| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 4.988354430379747, |
| "eval_steps": 500, |
| "global_step": 1230, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.004050632911392405, |
| "grad_norm": 8.650783065218267, |
| "learning_rate": 3.2520325203252037e-07, |
| "loss": 1.3913, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.00810126582278481, |
| "grad_norm": 8.714647912912307, |
| "learning_rate": 6.504065040650407e-07, |
| "loss": 1.3774, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.012151898734177215, |
| "grad_norm": 8.353177347578923, |
| "learning_rate": 9.75609756097561e-07, |
| "loss": 1.3575, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01620253164556962, |
| "grad_norm": 8.162241521948225, |
| "learning_rate": 1.3008130081300815e-06, |
| "loss": 1.349, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.020253164556962026, |
| "grad_norm": 8.004482633727692, |
| "learning_rate": 1.6260162601626018e-06, |
| "loss": 1.3472, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.02430379746835443, |
| "grad_norm": 7.798806851393027, |
| "learning_rate": 1.951219512195122e-06, |
| "loss": 1.3365, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.028354430379746835, |
| "grad_norm": 5.808865237205599, |
| "learning_rate": 2.2764227642276426e-06, |
| "loss": 1.2847, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.03240506329113924, |
| "grad_norm": 5.380815521518279, |
| "learning_rate": 2.601626016260163e-06, |
| "loss": 1.2818, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03645569620253165, |
| "grad_norm": 3.3570605809461673, |
| "learning_rate": 2.926829268292683e-06, |
| "loss": 1.2163, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.04050632911392405, |
| "grad_norm": 3.169911302714268, |
| "learning_rate": 3.2520325203252037e-06, |
| "loss": 1.2194, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.044556962025316456, |
| "grad_norm": 3.050444902160162, |
| "learning_rate": 3.577235772357724e-06, |
| "loss": 1.192, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.04860759493670886, |
| "grad_norm": 5.177287930649316, |
| "learning_rate": 3.902439024390244e-06, |
| "loss": 1.1752, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.052658227848101265, |
| "grad_norm": 5.479776150468036, |
| "learning_rate": 4.227642276422765e-06, |
| "loss": 1.1698, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05670886075949367, |
| "grad_norm": 5.163830492645947, |
| "learning_rate": 4.552845528455285e-06, |
| "loss": 1.1699, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.060759493670886074, |
| "grad_norm": 4.6478844520893885, |
| "learning_rate": 4.8780487804878055e-06, |
| "loss": 1.1234, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06481012658227848, |
| "grad_norm": 2.992358982228097, |
| "learning_rate": 5.203252032520326e-06, |
| "loss": 1.0782, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06886075949367089, |
| "grad_norm": 2.7806611145215343, |
| "learning_rate": 5.528455284552846e-06, |
| "loss": 1.0891, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0729113924050633, |
| "grad_norm": 2.432809679529271, |
| "learning_rate": 5.853658536585366e-06, |
| "loss": 1.0583, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0769620253164557, |
| "grad_norm": 2.234347328847241, |
| "learning_rate": 6.178861788617887e-06, |
| "loss": 1.0247, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.0810126582278481, |
| "grad_norm": 1.9572783738048747, |
| "learning_rate": 6.504065040650407e-06, |
| "loss": 1.025, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08506329113924051, |
| "grad_norm": 1.493007390667756, |
| "learning_rate": 6.829268292682928e-06, |
| "loss": 1.0083, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08911392405063291, |
| "grad_norm": 1.3206963982534945, |
| "learning_rate": 7.154471544715448e-06, |
| "loss": 1.0093, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.09316455696202532, |
| "grad_norm": 1.6292425252956484, |
| "learning_rate": 7.4796747967479676e-06, |
| "loss": 1.0046, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09721518987341772, |
| "grad_norm": 1.6060430969874047, |
| "learning_rate": 7.804878048780489e-06, |
| "loss": 0.9871, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.10126582278481013, |
| "grad_norm": 1.2270796490607963, |
| "learning_rate": 8.130081300813009e-06, |
| "loss": 0.9847, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10531645569620253, |
| "grad_norm": 1.1719994702377685, |
| "learning_rate": 8.45528455284553e-06, |
| "loss": 0.9885, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10936708860759493, |
| "grad_norm": 1.261993352596166, |
| "learning_rate": 8.78048780487805e-06, |
| "loss": 0.9681, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.11341772151898734, |
| "grad_norm": 1.2127018533449803, |
| "learning_rate": 9.10569105691057e-06, |
| "loss": 0.971, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11746835443037974, |
| "grad_norm": 1.004539974065213, |
| "learning_rate": 9.43089430894309e-06, |
| "loss": 0.9615, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.12151898734177215, |
| "grad_norm": 1.1967945256419181, |
| "learning_rate": 9.756097560975611e-06, |
| "loss": 0.9567, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12556962025316457, |
| "grad_norm": 1.108396427729017, |
| "learning_rate": 1.008130081300813e-05, |
| "loss": 0.9608, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12962025316455697, |
| "grad_norm": 1.0661215035473321, |
| "learning_rate": 1.0406504065040652e-05, |
| "loss": 0.9359, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.13367088607594937, |
| "grad_norm": 1.085861527059889, |
| "learning_rate": 1.0731707317073172e-05, |
| "loss": 0.9498, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13772151898734178, |
| "grad_norm": 1.021364986525048, |
| "learning_rate": 1.1056910569105692e-05, |
| "loss": 0.9436, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.14177215189873418, |
| "grad_norm": 0.9876393194556478, |
| "learning_rate": 1.1382113821138213e-05, |
| "loss": 0.953, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.1458227848101266, |
| "grad_norm": 1.1203257568285983, |
| "learning_rate": 1.1707317073170731e-05, |
| "loss": 0.9387, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.149873417721519, |
| "grad_norm": 1.051864298805972, |
| "learning_rate": 1.2032520325203254e-05, |
| "loss": 0.9416, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.1539240506329114, |
| "grad_norm": 0.8563328640211689, |
| "learning_rate": 1.2357723577235774e-05, |
| "loss": 0.9368, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.1579746835443038, |
| "grad_norm": 0.9629356844227813, |
| "learning_rate": 1.2682926829268294e-05, |
| "loss": 0.9324, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.1620253164556962, |
| "grad_norm": 0.9799350895554113, |
| "learning_rate": 1.3008130081300815e-05, |
| "loss": 0.9082, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1660759493670886, |
| "grad_norm": 0.854309973842661, |
| "learning_rate": 1.3333333333333333e-05, |
| "loss": 0.919, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.17012658227848101, |
| "grad_norm": 1.0025074114831438, |
| "learning_rate": 1.3658536585365855e-05, |
| "loss": 0.9175, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.17417721518987342, |
| "grad_norm": 0.8006683611542198, |
| "learning_rate": 1.3983739837398376e-05, |
| "loss": 0.9127, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17822784810126582, |
| "grad_norm": 0.832789529295091, |
| "learning_rate": 1.4308943089430896e-05, |
| "loss": 0.9188, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.18227848101265823, |
| "grad_norm": 0.981451492482887, |
| "learning_rate": 1.4634146341463415e-05, |
| "loss": 0.9106, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.18632911392405063, |
| "grad_norm": 0.8518466414584815, |
| "learning_rate": 1.4959349593495935e-05, |
| "loss": 0.9039, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.19037974683544304, |
| "grad_norm": 0.8901241698185981, |
| "learning_rate": 1.528455284552846e-05, |
| "loss": 0.9027, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.19443037974683544, |
| "grad_norm": 0.9273531258094387, |
| "learning_rate": 1.5609756097560978e-05, |
| "loss": 0.9246, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19848101265822785, |
| "grad_norm": 0.846020073542497, |
| "learning_rate": 1.5934959349593496e-05, |
| "loss": 0.9087, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.20253164556962025, |
| "grad_norm": 0.8739582328229863, |
| "learning_rate": 1.6260162601626018e-05, |
| "loss": 0.9106, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.20658227848101265, |
| "grad_norm": 0.8631244044895554, |
| "learning_rate": 1.6585365853658537e-05, |
| "loss": 0.9152, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.21063291139240506, |
| "grad_norm": 0.8501163721126644, |
| "learning_rate": 1.691056910569106e-05, |
| "loss": 0.9145, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.21468354430379746, |
| "grad_norm": 0.922549606515074, |
| "learning_rate": 1.7235772357723578e-05, |
| "loss": 0.8992, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21873417721518987, |
| "grad_norm": 1.0166007899378562, |
| "learning_rate": 1.75609756097561e-05, |
| "loss": 0.9043, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.22278481012658227, |
| "grad_norm": 1.0195781948664562, |
| "learning_rate": 1.788617886178862e-05, |
| "loss": 0.906, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.22683544303797468, |
| "grad_norm": 1.0672054360457455, |
| "learning_rate": 1.821138211382114e-05, |
| "loss": 0.9069, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.23088607594936708, |
| "grad_norm": 0.9814066612784734, |
| "learning_rate": 1.8536585365853663e-05, |
| "loss": 0.8965, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.23493670886075949, |
| "grad_norm": 0.9991426176254492, |
| "learning_rate": 1.886178861788618e-05, |
| "loss": 0.8991, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.2389873417721519, |
| "grad_norm": 0.937159550562, |
| "learning_rate": 1.91869918699187e-05, |
| "loss": 0.8946, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.2430379746835443, |
| "grad_norm": 0.9857585097402551, |
| "learning_rate": 1.9512195121951222e-05, |
| "loss": 0.893, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2470886075949367, |
| "grad_norm": 0.9676971799688627, |
| "learning_rate": 1.983739837398374e-05, |
| "loss": 0.906, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.25113924050632913, |
| "grad_norm": 1.0501588138843359, |
| "learning_rate": 2.016260162601626e-05, |
| "loss": 0.8917, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.25518987341772154, |
| "grad_norm": 0.956848643278113, |
| "learning_rate": 2.048780487804878e-05, |
| "loss": 0.9023, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.25924050632911394, |
| "grad_norm": 1.0230894917736468, |
| "learning_rate": 2.0813008130081303e-05, |
| "loss": 0.8926, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.26329113924050634, |
| "grad_norm": 1.1449566316125528, |
| "learning_rate": 2.1138211382113822e-05, |
| "loss": 0.8815, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.26734177215189875, |
| "grad_norm": 1.0955143356323889, |
| "learning_rate": 2.1463414634146344e-05, |
| "loss": 0.8836, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.27139240506329115, |
| "grad_norm": 0.9366400517927678, |
| "learning_rate": 2.1788617886178863e-05, |
| "loss": 0.8863, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.27544303797468356, |
| "grad_norm": 1.0773560614968867, |
| "learning_rate": 2.2113821138211385e-05, |
| "loss": 0.8952, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.27949367088607596, |
| "grad_norm": 0.8858104295711461, |
| "learning_rate": 2.2439024390243907e-05, |
| "loss": 0.8794, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.28354430379746837, |
| "grad_norm": 0.961041040234339, |
| "learning_rate": 2.2764227642276426e-05, |
| "loss": 0.8917, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.28759493670886077, |
| "grad_norm": 1.3418196681475556, |
| "learning_rate": 2.3089430894308948e-05, |
| "loss": 0.899, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.2916455696202532, |
| "grad_norm": 1.1199127135575435, |
| "learning_rate": 2.3414634146341463e-05, |
| "loss": 0.8915, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.2956962025316456, |
| "grad_norm": 1.0934877141967034, |
| "learning_rate": 2.3739837398373985e-05, |
| "loss": 0.9004, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.299746835443038, |
