | |
| "loss": 0.0001, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6483475677682882, | |
| "grad_norm": 0.0006587895331904292, | |
| "learning_rate": 2.9419863558700396e-05, | |
| "loss": 0.0001, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.649090233939844, | |
| "grad_norm": 0.0005403941031545401, | |
| "learning_rate": 2.9309463774743046e-05, | |
| "loss": 0.0001, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6498329001113999, | |
| "grad_norm": 0.001244431477971375, | |
| "learning_rate": 2.919918557054405e-05, | |
| "loss": 0.0001, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6505755662829558, | |
| "grad_norm": 0.004742435645312071, | |
| "learning_rate": 2.9089029594109683e-05, | |
| "loss": 0.0002, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6513182324545117, | |
| "grad_norm": 0.0021011109929531813, | |
| "learning_rate": 2.8978996492727994e-05, | |
| "loss": 0.0002, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6520608986260675, | |
| "grad_norm": 0.001175252953544259, | |
| "learning_rate": 2.886908691296504e-05, | |
| "loss": 0.0001, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.6528035647976235, | |
| "grad_norm": 0.0017059053061529994, | |
| "learning_rate": 2.8759301500660972e-05, | |
| "loss": 0.0001, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6535462309691794, | |
| "grad_norm": 0.0007660607225261629, | |
| "learning_rate": 2.864964090092644e-05, | |
| "loss": 0.0001, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6542888971407352, | |
| "grad_norm": 0.001221244689077139, | |
| "learning_rate": 2.854010575813856e-05, | |
| "loss": 0.0001, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.6550315633122912, | |
| "grad_norm": 0.0004929836140945554, | |
| "learning_rate": 2.8430696715937337e-05, | |
| "loss": 0.0001, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.655774229483847, | |
| "grad_norm": 0.0009183657239191234, | |
| "learning_rate": 2.8321414417221746e-05, | |
| "loss": 0.0001, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6565168956554029, | |
| "grad_norm": 0.0003201389918103814, | |
| "learning_rate": 2.8212259504146043e-05, | |
| "loss": 0.0001, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6572595618269588, | |
| "grad_norm": 0.0023057435173541307, | |
| "learning_rate": 2.8103232618115903e-05, | |
| "loss": 0.0001, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6580022279985147, | |
| "grad_norm": 0.0004039146879222244, | |
| "learning_rate": 2.7994334399784772e-05, | |
| "loss": 0.0001, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6587448941700705, | |
| "grad_norm": 0.001328380312770605, | |
| "learning_rate": 2.7885565489049946e-05, | |
| "loss": 0.0001, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.6594875603416265, | |
| "grad_norm": 0.0009456725674681365, | |
| "learning_rate": 2.777692652504895e-05, | |
| "loss": 0.0001, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6602302265131823, | |
| "grad_norm": 0.0006427007610909641, | |
| "learning_rate": 2.766841814615573e-05, | |
| "loss": 0.0001, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.6609728926847382, | |
| "grad_norm": 0.0006173664005473256, | |
| "learning_rate": 2.7560040989976892e-05, | |
| "loss": 0.0001, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6617155588562941, | |
| "grad_norm": 0.0005695072468370199, | |
| "learning_rate": 2.7451795693347936e-05, | |
| "loss": 0.0001, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.66245822502785, | |
| "grad_norm": 0.000993382534943521, | |
| "learning_rate": 2.734368289232959e-05, | |
| "loss": 0.0001, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.6632008911994058, | |
| "grad_norm": 0.00402169581502676, | |
| "learning_rate": 2.723570322220399e-05, | |
| "loss": 0.0001, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.6639435573709618, | |
| "grad_norm": 0.00042257923632860184, | |
| "learning_rate": 2.7127857317470968e-05, | |
| "loss": 0.0001, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.6646862235425176, | |
| "grad_norm": 0.0011959762778133154, | |
| "learning_rate": 2.702014581184439e-05, | |
| "loss": 0.0001, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.6654288897140735, | |
| "grad_norm": 0.0022542597725987434, | |
| "learning_rate": 2.6912569338248315e-05, | |
| "loss": 0.0001, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6661715558856294, | |
| "grad_norm": 0.0031469352543354034, | |
| "learning_rate": 2.6805128528813384e-05, | |
| "loss": 0.0002, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.6669142220571853, | |
| "grad_norm": 0.0017010547453537583, | |
| "learning_rate": 2.6697824014873075e-05, | |
| "loss": 0.0002, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6676568882287411, | |
| "grad_norm": 0.003644139738753438, | |
| "learning_rate": 2.6590656426959902e-05, | |
| "loss": 0.0002, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.6683995544002971, | |
| "grad_norm": 0.0016295432578772306, | |
| "learning_rate": 2.6483626394801873e-05, | |
| "loss": 0.0002, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6691422205718529, | |
| "grad_norm": 0.012327441945672035, | |
| "learning_rate": 2.637673454731868e-05, | |
| "loss": 0.0003, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.6698848867434088, | |
| "grad_norm": 0.2909511923789978, | |
| "learning_rate": 2.626998151261798e-05, | |
| "loss": 0.004, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6706275529149647, | |
| "grad_norm": 0.001305720885284245, | |
| "learning_rate": 2.6163367917991822e-05, | |
| "loss": 0.0001, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.6713702190865206, | |
| "grad_norm": 0.0014534946531057358, | |
| "learning_rate": 2.605689438991282e-05, | |
| "loss": 0.0001, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.6721128852580764, | |
| "grad_norm": 0.003072174033150077, | |
| "learning_rate": 2.595056155403063e-05, | |
| "loss": 0.0002, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.6728555514296324, | |
| "grad_norm": 0.004174454137682915, | |
| "learning_rate": 2.5844370035168073e-05, | |
| "loss": 0.0002, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.6735982176011883, | |
| "grad_norm": 0.001340761547908187, | |
| "learning_rate": 2.5738320457317698e-05, | |
| "loss": 0.0001, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.6743408837727441, | |
| "grad_norm": 0.5977171063423157, | |
| "learning_rate": 2.5632413443637888e-05, | |
| "loss": 0.0386, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6750835499443001, | |
| "grad_norm": 0.000558204366825521, | |
| "learning_rate": 2.552664961644936e-05, | |
| "loss": 0.0001, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.6758262161158559, | |
| "grad_norm": 0.0011802364606410265, | |
| "learning_rate": 2.5421029597231478e-05, | |
| "loss": 0.0001, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6765688822874119, | |
| "grad_norm": 0.000828778138384223, | |
| "learning_rate": 2.5315554006618485e-05, | |
| "loss": 0.0001, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.6773115484589677, | |
| "grad_norm": 0.0003953818522859365, | |
| "learning_rate": 2.5210223464396053e-05, | |
| "loss": 0.0001, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.6780542146305236, | |
| "grad_norm": 0.0005724510992877185, | |
| "learning_rate": 2.5105038589497433e-05, | |
| "loss": 0.0001, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.6787968808020794, | |
| "grad_norm": 0.0006651621079072356, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 0.0001, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.6795395469736354, | |
| "grad_norm": 0.0022599331568926573, | |
| "learning_rate": 2.4895108313121508e-05, | |
| "loss": 0.0002, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6802822131451912, | |
| "grad_norm": 0.0014916729414835572, | |
| "learning_rate": 2.4790364145216495e-05, | |
| "loss": 0.0001, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.6810248793167472, | |
| "grad_norm": 0.004133202601224184, | |
| "learning_rate": 2.4685768111772644e-05, | |
| "loss": 0.0002, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.681767545488303, | |
| "grad_norm": 0.06581906974315643, | |
| "learning_rate": 2.458132082740724e-05, | |
| "loss": 0.0024, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.6825102116598589, | |
| "grad_norm": 0.0005587110063061118, | |
| "learning_rate": 2.4477022905863427e-05, | |
| "loss": 0.0001, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.6832528778314148, | |
| "grad_norm": 0.0006018027779646218, | |
| "learning_rate": 2.4372874960006743e-05, | |
| "loss": 0.0001, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6839955440029707, | |
| "grad_norm": 0.0009498025756329298, | |
| "learning_rate": 2.4268877601821423e-05, | |
| "loss": 0.0001, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.6847382101745265, | |
| "grad_norm": 0.0018847642932087183, | |
| "learning_rate": 2.4165031442406855e-05, | |
| "loss": 0.0001, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.6854808763460825, | |
| "grad_norm": 0.0015400544507429004, | |
| "learning_rate": 2.4061337091973918e-05, | |
| "loss": 0.0001, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.6862235425176383, | |
| "grad_norm": 0.0018971724202856421, | |
| "learning_rate": 2.3957795159841517e-05, | |
| "loss": 0.0001, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.6869662086891942, | |
| "grad_norm": 0.004078839905560017, | |
| "learning_rate": 2.385440625443287e-05, | |
| "loss": 0.0002, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.68770887486075, | |
| "grad_norm": 0.0012903795577585697, | |
| "learning_rate": 2.3751170983272e-05, | |
| "loss": 0.0002, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.688451541032306, | |
| "grad_norm": 0.0016058724140748382, | |
| "learning_rate": 2.3648089952980202e-05, | |
| "loss": 0.0001, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.6891942072038618, | |
| "grad_norm": 0.002747948979958892, | |
| "learning_rate": 2.3545163769272383e-05, | |
| "loss": 0.0001, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.6899368733754178, | |
| "grad_norm": 0.0010071592405438423, | |
| "learning_rate": 2.3442393036953613e-05, | |
| "loss": 0.0001, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.6906795395469736, | |
| "grad_norm": 0.003560745855793357, | |
| "learning_rate": 2.333977835991545e-05, | |
| "loss": 0.0001, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6914222057185295, | |
| "grad_norm": 0.0034868186339735985, | |
| "learning_rate": 2.3237320341132497e-05, | |
| "loss": 0.0002, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.6921648718900854, | |