| "grad_norm": 1.3796357930581438, |
| "learning_rate": 2.4065040650406507e-05, |
| "loss": 0.8738, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.3037974683544304, |
| "grad_norm": 0.9541500327361558, |
| "learning_rate": 2.4390243902439026e-05, |
| "loss": 0.8883, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3078481012658228, |
| "grad_norm": 1.4149738983076507, |
| "learning_rate": 2.4715447154471548e-05, |
| "loss": 0.8906, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.3118987341772152, |
| "grad_norm": 1.0071301667690649, |
| "learning_rate": 2.5040650406504066e-05, |
| "loss": 0.8916, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.3159493670886076, |
| "grad_norm": 1.3236649738923665, |
| "learning_rate": 2.536585365853659e-05, |
| "loss": 0.8854, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 1.0526746243364173, |
| "learning_rate": 2.569105691056911e-05, |
| "loss": 0.8814, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.3240506329113924, |
| "grad_norm": 1.036014838205312, |
| "learning_rate": 2.601626016260163e-05, |
| "loss": 0.8832, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3281012658227848, |
| "grad_norm": 1.196391899738614, |
| "learning_rate": 2.634146341463415e-05, |
| "loss": 0.8823, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3321518987341772, |
| "grad_norm": 1.0634838781911455, |
| "learning_rate": 2.6666666666666667e-05, |
| "loss": 0.8784, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3362025316455696, |
| "grad_norm": 0.9752429745553766, |
| "learning_rate": 2.699186991869919e-05, |
| "loss": 0.8931, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.34025316455696203, |
| "grad_norm": 1.2854004301174213, |
| "learning_rate": 2.731707317073171e-05, |
| "loss": 0.8857, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.34430379746835443, |
| "grad_norm": 0.9469298240599222, |
| "learning_rate": 2.764227642276423e-05, |
| "loss": 0.9025, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.34835443037974684, |
| "grad_norm": 1.1394014304060927, |
| "learning_rate": 2.796747967479675e-05, |
| "loss": 0.8678, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.35240506329113924, |
| "grad_norm": 1.138589746089172, |
| "learning_rate": 2.829268292682927e-05, |
| "loss": 0.8601, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.35645569620253165, |
| "grad_norm": 1.267125473425037, |
| "learning_rate": 2.8617886178861792e-05, |
| "loss": 0.8768, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.36050632911392405, |
| "grad_norm": 1.1241472454225716, |
| "learning_rate": 2.8943089430894314e-05, |
| "loss": 0.8896, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.36455696202531646, |
| "grad_norm": 1.2930445657346719, |
| "learning_rate": 2.926829268292683e-05, |
| "loss": 0.891, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.36860759493670886, |
| "grad_norm": 1.1386378561967647, |
| "learning_rate": 2.959349593495935e-05, |
| "loss": 0.8888, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.37265822784810126, |
| "grad_norm": 1.2460041951108876, |
| "learning_rate": 2.991869918699187e-05, |
| "loss": 0.8653, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.37670886075949367, |
| "grad_norm": 1.1549570210491509, |
| "learning_rate": 3.0243902439024392e-05, |
| "loss": 0.8945, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3807594936708861, |
| "grad_norm": 1.2202030422838337, |
| "learning_rate": 3.056910569105692e-05, |
| "loss": 0.8786, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.3848101265822785, |
| "grad_norm": 1.0592951957176777, |
| "learning_rate": 3.089430894308943e-05, |
| "loss": 0.8763, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.3888607594936709, |
| "grad_norm": 1.018377865920856, |
| "learning_rate": 3.1219512195121955e-05, |
| "loss": 0.8763, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.3929113924050633, |
| "grad_norm": 1.1401287349093545, |
| "learning_rate": 3.154471544715447e-05, |
| "loss": 0.8756, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.3969620253164557, |
| "grad_norm": 0.9917555740850217, |
| "learning_rate": 3.186991869918699e-05, |
| "loss": 0.8648, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.4010126582278481, |
| "grad_norm": 1.1670249532819654, |
| "learning_rate": 3.2195121951219514e-05, |
| "loss": 0.877, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.4050632911392405, |
| "grad_norm": 1.1895422631432588, |
| "learning_rate": 3.2520325203252037e-05, |
| "loss": 0.8745, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4091139240506329, |
| "grad_norm": 1.2222663506580864, |
| "learning_rate": 3.284552845528456e-05, |
| "loss": 0.8938, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.4131645569620253, |
| "grad_norm": 1.072183866697836, |
| "learning_rate": 3.3170731707317074e-05, |
| "loss": 0.8784, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.4172151898734177, |
| "grad_norm": 1.1777948517820547, |
| "learning_rate": 3.3495934959349596e-05, |
| "loss": 0.8647, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.4212658227848101, |
| "grad_norm": 1.0589638016273495, |
| "learning_rate": 3.382113821138212e-05, |
| "loss": 0.8878, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4253164556962025, |
| "grad_norm": 1.483832248436783, |
| "learning_rate": 3.414634146341463e-05, |
| "loss": 0.8836, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4293670886075949, |
| "grad_norm": 1.0253028415304177, |
| "learning_rate": 3.4471544715447155e-05, |
| "loss": 0.8775, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.43341772151898733, |
| "grad_norm": 1.2485635116766391, |
| "learning_rate": 3.479674796747968e-05, |
| "loss": 0.8732, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.43746835443037974, |
| "grad_norm": 1.376252515590877, |
| "learning_rate": 3.51219512195122e-05, |
| "loss": 0.8855, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.44151898734177214, |
| "grad_norm": 1.2528248841054515, |
| "learning_rate": 3.544715447154472e-05, |
| "loss": 0.8717, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.44556962025316454, |
| "grad_norm": 1.312544709591596, |
| "learning_rate": 3.577235772357724e-05, |
| "loss": 0.8639, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.44962025316455695, |
| "grad_norm": 1.4187392243844432, |
| "learning_rate": 3.609756097560976e-05, |
| "loss": 0.8741, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.45367088607594935, |
| "grad_norm": 1.1013529396325934, |
| "learning_rate": 3.642276422764228e-05, |
| "loss": 0.8697, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.45772151898734176, |
| "grad_norm": 1.738351327969621, |
| "learning_rate": 3.67479674796748e-05, |
| "loss": 0.8768, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.46177215189873416, |
| "grad_norm": 1.4578562490506475, |
| "learning_rate": 3.7073170731707325e-05, |
| "loss": 0.8708, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.46582278481012657, |
| "grad_norm": 1.5622273525859154, |
| "learning_rate": 3.739837398373984e-05, |
| "loss": 0.8715, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.46987341772151897, |
| "grad_norm": 1.4076948468024415, |
| "learning_rate": 3.772357723577236e-05, |
| "loss": 0.8693, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.4739240506329114, |
| "grad_norm": 1.4637010194614586, |
| "learning_rate": 3.804878048780488e-05, |
| "loss": 0.8764, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.4779746835443038, |
| "grad_norm": 1.1724466711698687, |
| "learning_rate": 3.83739837398374e-05, |
| "loss": 0.8654, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.4820253164556962, |
| "grad_norm": 1.3636551721780057, |
| "learning_rate": 3.869918699186992e-05, |
| "loss": 0.873, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.4860759493670886, |
| "grad_norm": 0.9608070827844469, |
| "learning_rate": 3.9024390243902444e-05, |
| "loss": 0.8619, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.490126582278481, |
| "grad_norm": 1.4426232706556092, |
| "learning_rate": 3.9349593495934966e-05, |
| "loss": 0.864, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.4941772151898734, |
| "grad_norm": 0.9714070888051659, |
| "learning_rate": 3.967479674796748e-05, |
| "loss": 0.8738, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4982278481012658, |
| "grad_norm": 1.4302540897697196, |
| "learning_rate": 4e-05, |
| "loss": 0.8693, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.5022784810126583, |
| "grad_norm": 0.9108458519702944, |
| "learning_rate": 3.999991946137476e-05, |
| "loss": 0.8609, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.5063291139240507, |
| "grad_norm": 1.224679470659387, |
| "learning_rate": 3.999967784614766e-05, |
| "loss": 0.8737, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5103797468354431, |
| "grad_norm": 0.9011485523187657, |
| "learning_rate": 3.9999275156264656e-05, |
| "loss": 0.8839, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.5144303797468355, |
| "grad_norm": 1.0797263479415513, |
| "learning_rate": 3.999871139496895e-05, |
| "loss": 0.8783, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.5184810126582279, |
| "grad_norm": 1.1570731008070738, |
| "learning_rate": 3.9997986566800995e-05, |
| "loss": 0.8556, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.5225316455696203, |
| "grad_norm": 1.2728860010244547, |
| "learning_rate": 3.999710067759846e-05, |
| "loss": 0.8772, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.5265822784810127, |
| "grad_norm": 1.121453377089669, |
| "learning_rate": 3.999605373449617e-05, |
| "loss": 0.8626, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5306329113924051, |
| "grad_norm": 1.335361633394942, |
| "learning_rate": 3.9994845745926075e-05, |
| "loss": 0.871, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.5346835443037975, |
| "grad_norm": 1.1138697981389432, |
| "learning_rate": 3.999347672161713e-05, |
| "loss": 0.8552, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5387341772151899, |
| "grad_norm": 0.9115408383886785, |
| "learning_rate": 3.999194667259528e-05, |
| "loss": 0.867, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5427848101265823, |
| "grad_norm": 1.2413888524648395, |
| "learning_rate": 3.999025561118334e-05, |
| "loss": 0.8603, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5468354430379747, |
| "grad_norm": 0.89538013139761, |
| "learning_rate": 3.998840355100086e-05, |
| "loss": 0.8696, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5508860759493671, |
| "grad_norm": 1.0298510378266852, |
| "learning_rate": 3.998639050696409e-05, |
| "loss": 0.8662, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5549367088607595, |
| "grad_norm": 1.074990273032981, |
| "learning_rate": 3.998421649528582e-05, |
| "loss": 0.8498, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5589873417721519, |
| "grad_norm": 0.8885879506724239, |
| "learning_rate": 3.9981881533475234e-05, |
| "loss": 0.8626, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.5630379746835443, |
| "grad_norm": 1.0063958768189707, |
| "learning_rate": 3.997938564033779e-05, |
| "loss": 0.8487, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5670886075949367, |
| "grad_norm": 0.9405172650010258, |
| "learning_rate": 3.9976728835975064e-05, |
| "loss": 0.8634, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5711392405063291, |
| "grad_norm": 1.0844569867985883, |
| "learning_rate": 3.9973911141784605e-05, |
| "loss": 0.8641, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.5751898734177215, |
| "grad_norm": 1.0417382594986184, |
| "learning_rate": 3.997093258045973e-05, |
| "loss": 0.8619, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.579240506329114, |
| "grad_norm": 0.9020695973921515, |
| "learning_rate": 3.996779317598936e-05, |
| "loss": 0.8704, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.5832911392405064, |
| "grad_norm": 1.0077231703137328, |
| "learning_rate": 3.996449295365782e-05, |
| "loss": 0.8509, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5873417721518988, |
| "grad_norm": 1.030836074272146, |
| "learning_rate": 3.996103194004467e-05, |
| "loss": 0.8609, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5913924050632912, |
| "grad_norm": 0.8611224590359324, |
| "learning_rate": 3.995741016302441e-05, |
| "loss": 0.8625, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5954430379746836, |
| "grad_norm": 1.017335992615227, |
| "learning_rate": 3.9953627651766364e-05, |
| "loss": 0.8655, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.599493670886076, |
| "grad_norm": 0.8849168474814954, |
| "learning_rate": 3.9949684436734325e-05, |
| "loss": 0.8615, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.6035443037974684, |
| "grad_norm": 1.0195525504963265, |
| "learning_rate": 3.994558054968643e-05, |
| "loss": 0.8514, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.6075949367088608, |
| "grad_norm": 0.9069800435971621, |
| "learning_rate": 3.994131602367481e-05, |
| "loss": 0.8804, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6116455696202532, |
| "grad_norm": 1.1550432782341762, |
| "learning_rate": 3.9936890893045376e-05, |
| "loss": 0.8578, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.6156962025316456, |
| "grad_norm": 0.9230263098392637, |
| "learning_rate": 3.993230519343752e-05, |
| "loss": 0.8733, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.619746835443038, |
| "grad_norm": 0.9995126567198171, |
| "learning_rate": 3.992755896178383e-05, |
| "loss": 0.8635, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.6237974683544304, |
| "grad_norm": 0.8400293937524412, |
| "learning_rate": 3.992265223630981e-05, |
| "loss": 0.8744, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.6278481012658228, |
| "grad_norm": 1.0811956000166263, |
| "learning_rate": 3.991758505653355e-05, |
| "loss": 0.8566, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.6318987341772152, |
| "grad_norm": 0.8767437610594592, |
| "learning_rate": 3.991235746326543e-05, |
| "loss": 0.8512, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6359493670886076, |
| "grad_norm": 0.9594610984978277, |
| "learning_rate": 3.9906969498607745e-05, |
| "loss": 0.863, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 0.9301002229731159, |
| "learning_rate": 3.990142120595444e-05, |
| "loss": 0.8577, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.6440506329113924, |
| "grad_norm": 1.0210785055648206, |
| "learning_rate": 3.98957126299907e-05, |
| "loss": 0.8671, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6481012658227848, |
| "grad_norm": 1.0012090147545056, |
| "learning_rate": 3.9889843816692596e-05, |
| "loss": 0.8728, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6521518987341772, |
| "grad_norm": 0.8526130938904883, |
| "learning_rate": 3.9883814813326766e-05, |
| "loss": 0.8517, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6562025316455696, |
| "grad_norm": 1.0602380461445857, |
| "learning_rate": 3.9877625668449956e-05, |
| "loss": 0.864, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.660253164556962, |
| "grad_norm": 0.8696928142801702, |
| "learning_rate": 3.98712764319087e-05, |
| "loss": 0.8481, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.6643037974683544, |
| "grad_norm": 0.9613920419422142, |
| "learning_rate": 3.9864767154838864e-05, |
| "loss": 0.8564, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6683544303797468, |
| "grad_norm": 0.982528746686371, |
| "learning_rate": 3.9858097889665277e-05, |
| "loss": 0.8626, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6724050632911392, |
| "grad_norm": 0.817050257587898, |
| "learning_rate": 3.985126869010129e-05, |
| "loss": 0.863, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6764556962025317, |
| "grad_norm": 0.986087213113688, |
| "learning_rate": 3.984427961114833e-05, |
| "loss": 0.8704, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.6805063291139241, |
| "grad_norm": 0.8666096262886734, |
| "learning_rate": 3.9837130709095475e-05, |
| "loss": 0.854, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.6845569620253165, |
| "grad_norm": 0.9465129475219728, |
| "learning_rate": 3.982982204151901e-05, |
| "loss": 0.8595, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6886075949367089, |
| "grad_norm": 0.9524330410161248, |
| "learning_rate": 3.982235366728193e-05, |
| "loss": 0.8602, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6926582278481013, |
| "grad_norm": 1.104498693989606, |
| "learning_rate": 3.9814725646533505e-05, |
| "loss": 0.8515, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6967088607594937, |
| "grad_norm": 0.9981786644465908, |
| "learning_rate": 3.9806938040708746e-05, |
| "loss": 0.8677, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.7007594936708861, |
| "grad_norm": 0.9239300497179832, |
| "learning_rate": 3.9798990912527976e-05, |
| "loss": 0.8571, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.7048101265822785, |
| "grad_norm": 1.0181648763287732, |
| "learning_rate": 3.979088432599627e-05, |
| "loss": 0.8487, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.7088607594936709, |
| "grad_norm": 0.8176363484502193, |
| "learning_rate": 3.9782618346402964e-05, |
| "loss": 0.8583, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7129113924050633, |
| "grad_norm": 0.8839501403174047, |
| "learning_rate": 3.977419304032111e-05, |
| "loss": 0.8522, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.7169620253164557, |
| "grad_norm": 0.8769295713099396, |
| "learning_rate": 3.976560847560697e-05, |
| "loss": 0.8686, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.7210126582278481, |
| "grad_norm": 0.7253185219148086, |
| "learning_rate": 3.9756864721399456e-05, |
| "loss": 0.8682, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.7250632911392405, |
| "grad_norm": 0.8154668233806455, |
| "learning_rate": 3.974796184811956e-05, |
| "loss": 0.8551, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.7291139240506329, |
| "grad_norm": 0.8058621183930925, |
| "learning_rate": 3.973889992746979e-05, |
| "loss": 0.8502, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7331645569620253, |
| "grad_norm": 0.7625048428820019, |
| "learning_rate": 3.972967903243361e-05, |
| "loss": 0.8665, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.7372151898734177, |
| "grad_norm": 0.7945922791052847, |
| "learning_rate": 3.972029923727486e-05, |
| "loss": 0.8505, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7412658227848101, |
| "grad_norm": 0.7828455173122494, |
| "learning_rate": 3.971076061753709e-05, |
| "loss": 0.8568, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7453164556962025, |
| "grad_norm": 0.7441126826495005, |
| "learning_rate": 3.9701063250043066e-05, |
| "loss": 0.8459, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7493670886075949, |
| "grad_norm": 0.8056011965714972, |
| "learning_rate": 3.969120721289402e-05, |
| "loss": 0.8573, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.7534177215189873, |
| "grad_norm": 0.8693215938577931, |
| "learning_rate": 3.9681192585469146e-05, |
| "loss": 0.8579, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.7574683544303797, |
| "grad_norm": 0.7887322799748502, |
| "learning_rate": 3.9671019448424865e-05, |
| "loss": 0.8666, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7615189873417721, |
| "grad_norm": 0.7983154665119131, |
| "learning_rate": 3.966068788369422e-05, |
| "loss": 0.8489, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7655696202531646, |
| "grad_norm": 0.7964418630462047, |
| "learning_rate": 3.965019797448622e-05, |
| "loss": 0.8538, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.769620253164557, |
| "grad_norm": 0.840740150870507, |
| "learning_rate": 3.963954980528515e-05, |
| "loss": 0.8587, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.7736708860759494, |
| "grad_norm": 0.7646465452623269, |
| "learning_rate": 3.9628743461849905e-05, |
| "loss": 0.85, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7777215189873418, |
| "grad_norm": 0.8708675193464155, |
| "learning_rate": 3.961777903121329e-05, |
| "loss": 0.8542, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7817721518987342, |
| "grad_norm": 0.9048944568960345, |
| "learning_rate": 3.960665660168131e-05, |
| "loss": 0.8427, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7858227848101266, |
| "grad_norm": 0.8876587546151471, |
| "learning_rate": 3.9595376262832485e-05, |
| "loss": 0.853, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.789873417721519, |
| "grad_norm": 0.7738223659273141, |
| "learning_rate": 3.9583938105517127e-05, |
| "loss": 0.8526, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7939240506329114, |
| "grad_norm": 0.8039537499772235, |
| "learning_rate": 3.957234222185657e-05, |
| "loss": 0.842, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7979746835443038, |
| "grad_norm": 0.826076161814544, |
| "learning_rate": 3.9560588705242474e-05, |
| "loss": 0.8431, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.8020253164556962, |
| "grad_norm": 0.7058582454186388, |
| "learning_rate": 3.954867765033605e-05, |
| "loss": 0.8708, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.8060759493670886, |
| "grad_norm": 0.8658397475225938, |
| "learning_rate": 3.953660915306728e-05, |
| "loss": 0.8593, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.810126582278481, |
| "grad_norm": 0.7601283495767898, |
| "learning_rate": 3.952438331063419e-05, |
| "loss": 0.8469, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8141772151898734, |
| "grad_norm": 0.7907070519156825, |
| "learning_rate": 3.951200022150205e-05, |
| "loss": 0.8384, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.8182278481012658, |
| "grad_norm": 0.7140891551854461, |
| "learning_rate": 3.949945998540253e-05, |
| "loss": 0.8398, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.8222784810126582, |
| "grad_norm": 0.7735399267017338, |
| "learning_rate": 3.9486762703332993e-05, |
| "loss": 0.8511, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.8263291139240506, |
| "grad_norm": 0.7421070484157085, |
| "learning_rate": 3.947390847755559e-05, |
| "loss": 0.8253, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.830379746835443, |
| "grad_norm": 0.7604837479209882, |
| "learning_rate": 3.946089741159648e-05, |
| "loss": 0.8454, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8344303797468354, |
| "grad_norm": 0.9348730287890431, |
| "learning_rate": 3.944772961024501e-05, |
| "loss": 0.8545, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.8384810126582278, |
| "grad_norm": 0.8535649004197571, |
| "learning_rate": 3.943440517955285e-05, |
| "loss": 0.839, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.8425316455696202, |
| "grad_norm": 0.7653598686279268, |
| "learning_rate": 3.9420924226833126e-05, |
| "loss": 0.8365, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8465822784810126, |
| "grad_norm": 0.902581413514666, |
| "learning_rate": 3.9407286860659566e-05, |
| "loss": 0.8386, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.850632911392405, |
| "grad_norm": 0.9175585066209366, |
| "learning_rate": 3.9393493190865657e-05, |
| "loss": 0.8456, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8546835443037974, |
| "grad_norm": 0.9023252070989161, |
| "learning_rate": 3.937954332854371e-05, |
| "loss": 0.8453, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8587341772151899, |
| "grad_norm": 0.8137365014917516, |
| "learning_rate": 3.9365437386044016e-05, |
| "loss": 0.8544, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.8627848101265823, |
| "grad_norm": 0.8704534006239567, |
| "learning_rate": 3.935117547697387e-05, |
| "loss": 0.84, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8668354430379747, |
| "grad_norm": 0.9079644653765906, |
| "learning_rate": 3.933675771619675e-05, |
| "loss": 0.842, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8708860759493671, |
| "grad_norm": 0.7005524260362889, |
| "learning_rate": 3.932218421983131e-05, |
| "loss": 0.8311, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8749367088607595, |
| "grad_norm": 0.8323059635750771, |
| "learning_rate": 3.9307455105250484e-05, |
| "loss": 0.8487, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8789873417721519, |
| "grad_norm": 0.7717466243752668, |
| "learning_rate": 3.929257049108054e-05, |
| "loss": 0.841, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8830379746835443, |
| "grad_norm": 0.7581169543485792, |
| "learning_rate": 3.927753049720011e-05, |
| "loss": 0.8295, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8870886075949367, |
| "grad_norm": 0.7779852048756503, |
| "learning_rate": 3.9262335244739234e-05, |
| "loss": 0.8506, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8911392405063291, |
| "grad_norm": 0.7374338842143232, |
| "learning_rate": 3.92469848560784e-05, |
| "loss": 0.8281, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8951898734177215, |
| "grad_norm": 0.7514393532595064, |
| "learning_rate": 3.923147945484751e-05, |
| "loss": 0.8367, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8992405063291139, |
| "grad_norm": 0.7514479853012387, |
| "learning_rate": 3.9215819165924956e-05, |
| "loss": 0.842, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.9032911392405063, |