| "grad_norm": 0.0018978634616360068, | |
| "learning_rate": 2.3135019582658802e-05, | |
| "loss": 0.0002, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.6929075380616413, | |
| "grad_norm": 0.000817135616671294, | |
| "learning_rate": 2.3032876685624367e-05, | |
| "loss": 0.0001, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.6936502042331972, | |
| "grad_norm": 0.002365557011216879, | |
| "learning_rate": 2.2930892250231516e-05, | |
| "loss": 0.0001, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.6943928704047531, | |
| "grad_norm": 0.0016304274322465062, | |
| "learning_rate": 2.282906687575151e-05, | |
| "loss": 0.0001, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.695135536576309, | |
| "grad_norm": 0.004060409963130951, | |
| "learning_rate": 2.2727401160520876e-05, | |
| "loss": 0.0002, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.6958782027478648, | |
| "grad_norm": 0.0014790298882871866, | |
| "learning_rate": 2.262589570193805e-05, | |
| "loss": 0.0001, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.6966208689194208, | |
| "grad_norm": 0.0010480373166501522, | |
| "learning_rate": 2.25245510964597e-05, | |
| "loss": 0.0001, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.6973635350909766, | |
| "grad_norm": 0.0022476809099316597, | |
| "learning_rate": 2.2423367939597385e-05, | |
| "loss": 0.0002, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.6981062012625325, | |
| "grad_norm": 0.0014067289885133505, | |
| "learning_rate": 2.2322346825913886e-05, | |
| "loss": 0.0001, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6988488674340884, | |
| "grad_norm": 0.002156575210392475, | |
| "learning_rate": 2.2221488349019903e-05, | |
| "loss": 0.0001, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.6995915336056443, | |
| "grad_norm": 0.0006094371201470494, | |
| "learning_rate": 2.2120793101570364e-05, | |
| "loss": 0.0001, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.7003341997772001, | |
| "grad_norm": 0.0005834063631482422, | |
| "learning_rate": 2.202026167526111e-05, | |
| "loss": 0.0001, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.7010768659487561, | |
| "grad_norm": 0.00110190873965621, | |
| "learning_rate": 2.1919894660825365e-05, | |
| "loss": 0.0001, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.7018195321203119, | |
| "grad_norm": 0.06204232573509216, | |
| "learning_rate": 2.1819692648030165e-05, | |
| "loss": 0.0018, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7025621982918678, | |
| "grad_norm": 0.002225680975243449, | |
| "learning_rate": 2.171965622567308e-05, | |
| "loss": 0.0001, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.7033048644634237, | |
| "grad_norm": 0.005498223472386599, | |
| "learning_rate": 2.1619785981578573e-05, | |
| "loss": 0.0002, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.7040475306349796, | |
| "grad_norm": 0.2603996992111206, | |
| "learning_rate": 2.1520082502594685e-05, | |
| "loss": 0.0036, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.7047901968065354, | |
| "grad_norm": 0.0005693411221727729, | |
| "learning_rate": 2.1420546374589484e-05, | |
| "loss": 0.0001, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.7055328629780914, | |
| "grad_norm": 0.0005677467561326921, | |
| "learning_rate": 2.132117818244771e-05, | |
| "loss": 0.0001, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7062755291496472, | |
| "grad_norm": 0.019135845825076103, | |
| "learning_rate": 2.1221978510067236e-05, | |
| "loss": 0.0006, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.7070181953212031, | |
| "grad_norm": 0.3902997672557831, | |
| "learning_rate": 2.1122947940355747e-05, | |
| "loss": 0.0052, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.707760861492759, | |
| "grad_norm": 0.0012116495054215193, | |
| "learning_rate": 2.1024087055227248e-05, | |
| "loss": 0.0001, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.7085035276643149, | |
| "grad_norm": 0.3175068795681, | |
| "learning_rate": 2.0925396435598664e-05, | |
| "loss": 0.0045, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.7092461938358707, | |
| "grad_norm": 0.015600265935063362, | |
| "learning_rate": 2.0826876661386395e-05, | |
| "loss": 0.0002, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.7099888600074267, | |
| "grad_norm": 0.0010375302517786622, | |
| "learning_rate": 2.0728528311502976e-05, | |
| "loss": 0.0001, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.7107315261789825, | |
| "grad_norm": 0.0014974784571677446, | |
| "learning_rate": 2.0630351963853587e-05, | |
| "loss": 0.0001, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.7114741923505384, | |
| "grad_norm": 0.0019292763900011778, | |
| "learning_rate": 2.053234819533276e-05, | |
| "loss": 0.0001, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.7122168585220943, | |
| "grad_norm": 0.0008647122303955257, | |
| "learning_rate": 2.0434517581820896e-05, | |
| "loss": 0.0001, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.7129595246936502, | |
| "grad_norm": 0.0007554054609499872, | |
| "learning_rate": 2.0336860698180905e-05, | |
| "loss": 0.0001, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7137021908652061, | |
| "grad_norm": 0.01203538291156292, | |
| "learning_rate": 2.0239378118254905e-05, | |
| "loss": 0.0004, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.714444857036762, | |
| "grad_norm": 0.005261108744889498, | |
| "learning_rate": 2.0142070414860704e-05, | |
| "loss": 0.0003, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.7151875232083179, | |
| "grad_norm": 0.006285266485065222, | |
| "learning_rate": 2.0044938159788584e-05, | |
| "loss": 0.0002, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.7159301893798737, | |
| "grad_norm": 0.0008547162869945168, | |
| "learning_rate": 1.9947981923797853e-05, | |
| "loss": 0.0001, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.7166728555514297, | |
| "grad_norm": 0.00040477412403561175, | |
| "learning_rate": 1.9851202276613523e-05, | |
| "loss": 0.0001, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.7174155217229855, | |
| "grad_norm": 0.0028770139906555414, | |
| "learning_rate": 1.975459978692291e-05, | |
| "loss": 0.0003, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7181581878945414, | |
| "grad_norm": 0.001716628554277122, | |
| "learning_rate": 1.96581750223724e-05, | |
| "loss": 0.0002, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7189008540660973, | |
| "grad_norm": 0.0005640503368340433, | |
| "learning_rate": 1.9561928549563968e-05, | |
| "loss": 0.0001, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.7196435202376532, | |
| "grad_norm": 0.001024580909870565, | |
| "learning_rate": 1.9465860934052006e-05, | |
| "loss": 0.0001, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.720386186409209, | |
| "grad_norm": 0.0007752843666821718, | |
| "learning_rate": 1.936997274033986e-05, | |
| "loss": 0.0001, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.721128852580765, | |
| "grad_norm": 0.0008537242538295686, | |
| "learning_rate": 1.927426453187663e-05, | |
| "loss": 0.0001, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7218715187523208, | |
| "grad_norm": 0.0009459343855269253, | |
| "learning_rate": 1.9178736871053736e-05, | |
| "loss": 0.0001, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7226141849238767, | |
| "grad_norm": 0.0008696299046278, | |
| "learning_rate": 1.9083390319201764e-05, | |
| "loss": 0.0001, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7233568510954326, | |
| "grad_norm": 0.0020470714662224054, | |
| "learning_rate": 1.8988225436587002e-05, | |
| "loss": 0.0002, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.7240995172669885, | |
| "grad_norm": 0.0032244205940514803, | |
| "learning_rate": 1.889324278240829e-05, | |
| "loss": 0.0002, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7248421834385443, | |
| "grad_norm": 0.007482845336198807, | |
| "learning_rate": 1.8798442914793663e-05, | |
| "loss": 0.0002, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.7255848496101003, | |
| "grad_norm": 0.25979605317115784, | |
| "learning_rate": 1.8703826390797048e-05, | |
| "loss": 0.0077, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.7263275157816561, | |
| "grad_norm": 0.0036151015665382147, | |
| "learning_rate": 1.8609393766395085e-05, | |
| "loss": 0.0002, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.727070181953212, | |
| "grad_norm": 1.0644394159317017, | |
| "learning_rate": 1.8515145596483723e-05, | |
| "loss": 0.0138, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7278128481247679, | |
| "grad_norm": 0.0007425799267366529, | |
| "learning_rate": 1.842108243487513e-05, | |
| "loss": 0.0001, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7285555142963238, | |
| "grad_norm": 0.0005094231455586851, | |
| "learning_rate": 1.832720483429426e-05, | |
| "loss": 0.0001, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.7292981804678796, | |
| "grad_norm": 0.0007252280483953655, | |
| "learning_rate": 1.8233513346375756e-05, | |
| "loss": 0.0001, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7300408466394356, | |
| "grad_norm": 0.0005672296392731369, | |
| "learning_rate": 1.814000852166059e-05, | |
| "loss": 0.0001, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.7307835128109914, | |
| "grad_norm": 0.9842846989631653, | |
| "learning_rate": 1.8046690909592917e-05, | |
| "loss": 0.1766, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7315261789825473, | |
| "grad_norm": 0.000559376843739301, | |
| "learning_rate": 1.7953561058516828e-05, | |
| "loss": 0.0001, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7322688451541032, | |
| "grad_norm": 0.0007592244073748589, | |
| "learning_rate": 1.7860619515673033e-05, | |
| "loss": 0.0001, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.7330115113256591, | |
| "grad_norm": 0.0009743353584781289, | |
| "learning_rate": 1.7767866827195805e-05, | |
| "loss": 0.0001, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.733754177497215, | |
| "grad_norm": 0.0004740802978631109, | |
| "learning_rate": 1.767530353810968e-05, | |
| "loss": 0.0001, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.7344968436687709, | |
| "grad_norm": 0.001709680538624525, | |
| "learning_rate": 1.7582930192326186e-05, | |
| "loss": 0.0001, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.7352395098403268, | |
| "grad_norm": 0.004645473323762417, | |
| "learning_rate": 1.7490747332640833e-05, | |
| "loss": 0.0002, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7359821760118826, | |
| "grad_norm": 0.002532870974391699, | |
| "learning_rate": 1.7398755500729734e-05, | |