| "grad_norm": 0.773756683172937, |
| "learning_rate": 3.920000411543654e-05, |
| "loss": 0.8448, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.9073417721518987, |
| "grad_norm": 0.7272193603130684, |
| "learning_rate": 3.9184034430754495e-05, |
| "loss": 0.8169, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.9113924050632911, |
| "grad_norm": 0.8429926873107061, |
| "learning_rate": 3.916791024049648e-05, |
| "loss": 0.8431, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9154430379746835, |
| "grad_norm": 0.7850768337257563, |
| "learning_rate": 3.91516316745245e-05, |
| "loss": 0.8447, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.9194936708860759, |
| "grad_norm": 0.8490466724317715, |
| "learning_rate": 3.913519886394389e-05, |
| "loss": 0.8576, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.9235443037974683, |
| "grad_norm": 0.6808863404030197, |
| "learning_rate": 3.911861194110225e-05, |
| "loss": 0.8334, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.9275949367088607, |
| "grad_norm": 0.9365751871865775, |
| "learning_rate": 3.910187103958837e-05, |
| "loss": 0.8518, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.9316455696202531, |
| "grad_norm": 0.8569260755536936, |
| "learning_rate": 3.908497629423117e-05, |
| "loss": 0.8341, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9356962025316455, |
| "grad_norm": 0.8866802829538432, |
| "learning_rate": 3.9067927841098614e-05, |
| "loss": 0.8314, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.9397468354430379, |
| "grad_norm": 0.7648595251560635, |
| "learning_rate": 3.9050725817496594e-05, |
| "loss": 0.8334, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.9437974683544303, |
| "grad_norm": 0.8096037626691264, |
| "learning_rate": 3.9033370361967844e-05, |
| "loss": 0.8357, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.9478481012658228, |
| "grad_norm": 0.7478638535632488, |
| "learning_rate": 3.901586161429081e-05, |
| "loss": 0.8189, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.9518987341772152, |
| "grad_norm": 0.7434831179999241, |
| "learning_rate": 3.8998199715478545e-05, |
| "loss": 0.83, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9559493670886076, |
| "grad_norm": 0.8045161142464332, |
| "learning_rate": 3.8980384807777564e-05, |
| "loss": 0.826, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 0.6883425019017421, |
| "learning_rate": 3.896241703466667e-05, |
| "loss": 0.8308, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9640506329113924, |
| "grad_norm": 0.8173612006473409, |
| "learning_rate": 3.894429654085585e-05, |
| "loss": 0.8336, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9681012658227848, |
| "grad_norm": 0.7210556035967629, |
| "learning_rate": 3.892602347228505e-05, |
| "loss": 0.8439, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.9721518987341772, |
| "grad_norm": 0.8205453611237419, |
| "learning_rate": 3.890759797612307e-05, |
| "loss": 0.837, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9762025316455696, |
| "grad_norm": 0.7737692361348217, |
| "learning_rate": 3.888902020076632e-05, |
| "loss": 0.8455, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.980253164556962, |
| "grad_norm": 0.7661820721915703, |
| "learning_rate": 3.887029029583764e-05, |
| "loss": 0.8352, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9843037974683544, |
| "grad_norm": 0.7868417666757117, |
| "learning_rate": 3.8851408412185125e-05, |
| "loss": 0.8411, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9883544303797468, |
| "grad_norm": 0.7670742798327198, |
| "learning_rate": 3.8832374701880855e-05, |
| "loss": 0.8406, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9924050632911392, |
| "grad_norm": 0.7279174099763203, |
| "learning_rate": 3.881318931821972e-05, |
| "loss": 0.8342, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9964556962025316, |
| "grad_norm": 0.8231301571643947, |
| "learning_rate": 3.879385241571817e-05, |
| "loss": 0.8289, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0020253164556963, |
| "grad_norm": 0.7449662499634566, |
| "learning_rate": 3.8774364150112955e-05, |
| "loss": 0.7342, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.0060759493670886, |
| "grad_norm": 0.9634426049295803, |
| "learning_rate": 3.8754724678359884e-05, |
| "loss": 0.649, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.010126582278481, |
| "grad_norm": 0.9094222211144166, |
| "learning_rate": 3.873493415863256e-05, |
| "loss": 0.6296, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.0141772151898734, |
| "grad_norm": 1.0914099495542213, |
| "learning_rate": 3.871499275032111e-05, |
| "loss": 0.6296, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0182278481012659, |
| "grad_norm": 0.8019702066747577, |
| "learning_rate": 3.869490061403091e-05, |
| "loss": 0.6128, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.0222784810126582, |
| "grad_norm": 0.7490244399307664, |
| "learning_rate": 3.867465791158124e-05, |
| "loss": 0.6153, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.0263291139240507, |
| "grad_norm": 0.9016270401121719, |
| "learning_rate": 3.865426480600407e-05, |
| "loss": 0.6131, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.030379746835443, |
| "grad_norm": 0.7418799051857413, |
| "learning_rate": 3.863372146154264e-05, |
| "loss": 0.6196, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.0344303797468355, |
| "grad_norm": 0.8543761195367882, |
| "learning_rate": 3.861302804365024e-05, |
| "loss": 0.6159, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.0384810126582278, |
| "grad_norm": 0.7984968104292919, |
| "learning_rate": 3.85921847189888e-05, |
| "loss": 0.6161, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.0425316455696203, |
| "grad_norm": 0.9664154711378431, |
| "learning_rate": 3.85711916554276e-05, |
| "loss": 0.6278, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.0465822784810126, |
| "grad_norm": 0.8369524784993899, |
| "learning_rate": 3.85500490220419e-05, |
| "loss": 0.6138, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.0506329113924051, |
| "grad_norm": 0.763789788683521, |
| "learning_rate": 3.852875698911154e-05, |
| "loss": 0.6167, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.0546835443037974, |
| "grad_norm": 0.8300242637400909, |
| "learning_rate": 3.850731572811963e-05, |
| "loss": 0.6272, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.05873417721519, |
| "grad_norm": 0.8045655627085616, |
| "learning_rate": 3.848572541175116e-05, |
| "loss": 0.6169, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.0627848101265822, |
| "grad_norm": 0.7317961333086007, |
| "learning_rate": 3.846398621389154e-05, |
| "loss": 0.6093, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.0668354430379747, |
| "grad_norm": 0.7614034317649729, |
| "learning_rate": 3.84420983096253e-05, |
| "loss": 0.6124, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.070886075949367, |
| "grad_norm": 0.6926018279374617, |
| "learning_rate": 3.8420061875234606e-05, |
| "loss": 0.6101, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.0749367088607595, |
| "grad_norm": 0.7375902088518602, |
| "learning_rate": 3.839787708819787e-05, |
| "loss": 0.6401, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.0789873417721518, |
| "grad_norm": 0.7124881044184483, |
| "learning_rate": 3.8375544127188325e-05, |
| "loss": 0.6114, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.0830379746835443, |
| "grad_norm": 0.726972432197345, |
| "learning_rate": 3.8353063172072564e-05, |
| "loss": 0.6302, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.0870886075949366, |
| "grad_norm": 0.6886723089112419, |
| "learning_rate": 3.8330434403909105e-05, |
| "loss": 0.6115, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.0911392405063292, |
| "grad_norm": 0.76684997945864, |
| "learning_rate": 3.8307658004946934e-05, |
| "loss": 0.6115, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.0951898734177214, |
| "grad_norm": 0.8253717313773576, |
| "learning_rate": 3.8284734158624046e-05, |
| "loss": 0.6102, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.099240506329114, |
| "grad_norm": 0.7911596758498968, |
| "learning_rate": 3.826166304956594e-05, |
| "loss": 0.6168, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.1032911392405063, |
| "grad_norm": 0.7600991659271169, |
| "learning_rate": 3.8238444863584164e-05, |
| "loss": 0.6357, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.1073417721518988, |
| "grad_norm": 0.8819116134973256, |
| "learning_rate": 3.821507978767479e-05, |
| "loss": 0.6084, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.111392405063291, |
| "grad_norm": 0.7746644747740855, |
| "learning_rate": 3.819156801001693e-05, |
| "loss": 0.6152, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.1154430379746836, |
| "grad_norm": 0.8219029852053167, |
| "learning_rate": 3.816790971997121e-05, |
| "loss": 0.6025, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.1194936708860759, |
| "grad_norm": 0.8410819601199923, |
| "learning_rate": 3.8144105108078246e-05, |
| "loss": 0.6041, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.1235443037974684, |
| "grad_norm": 0.7935390725943551, |
| "learning_rate": 3.81201543660571e-05, |
| "loss": 0.6115, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.1275949367088607, |
| "grad_norm": 0.7793971105656661, |
| "learning_rate": 3.809605768680377e-05, |
| "loss": 0.6292, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.1316455696202532, |
| "grad_norm": 0.7749391860357793, |
| "learning_rate": 3.807181526438958e-05, |
| "loss": 0.6068, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.1356962025316455, |
| "grad_norm": 0.9232864349449538, |
| "learning_rate": 3.8047427294059697e-05, |
| "loss": 0.6237, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.139746835443038, |
| "grad_norm": 0.7485020575044041, |
| "learning_rate": 3.802289397223145e-05, |
| "loss": 0.6042, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.1437974683544303, |
| "grad_norm": 0.7931445987204211, |
| "learning_rate": 3.7998215496492854e-05, |
| "loss": 0.6224, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.1478481012658228, |
| "grad_norm": 0.7376487416518286, |
| "learning_rate": 3.797339206560096e-05, |
| "loss": 0.6235, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.1518987341772151, |
| "grad_norm": 0.8211687747606675, |
| "learning_rate": 3.794842387948027e-05, |
| "loss": 0.6045, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.1559493670886076, |
| "grad_norm": 0.7857930612375194, |
| "learning_rate": 3.7923311139221114e-05, |
| "loss": 0.6146, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.16, |
| "grad_norm": 0.825817271934582, |
| "learning_rate": 3.7898054047078054e-05, |
| "loss": 0.6513, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.1640506329113924, |
| "grad_norm": 0.7064257922760896, |
| "learning_rate": 3.787265280646825e-05, |
| "loss": 0.6179, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.1681012658227847, |
| "grad_norm": 0.7854681198893417, |
| "learning_rate": 3.7847107621969786e-05, |
| "loss": 0.6151, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.1721518987341772, |
| "grad_norm": 0.7148291766495133, |
| "learning_rate": 3.7821418699320064e-05, |
| "loss": 0.6243, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.1762025316455695, |
| "grad_norm": 0.7850327256603525, |
| "learning_rate": 3.7795586245414145e-05, |
| "loss": 0.6274, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.180253164556962, |
| "grad_norm": 0.679384658803092, |
| "learning_rate": 3.776961046830306e-05, |
| "loss": 0.6179, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.1843037974683543, |
| "grad_norm": 0.7340266042575074, |
| "learning_rate": 3.774349157719215e-05, |
| "loss": 0.625, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.1883544303797469, |
| "grad_norm": 0.7307095279937353, |
| "learning_rate": 3.7717229782439365e-05, |
| "loss": 0.6195, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.1924050632911392, |
| "grad_norm": 0.6509581253630271, |
| "learning_rate": 3.769082529555359e-05, |
| "loss": 0.6067, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.1964556962025317, |
| "grad_norm": 0.7176051474980809, |
| "learning_rate": 3.766427832919294e-05, |
| "loss": 0.6146, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.200506329113924, |
| "grad_norm": 0.6689198165977466, |
| "learning_rate": 3.7637589097163024e-05, |
| "loss": 0.6258, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.2045569620253165, |
| "grad_norm": 0.6806204268691518, |
| "learning_rate": 3.761075781441526e-05, |