| "loss": 0.0001, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.7367248421834386, | |
| "grad_norm": 0.0024906517937779427, | |
| "learning_rate": 1.730695523714652e-05, | |
| "loss": 0.0001, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7374675083549944, | |
| "grad_norm": 0.006111454218626022, | |
| "learning_rate": 1.7215347081319188e-05, | |
| "loss": 0.0003, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7382101745265504, | |
| "grad_norm": 0.004187372513115406, | |
| "learning_rate": 1.7123931571546827e-05, | |
| "loss": 0.0002, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7389528406981062, | |
| "grad_norm": 0.006407550070434809, | |
| "learning_rate": 1.703270924499656e-05, | |
| "loss": 0.0004, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.7396955068696621, | |
| "grad_norm": 0.01187540590763092, | |
| "learning_rate": 1.6941680637700334e-05, | |
| "loss": 0.0005, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.740438173041218, | |
| "grad_norm": 0.008408035151660442, | |
| "learning_rate": 1.68508462845518e-05, | |
| "loss": 0.0003, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7411808392127739, | |
| "grad_norm": 0.00872902199625969, | |
| "learning_rate": 1.6760206719303105e-05, | |
| "loss": 0.0003, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7419235053843297, | |
| "grad_norm": 0.013007326982915401, | |
| "learning_rate": 1.6669762474561857e-05, | |
| "loss": 0.0005, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7426661715558857, | |
| "grad_norm": 0.044673044234514236, | |
| "learning_rate": 1.657951408178788e-05, | |
| "loss": 0.0013, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7434088377274415, | |
| "grad_norm": 0.017271386459469795, | |
| "learning_rate": 1.6489462071290214e-05, | |
| "loss": 0.0006, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7441515038989974, | |
| "grad_norm": 0.03663730248808861, | |
| "learning_rate": 1.639960697222388e-05, | |
| "loss": 0.001, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.7448941700705533, | |
| "grad_norm": 0.01725134626030922, | |
| "learning_rate": 1.630994931258688e-05, | |
| "loss": 0.0005, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7456368362421092, | |
| "grad_norm": 0.0050296480767428875, | |
| "learning_rate": 1.622048961921699e-05, | |
| "loss": 0.0003, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.746379502413665, | |
| "grad_norm": 0.013799471780657768, | |
| "learning_rate": 1.6131228417788765e-05, | |
| "loss": 0.0004, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.747122168585221, | |
| "grad_norm": 0.005248996429145336, | |
| "learning_rate": 1.6042166232810347e-05, | |
| "loss": 0.0003, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7478648347567768, | |
| "grad_norm": 0.0058412919752299786, | |
| "learning_rate": 1.595330358762047e-05, | |
| "loss": 0.0005, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.7486075009283327, | |
| "grad_norm": 0.006610256619751453, | |
| "learning_rate": 1.5864641004385384e-05, | |
| "loss": 0.0003, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7493501670998886, | |
| "grad_norm": 0.004714768845587969, | |
| "learning_rate": 1.577617900409566e-05, | |
| "loss": 0.0002, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.7500928332714445, | |
| "grad_norm": 0.005967407487332821, | |
| "learning_rate": 1.5687918106563326e-05, | |
| "loss": 0.0003, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7508354994430003, | |
| "grad_norm": 0.004380329977720976, | |
| "learning_rate": 1.5599858830418628e-05, | |
| "loss": 0.0003, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7508354994430003, | |
| "eval_loss": 9.535554272588342e-05, | |
| "eval_runtime": 187.7576, | |
| "eval_samples_per_second": 6.04, | |
| "eval_steps_per_second": 3.02, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7515781656145563, | |
| "grad_norm": 0.010244094766676426, | |
| "learning_rate": 1.5512001693107124e-05, | |
| "loss": 0.0003, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7523208317861121, | |
| "grad_norm": 0.003399641951546073, | |
| "learning_rate": 1.5424347210886538e-05, | |
| "loss": 0.0002, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.753063497957668, | |
| "grad_norm": 0.004665118642151356, | |
| "learning_rate": 1.53368958988238e-05, | |
| "loss": 0.0003, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.753806164129224, | |
| "grad_norm": 0.007897808216512203, | |
| "learning_rate": 1.5249648270791966e-05, | |
| "loss": 0.0004, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.7545488303007798, | |
| "grad_norm": 0.005702964495867491, | |
| "learning_rate": 1.5162604839467265e-05, | |
| "loss": 0.0003, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.7552914964723357, | |
| "grad_norm": 0.1535305380821228, | |
| "learning_rate": 1.5075766116325985e-05, | |
| "loss": 0.0013, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.7560341626438916, | |
| "grad_norm": 0.0065938313491642475, | |
| "learning_rate": 1.4989132611641576e-05, | |
| "loss": 0.0003, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.7567768288154475, | |
| "grad_norm": 0.011125356890261173, | |
| "learning_rate": 1.4902704834481584e-05, | |
| "loss": 0.0004, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.7575194949870033, | |
| "grad_norm": 0.010162554681301117, | |
| "learning_rate": 1.4816483292704708e-05, | |
| "loss": 0.0004, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7582621611585593, | |
| "grad_norm": 0.012406081892549992, | |
| "learning_rate": 1.4730468492957722e-05, | |
| "loss": 0.0004, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.7590048273301151, | |
| "grad_norm": 0.010129008442163467, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 0.0004, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.759747493501671, | |
| "grad_norm": 0.018571769818663597, | |
| "learning_rate": 1.4559061140063578e-05, | |
| "loss": 0.0006, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.7604901596732269, | |
| "grad_norm": 0.01117369718849659, | |
| "learning_rate": 1.4473669594123956e-05, | |
| "loss": 0.0004, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.7612328258447828, | |
| "grad_norm": 0.011828822083771229, | |
| "learning_rate": 1.4388486804623464e-05, | |
| "loss": 0.0005, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7619754920163386, | |
| "grad_norm": 0.006083609536290169, | |
| "learning_rate": 1.4303513272105057e-05, | |
| "loss": 0.0003, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.7627181581878946, | |
| "grad_norm": 0.0038173352368175983, | |
| "learning_rate": 1.4218749495882133e-05, | |
| "loss": 0.0002, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.7634608243594504, | |
| "grad_norm": 0.003378799417987466, | |
| "learning_rate": 1.4134195974035525e-05, | |
| "loss": 0.0002, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7642034905310063, | |
| "grad_norm": 0.0023517953231930733, | |
| "learning_rate": 1.4049853203410596e-05, | |
| "loss": 0.0002, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.7649461567025622, | |
| "grad_norm": 0.005170909222215414, | |
| "learning_rate": 1.396572167961427e-05, | |
| "loss": 0.0002, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7656888228741181, | |
| "grad_norm": 0.003187798196449876, | |
| "learning_rate": 1.3881801897012225e-05, | |
| "loss": 0.0002, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.7664314890456739, | |
| "grad_norm": 0.0076558575965464115, | |
| "learning_rate": 1.3798094348725871e-05, | |
| "loss": 0.0002, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.7671741552172299, | |
| "grad_norm": 0.002008821815252304, | |
| "learning_rate": 1.3714599526629551e-05, | |
| "loss": 0.0002, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.7679168213887857, | |
| "grad_norm": 0.011446031741797924, | |
| "learning_rate": 1.3631317921347563e-05, | |
| "loss": 0.0003, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7686594875603416, | |
| "grad_norm": 0.0025135111063718796, | |
| "learning_rate": 1.3548250022251375e-05, | |
| "loss": 0.0002, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7694021537318975, | |
| "grad_norm": 0.0011204791953787208, | |
| "learning_rate": 1.346539631745663e-05, | |
| "loss": 0.0001, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.7701448199034534, | |
| "grad_norm": 0.002433129120618105, | |
| "learning_rate": 1.3382757293820409e-05, | |
| "loss": 0.0002, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.7708874860750092, | |
| "grad_norm": 0.004581818822771311, | |
| "learning_rate": 1.330033343693824e-05, | |
| "loss": 0.0003, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7716301522465652, | |
| "grad_norm": 0.001932491664774716, | |
| "learning_rate": 1.3218125231141354e-05, | |
| "loss": 0.0002, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.7723728184181211, | |
| "grad_norm": 0.001757967984303832, | |
| "learning_rate": 1.3136133159493802e-05, | |
| "loss": 0.0002, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7731154845896769, | |
| "grad_norm": 0.0024122344329953194, | |
| "learning_rate": 1.3054357703789554e-05, | |
| "loss": 0.0001, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.7738581507612329, | |
| "grad_norm": 0.003989276476204395, | |
| "learning_rate": 1.297279934454978e-05, | |
| "loss": 0.0003, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.7746008169327887, | |
| "grad_norm": 0.10799296200275421, | |
| "learning_rate": 1.2891458561019915e-05, | |
| "loss": 0.0017, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.7753434831043446, | |
| "grad_norm": 0.005773243959993124, | |
| "learning_rate": 1.2810335831166954e-05, | |
| "loss": 0.0002, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.7760861492759005, | |
| "grad_norm": 0.003262764774262905, | |
| "learning_rate": 1.2729431631676525e-05, | |
| "loss": 0.0002, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.7768288154474564, | |
| "grad_norm": 0.0027453494258224964, | |
| "learning_rate": 1.264874643795021e-05, | |
| "loss": 0.0002, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.7775714816190122, | |
| "grad_norm": 0.030763821676373482, | |
| "learning_rate": 1.2568280724102632e-05, | |
| "loss": 0.001, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.7783141477905682, | |
| "grad_norm": 0.004409801680594683, | |
| "learning_rate": 1.2488034962958789e-05, | |
| "loss": 0.0002, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.779056813962124, | |
| "grad_norm": 0.004938923753798008, | |
| "learning_rate": 1.2408009626051137e-05, | |
| "loss": 0.0003, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.77979948013368, | |