| "loss": 0.6304, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.2086075949367088, |
| "grad_norm": 0.7350715762844463, |
| "learning_rate": 3.75837846970451e-05, |
| "loss": 0.5916, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.2126582278481013, |
| "grad_norm": 0.6993849373689153, |
| "learning_rate": 3.755666996229032e-05, |
| "loss": 0.6248, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.2167088607594936, |
| "grad_norm": 0.7421473645316573, |
| "learning_rate": 3.752941382852927e-05, |
| "loss": 0.6426, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.220759493670886, |
| "grad_norm": 0.7423406018007634, |
| "learning_rate": 3.7502016515279115e-05, |
| "loss": 0.6217, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.2248101265822784, |
| "grad_norm": 0.7004318058395987, |
| "learning_rate": 3.7474478243194043e-05, |
| "loss": 0.6436, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.228860759493671, |
| "grad_norm": 0.7387962265938995, |
| "learning_rate": 3.744679923406351e-05, |
| "loss": 0.6193, |
| "step": 303 |
| }, |
| { |
| "epoch": 1.2329113924050632, |
| "grad_norm": 0.7196745700319443, |
| "learning_rate": 3.741897971081043e-05, |
| "loss": 0.6374, |
| "step": 304 |
| }, |
| { |
| "epoch": 1.2369620253164557, |
| "grad_norm": 0.6930736535272889, |
| "learning_rate": 3.739101989748946e-05, |
| "loss": 0.6293, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.241012658227848, |
| "grad_norm": 0.7171481658131652, |
| "learning_rate": 3.7362920019285066e-05, |
| "loss": 0.63, |
| "step": 306 |
| }, |
| { |
| "epoch": 1.2450632911392405, |
| "grad_norm": 0.6527542781250701, |
| "learning_rate": 3.73346803025098e-05, |
| "loss": 0.6176, |
| "step": 307 |
| }, |
| { |
| "epoch": 1.2491139240506328, |
| "grad_norm": 0.7260808527224281, |
| "learning_rate": 3.730630097460247e-05, |
| "loss": 0.6174, |
| "step": 308 |
| }, |
| { |
| "epoch": 1.2531645569620253, |
| "grad_norm": 0.7375840750825797, |
| "learning_rate": 3.727778226412628e-05, |
| "loss": 0.6355, |
| "step": 309 |
| }, |
| { |
| "epoch": 1.2572151898734178, |
| "grad_norm": 0.7111025626160066, |
| "learning_rate": 3.7249124400767006e-05, |
| "loss": 0.63, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.2612658227848101, |
| "grad_norm": 0.7092612078729401, |
| "learning_rate": 3.722032761533114e-05, |
| "loss": 0.6188, |
| "step": 311 |
| }, |
| { |
| "epoch": 1.2653164556962024, |
| "grad_norm": 0.7934818385998761, |
| "learning_rate": 3.719139213974403e-05, |
| "loss": 0.6336, |
| "step": 312 |
| }, |
| { |
| "epoch": 1.269367088607595, |
| "grad_norm": 0.6676551548725619, |
| "learning_rate": 3.7162318207048006e-05, |
| "loss": 0.6195, |
| "step": 313 |
| }, |
| { |
| "epoch": 1.2734177215189875, |
| "grad_norm": 0.8299110767458143, |
| "learning_rate": 3.713310605140055e-05, |
| "loss": 0.6313, |
| "step": 314 |
| }, |
| { |
| "epoch": 1.2774683544303798, |
| "grad_norm": 0.6867485386380483, |
| "learning_rate": 3.710375590807233e-05, |
| "loss": 0.6264, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.281518987341772, |
| "grad_norm": 0.7607869121709716, |
| "learning_rate": 3.7074268013445365e-05, |
| "loss": 0.6353, |
| "step": 316 |
| }, |
| { |
| "epoch": 1.2855696202531646, |
| "grad_norm": 0.7139079688015018, |
| "learning_rate": 3.7044642605011114e-05, |
| "loss": 0.6086, |
| "step": 317 |
| }, |
| { |
| "epoch": 1.289620253164557, |
| "grad_norm": 0.7628900113144564, |
| "learning_rate": 3.701487992136854e-05, |
| "loss": 0.6069, |
| "step": 318 |
| }, |
| { |
| "epoch": 1.2936708860759494, |
| "grad_norm": 0.7271475440943644, |
| "learning_rate": 3.69849802022222e-05, |
| "loss": 0.6146, |
| "step": 319 |
| }, |
| { |
| "epoch": 1.2977215189873417, |
| "grad_norm": 0.7352570009606743, |
| "learning_rate": 3.6954943688380334e-05, |
| "loss": 0.616, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.3017721518987342, |
| "grad_norm": 0.7712623748350835, |
| "learning_rate": 3.692477062175289e-05, |
| "loss": 0.6408, |
| "step": 321 |
| }, |
| { |
| "epoch": 1.3058227848101267, |
| "grad_norm": 0.738687017795699, |
| "learning_rate": 3.689446124534958e-05, |
| "loss": 0.6491, |
| "step": 322 |
| }, |
| { |
| "epoch": 1.309873417721519, |
| "grad_norm": 0.7978223766275364, |
| "learning_rate": 3.686401580327799e-05, |
| "loss": 0.6282, |
| "step": 323 |
| }, |
| { |
| "epoch": 1.3139240506329113, |
| "grad_norm": 0.6844648810432086, |
| "learning_rate": 3.683343454074149e-05, |
| "loss": 0.633, |
| "step": 324 |
| }, |
| { |
| "epoch": 1.3179746835443038, |
| "grad_norm": 0.7263551000587974, |
| "learning_rate": 3.6802717704037386e-05, |
| "loss": 0.6128, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.3220253164556963, |
| "grad_norm": 0.7168459248256762, |
| "learning_rate": 3.6771865540554855e-05, |
| "loss": 0.6468, |
| "step": 326 |
| }, |
| { |
| "epoch": 1.3260759493670886, |
| "grad_norm": 0.7331321397673953, |
| "learning_rate": 3.674087829877297e-05, |
| "loss": 0.6341, |
| "step": 327 |
| }, |
| { |
| "epoch": 1.330126582278481, |
| "grad_norm": 0.7488622657092525, |
| "learning_rate": 3.6709756228258735e-05, |
| "loss": 0.6147, |
| "step": 328 |
| }, |
| { |
| "epoch": 1.3341772151898734, |
| "grad_norm": 0.8691552110989047, |
| "learning_rate": 3.667849957966501e-05, |
| "loss": 0.6127, |
| "step": 329 |
| }, |
| { |
| "epoch": 1.338227848101266, |
| "grad_norm": 0.7071151713655003, |
| "learning_rate": 3.6647108604728546e-05, |
| "loss": 0.6305, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3422784810126582, |
| "grad_norm": 0.8445122730095403, |
| "learning_rate": 3.661558355626795e-05, |
| "loss": 0.6367, |
| "step": 331 |
| }, |
| { |
| "epoch": 1.3463291139240505, |
| "grad_norm": 0.799326320691731, |
| "learning_rate": 3.658392468818163e-05, |
| "loss": 0.6324, |
| "step": 332 |
| }, |
| { |
| "epoch": 1.350379746835443, |
| "grad_norm": 0.6848818841288423, |
| "learning_rate": 3.655213225544574e-05, |
| "loss": 0.6321, |
| "step": 333 |
| }, |
| { |
| "epoch": 1.3544303797468356, |
| "grad_norm": 0.7534108061709844, |
| "learning_rate": 3.652020651411218e-05, |
| "loss": 0.6065, |
| "step": 334 |
| }, |
| { |
| "epoch": 1.3584810126582278, |
| "grad_norm": 0.6736216086271144, |
| "learning_rate": 3.6488147721306474e-05, |
| "loss": 0.625, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.3625316455696201, |
| "grad_norm": 0.7949342295738299, |
| "learning_rate": 3.645595613522574e-05, |
| "loss": 0.6322, |
| "step": 336 |
| }, |
| { |
| "epoch": 1.3665822784810127, |
| "grad_norm": 0.7226895209640531, |
| "learning_rate": 3.642363201513657e-05, |
| "loss": 0.6308, |
| "step": 337 |
| }, |
| { |
| "epoch": 1.3706329113924052, |
| "grad_norm": 0.8158201776150257, |
| "learning_rate": 3.6391175621373006e-05, |
| "loss": 0.6148, |
| "step": 338 |
| }, |
| { |
| "epoch": 1.3746835443037975, |
| "grad_norm": 0.8305828723910289, |
| "learning_rate": 3.6358587215334355e-05, |
| "loss": 0.6129, |
| "step": 339 |
| }, |
| { |
| "epoch": 1.3787341772151898, |
| "grad_norm": 0.7030432723603098, |
| "learning_rate": 3.632586705948318e-05, |
| "loss": 0.6172, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.3827848101265823, |
| "grad_norm": 0.7848326886618123, |
| "learning_rate": 3.629301541734311e-05, |
| "loss": 0.6281, |
| "step": 341 |
| }, |
| { |
| "epoch": 1.3868354430379748, |
| "grad_norm": 0.8137528017494025, |
| "learning_rate": 3.626003255349676e-05, |
| "loss": 0.6489, |
| "step": 342 |
| }, |
| { |
| "epoch": 1.390886075949367, |
| "grad_norm": 0.7123984171068289, |
| "learning_rate": 3.622691873358357e-05, |
| "loss": 0.643, |
| "step": 343 |
| }, |
| { |
| "epoch": 1.3949367088607594, |
| "grad_norm": 0.8276538913553455, |
| "learning_rate": 3.61936742242977e-05, |
| "loss": 0.6198, |
| "step": 344 |
| }, |
| { |
| "epoch": 1.398987341772152, |
| "grad_norm": 0.8536274460439136, |
| "learning_rate": 3.6160299293385864e-05, |
| "loss": 0.6394, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4030379746835444, |
| "grad_norm": 0.7594030519318117, |
| "learning_rate": 3.612679420964516e-05, |
| "loss": 0.6383, |
| "step": 346 |
| }, |
| { |
| "epoch": 1.4070886075949367, |
| "grad_norm": 0.8733744392208227, |
| "learning_rate": 3.609315924292092e-05, |
| "loss": 0.6287, |
| "step": 347 |
| }, |
| { |
| "epoch": 1.411139240506329, |
| "grad_norm": 0.7211642701732904, |
| "learning_rate": 3.6059394664104554e-05, |
| "loss": 0.6221, |
| "step": 348 |
| }, |
| { |
| "epoch": 1.4151898734177215, |
| "grad_norm": 0.8030806757734419, |
| "learning_rate": 3.602550074513133e-05, |
| "loss": 0.6349, |
| "step": 349 |
| }, |
| { |
| "epoch": 1.419240506329114, |
| "grad_norm": 0.7093133107251683, |
| "learning_rate": 3.599147775897822e-05, |
| "loss": 0.6356, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4232911392405063, |
| "grad_norm": 0.7304263321068867, |
| "learning_rate": 3.595732597966167e-05, |
| "loss": 0.6357, |
| "step": 351 |
| }, |
| { |
| "epoch": 1.4273417721518986, |
| "grad_norm": 0.7080321758715589, |
| "learning_rate": 3.592304568223542e-05, |
| "loss": 0.6211, |
| "step": 352 |
| }, |
| { |
| "epoch": 1.4313924050632911, |
| "grad_norm": 0.7096421007090666, |
| "learning_rate": 3.588863714278826e-05, |
| "loss": 0.6137, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.4354430379746836, |
| "grad_norm": 0.8142665528168903, |
| "learning_rate": 3.585410063844186e-05, |
| "loss": 0.6161, |
| "step": 354 |
| }, |
| { |
| "epoch": 1.439493670886076, |
| "grad_norm": 0.7043816117814798, |
| "learning_rate": 3.581943644734846e-05, |
| "loss": 0.6353, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.4435443037974682, |
| "grad_norm": 0.7671874761449864, |
| "learning_rate": 3.578464484868869e-05, |
| "loss": 0.6168, |
| "step": 356 |
| }, |
| { |
| "epoch": 1.4475949367088607, |
| "grad_norm": 0.7474363218669201, |
| "learning_rate": 3.5749726122669316e-05, |
| "loss": 0.6339, |
| "step": 357 |
| }, |
| { |
| "epoch": 1.4516455696202533, |
| "grad_norm": 0.6750934107720682, |
| "learning_rate": 3.5714680550520943e-05, |
| "loss": 0.6262, |
| "step": 358 |
| }, |
| { |
| "epoch": 1.4556962025316456, |
| "grad_norm": 0.6845317007089304, |
| "learning_rate": 3.5679508414495794e-05, |
| "loss": 0.6307, |
| "step": 359 |
| }, |
| { |
| "epoch": 1.4597468354430378, |
| "grad_norm": 0.7193050731779733, |
| "learning_rate": 3.564420999786543e-05, |
| "loss": 0.6303, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.4637974683544304, |
| "grad_norm": 0.6655894977012052, |
| "learning_rate": 3.560878558491842e-05, |
| "loss": 0.6208, |
| "step": 361 |
| }, |
| { |
| "epoch": 1.4678481012658229, |
| "grad_norm": 0.7481642292478164, |
| "learning_rate": 3.5573235460958145e-05, |
| "loss": 0.6255, |
| "step": 362 |
| }, |
| { |
| "epoch": 1.4718987341772152, |
| "grad_norm": 0.6954256570242934, |
| "learning_rate": 3.553755991230039e-05, |
| "loss": 0.6445, |
| "step": 363 |
| }, |
| { |
| "epoch": 1.4759493670886075, |
| "grad_norm": 0.6742305921106254, |
| "learning_rate": 3.5501759226271144e-05, |
| "loss": 0.6288, |
| "step": 364 |
| }, |
| { |
| "epoch": 1.48, |
| "grad_norm": 0.7747773500203502, |
| "learning_rate": 3.546583369120419e-05, |
| "loss": 0.631, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.4840506329113925, |
| "grad_norm": 0.7493909267529125, |
| "learning_rate": 3.5429783596438864e-05, |
| "loss": 0.6303, |
| "step": 366 |
| }, |
| { |
| "epoch": 1.4881012658227848, |
| "grad_norm": 0.7246164833902315, |
| "learning_rate": 3.539360923231766e-05, |
| "loss": 0.6194, |
| "step": 367 |
| }, |
| { |
| "epoch": 1.492151898734177, |
| "grad_norm": 0.7260252416230841, |
| "learning_rate": 3.535731089018394e-05, |
| "loss": 0.6232, |
| "step": 368 |
| }, |
| { |
| "epoch": 1.4962025316455696, |
| "grad_norm": 0.6959655817962439, |
| "learning_rate": 3.532088886237956e-05, |
| "loss": 0.6298, |
| "step": 369 |
| }, |
| { |
| "epoch": 1.5002531645569621, |
| "grad_norm": 0.7678446925659274, |
| "learning_rate": 3.528434344224253e-05, |
| "loss": 0.6277, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5043037974683544, |
| "grad_norm": 0.6952522583154441, |
| "learning_rate": 3.524767492410464e-05, |
| "loss": 0.6424, |
| "step": 371 |
| }, |
| { |
| "epoch": 1.5083544303797467, |
| "grad_norm": 0.770666662132528, |
| "learning_rate": 3.521088360328908e-05, |
| "loss": 0.6305, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.5124050632911392, |
| "grad_norm": 0.7744864539118026, |
| "learning_rate": 3.517396977610811e-05, |
| "loss": 0.6405, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.5164556962025317, |
| "grad_norm": 0.6618354809725653, |
| "learning_rate": 3.5136933739860595e-05, |
| "loss": 0.6387, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.520506329113924, |