| "grad_norm": 0.007012166548520327, | |
| "learning_rate": 1.2328205183616965e-05, | |
| "loss": 0.0003, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7805421463052358, | |
| "grad_norm": 0.004935544449836016, | |
| "learning_rate": 1.2248622104595532e-05, | |
| "loss": 0.0002, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.7812848124767917, | |
| "grad_norm": 0.002522197552025318, | |
| "learning_rate": 1.216926085662536e-05, | |
| "loss": 0.0002, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.7820274786483475, | |
| "grad_norm": 0.001708640600554645, | |
| "learning_rate": 1.209012190604143e-05, | |
| "loss": 0.0002, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.7827701448199035, | |
| "grad_norm": 0.001268695225007832, | |
| "learning_rate": 1.2011205717872537e-05, | |
| "loss": 0.0001, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.7835128109914593, | |
| "grad_norm": 0.0013489356497302651, | |
| "learning_rate": 1.1932512755838449e-05, | |
| "loss": 0.0001, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.7842554771630152, | |
| "grad_norm": 0.002102524507790804, | |
| "learning_rate": 1.1854043482347248e-05, | |
| "loss": 0.0001, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.7849981433345711, | |
| "grad_norm": 0.0030939369462430477, | |
| "learning_rate": 1.1775798358492629e-05, | |
| "loss": 0.0002, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.785740809506127, | |
| "grad_norm": 0.001534225302748382, | |
| "learning_rate": 1.1697777844051105e-05, | |
| "loss": 0.0001, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.7864834756776828, | |
| "grad_norm": 0.008333349600434303, | |
| "learning_rate": 1.1619982397479412e-05, | |
| "loss": 0.0002, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.7872261418492388, | |
| "grad_norm": 0.2474621683359146, | |
| "learning_rate": 1.1542412475911758e-05, | |
| "loss": 0.0036, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7879688080207946, | |
| "grad_norm": 0.0020573630463331938, | |
| "learning_rate": 1.1465068535157097e-05, | |
| "loss": 0.0002, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.7887114741923505, | |
| "grad_norm": 0.009289558976888657, | |
| "learning_rate": 1.1387951029696542e-05, | |
| "loss": 0.0003, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.7894541403639064, | |
| "grad_norm": 0.0011494681239128113, | |
| "learning_rate": 1.1311060412680647e-05, | |
| "loss": 0.0001, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.7901968065354623, | |
| "grad_norm": 0.002744733588770032, | |
| "learning_rate": 1.1234397135926705e-05, | |
| "loss": 0.0002, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.7909394727070181, | |
| "grad_norm": 0.0026154378429055214, | |
| "learning_rate": 1.1157961649916194e-05, | |
| "loss": 0.0001, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.7916821388785741, | |
| "grad_norm": 0.002585667185485363, | |
| "learning_rate": 1.1081754403791999e-05, | |
| "loss": 0.0002, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.79242480505013, | |
| "grad_norm": 0.0019874395802617073, | |
| "learning_rate": 1.100577584535592e-05, | |
| "loss": 0.0002, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.7931674712216858, | |
| "grad_norm": 0.0020060345996171236, | |
| "learning_rate": 1.0930026421065887e-05, | |
| "loss": 0.0002, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.7939101373932418, | |
| "grad_norm": 0.0012965823989361525, | |
| "learning_rate": 1.0854506576033491e-05, | |
| "loss": 0.0001, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.7946528035647976, | |
| "grad_norm": 0.0032621456775814295, | |
| "learning_rate": 1.0779216754021215e-05, | |
| "loss": 0.0002, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7953954697363536, | |
| "grad_norm": 0.002112373011186719, | |
| "learning_rate": 1.0704157397439962e-05, | |
| "loss": 0.0001, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.7961381359079094, | |
| "grad_norm": 0.00291498564183712, | |
| "learning_rate": 1.062932894734639e-05, | |
| "loss": 0.0002, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.7968808020794653, | |
| "grad_norm": 0.005537567660212517, | |
| "learning_rate": 1.0554731843440275e-05, | |
| "loss": 0.0002, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.7976234682510212, | |
| "grad_norm": 0.002763421507552266, | |
| "learning_rate": 1.0480366524062042e-05, | |
| "loss": 0.0002, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.7983661344225771, | |
| "grad_norm": 0.007374095730483532, | |
| "learning_rate": 1.0406233426190076e-05, | |
| "loss": 0.0003, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.7991088005941329, | |
| "grad_norm": 0.009125984273850918, | |
| "learning_rate": 1.0332332985438248e-05, | |
| "loss": 0.0002, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.7998514667656889, | |
| "grad_norm": 0.004513624124228954, | |
| "learning_rate": 1.0258665636053272e-05, | |
| "loss": 0.0003, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.8005941329372447, | |
| "grad_norm": 0.002031454350799322, | |
| "learning_rate": 1.0185231810912222e-05, | |
| "loss": 0.0002, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.8013367991088006, | |
| "grad_norm": 0.0018436467507854104, | |
| "learning_rate": 1.0112031941519934e-05, | |
| "loss": 0.0002, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.8020794652803565, | |
| "grad_norm": 0.02794628217816353, | |
| "learning_rate": 1.003906645800653e-05, | |
| "loss": 0.0003, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8028221314519124, | |
| "grad_norm": 0.0030993130058050156, | |
| "learning_rate": 9.966335789124809e-06, | |
| "loss": 0.0002, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.8035647976234682, | |
| "grad_norm": 0.01633264310657978, | |
| "learning_rate": 9.893840362247809e-06, | |
| "loss": 0.0004, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.8043074637950242, | |
| "grad_norm": 0.002037476282566786, | |
| "learning_rate": 9.821580603366259e-06, | |
| "loss": 0.0001, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.80505012996658, | |
| "grad_norm": 0.0021626856178045273, | |
| "learning_rate": 9.749556937086085e-06, | |
| "loss": 0.0001, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.8057927961381359, | |
| "grad_norm": 0.0009408566984347999, | |
| "learning_rate": 9.677769786625867e-06, | |
| "loss": 0.0001, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.8065354623096918, | |
| "grad_norm": 0.0013037261087447405, | |
| "learning_rate": 9.606219573814445e-06, | |
| "loss": 0.0001, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.8072781284812477, | |
| "grad_norm": 0.0014423932880163193, | |
| "learning_rate": 9.534906719088355e-06, | |
| "loss": 0.0002, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.8080207946528035, | |
| "grad_norm": 0.0028046793304383755, | |
| "learning_rate": 9.46383164148939e-06, | |
| "loss": 0.0002, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.8087634608243595, | |
| "grad_norm": 0.0011104003060609102, | |
| "learning_rate": 9.392994758662178e-06, | |
| "loss": 0.0001, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.8095061269959153, | |
| "grad_norm": 0.0008305896772071719, | |
| "learning_rate": 9.322396486851626e-06, | |
| "loss": 0.0001, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8102487931674712, | |
| "grad_norm": 0.049922533333301544, | |
| "learning_rate": 9.252037240900619e-06, | |
| "loss": 0.0017, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.810991459339027, | |
| "grad_norm": 0.0017735987203195691, | |
| "learning_rate": 9.181917434247416e-06, | |
| "loss": 0.0002, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.811734125510583, | |
| "grad_norm": 0.0012938451254740357, | |
| "learning_rate": 9.112037478923363e-06, | |
| "loss": 0.0002, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.8124767916821389, | |
| "grad_norm": 0.0025100386701524258, | |
| "learning_rate": 9.042397785550405e-06, | |
| "loss": 0.0002, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.8132194578536948, | |
| "grad_norm": 2.1535964012145996, | |
| "learning_rate": 8.972998763338685e-06, | |
| "loss": 0.291, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.8139621240252507, | |
| "grad_norm": 0.02024056576192379, | |
| "learning_rate": 8.903840820084096e-06, | |
| "loss": 0.0004, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.8147047901968065, | |
| "grad_norm": 0.002208460820838809, | |
| "learning_rate": 8.834924362165992e-06, | |
| "loss": 0.0002, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.8154474563683625, | |
| "grad_norm": 0.0023833250161260366, | |
| "learning_rate": 8.766249794544662e-06, | |
| "loss": 0.0002, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.8161901225399183, | |
| "grad_norm": 0.003369487589225173, | |
| "learning_rate": 8.697817520759093e-06, | |
| "loss": 0.0002, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.8169327887114742, | |
| "grad_norm": 0.003595678135752678, | |
| "learning_rate": 8.629627942924473e-06, | |
| "loss": 0.0002, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8176754548830301, | |
| "grad_norm": 0.011070462875068188, | |
| "learning_rate": 8.561681461729926e-06, | |
| "loss": 0.0003, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.818418121054586, | |
| "grad_norm": 0.000978004070930183, | |
| "learning_rate": 8.49397847643606e-06, | |
| "loss": 0.0001, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.8191607872261418, | |
| "grad_norm": 0.0015130506362766027, | |
| "learning_rate": 8.426519384872733e-06, | |
| "loss": 0.0001, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.8199034533976978, | |
| "grad_norm": 0.0014117389218881726, | |
| "learning_rate": 8.359304583436644e-06, | |
| "loss": 0.0002, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8206461195692536, | |
| "grad_norm": 0.0007138338405638933, | |
| "learning_rate": 8.29233446708898e-06, | |
| "loss": 0.0001, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8213887857408095, | |
| "grad_norm": 0.0014467401197180152, | |
| "learning_rate": 8.225609429353187e-06, | |
| "loss": 0.0001, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.8221314519123654, | |
| "grad_norm": 0.0005772042204625905, | |
| "learning_rate": 8.159129862312548e-06, | |
| "loss": 0.0001, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8228741180839213, | |
| "grad_norm": 0.0007005248917266726, | |
| "learning_rate": 8.092896156607987e-06, | |
| "loss": 0.0001, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8236167842554771, | |
| "grad_norm": 0.0023481727112084627, | |
| "learning_rate": 8.026908701435681e-06, | |
| "loss": 0.0001, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.8243594504270331, | |
| "grad_norm": 0.0042024110443890095, | |
| "learning_rate": 7.96116788454485e-06, | |
| "loss": 0.0002, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8251021165985889, | |
| "grad_norm": 0.001470068353228271, | |
| "learning_rate": 7.895674092235395e-06, | |
| "loss": 0.0001, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.8258447827701448, | |
| "grad_norm": 0.0019429409876465797, | |
| "learning_rate": 7.830427709355725e-06, | |
| "loss": 0.0002, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.8265874489417007, | |
| "grad_norm": 0.006123952101916075, | |
| "learning_rate": 7.765429119300394e-06, | |
| "loss": 0.0003, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.8273301151132566, | |
| "grad_norm": 0.0007870638510212302, | |
| "learning_rate": 7.700678704007947e-06, | |
| "loss": 0.0001, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8280727812848124, | |
| "grad_norm": 0.0008543855510652065, | |
| "learning_rate": 7.636176843958598e-06, | |
| "loss": 0.0001, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8288154474563684, | |
| "grad_norm": 0.010817396454513073, | |
| "learning_rate": 7.571923918172053e-06, | |
| "loss": 0.0002, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8295581136279242, | |
| "grad_norm": 0.0015762196853756905, | |
| "learning_rate": 7.507920304205202e-06, | |
| "loss": 0.0002, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8303007797994801, | |
| "grad_norm": 0.0009204100351780653, | |
| "learning_rate": 7.444166378150013e-06, | |
| "loss": 0.0001, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.831043445971036, | |
| "grad_norm": 0.0015323157422244549, | |
| "learning_rate": 7.3806625146312205e-06, | |
| "loss": 0.0002, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8317861121425919, | |
| "grad_norm": 0.001088933553546667, | |
| "learning_rate": 7.317409086804189e-06, | |
| "loss": 0.0001, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8325287783141478, | |
| "grad_norm": 0.002421481069177389, | |
| "learning_rate": 7.2544064663526815e-06, | |
| "loss": 0.0002, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8332714444857037, | |
| "grad_norm": 0.0009983620839193463, | |
| "learning_rate": 7.191655023486682e-06, | |
| "loss": 0.0001, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8340141106572596, | |
| "grad_norm": 0.003855592804029584, | |
| "learning_rate": 7.129155126940268e-06, | |
| "loss": 0.0002, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.8347567768288154, | |
| "grad_norm": 0.003245392581447959, | |
| "learning_rate": 7.066907143969353e-06, | |
| "loss": 0.0002, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8354994430003714, | |
| "grad_norm": 0.005832053255289793, | |
| "learning_rate": 7.004911440349615e-06, | |
| "loss": 0.0003, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8362421091719272, | |
| "grad_norm": 0.0075843255035579205, | |
| "learning_rate": 6.94316838037431e-06, | |
| "loss": 0.0002, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8369847753434831, | |
| "grad_norm": 0.019108982756733894, | |
| "learning_rate": 6.881678326852136e-06, | |
| "loss": 0.0005, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.837727441515039, | |
| "grad_norm": 0.0029295405838638544, | |
| "learning_rate": 6.820441641105063e-06, | |
| "loss": 0.0003, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8384701076865949, | |
| "grad_norm": 0.0009995042346417904, | |
| "learning_rate": 6.759458682966296e-06, | |
| "loss": 0.0001, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8392127738581507, | |
| "grad_norm": 0.0011027619475498796, | |
| "learning_rate": 6.698729810778065e-06, | |
| "loss": 0.0002, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8399554400297067, | |
| "grad_norm": 0.0015378465177491307, | |
| "learning_rate": 6.6382553813896075e-06, | |
| "loss": 0.0001, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.8406981062012625, | |
| "grad_norm": 0.0011789267882704735, | |
| "learning_rate": 6.578035750154976e-06, | |
| "loss": 0.0001, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.8414407723728184, | |
| "grad_norm": 0.003139357315376401, | |
| "learning_rate": 6.518071270931059e-06, | |
| "loss": 0.0002, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.8421834385443743, | |
| "grad_norm": 0.004962114151567221, | |
| "learning_rate": 6.458362296075399e-06, | |
| "loss": 0.0003, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8429261047159302, | |
| "grad_norm": 0.0013825440546497703, | |
| "learning_rate": 6.398909176444207e-06, | |
| "loss": 0.0001, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.843668770887486, | |
| "grad_norm": 0.0008226705831475556, | |
| "learning_rate": 6.339712261390213e-06, | |
| "loss": 0.0001, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.844411437059042, | |
| "grad_norm": 0.00134433398488909, | |
| "learning_rate": 6.280771898760718e-06, | |
| "loss": 0.0001, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.8451541032305978, | |
| "grad_norm": 0.0013929035048931837, | |
| "learning_rate": 6.222088434895462e-06, | |
| "loss": 0.0001, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.8458967694021537, | |
| "grad_norm": 0.00138406315818429, | |
| "learning_rate": 6.163662214624616e-06, | |
| "loss": 0.0002, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.8466394355737096, | |
| "grad_norm": 0.0019744529854506254, | |
| "learning_rate": 6.1054935812667855e-06, | |
| "loss": 0.0002, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.8473821017452655, | |
| "grad_norm": 0.0007763861794956028, | |
| "learning_rate": 6.047582876626934e-06, | |
| "loss": 0.0001, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.8481247679168213, | |
| "grad_norm": 0.0008731714333407581, | |
| "learning_rate": 5.989930440994451e-06, | |
| "loss": 0.0001, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.8488674340883773, | |
| "grad_norm": 0.0010884518269449472, | |
| "learning_rate": 5.932536613141065e-06, | |
| "loss": 0.0001, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.8496101002599331, | |
| "grad_norm": 0.004011042881757021, | |
| "learning_rate": 5.875401730318947e-06, | |
| "loss": 0.0002, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.850352766431489, | |
| "grad_norm": 0.0016273773508146405, | |
| "learning_rate": 5.818526128258622e-06, | |
| "loss": 0.0002, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.8510954326030449, | |
| "grad_norm": 0.001861872267909348, | |
| "learning_rate": 5.7619101411671095e-06, | |
| "loss": 0.0001, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.8518380987746008, | |
| "grad_norm": 0.0037269063759595156, | |
| "learning_rate": 5.7055541017258855e-06, | |
| "loss": 0.0003, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.8525807649461568, | |
| "grad_norm": 0.0021908970084041357, | |
| "learning_rate": 5.649458341088915e-06, | |
| "loss": 0.0002, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.8533234311177126, | |
| "grad_norm": 0.002067559864372015, | |
| "learning_rate": 5.593623188880781e-06, | |
| "loss": 0.0002, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.8540660972892685, | |
| "grad_norm": 0.004860437475144863, | |
| "learning_rate": 5.538048973194698e-06, | |
| "loss": 0.0002, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8548087634608244, | |
| "grad_norm": 0.00816015899181366, | |
| "learning_rate": 5.482736020590551e-06, | |
| "loss": 0.0002, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.8555514296323803, | |
| "grad_norm": 0.23404280841350555, | |
| "learning_rate": 5.427684656093074e-06, | |
| "loss": 0.0006, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.8562940958039361, | |
| "grad_norm": 0.0022261200938373804, | |
| "learning_rate": 5.372895203189837e-06, | |
| "loss": 0.0002, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.8570367619754921, | |
| "grad_norm": 0.001256439252756536, | |
| "learning_rate": 5.318367983829392e-06, | |
| "loss": 0.0001, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.8577794281470479, | |
| "grad_norm": 0.0017608028138056397, | |
| "learning_rate": 5.2641033184194165e-06, | |
| "loss": 0.0002, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.8585220943186038, | |
| "grad_norm": 0.0018713950412347913, | |
| "learning_rate": 5.210101525824746e-06, | |
| "loss": 0.0001, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.8592647604901597, | |
| "grad_norm": 0.0007540153455920517, | |
| "learning_rate": 5.156362923365588e-06, | |
| "loss": 0.0001, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.8600074266617156, | |
| "grad_norm": 0.0012859681155532598, | |
| "learning_rate": 5.102887826815589e-06, | |
| "loss": 0.0001, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.8607500928332714, | |
| "grad_norm": 0.0008075116202235222, | |
| "learning_rate": 5.049676550400034e-06, | |
| "loss": 0.0001, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.8614927590048274, | |
| "grad_norm": 0.0007994707557372749, | |
| "learning_rate": 4.996729406793943e-06, | |
| "loss": 0.0001, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.8622354251763832, | |
| "grad_norm": 0.004248554352670908, | |
| "learning_rate": 4.944046707120287e-06, | |
| "loss": 0.0002, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.8629780913479391, | |
| "grad_norm": 0.0005359674105420709, | |
| "learning_rate": 4.891628760948114e-06, | |
| "loss": 0.0001, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.863720757519495, | |
| "grad_norm": 0.0008803227683529258, | |
| "learning_rate": 4.839475876290789e-06, | |
| "loss": 0.0001, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.8644634236910509, | |
| "grad_norm": 0.0011372326407581568, | |
| "learning_rate": 4.787588359604106e-06, | |
| "loss": 0.0002, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.8652060898626067, | |
| "grad_norm": 0.004412113688886166, | |
| "learning_rate": 4.735966515784562e-06, | |
| "loss": 0.0002, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.8659487560341627, | |
| "grad_norm": 0.0014664451591670513, | |
| "learning_rate": 4.684610648167503e-06, | |
| "loss": 0.0002, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.8666914222057185, | |
| "grad_norm": 0.00369854923337698, | |
| "learning_rate": 4.633521058525408e-06, | |
| "loss": 0.0002, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.8674340883772744, | |
| "grad_norm": 0.0015638495096936822, | |
| "learning_rate": 4.582698047066036e-06, | |
| "loss": 0.0001, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.8681767545488303, | |
| "grad_norm": 0.026589730754494667, | |