| "grad_norm": 0.7610564821690794, |
| "learning_rate": 3.509977579282971e-05, |
| "loss": 0.6322, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.5245569620253163, |
| "grad_norm": 0.6909616569191669, |
| "learning_rate": 3.5062496234280424e-05, |
| "loss": 0.6297, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.5286075949367088, |
| "grad_norm": 0.7132011507091318, |
| "learning_rate": 3.502509536445719e-05, |
| "loss": 0.6441, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.5326582278481014, |
| "grad_norm": 0.7247111834189273, |
| "learning_rate": 3.498757348458147e-05, |
| "loss": 0.6161, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.5367088607594936, |
| "grad_norm": 0.637860727062116, |
| "learning_rate": 3.4949930896849324e-05, |
| "loss": 0.6346, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.540759493670886, |
| "grad_norm": 0.7311136907184047, |
| "learning_rate": 3.491216790442899e-05, |
| "loss": 0.6201, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.5448101265822785, |
| "grad_norm": 0.6611816895391148, |
| "learning_rate": 3.487428481145839e-05, |
| "loss": 0.6164, |
| "step": 381 |
| }, |
| { |
| "epoch": 1.548860759493671, |
| "grad_norm": 0.6886033092922661, |
| "learning_rate": 3.483628192304278e-05, |
| "loss": 0.6112, |
| "step": 382 |
| }, |
| { |
| "epoch": 1.5529113924050633, |
| "grad_norm": 0.6504922813675613, |
| "learning_rate": 3.479815954525219e-05, |
| "loss": 0.6486, |
| "step": 383 |
| }, |
| { |
| "epoch": 1.5569620253164556, |
| "grad_norm": 0.7155451046430811, |
| "learning_rate": 3.475991798511899e-05, |
| "loss": 0.631, |
| "step": 384 |
| }, |
| { |
| "epoch": 1.561012658227848, |
| "grad_norm": 0.6943736767001474, |
| "learning_rate": 3.4721557550635464e-05, |
| "loss": 0.632, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.5650632911392406, |
| "grad_norm": 0.6988530320691505, |
| "learning_rate": 3.468307855075128e-05, |
| "loss": 0.6316, |
| "step": 386 |
| }, |
| { |
| "epoch": 1.5691139240506329, |
| "grad_norm": 0.7374721116305034, |
| "learning_rate": 3.4644481295371005e-05, |
| "loss": 0.6249, |
| "step": 387 |
| }, |
| { |
| "epoch": 1.5731645569620252, |
| "grad_norm": 0.7217496261870702, |
| "learning_rate": 3.460576609535163e-05, |
| "loss": 0.6459, |
| "step": 388 |
| }, |
| { |
| "epoch": 1.5772151898734177, |
| "grad_norm": 0.7146167847515351, |
| "learning_rate": 3.456693326250006e-05, |
| "loss": 0.6256, |
| "step": 389 |
| }, |
| { |
| "epoch": 1.5812658227848102, |
| "grad_norm": 0.6804597736940884, |
| "learning_rate": 3.452798310957058e-05, |
| "loss": 0.6469, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.5853164556962025, |
| "grad_norm": 0.683244559713358, |
| "learning_rate": 3.4488915950262386e-05, |
| "loss": 0.6215, |
| "step": 391 |
| }, |
| { |
| "epoch": 1.5893670886075948, |
| "grad_norm": 0.7057139748126431, |
| "learning_rate": 3.4449732099216985e-05, |
| "loss": 0.6336, |
| "step": 392 |
| }, |
| { |
| "epoch": 1.5934177215189873, |
| "grad_norm": 0.6395466820372433, |
| "learning_rate": 3.441043187201574e-05, |
| "loss": 0.6172, |
| "step": 393 |
| }, |
| { |
| "epoch": 1.5974683544303798, |
| "grad_norm": 0.7068014839982861, |
| "learning_rate": 3.437101558517728e-05, |
| "loss": 0.622, |
| "step": 394 |
| }, |
| { |
| "epoch": 1.6015189873417721, |
| "grad_norm": 0.6448097964299728, |
| "learning_rate": 3.433148355615496e-05, |
| "loss": 0.6393, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.6055696202531644, |
| "grad_norm": 0.6947297618087565, |
| "learning_rate": 3.4291836103334294e-05, |
| "loss": 0.6414, |
| "step": 396 |
| }, |
| { |
| "epoch": 1.609620253164557, |
| "grad_norm": 0.6915861151684243, |
| "learning_rate": 3.425207354603043e-05, |
| "loss": 0.6389, |
| "step": 397 |
| }, |
| { |
| "epoch": 1.6136708860759494, |
| "grad_norm": 0.6860026495353657, |
| "learning_rate": 3.421219620448553e-05, |
| "loss": 0.6059, |
| "step": 398 |
| }, |
| { |
| "epoch": 1.6177215189873417, |
| "grad_norm": 0.7054083564637496, |
| "learning_rate": 3.417220439986623e-05, |
| "loss": 0.6328, |
| "step": 399 |
| }, |
| { |
| "epoch": 1.621772151898734, |
| "grad_norm": 0.7261935278691993, |
| "learning_rate": 3.4132098454261024e-05, |
| "loss": 0.6205, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.6258227848101265, |
| "grad_norm": 0.679356261147221, |
| "learning_rate": 3.4091878690677676e-05, |
| "loss": 0.6384, |
| "step": 401 |
| }, |
| { |
| "epoch": 1.629873417721519, |
| "grad_norm": 0.7049566057966344, |
| "learning_rate": 3.405154543304065e-05, |
| "loss": 0.628, |
| "step": 402 |
| }, |
| { |
| "epoch": 1.6339240506329114, |
| "grad_norm": 0.6987372637388837, |
| "learning_rate": 3.401109900618843e-05, |
| "loss": 0.6325, |
| "step": 403 |
| }, |
| { |
| "epoch": 1.6379746835443036, |
| "grad_norm": 0.7142503988417289, |
| "learning_rate": 3.3970539735870996e-05, |
| "loss": 0.6463, |
| "step": 404 |
| }, |
| { |
| "epoch": 1.6420253164556962, |
| "grad_norm": 0.6738352276339201, |
| "learning_rate": 3.392986794874714e-05, |
| "loss": 0.6224, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.6460759493670887, |
| "grad_norm": 0.7173491532031494, |
| "learning_rate": 3.388908397238184e-05, |
| "loss": 0.6263, |
| "step": 406 |
| }, |
| { |
| "epoch": 1.650126582278481, |
| "grad_norm": 0.6760466105304832, |
| "learning_rate": 3.384818813524362e-05, |
| "loss": 0.6296, |
| "step": 407 |
| }, |
| { |
| "epoch": 1.6541772151898733, |
| "grad_norm": 0.7415919770414472, |
| "learning_rate": 3.380718076670195e-05, |
| "loss": 0.608, |
| "step": 408 |
| }, |
| { |
| "epoch": 1.6582278481012658, |
| "grad_norm": 0.6967386732084805, |
| "learning_rate": 3.376606219702454e-05, |
| "loss": 0.6313, |
| "step": 409 |
| }, |
| { |
| "epoch": 1.6622784810126583, |
| "grad_norm": 0.6713367519759615, |
| "learning_rate": 3.372483275737468e-05, |
| "loss": 0.6262, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.6663291139240506, |
| "grad_norm": 0.6946116417322848, |
| "learning_rate": 3.368349277980861e-05, |
| "loss": 0.6268, |
| "step": 411 |
| }, |
| { |
| "epoch": 1.6703797468354429, |
| "grad_norm": 0.6587102977087628, |
| "learning_rate": 3.3642042597272844e-05, |
| "loss": 0.6277, |
| "step": 412 |
| }, |
| { |
| "epoch": 1.6744303797468354, |
| "grad_norm": 0.6282186719362767, |
| "learning_rate": 3.360048254360144e-05, |
| "loss": 0.6209, |
| "step": 413 |
| }, |
| { |
| "epoch": 1.678481012658228, |
| "grad_norm": 0.6727762176727695, |
| "learning_rate": 3.355881295351336e-05, |
| "loss": 0.6364, |
| "step": 414 |
| }, |
| { |
| "epoch": 1.6825316455696202, |
| "grad_norm": 0.6311536935775973, |
| "learning_rate": 3.351703416260975e-05, |
| "loss": 0.6316, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.6865822784810125, |
| "grad_norm": 0.6612795305902537, |
| "learning_rate": 3.347514650737126e-05, |
| "loss": 0.6303, |
| "step": 416 |
| }, |
| { |
| "epoch": 1.690632911392405, |
| "grad_norm": 0.6713363580347264, |
| "learning_rate": 3.3433150325155295e-05, |
| "loss": 0.6333, |
| "step": 417 |
| }, |
| { |
| "epoch": 1.6946835443037975, |
| "grad_norm": 0.6791069711451984, |
| "learning_rate": 3.339104595419334e-05, |
| "loss": 0.6189, |
| "step": 418 |
| }, |
| { |
| "epoch": 1.6987341772151898, |
| "grad_norm": 0.6240582472556034, |
| "learning_rate": 3.3348833733588204e-05, |
| "loss": 0.6274, |
| "step": 419 |
| }, |
| { |
| "epoch": 1.7027848101265821, |
| "grad_norm": 0.6703449818857061, |
| "learning_rate": 3.3306514003311305e-05, |
| "loss": 0.6086, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7068354430379746, |
| "grad_norm": 0.6255468808568768, |
| "learning_rate": 3.326408710419996e-05, |
| "loss": 0.6172, |
| "step": 421 |
| }, |
| { |
| "epoch": 1.7108860759493671, |
| "grad_norm": 0.6986367998709774, |
| "learning_rate": 3.322155337795454e-05, |
| "loss": 0.6222, |
| "step": 422 |
| }, |
| { |
| "epoch": 1.7149367088607594, |
| "grad_norm": 0.650574513207251, |
| "learning_rate": 3.317891316713587e-05, |
| "loss": 0.6394, |
| "step": 423 |
| }, |
| { |
| "epoch": 1.7189873417721517, |
| "grad_norm": 0.6670694820214249, |
| "learning_rate": 3.313616681516231e-05, |
| "loss": 0.6363, |
| "step": 424 |
| }, |
| { |
| "epoch": 1.7230379746835442, |
| "grad_norm": 0.6674381486305097, |
| "learning_rate": 3.309331466630713e-05, |
| "loss": 0.6379, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.7270886075949368, |
| "grad_norm": 0.7230618841694668, |
| "learning_rate": 3.305035706569563e-05, |
| "loss": 0.6199, |
| "step": 426 |
| }, |
| { |
| "epoch": 1.7311392405063293, |
| "grad_norm": 0.6490486190408545, |
| "learning_rate": 3.3007294359302433e-05, |
| "loss": 0.6251, |
| "step": 427 |
| }, |
| { |
| "epoch": 1.7351898734177216, |
| "grad_norm": 0.7423194648843672, |
| "learning_rate": 3.296412689394864e-05, |
| "loss": 0.6424, |
| "step": 428 |
| }, |
| { |
| "epoch": 1.7392405063291139, |
| "grad_norm": 0.6465398412959641, |
| "learning_rate": 3.292085501729909e-05, |
| "loss": 0.6163, |
| "step": 429 |
| }, |
| { |
| "epoch": 1.7432911392405064, |
| "grad_norm": 0.6124662889650322, |
| "learning_rate": 3.2877479077859534e-05, |
| "loss": 0.6312, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.747341772151899, |
| "grad_norm": 0.6868433166204331, |
| "learning_rate": 3.283399942497381e-05, |
| "loss": 0.6224, |
| "step": 431 |
| }, |
| { |
| "epoch": 1.7513924050632912, |
| "grad_norm": 0.6811452925819548, |
| "learning_rate": 3.279041640882108e-05, |
| "loss": 0.6169, |
| "step": 432 |
| }, |
| { |
| "epoch": 1.7554430379746835, |
| "grad_norm": 0.628192416745338, |
| "learning_rate": 3.2746730380412964e-05, |
| "loss": 0.6128, |
| "step": 433 |
| }, |
| { |
| "epoch": 1.759493670886076, |
| "grad_norm": 0.694042249984295, |
| "learning_rate": 3.2702941691590726e-05, |
| "loss": 0.6253, |
| "step": 434 |
| }, |
| { |
| "epoch": 1.7635443037974685, |
| "grad_norm": 0.6372850296016521, |
| "learning_rate": 3.265905069502244e-05, |
| "loss": 0.627, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.7675949367088608, |
| "grad_norm": 0.6650523997839635, |
| "learning_rate": 3.261505774420016e-05, |
| "loss": 0.6182, |
| "step": 436 |
| }, |
| { |
| "epoch": 1.771645569620253, |
| "grad_norm": 0.6451378443389371, |
| "learning_rate": 3.257096319343707e-05, |
| "loss": 0.6399, |
| "step": 437 |
| }, |
| { |
| "epoch": 1.7756962025316456, |
| "grad_norm": 0.7271512414464208, |
| "learning_rate": 3.2526767397864614e-05, |
| "loss": 0.6316, |
| "step": 438 |
| }, |
| { |
| "epoch": 1.7797468354430381, |
| "grad_norm": 0.6447729683130268, |
| "learning_rate": 3.248247071342966e-05, |
| "loss": 0.6181, |
| "step": 439 |
| }, |
| { |
| "epoch": 1.7837974683544304, |
| "grad_norm": 0.6970881584430296, |
| "learning_rate": 3.243807349689161e-05, |
| "loss": 0.6161, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.7878481012658227, |
| "grad_norm": 0.7754271431106726, |
| "learning_rate": 3.2393576105819544e-05, |
| "loss": 0.6315, |
| "step": 441 |
| }, |
| { |
| "epoch": 1.7918987341772152, |
| "grad_norm": 0.6685043923629355, |
| "learning_rate": 3.2348978898589333e-05, |
| "loss": 0.627, |
| "step": 442 |
| }, |
| { |
| "epoch": 1.7959493670886078, |
| "grad_norm": 0.734671286599649, |
| "learning_rate": 3.230428223438075e-05, |
| "loss": 0.6133, |
| "step": 443 |
| }, |
| { |
| "epoch": 1.8, |
| "grad_norm": 0.6876316806439512, |
| "learning_rate": 3.225948647317459e-05, |
| "loss": 0.6438, |
| "step": 444 |
| }, |
| { |
| "epoch": 1.8040506329113923, |
| "grad_norm": 0.7903399614851329, |
| "learning_rate": 3.2214591975749745e-05, |
| "loss": 0.6148, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.8081012658227849, |
| "grad_norm": 0.6737280509502993, |
| "learning_rate": 3.216959910368034e-05, |
| "loss": 0.6384, |
| "step": 446 |
| }, |
| { |
| "epoch": 1.8121518987341774, |
| "grad_norm": 0.73234658674552, |
| "learning_rate": 3.212450821933277e-05, |
| "loss": 0.6379, |
| "step": 447 |
| }, |
| { |
| "epoch": 1.8162025316455697, |
| "grad_norm": 0.6688668683752333, |
| "learning_rate": 3.207931968586281e-05, |
| "loss": 0.6315, |
| "step": 448 |
| }, |
| { |
| "epoch": 1.820253164556962, |
| "grad_norm": 0.6849526761915837, |
| "learning_rate": 3.203403386721272e-05, |
| "loss": 0.6243, |
| "step": 449 |
| }, |
| { |
| "epoch": 1.8243037974683545, |
| "grad_norm": 0.6927140685835738, |
| "learning_rate": 3.1988651128108245e-05, |
| "loss": 0.6125, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.828354430379747, |
| "grad_norm": 0.653545827314719, |
| "learning_rate": 3.194317183405573e-05, |
| "loss": 0.6286, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.8324050632911393, |
| "grad_norm": 0.6759532610314132, |