| "learning_rate": 4.532141912430743e-06, | |
| "loss": 0.0005, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.8689194207203862, | |
| "grad_norm": 0.0014144079759716988, | |
| "learning_rate": 4.4818529516926726e-06, | |
| "loss": 0.0001, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.869662086891942, | |
| "grad_norm": 0.0023864826653152704, | |
| "learning_rate": 4.4318314603550074e-06, | |
| "loss": 0.0002, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.870404753063498, | |
| "grad_norm": 0.00443720119073987, | |
| "learning_rate": 4.382077732349299e-06, | |
| "loss": 0.0002, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.8711474192350538, | |
| "grad_norm": 0.0019203081028535962, | |
| "learning_rate": 4.332592060033636e-06, | |
| "loss": 0.0001, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.8718900854066097, | |
| "grad_norm": 0.0016560673248022795, | |
| "learning_rate": 4.283374734191037e-06, | |
| "loss": 0.0002, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.8726327515781657, | |
| "grad_norm": 0.0018693654565140605, | |
| "learning_rate": 4.234426044027645e-06, | |
| "loss": 0.0001, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8733754177497215, | |
| "grad_norm": 0.041886404156684875, | |
| "learning_rate": 4.18574627717111e-06, | |
| "loss": 0.0008, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.8741180839212774, | |
| "grad_norm": 0.0033249545376747847, | |
| "learning_rate": 4.137335719668822e-06, | |
| "loss": 0.0001, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.8748607500928333, | |
| "grad_norm": 0.0029907547868788242, | |
| "learning_rate": 4.089194655986306e-06, | |
| "loss": 0.0002, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.8756034162643892, | |
| "grad_norm": 0.0019523641094565392, | |
| "learning_rate": 4.041323369005479e-06, | |
| "loss": 0.0002, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.876346082435945, | |
| "grad_norm": 0.0027865059673786163, | |
| "learning_rate": 3.9937221400230484e-06, | |
| "loss": 0.0001, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.877088748607501, | |
| "grad_norm": 0.0016474600415676832, | |
| "learning_rate": 3.946391248748821e-06, | |
| "loss": 0.0002, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.8778314147790568, | |
| "grad_norm": 0.0016686657909303904, | |
| "learning_rate": 3.899330973304083e-06, | |
| "loss": 0.0001, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.8785740809506127, | |
| "grad_norm": 0.03477117419242859, | |
| "learning_rate": 3.8525415902199154e-06, | |
| "loss": 0.0013, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.8793167471221686, | |
| "grad_norm": 0.000568350194953382, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 0.0001, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.8800594132937245, | |
| "grad_norm": 0.0022050561383366585, | |
| "learning_rate": 3.7597765992972267e-06, | |
| "loss": 0.0002, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.8808020794652803, | |
| "grad_norm": 0.0006907075876370072, | |
| "learning_rate": 3.7138015365554833e-06, | |
| "loss": 0.0001, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.8815447456368363, | |
| "grad_norm": 0.0008904426940716803, | |
| "learning_rate": 3.6680984563647434e-06, | |
| "loss": 0.0001, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.8822874118083921, | |
| "grad_norm": 0.0031244743149727583, | |
| "learning_rate": 3.6226676272810735e-06, | |
| "loss": 0.0002, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.883030077979948, | |
| "grad_norm": 0.0016301183495670557, | |
| "learning_rate": 3.5775093162607963e-06, | |
| "loss": 0.0001, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.8837727441515039, | |
| "grad_norm": 0.006416819058358669, | |
| "learning_rate": 3.5326237886588732e-06, | |
| "loss": 0.0003, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8845154103230598, | |
| "grad_norm": 0.0009310771129094064, | |
| "learning_rate": 3.488011308227379e-06, | |
| "loss": 0.0001, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.8852580764946156, | |
| "grad_norm": 0.00095487164799124, | |
| "learning_rate": 3.443672137113901e-06, | |
| "loss": 0.0001, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.8860007426661716, | |
| "grad_norm": 0.0011116062523797154, | |
| "learning_rate": 3.3996065358600782e-06, | |
| "loss": 0.0001, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.8867434088377274, | |
| "grad_norm": 0.0024930539075285196, | |
| "learning_rate": 3.3558147633999728e-06, | |
| "loss": 0.0002, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.8874860750092833, | |
| "grad_norm": 0.0017195651307702065, | |
| "learning_rate": 3.3122970770586514e-06, | |
| "loss": 0.0001, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.8882287411808392, | |
| "grad_norm": 0.0015204905066639185, | |
| "learning_rate": 3.269053732550581e-06, | |
| "loss": 0.0002, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.8889714073523951, | |
| "grad_norm": 0.005054555833339691, | |
| "learning_rate": 3.226084983978195e-06, | |
| "loss": 0.0003, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.8897140735239509, | |
| "grad_norm": 0.0018111229874193668, | |
| "learning_rate": 3.183391083830345e-06, | |
| "loss": 0.0001, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.8904567396955069, | |
| "grad_norm": 0.026347478851675987, | |
| "learning_rate": 3.1409722829808863e-06, | |
| "loss": 0.0005, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.8911994058670628, | |
| "grad_norm": 0.002853583311662078, | |
| "learning_rate": 3.0988288306871115e-06, | |
| "loss": 0.0002, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8919420720386186, | |
| "grad_norm": 0.0017346449894830585, | |
| "learning_rate": 3.0569609745883743e-06, | |
| "loss": 0.0001, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.8926847382101746, | |
| "grad_norm": 0.31705546379089355, | |
| "learning_rate": 3.0153689607045845e-06, | |
| "loss": 0.0023, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.8934274043817304, | |
| "grad_norm": 0.2009972184896469, | |
| "learning_rate": 2.9740530334347594e-06, | |
| "loss": 0.0012, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.8941700705532863, | |
| "grad_norm": 0.009426708333194256, | |
| "learning_rate": 2.933013435555637e-06, | |
| "loss": 0.0003, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.8949127367248422, | |
| "grad_norm": 0.0008448066073469818, | |
| "learning_rate": 2.8922504082201764e-06, | |
| "loss": 0.0001, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.8956554028963981, | |
| "grad_norm": 0.0007227634778246284, | |
| "learning_rate": 2.8517641909562077e-06, | |
| "loss": 0.0001, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.8963980690679539, | |
| "grad_norm": 0.007234448567032814, | |
| "learning_rate": 2.811555021664969e-06, | |
| "loss": 0.0003, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.8971407352395099, | |
| "grad_norm": 0.003920075949281454, | |
| "learning_rate": 2.7716231366197663e-06, | |
| "loss": 0.0002, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.8978834014110657, | |
| "grad_norm": 0.0014649624936282635, | |
| "learning_rate": 2.7319687704645326e-06, | |
| "loss": 0.0001, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.8986260675826216, | |
| "grad_norm": 0.0005429857992567122, | |
| "learning_rate": 2.692592156212487e-06, | |
| "loss": 0.0001, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.8993687337541775, | |
| "grad_norm": 0.004393384791910648, | |
| "learning_rate": 2.653493525244721e-06, | |
| "loss": 0.0002, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.9001113999257334, | |
| "grad_norm": 0.0011112246429547668, | |
| "learning_rate": 2.614673107308896e-06, | |
| "loss": 0.0001, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.9008540660972892, | |
| "grad_norm": 0.0010931448778137565, | |
| "learning_rate": 2.5761311305178614e-06, | |
| "loss": 0.0001, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.9015967322688452, | |
| "grad_norm": 0.0010206797160208225, | |
| "learning_rate": 2.5378678213483054e-06, | |
| "loss": 0.0001, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.902339398440401, | |
| "grad_norm": 0.0024138889275491238, | |
| "learning_rate": 2.499883404639436e-06, | |
| "loss": 0.0002, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.903082064611957, | |
| "grad_norm": 0.0011341659119352698, | |
| "learning_rate": 2.462178103591678e-06, | |
| "loss": 0.0001, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.9038247307835128, | |
| "grad_norm": 0.0011666857171803713, | |
| "learning_rate": 2.424752139765318e-06, | |
| "loss": 0.0001, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.9045673969550687, | |
| "grad_norm": 0.0019645995926111937, | |
| "learning_rate": 2.3876057330792346e-06, | |
| "loss": 0.0002, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.9053100631266245, | |
| "grad_norm": 0.0026836565230041742, | |
| "learning_rate": 2.350739101809624e-06, | |
| "loss": 0.0003, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.9060527292981805, | |
| "grad_norm": 0.0015338718658313155, | |
| "learning_rate": 2.314152462588659e-06, | |
| "loss": 0.0001, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9067953954697363, | |
| "grad_norm": 0.007026324979960918, | |
| "learning_rate": 2.2778460304032634e-06, | |
| "loss": 0.0003, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.9075380616412922, | |
| "grad_norm": 0.0013827058719471097, | |
| "learning_rate": 2.2418200185938487e-06, | |
| "loss": 0.0002, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.9082807278128481, | |
| "grad_norm": 0.0026905061677098274, | |
| "learning_rate": 2.206074638853023e-06, | |
| "loss": 0.0002, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.909023393984404, | |
| "grad_norm": 0.001551009714603424, | |
| "learning_rate": 2.170610101224385e-06, | |
| "loss": 0.0001, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.9097660601559598, | |
| "grad_norm": 0.004318626597523689, | |
| "learning_rate": 2.1354266141012824e-06, | |
| "loss": 0.0002, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.9105087263275158, | |
| "grad_norm": 0.002215152606368065, | |
| "learning_rate": 2.100524384225555e-06, | |
| "loss": 0.0001, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.9112513924990717, | |
| "grad_norm": 0.011130552738904953, | |
| "learning_rate": 2.0659036166863766e-06, | |
| "loss": 0.0004, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.9119940586706275, | |
| "grad_norm": 0.0008930191397666931, | |
| "learning_rate": 2.0315645149189935e-06, | |
| "loss": 0.0001, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.9127367248421835, | |
| "grad_norm": 0.0026403944939374924, | |
| "learning_rate": 1.99750728070357e-06, | |
| "loss": 0.0002, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.9134793910137393, | |
| "grad_norm": 0.0009700857335701585, | |
| "learning_rate": 1.9637321141639743e-06, | |
| "loss": 0.0001, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9142220571852953, | |
| "grad_norm": 0.045662786811590195, | |
| "learning_rate": 1.930239213766638e-06, | |
| "loss": 0.0009, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.9149647233568511, | |
| "grad_norm": 0.02511802315711975, | |
| "learning_rate": 1.8970287763193429e-06, | |
| "loss": 0.0007, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.915707389528407, | |
| "grad_norm": 0.001158213010057807, | |
| "learning_rate": 1.8641009969700917e-06, | |
| "loss": 0.0001, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.9164500556999629, | |
| "grad_norm": 0.00307093164883554, | |
| "learning_rate": 1.8314560692059835e-06, | |
| "loss": 0.0002, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.9171927218715188, | |
| "grad_norm": 0.0006863306043669581, | |
| "learning_rate": 1.799094184852018e-06, | |
| "loss": 0.0001, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.9179353880430746, | |
| "grad_norm": 0.001991459634155035, | |
| "learning_rate": 1.7670155340700323e-06, | |
| "loss": 0.0002, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.9186780542146306, | |
| "grad_norm": 0.0033494269009679556, | |
| "learning_rate": 1.7352203053575157e-06, | |
| "loss": 0.0003, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.9194207203861864, | |
| "grad_norm": 0.0011704204371199012, | |
| "learning_rate": 1.70370868554659e-06, | |
| "loss": 0.0001, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.9201633865577423, | |
| "grad_norm": 0.000724752142559737, | |
| "learning_rate": 1.6724808598028108e-06, | |
| "loss": 0.0001, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.9209060527292982, | |
| "grad_norm": 0.002084789564833045, | |
| "learning_rate": 1.6415370116241724e-06, | |
| "loss": 0.0001, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9216487189008541, | |
| "grad_norm": 0.0011653484543785453, | |
| "learning_rate": 1.6108773228399543e-06, | |
| "loss": 0.0001, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.9223913850724099, | |
| "grad_norm": 0.001394697930663824, | |
| "learning_rate": 1.5805019736097104e-06, | |
| "loss": 0.0002, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.9231340512439659, | |
| "grad_norm": 0.0010144009720534086, | |
| "learning_rate": 1.5504111424221646e-06, | |
| "loss": 0.0001, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9238767174155217, | |
| "grad_norm": 0.012748640961945057, | |
| "learning_rate": 1.5206050060942e-06, | |
| "loss": 0.0001, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9246193835870776, | |
| "grad_norm": 0.0012531749671325088, | |
| "learning_rate": 1.4910837397697886e-06, | |
| "loss": 0.0001, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9253620497586335, | |
| "grad_norm": 0.0027611353434622288, | |
| "learning_rate": 1.4618475169190017e-06, | |
| "loss": 0.0002, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.9261047159301894, | |
| "grad_norm": 0.0028265246655792, | |
| "learning_rate": 1.4328965093369283e-06, | |
| "loss": 0.0001, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9268473821017452, | |
| "grad_norm": 0.0014653587713837624, | |
| "learning_rate": 1.4042308871427324e-06, | |
| "loss": 0.0001, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9275900482733012, | |
| "grad_norm": 0.0027727442793548107, | |
| "learning_rate": 1.3758508187786068e-06, | |
| "loss": 0.0002, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.928332714444857, | |
| "grad_norm": 0.003438390791416168, | |
| "learning_rate": 1.3477564710088098e-06, | |
| "loss": 0.0002, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9290753806164129, | |
| "grad_norm": 0.0032505146227777004, | |
| "learning_rate": 1.3199480089186756e-06, | |
| "loss": 0.0002, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9298180467879688, | |
| "grad_norm": 0.005036122631281614, | |
| "learning_rate": 1.2924255959136266e-06, | |
| "loss": 0.0002, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.9305607129595247, | |
| "grad_norm": 0.011345870792865753, | |
| "learning_rate": 1.2651893937182635e-06, | |
| "loss": 0.0003, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.9313033791310806, | |
| "grad_norm": 0.0007504152599722147, | |
| "learning_rate": 1.2382395623753485e-06, | |
| "loss": 0.0001, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.9320460453026365, | |
| "grad_norm": 0.0009204870439134538, | |
| "learning_rate": 1.2115762602449177e-06, | |
| "loss": 0.0001, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.9327887114741924, | |
| "grad_norm": 0.021037423983216286, | |
| "learning_rate": 1.1851996440033319e-06, | |
| "loss": 0.0007, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.9335313776457482, | |
| "grad_norm": 0.05482451990246773, | |
| "learning_rate": 1.1591098686423496e-06, | |
| "loss": 0.0006, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9342740438173042, | |
| "grad_norm": 0.0006609958363696933, | |
| "learning_rate": 1.1333070874682216e-06, | |
| "loss": 0.0001, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.93501670998886, | |
| "grad_norm": 0.0011097380192950368, | |
| "learning_rate": 1.1077914521007981e-06, | |
| "loss": 0.0002, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.9357593761604159, | |
| "grad_norm": 0.0019438870949670672, | |
| "learning_rate": 1.0825631124726232e-06, | |
| "loss": 0.0001, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9365020423319718, | |
| "grad_norm": 0.0007544804248027503, | |
| "learning_rate": 1.0576222168280636e-06, | |
| "loss": 0.0001, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9372447085035277, | |
| "grad_norm": 0.007262110244482756, | |
| "learning_rate": 1.0329689117224262e-06, | |
| "loss": 0.0004, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.9379873746750835, | |
| "grad_norm": 0.0008670915849506855, | |
| "learning_rate": 1.008603342021125e-06, | |
| "loss": 0.0001, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.9387300408466395, | |
| "grad_norm": 0.0009119977476075292, | |
| "learning_rate": 9.845256508987989e-07, | |
| "loss": 0.0001, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9394727070181953, | |
| "grad_norm": 0.00047243438893929124, | |
| "learning_rate": 9.607359798384785e-07, | |
| "loss": 0.0001, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9402153731897512, | |
| "grad_norm": 0.0028327691834419966, | |
| "learning_rate": 9.372344686307655e-07, | |
| "loss": 0.0002, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.9409580393613071, | |
| "grad_norm": 0.044161271303892136, | |
| "learning_rate": 9.140212553730043e-07, | |
| "loss": 0.0016, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.941700705532863, | |
| "grad_norm": 0.3684042692184448, | |
| "learning_rate": 8.910964764684671e-07, | |
| "loss": 0.0089, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.9424433717044188, | |
| "grad_norm": 0.002094242488965392, | |
| "learning_rate": 8.684602666255481e-07, | |
| "loss": 0.0001, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.9431860378759748, | |
| "grad_norm": 0.0017989203333854675, | |
| "learning_rate": 8.46112758857004e-07, | |
| "loss": 0.0002, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.9439287040475306, | |
| "grad_norm": 0.0011526981834322214, | |
| "learning_rate": 8.240540844791145e-07, | |
| "loss": 0.0001, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.9446713702190865, | |
| "grad_norm": 0.0008104105945676565, | |
| "learning_rate": 8.022843731109674e-07, | |
| "loss": 0.0001, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.9454140363906424, | |
| "grad_norm": 0.0018936687847599387, | |
| "learning_rate": 7.8080375267367e-07, | |
| "loss": 0.0002, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.9461567025621983, | |
| "grad_norm": 0.0015682062366977334, | |
| "learning_rate": 7.596123493895991e-07, | |
| "loss": 0.0001, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.9468993687337541, | |
| "grad_norm": 0.0024748966097831726, | |
| "learning_rate": 7.387102877816521e-07, | |
| "loss": 0.0002, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.9476420349053101, | |
| "grad_norm": 0.0018025932367891073, | |
| "learning_rate": 7.180976906725423e-07, | |
| "loss": 0.0001, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.9483847010768659, | |
| "grad_norm": 0.0015493928221985698, | |
| "learning_rate": 6.977746791840434e-07, | |
| "loss": 0.0002, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.9491273672484218, | |
| "grad_norm": 0.004556954838335514, | |
| "learning_rate": 6.777413727363069e-07, | |
| "loss": 0.0002, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.9498700334199777, | |
| "grad_norm": 0.018575742840766907, | |
| "learning_rate": 6.579978890471294e-07, | |
| "loss": 0.0003, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.9506126995915336, | |
| "grad_norm": 0.002247869735583663, | |
| "learning_rate": 6.385443441312978e-07, | |
| "loss": 0.0002, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.9513553657630895, | |
| "grad_norm": 0.0014160757418721914, | |
| "learning_rate": 6.193808522998723e-07, | |
| "loss": 0.0001, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.9520980319346454, | |
| "grad_norm": 0.0015209565171971917, | |
| "learning_rate": 6.005075261595494e-07, | |
| "loss": 0.0002, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.9528406981062013, | |
| "grad_norm": 0.0009286428103223443, | |
| "learning_rate": 5.81924476611967e-07, | |
| "loss": 0.0001, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.9535833642777571, | |
| "grad_norm": 0.013380018062889576, | |
| "learning_rate": 5.63631812853066e-07, | |
| "loss": 0.0002, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.9543260304493131, | |
| "grad_norm": 0.00046919341548345983, | |
| "learning_rate": 5.456296423724527e-07, | |
| "loss": 0.0001, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.9550686966208689, | |
| "grad_norm": 0.001202006940729916, | |
| "learning_rate": 5.279180709527765e-07, | |
| "loss": 0.0001, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.9558113627924248, | |
| "grad_norm": 0.0014856454217806458, | |
| "learning_rate": 5.104972026690802e-07, | |
| "loss": 0.0001, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.9565540289639807, | |
| "grad_norm": 0.0008986307075247169, | |
| "learning_rate": 4.933671398882067e-07, | |