| "learning_rate": 3.189759635133914e-05, |
| "loss": 0.62, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.8364556962025316, |
| "grad_norm": 0.6606131379110115, |
| "learning_rate": 3.185192504701718e-05, |
| "loss": 0.6358, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.840506329113924, |
| "grad_norm": 0.653286715776416, |
| "learning_rate": 3.1806158288920234e-05, |
| "loss": 0.637, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.8445569620253166, |
| "grad_norm": 0.6413118507457694, |
| "learning_rate": 3.1760296445647477e-05, |
| "loss": 0.6346, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.848607594936709, |
| "grad_norm": 0.7051211851307787, |
| "learning_rate": 3.1714339886563896e-05, |
| "loss": 0.6383, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.8526582278481012, |
| "grad_norm": 0.6118483642291277, |
| "learning_rate": 3.166828898179731e-05, |
| "loss": 0.6281, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.8567088607594937, |
| "grad_norm": 0.678666207787659, |
| "learning_rate": 3.162214410223536e-05, |
| "loss": 0.6232, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.8607594936708862, |
| "grad_norm": 0.6346324179713924, |
| "learning_rate": 3.157590561952257e-05, |
| "loss": 0.6002, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.8648101265822785, |
| "grad_norm": 0.6956428324569127, |
| "learning_rate": 3.152957390605732e-05, |
| "loss": 0.6271, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.8688607594936708, |
| "grad_norm": 0.6494142597123529, |
| "learning_rate": 3.148314933498886e-05, |
| "loss": 0.606, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.8729113924050633, |
| "grad_norm": 0.6711456997726617, |
| "learning_rate": 3.143663228021431e-05, |
| "loss": 0.6121, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.8769620253164558, |
| "grad_norm": 0.6652575641576396, |
| "learning_rate": 3.1390023116375624e-05, |
| "loss": 0.6292, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.8810126582278481, |
| "grad_norm": 0.6685347085351938, |
| "learning_rate": 3.134332221885661e-05, |
| "loss": 0.6328, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.8850632911392404, |
| "grad_norm": 0.6574153669313453, |
| "learning_rate": 3.129652996377987e-05, |
| "loss": 0.6254, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.889113924050633, |
| "grad_norm": 0.7037290617156495, |
| "learning_rate": 3.12496467280038e-05, |
| "loss": 0.6156, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.8931645569620255, |
| "grad_norm": 0.6422355704084896, |
| "learning_rate": 3.120267288911952e-05, |
| "loss": 0.6245, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.8972151898734178, |
| "grad_norm": 0.6474637560379554, |
| "learning_rate": 3.11556088254479e-05, |
| "loss": 0.6327, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.90126582278481, |
| "grad_norm": 0.6896181871414927, |
| "learning_rate": 3.11084549160364e-05, |
| "loss": 0.6038, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.9053164556962026, |
| "grad_norm": 0.6566737088298771, |
| "learning_rate": 3.106121154065615e-05, |
| "loss": 0.6187, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.909367088607595, |
| "grad_norm": 0.7058288915618445, |
| "learning_rate": 3.1013879079798805e-05, |
| "loss": 0.6147, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.9134177215189874, |
| "grad_norm": 0.7142696593269809, |
| "learning_rate": 3.096645791467348e-05, |
| "loss": 0.6147, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.9174683544303797, |
| "grad_norm": 0.6500506433238104, |
| "learning_rate": 3.091894842720373e-05, |
| "loss": 0.6329, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.9215189873417722, |
| "grad_norm": 0.7409791219202984, |
| "learning_rate": 3.0871351000024425e-05, |
| "loss": 0.6331, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.9255696202531647, |
| "grad_norm": 0.6394757119175072, |
| "learning_rate": 3.0823666016478716e-05, |
| "loss": 0.6182, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.929620253164557, |
| "grad_norm": 0.6898376509809921, |
| "learning_rate": 3.0775893860614896e-05, |
| "loss": 0.6262, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.9336708860759493, |
| "grad_norm": 0.6821178841710213, |
| "learning_rate": 3.0728034917183336e-05, |
| "loss": 0.6147, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.9377215189873418, |
| "grad_norm": 0.6704170435635483, |
| "learning_rate": 3.06800895716334e-05, |
| "loss": 0.6124, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.9417721518987343, |
| "grad_norm": 0.697601429763175, |
| "learning_rate": 3.063205821011029e-05, |
| "loss": 0.6275, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.9458227848101266, |
| "grad_norm": 0.6900680025697885, |
| "learning_rate": 3.0583941219452016e-05, |
| "loss": 0.634, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.949873417721519, |
| "grad_norm": 0.6960200473663498, |
| "learning_rate": 3.053573898718618e-05, |
| "loss": 0.6171, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.9539240506329114, |
| "grad_norm": 0.6718010652456085, |
| "learning_rate": 3.0487451901526956e-05, |
| "loss": 0.6167, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.957974683544304, |
| "grad_norm": 0.7184378794570101, |
| "learning_rate": 3.0439080351371875e-05, |
| "loss": 0.6324, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.9620253164556962, |
| "grad_norm": 0.6426537941094919, |
| "learning_rate": 3.0390624726298764e-05, |
| "loss": 0.6225, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.9660759493670885, |
| "grad_norm": 0.6904702813324076, |
| "learning_rate": 3.034208541656255e-05, |
| "loss": 0.6376, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.970126582278481, |
| "grad_norm": 0.6427893500738635, |
| "learning_rate": 3.029346281309218e-05, |
| "loss": 0.6215, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.9741772151898735, |
| "grad_norm": 0.6298192147825805, |
| "learning_rate": 3.0244757307487415e-05, |
| "loss": 0.6156, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.9782278481012658, |
| "grad_norm": 0.6795042945639659, |
| "learning_rate": 3.019596929201569e-05, |
| "loss": 0.612, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.9822784810126581, |
| "grad_norm": 0.6916457871159036, |
| "learning_rate": 3.0147099159608985e-05, |
| "loss": 0.6314, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.9863291139240506, |
| "grad_norm": 0.6310458255490643, |
| "learning_rate": 3.0098147303860616e-05, |
| "loss": 0.6042, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.9903797468354432, |
| "grad_norm": 0.7022808118411401, |
| "learning_rate": 3.0049114119022117e-05, |
| "loss": 0.624, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.9944303797468355, |
| "grad_norm": 0.6285895164471746, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.6331, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.9984810126582278, |
| "grad_norm": 0.6510675770362796, |
| "learning_rate": 2.995080534235264e-05, |
| "loss": 0.6312, |
| "step": 493 |
| }, |
| { |
| "epoch": 2.0040506329113925, |
| "grad_norm": 1.2129812077190039, |
| "learning_rate": 2.9901530542287044e-05, |
| "loss": 0.382, |
| "step": 494 |
| }, |
| { |
| "epoch": 2.008101265822785, |
| "grad_norm": 0.8309313759459208, |
| "learning_rate": 2.9852175996655676e-05, |
| "loss": 0.3711, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.012151898734177, |
| "grad_norm": 1.1842967528606996, |
| "learning_rate": 2.980274210295326e-05, |
| "loss": 0.3457, |
| "step": 496 |
| }, |
| { |
| "epoch": 2.0162025316455696, |
| "grad_norm": 1.1467194887051795, |
| "learning_rate": 2.9753229259313578e-05, |
| "loss": 0.3506, |
| "step": 497 |
| }, |
| { |
| "epoch": 2.020253164556962, |
| "grad_norm": 0.7174007035756387, |
| "learning_rate": 2.9703637864506274e-05, |
| "loss": 0.343, |
| "step": 498 |
| }, |
| { |
| "epoch": 2.0243037974683546, |
| "grad_norm": 0.6609001133965249, |
| "learning_rate": 2.965396831793362e-05, |
| "loss": 0.3414, |
| "step": 499 |
| }, |
| { |
| "epoch": 2.0283544303797467, |
| "grad_norm": 0.6491807540445887, |
| "learning_rate": 2.9604221019627316e-05, |
| "loss": 0.3529, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0324050632911392, |
| "grad_norm": 0.6632487979011855, |
| "learning_rate": 2.955439637024526e-05, |
| "loss": 0.3358, |
| "step": 501 |
| }, |
| { |
| "epoch": 2.0364556962025318, |
| "grad_norm": 0.6132745177548243, |
| "learning_rate": 2.9504494771068334e-05, |
| "loss": 0.3432, |
| "step": 502 |
| }, |
| { |
| "epoch": 2.0405063291139243, |
| "grad_norm": 0.677419854553167, |
| "learning_rate": 2.9454516623997156e-05, |
| "loss": 0.3499, |
| "step": 503 |
| }, |
| { |
| "epoch": 2.0445569620253163, |
| "grad_norm": 0.635039805524588, |
| "learning_rate": 2.9404462331548847e-05, |
| "loss": 0.3582, |
| "step": 504 |
| }, |
| { |
| "epoch": 2.048607594936709, |
| "grad_norm": 0.6470598182443104, |
| "learning_rate": 2.93543322968538e-05, |
| "loss": 0.3406, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.0526582278481014, |
| "grad_norm": 0.6904103784346265, |
| "learning_rate": 2.9304126923652428e-05, |
| "loss": 0.3533, |
| "step": 506 |
| }, |
| { |
| "epoch": 2.056708860759494, |
| "grad_norm": 0.6233579612641248, |
| "learning_rate": 2.9253846616291896e-05, |
| "loss": 0.3342, |
| "step": 507 |
| }, |
| { |
| "epoch": 2.060759493670886, |
| "grad_norm": 0.661182241364242, |
| "learning_rate": 2.9203491779722896e-05, |
| "loss": 0.3386, |
| "step": 508 |
| }, |
| { |
| "epoch": 2.0648101265822785, |
| "grad_norm": 0.6490389770932777, |
| "learning_rate": 2.9153062819496357e-05, |
| "loss": 0.3403, |
| "step": 509 |
| }, |
| { |
| "epoch": 2.068860759493671, |
| "grad_norm": 0.6402456519030116, |
| "learning_rate": 2.9102560141760178e-05, |
| "loss": 0.3487, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.0729113924050635, |
| "grad_norm": 0.6084525307477499, |
| "learning_rate": 2.9051984153256004e-05, |
| "loss": 0.349, |
| "step": 511 |
| }, |
| { |
| "epoch": 2.0769620253164556, |
| "grad_norm": 0.6657489949168601, |
| "learning_rate": 2.900133526131588e-05, |
| "loss": 0.3413, |
| "step": 512 |
| }, |
| { |
| "epoch": 2.081012658227848, |
| "grad_norm": 0.6100147889230164, |
| "learning_rate": 2.8950613873859025e-05, |
| "loss": 0.3332, |
| "step": 513 |
| }, |
| { |
| "epoch": 2.0850632911392406, |
| "grad_norm": 0.6473946837068787, |
| "learning_rate": 2.8899820399388515e-05, |
| "loss": 0.3376, |
| "step": 514 |
| }, |
| { |
| "epoch": 2.089113924050633, |
| "grad_norm": 0.6086402321009835, |
| "learning_rate": 2.8848955246988012e-05, |
| "loss": 0.3449, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.093164556962025, |
| "grad_norm": 0.6344466538087099, |
| "learning_rate": 2.879801882631847e-05, |
| "loss": 0.3602, |
| "step": 516 |
| }, |
| { |
| "epoch": 2.0972151898734177, |
| "grad_norm": 0.6102395189402726, |
| "learning_rate": 2.8747011547614808e-05, |
| "loss": 0.3355, |
| "step": 517 |
| }, |
| { |
| "epoch": 2.1012658227848102, |
| "grad_norm": 0.604423167279989, |
| "learning_rate": 2.8695933821682635e-05, |
| "loss": 0.3446, |
| "step": 518 |
| }, |
| { |
| "epoch": 2.1053164556962027, |
| "grad_norm": 0.6166389802803812, |
| "learning_rate": 2.864478605989494e-05, |
| "loss": 0.3394, |
| "step": 519 |
| }, |
| { |
| "epoch": 2.109367088607595, |
| "grad_norm": 0.5872272050001399, |
| "learning_rate": 2.8593568674188765e-05, |
| "loss": 0.3429, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.1134177215189873, |
| "grad_norm": 0.619353786548746, |
| "learning_rate": 2.8542282077061892e-05, |
| "loss": 0.3195, |
| "step": 521 |
| }, |
| { |
| "epoch": 2.11746835443038, |
| "grad_norm": 0.6527045598917862, |
| "learning_rate": 2.8490926681569523e-05, |
| "loss": 0.3447, |
| "step": 522 |
| }, |
| { |
| "epoch": 2.1215189873417724, |
| "grad_norm": 0.6344473495391194, |
| "learning_rate": 2.8439502901320956e-05, |
| "loss": 0.3285, |
| "step": 523 |
| }, |
| { |
| "epoch": 2.1255696202531644, |
| "grad_norm": 0.6066329139593923, |
| "learning_rate": 2.8388011150476237e-05, |
| "loss": 0.3343, |
| "step": 524 |
| }, |
| { |
| "epoch": 2.129620253164557, |
| "grad_norm": 0.605176575264733, |
| "learning_rate": 2.8336451843742866e-05, |
| "loss": 0.3385, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.1336708860759495, |
| "grad_norm": 0.6175501889891015, |
| "learning_rate": 2.8284825396372387e-05, |
| "loss": 0.3298, |
| "step": 526 |
| }, |
| { |
| "epoch": 2.137721518987342, |
| "grad_norm": 0.5991550075520566, |
| "learning_rate": 2.8233132224157132e-05, |
| "loss": 0.325, |
| "step": 527 |
| }, |
| { |
| "epoch": 2.141772151898734, |
| "grad_norm": 0.653822123786142, |
| "learning_rate": 2.8181372743426805e-05, |
| "loss": 0.3398, |
| "step": 528 |
| }, |
| { |
| "epoch": 2.1458227848101266, |
| "grad_norm": 0.6087403465902292, |
| "learning_rate": 2.8129547371045128e-05, |
| "loss": 0.3446, |
| "step": 529 |
| }, |
| { |
| "epoch": 2.149873417721519, |
| "grad_norm": 0.6441894069791443, |
| "learning_rate": 2.8077656524406534e-05, |
| "loss": 0.3349, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.1539240506329116, |
| "grad_norm": 0.610178495200193, |
| "learning_rate": 2.802570062143278e-05, |
| "loss": 0.3496, |
| "step": 531 |
| }, |
| { |
| "epoch": 2.1579746835443037, |
| "grad_norm": 0.6135917083746116, |
| "learning_rate": 2.7973680080569555e-05, |
| "loss": 0.3373, |
| "step": 532 |
| }, |
| { |
| "epoch": 2.162025316455696, |
| "grad_norm": 0.626850258133874, |
| "learning_rate": 2.792159532078314e-05, |
| "loss": 0.3469, |
| "step": 533 |
| }, |
| { |
| "epoch": 2.1660759493670887, |
| "grad_norm": 0.6303176619268304, |
| "learning_rate": 2.7869446761557033e-05, |
| "loss": 0.3535, |
| "step": 534 |
| }, |
| { |
| "epoch": 2.170126582278481, |
| "grad_norm": 0.6130734372202532, |
| "learning_rate": 2.781723482288857e-05, |
| "loss": 0.3276, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.1741772151898733, |
| "grad_norm": 0.5832139829863382, |
| "learning_rate": 2.7764959925285517e-05, |
| "loss": 0.3381, |
| "step": 536 |
| }, |
| { |
| "epoch": 2.178227848101266, |
| "grad_norm": 18.08758092899207, |
| "learning_rate": 2.771262248976272e-05, |
| "loss": 0.339, |
| "step": 537 |
| }, |
| { |
| "epoch": 2.1822784810126583, |
| "grad_norm": 1.0979276305384145, |
| "learning_rate": 2.7660222937838677e-05, |
| "loss": 0.3528, |
| "step": 538 |
| }, |
| { |
| "epoch": 2.186329113924051, |
| "grad_norm": 4.62947647429476, |
| "learning_rate": 2.7607761691532186e-05, |
| "loss": 0.3394, |
| "step": 539 |
| }, |
| { |
| "epoch": 2.190379746835443, |
| "grad_norm": 5.320259781526249, |
| "learning_rate": 2.7555239173358916e-05, |
| "loss": 0.3532, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.1944303797468354, |
| "grad_norm": 6.527552822870686, |
| "learning_rate": 2.7502655806328e-05, |
| "loss": 0.3685, |
| "step": 541 |
| }, |
| { |
| "epoch": 2.198481012658228, |
| "grad_norm": 1.198204093898129, |
| "learning_rate": 2.7450012013938648e-05, |
| "loss": 0.3567, |
| "step": 542 |
| }, |
| { |
| "epoch": 2.2025316455696204, |
| "grad_norm": 0.662935630112486, |
| "learning_rate": 2.739730822017673e-05, |
| "loss": 0.339, |
| "step": 543 |
| }, |
| { |
| "epoch": 2.2065822784810125, |
| "grad_norm": 0.892422857663448, |
| "learning_rate": 2.7344544849511355e-05, |
| "loss": 0.3624, |
| "step": 544 |
| }, |
| { |
| "epoch": 2.210632911392405, |
| "grad_norm": 0.7978995580917061, |
| "learning_rate": 2.7291722326891456e-05, |
| "loss": 0.3619, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.2146835443037975, |
| "grad_norm": 0.736815467391293, |
| "learning_rate": 2.723884107774236e-05, |
| "loss": 0.3512, |
| "step": 546 |
| }, |
| { |
| "epoch": 2.21873417721519, |
| "grad_norm": 0.8144574354453488, |
| "learning_rate": 2.718590152796239e-05, |
| "loss": 0.3583, |
| "step": 547 |
| }, |
| { |
| "epoch": 2.222784810126582, |
| "grad_norm": 0.7046434758251732, |
| "learning_rate": 2.71329041039194e-05, |
| "loss": 0.3564, |
| "step": 548 |
| }, |
| { |
| "epoch": 2.2268354430379746, |
| "grad_norm": 0.6948367681868699, |
| "learning_rate": 2.7079849232447357e-05, |
| "loss": 0.3533, |
| "step": 549 |
| }, |
| { |
| "epoch": 2.230886075949367, |
| "grad_norm": 0.7516422564528196, |
| "learning_rate": 2.7026737340842895e-05, |
| "loss": 0.3484, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.2349367088607597, |
| "grad_norm": 0.6380137529200857, |
| "learning_rate": 2.697356885686189e-05, |
| "loss": 0.3399, |
| "step": 551 |
| }, |
| { |
| "epoch": 2.2389873417721518, |
| "grad_norm": 0.6978010018856712, |
| "learning_rate": 2.6920344208716014e-05, |
| "loss": 0.346, |
| "step": 552 |
| }, |
| { |
| "epoch": 2.2430379746835443, |
| "grad_norm": 0.6589749524353324, |
| "learning_rate": 2.6867063825069252e-05, |
| "loss": 0.3473, |
| "step": 553 |
| }, |
| { |
| "epoch": 2.247088607594937, |
| "grad_norm": 0.6966849315520764, |
| "learning_rate": 2.6813728135034494e-05, |
| "loss": 0.3603, |
| "step": 554 |
| }, |
| { |
| "epoch": 2.2511392405063293, |
| "grad_norm": 0.6525865395928766, |
| "learning_rate": 2.6760337568170056e-05, |
| "loss": 0.355, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.2551898734177214, |
| "grad_norm": 0.6594144690532489, |
| "learning_rate": 2.6706892554476226e-05, |
| "loss": 0.3426, |
| "step": 556 |
| }, |
| { |
| "epoch": 2.259240506329114, |
| "grad_norm": 0.6408490941800271, |
| "learning_rate": 2.6653393524391795e-05, |
| "loss": 0.3324, |
| "step": 557 |
| }, |
| { |
| "epoch": 2.2632911392405064, |
| "grad_norm": 0.6519504153112066, |
| "learning_rate": 2.6599840908790592e-05, |
| "loss": 0.3558, |
| "step": 558 |
| }, |
| { |
| "epoch": 2.267341772151899, |
| "grad_norm": 0.6389040010547141, |
| "learning_rate": 2.6546235138978028e-05, |
| "loss": 0.353, |
| "step": 559 |
| }, |
| { |
| "epoch": 2.271392405063291, |
| "grad_norm": 0.6507279457945371, |
| "learning_rate": 2.6492576646687597e-05, |
| "loss": 0.352, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.2754430379746835, |
| "grad_norm": 0.6507101700682499, |
| "learning_rate": 2.6438865864077425e-05, |
| "loss": 0.355, |
| "step": 561 |
| }, |
| { |
| "epoch": 2.279493670886076, |
| "grad_norm": 0.6272377498803237, |
| "learning_rate": 2.6385103223726766e-05, |
| "loss": 0.3441, |
| "step": 562 |
| }, |
| { |
| "epoch": 2.2835443037974685, |
| "grad_norm": 0.6309390716624587, |
| "learning_rate": 2.6331289158632537e-05, |
| "loss": 0.3462, |
| "step": 563 |
| }, |
| { |
| "epoch": 2.2875949367088606, |
| "grad_norm": 0.6291650951052353, |
| "learning_rate": 2.6277424102205817e-05, |
| "loss": 0.3547, |
| "step": 564 |
| }, |
| { |
| "epoch": 2.291645569620253, |
| "grad_norm": 0.6277305229884153, |
| "learning_rate": 2.6223508488268374e-05, |
| "loss": 0.3332, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.2956962025316456, |
| "grad_norm": 0.6397566643316039, |
| "learning_rate": 2.6169542751049148e-05, |
| "loss": 0.3436, |
| "step": 566 |
| }, |
| { |
| "epoch": 2.299746835443038, |
| "grad_norm": 0.6287917368458666, |
| "learning_rate": 2.6115527325180754e-05, |
| "loss": 0.3542, |
| "step": 567 |
| }, |
| { |
| "epoch": 2.3037974683544302, |
| "grad_norm": 0.599067625573669, |
| "learning_rate": 2.606146264569603e-05, |
| "loss": 0.3497, |
| "step": 568 |
| }, |
| { |
| "epoch": 2.3078481012658227, |
| "grad_norm": 0.6300013279509861, |
| "learning_rate": 2.6007349148024447e-05, |
| "loss": 0.3575, |
| "step": 569 |
| }, |
| { |
| "epoch": 2.3118987341772153, |
| "grad_norm": 0.608763641222871, |
| "learning_rate": 2.5953187267988694e-05, |
| "loss": 0.3404, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.3159493670886078, |
| "grad_norm": 0.7157768188647274, |
| "learning_rate": 2.5898977441801097e-05, |
| "loss": 0.3526, |
| "step": 571 |
| }, |
| { |
| "epoch": 2.32, |
| "grad_norm": 0.5899944534855082, |
| "learning_rate": 2.584472010606015e-05, |
| "loss": 0.3354, |
| "step": 572 |
| }, |
| { |
| "epoch": 2.3240506329113924, |
| "grad_norm": 0.6744366011488463, |
| "learning_rate": 2.5790415697746976e-05, |
| "loss": 0.3441, |
| "step": 573 |
| }, |
| { |
| "epoch": 2.328101265822785, |
| "grad_norm": 0.5973069599746402, |
| "learning_rate": 2.5736064654221808e-05, |
| "loss": 0.3377, |
| "step": 574 |
| }, |
| { |
| "epoch": 2.3321518987341774, |
| "grad_norm": 0.6231521589349257, |
| "learning_rate": 2.568166741322048e-05, |
| "loss": 0.3621, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.3362025316455695, |
| "grad_norm": 0.6205131143093275, |
| "learning_rate": 2.56272244128509e-05, |
| "loss": 0.3588, |
| "step": 576 |
| }, |
| { |
| "epoch": 2.340253164556962, |
| "grad_norm": 0.6221223004848307, |
| "learning_rate": 2.55727360915895e-05, |
| "loss": 0.3521, |
| "step": 577 |
| }, |
| { |
| "epoch": 2.3443037974683545, |
| "grad_norm": 0.6058741873553822, |
| "learning_rate": 2.5518202888277734e-05, |
| "loss": 0.3343, |
| "step": 578 |
| }, |
| { |
| "epoch": 2.348354430379747, |
| "grad_norm": 0.611121225500417, |
| "learning_rate": 2.5463625242118523e-05, |
| "loss": 0.3541, |
| "step": 579 |
| }, |
| { |
| "epoch": 2.352405063291139, |
| "grad_norm": 0.6223611309660897, |
| "learning_rate": 2.5409003592672723e-05, |
| "loss": 0.3604, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.3564556962025316, |
| "grad_norm": 0.5985915785586614, |
| "learning_rate": 2.535433837985559e-05, |
| "loss": 0.3512, |
| "step": 581 |
| }, |
| { |
| "epoch": 2.360506329113924, |
| "grad_norm": 0.614131702572587, |
| "learning_rate": 2.529963004393324e-05, |
| "loss": 0.3483, |
| "step": 582 |
| }, |
| { |
| "epoch": 2.3645569620253166, |
| "grad_norm": 0.6089185845987745, |
| "learning_rate": 2.524487902551908e-05, |
| "loss": 0.3346, |
| "step": 583 |
| }, |
| { |
| "epoch": 2.3686075949367087, |
| "grad_norm": 0.6138964691462223, |
| "learning_rate": 2.519008576557029e-05, |
| "loss": 0.3308, |
| "step": 584 |
| }, |
| { |
| "epoch": 2.372658227848101, |
| "grad_norm": 0.606830863740006, |
| "learning_rate": 2.5135250705384254e-05, |
| "loss": 0.3441, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.3767088607594937, |
| "grad_norm": 0.6177615873884438, |
| "learning_rate": 2.5080374286595007e-05, |
| "loss": 0.3426, |
| "step": 586 |
| }, |
| { |
| "epoch": 2.3807594936708862, |
| "grad_norm": 0.5948130663895523, |
| "learning_rate": 2.5025456951169677e-05, |
| "loss": 0.3652, |
| "step": 587 |
| }, |
| { |
| "epoch": 2.3848101265822783, |
| "grad_norm": 0.6406605587925153, |
| "learning_rate": 2.4970499141404942e-05, |
| "loss": 0.3466, |
| "step": 588 |
| }, |
| { |
| "epoch": 2.388860759493671, |
| "grad_norm": 0.5788474816226282, |
| "learning_rate": 2.491550129992345e-05, |
| "loss": 0.3621, |
| "step": 589 |
| }, |
| { |
| "epoch": 2.3929113924050633, |
| "grad_norm": 0.6521932588058904, |
| "learning_rate": 2.486046386967024e-05, |
| "loss": 0.357, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.396962025316456, |
| "grad_norm": 0.5855219505618974, |
| "learning_rate": 2.4805387293909214e-05, |
| "loss": 0.3467, |
| "step": 591 |
| }, |
| { |
| "epoch": 2.401012658227848, |
| "grad_norm": 0.5989105312102184, |
| "learning_rate": 2.4750272016219552e-05, |
| "loss": 0.3627, |
| "step": 592 |
| }, |
| { |
| "epoch": 2.4050632911392404, |
| "grad_norm": 0.6046780580599569, |
| "learning_rate": 2.4695118480492114e-05, |
| "loss": 0.3486, |
| "step": 593 |
| }, |
| { |
| "epoch": 2.409113924050633, |
| "grad_norm": 0.5933636822150639, |
| "learning_rate": 2.4639927130925898e-05, |
| "loss": 0.3485, |
| "step": 594 |
| }, |
| { |
| "epoch": 2.4131645569620255, |
| "grad_norm": 0.602456990423662, |
| "learning_rate": 2.458469841202444e-05, |
| "loss": 0.3552, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.4172151898734175, |
| "grad_norm": 0.6020047392708382, |
| "learning_rate": 2.452943276859226e-05, |
| "loss": 0.3532, |
| "step": 596 |
| }, |
| { |
| "epoch": 2.42126582278481, |
| "grad_norm": 0.5833572919921017, |
| "learning_rate": 2.447413064573125e-05, |
| "loss": 0.3596, |
| "step": 597 |
| }, |
| { |
| "epoch": 2.4253164556962026, |
| "grad_norm": 0.5827414657322204, |
| "learning_rate": 2.4418792488837095e-05, |
| "loss": 0.3532, |
| "step": 598 |
| }, |
| { |
| "epoch": 2.429367088607595, |
| "grad_norm": 0.6297884322228177, |
| "learning_rate": 2.4363418743595713e-05, |
| "loss": 0.354, |
| "step": 599 |
| }, |
| { |
| "epoch": 2.433417721518987, |
| "grad_norm": 0.5913158878953363, |
| "learning_rate": 2.430800985597963e-05, |
| "loss": 0.355, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4374683544303797, |
| "grad_norm": 0.5945668556897776, |
| "learning_rate": 2.4252566272244415e-05, |
| "loss": 0.3486, |
| "step": 601 |
| }, |
| { |
| "epoch": 2.441518987341772, |
| "grad_norm": 0.5985271705105742, |
| "learning_rate": 2.4197088438925063e-05, |
| "loss": 0.3498, |
| "step": 602 |
| }, |
| { |
| "epoch": 2.4455696202531647, |
| "grad_norm": 0.5727954331932286, |
| "learning_rate": 2.4141576802832417e-05, |
| "loss": 0.344, |
| "step": 603 |
| }, |
| { |
| "epoch": 2.449620253164557, |
| "grad_norm": 0.5859274961727603, |
| "learning_rate": 2.408603181104957e-05, |
| "loss": 0.3613, |
| "step": 604 |
| }, |
| { |
| "epoch": 2.4536708860759493, |
| "grad_norm": 0.5964290865552684, |
| "learning_rate": 2.4030453910928245e-05, |
| "loss": 