| "loss": 0.0001, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.9572966951355366, | |
| "grad_norm": 0.0012058410793542862, | |
| "learning_rate": 4.765279832682101e-07, | |
| "loss": 0.0001, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.9580393613070924, | |
| "grad_norm": 0.003975100349634886, | |
| "learning_rate": 4.5997983175773417e-07, | |
| "loss": 0.0002, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.9587820274786484, | |
| "grad_norm": 0.003919736947864294, | |
| "learning_rate": 4.4372278259545155e-07, | |
| "loss": 0.0003, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.9595246936502042, | |
| "grad_norm": 0.019034747034311295, | |
| "learning_rate": 4.277569313094809e-07, | |
| "loss": 0.0006, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.9602673598217601, | |
| "grad_norm": 0.004752116743475199, | |
| "learning_rate": 4.120823717168432e-07, | |
| "loss": 0.0002, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.961010025993316, | |
| "grad_norm": 0.0008414981421083212, | |
| "learning_rate": 3.9669919592288384e-07, | |
| "loss": 0.0001, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.9617526921648719, | |
| "grad_norm": 0.0018463104497641325, | |
| "learning_rate": 3.8160749432076815e-07, | |
| "loss": 0.0002, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.9624953583364277, | |
| "grad_norm": 0.0009904575999826193, | |
| "learning_rate": 3.6680735559090906e-07, | |
| "loss": 0.0001, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.9632380245079837, | |
| "grad_norm": 0.001538589014671743, | |
| "learning_rate": 3.5229886670046783e-07, | |
| "loss": 0.0002, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.9639806906795395, | |
| "grad_norm": 0.0009757834486663342, | |
| "learning_rate": 3.380821129028489e-07, | |
| "loss": 0.0001, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.9647233568510954, | |
| "grad_norm": 0.00236718263477087, | |
| "learning_rate": 3.241571777371888e-07, | |
| "loss": 0.0002, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.9654660230226513, | |
| "grad_norm": 0.00395069969817996, | |
| "learning_rate": 3.1052414302786823e-07, | |
| "loss": 0.0002, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9662086891942072, | |
| "grad_norm": 0.006602929905056953, | |
| "learning_rate": 2.971830888840177e-07, | |
| "loss": 0.0003, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.966951355365763, | |
| "grad_norm": 0.0023442876990884542, | |
| "learning_rate": 2.8413409369907886e-07, | |
| "loss": 0.0002, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.967694021537319, | |
| "grad_norm": 0.012941988185048103, | |
| "learning_rate": 2.7137723415030516e-07, | |
| "loss": 0.0003, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.9684366877088748, | |
| "grad_norm": 0.005490366835147142, | |
| "learning_rate": 2.589125851983509e-07, | |
| "loss": 0.0002, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.9691793538804307, | |
| "grad_norm": 0.0007061643409542739, | |
| "learning_rate": 2.46740220086783e-07, | |
| "loss": 0.0001, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.9699220200519866, | |
| "grad_norm": 0.0017634638352319598, | |
| "learning_rate": 2.3486021034170857e-07, | |
| "loss": 0.0002, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.9706646862235425, | |
| "grad_norm": 0.0028723692521452904, | |
| "learning_rate": 2.232726257713036e-07, | |
| "loss": 0.0002, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.9714073523950985, | |
| "grad_norm": 0.0016753628151491284, | |
| "learning_rate": 2.1197753446542957e-07, | |
| "loss": 0.0001, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.9721500185666543, | |
| "grad_norm": 0.02156355232000351, | |
| "learning_rate": 2.009750027952284e-07, | |
| "loss": 0.0007, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.9728926847382102, | |
| "grad_norm": 0.0010258769616484642, | |
| "learning_rate": 1.9026509541272275e-07, | |
| "loss": 0.0001, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.973635350909766, | |
| "grad_norm": 0.001431930111721158, | |
| "learning_rate": 1.798478752504551e-07, | |
| "loss": 0.0001, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.974378017081322, | |
| "grad_norm": 0.7691323161125183, | |
| "learning_rate": 1.6972340352110483e-07, | |
| "loss": 0.0318, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.9751206832528778, | |
| "grad_norm": 0.005196771118789911, | |
| "learning_rate": 1.598917397171218e-07, | |
| "loss": 0.0003, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.9758633494244338, | |
| "grad_norm": 0.0011321887141093612, | |
| "learning_rate": 1.503529416103988e-07, | |
| "loss": 0.0002, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.9766060155959896, | |
| "grad_norm": 0.0005147146875970066, | |
| "learning_rate": 1.4110706525190531e-07, | |
| "loss": 0.0001, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.9773486817675455, | |
| "grad_norm": 0.008908793330192566, | |
| "learning_rate": 1.3215416497138754e-07, | |
| "loss": 0.0003, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.9780913479391014, | |
| "grad_norm": 0.002536405110731721, | |
| "learning_rate": 1.234942933770189e-07, | |
| "loss": 0.0002, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.9788340141106573, | |
| "grad_norm": 0.0017400149954482913, | |
| "learning_rate": 1.1512750135511674e-07, | |
| "loss": 0.0001, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.9795766802822131, | |
| "grad_norm": 0.0011383086675778031, | |
| "learning_rate": 1.0705383806982606e-07, | |
| "loss": 0.0001, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.9803193464537691, | |
| "grad_norm": 0.0018386361189186573, | |
| "learning_rate": 9.927335096283629e-08, | |
| "loss": 0.0002, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9810620126253249, | |
| "grad_norm": 0.0018964792834594846, | |
| "learning_rate": 9.178608575310388e-08, | |
| "loss": 0.0001, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.9818046787968808, | |
| "grad_norm": 0.001377547043375671, | |
| "learning_rate": 8.459208643659122e-08, | |
| "loss": 0.0001, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.9825473449684367, | |
| "grad_norm": 0.0015313441399484873, | |
| "learning_rate": 7.769139528598368e-08, | |
| "loss": 0.0001, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.9832900111399926, | |
| "grad_norm": 0.00252739991992712, | |
| "learning_rate": 7.108405285046749e-08, | |
| "loss": 0.0002, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.9840326773115484, | |
| "grad_norm": 0.0026342032942920923, | |
| "learning_rate": 6.477009795547995e-08, | |
| "loss": 0.0002, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9847753434831044, | |
| "grad_norm": 0.0036653559654951096, | |
| "learning_rate": 5.8749567702481856e-08, | |
| "loss": 0.0002, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.9855180096546602, | |
| "grad_norm": 0.031000560149550438, | |
| "learning_rate": 5.3022497468735446e-08, | |
| "loss": 0.0012, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.9862606758262161, | |
| "grad_norm": 0.01462267991155386, | |
| "learning_rate": 4.7588920907110094e-08, | |
| "loss": 0.0005, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.987003341997772, | |
| "grad_norm": 0.0009502037428319454, | |
| "learning_rate": 4.244886994587138e-08, | |
| "loss": 0.0001, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.9877460081693279, | |
| "grad_norm": 0.0015973957488313317, | |
| "learning_rate": 3.760237478849793e-08, | |
| "loss": 0.0001, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9884886743408837, | |
| "grad_norm": 0.0013975318288430572, | |
| "learning_rate": 3.3049463913498166e-08, | |
| "loss": 0.0001, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.9892313405124397, | |
| "grad_norm": 0.06534797698259354, | |
| "learning_rate": 2.8790164074254943e-08, | |
| "loss": 0.0017, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.9899740066839955, | |
| "grad_norm": 0.000878207094501704, | |
| "learning_rate": 2.4824500298858967e-08, | |
| "loss": 0.0001, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.9907166728555514, | |
| "grad_norm": 0.0011413278989493847, | |
| "learning_rate": 2.1152495889970035e-08, | |
| "loss": 0.0001, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.9914593390271074, | |
| "grad_norm": 0.0010686512105166912, | |
| "learning_rate": 1.777417242467272e-08, | |
| "loss": 0.0001, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.9922020051986632, | |
| "grad_norm": 0.001142362249083817, | |
| "learning_rate": 1.4689549754337562e-08, | |
| "loss": 0.0001, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.9929446713702191, | |
| "grad_norm": 0.0008279301691800356, | |
| "learning_rate": 1.189864600454338e-08, | |
| "loss": 0.0001, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.993687337541775, | |
| "grad_norm": 0.0007734330138191581, | |
| "learning_rate": 9.401477574932926e-09, | |
| "loss": 0.0001, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.9944300037133309, | |
| "grad_norm": 0.004689542576670647, | |
| "learning_rate": 7.198059139118529e-09, | |
| "loss": 0.0002, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.9951726698848867, | |
| "grad_norm": 0.0011075050570070744, | |
| "learning_rate": 5.2884036446265714e-09, | |
| "loss": 0.0001, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9959153360564427, | |
| "grad_norm": 0.0022147931158542633, | |
| "learning_rate": 3.672522312786475e-09, | |
| "loss": 0.0002, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.9966580022279985, | |
| "grad_norm": 0.0008529072511009872, | |
| "learning_rate": 2.3504246386918393e-09, | |
| "loss": 0.0001, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.9974006683995544, | |
| "grad_norm": 0.0028021601028740406, | |
| "learning_rate": 1.3221183911282798e-09, | |
| "loss": 0.0002, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.9981433345711103, | |
| "grad_norm": 0.0028829630464315414, | |
| "learning_rate": 5.876096125234653e-10, | |
| "loss": 0.0001, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.9988860007426662, | |
| "grad_norm": 0.002654945245012641, | |
| "learning_rate": 1.4690261893601786e-10, | |
| "loss": 0.0002, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.999628666914222, | |
| "grad_norm": 0.0032625223975628614, | |
| "learning_rate": 0.0, | |
| "loss": 0.0001, | |
| "step": 1346 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1346, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 337, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.7591268226010972e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |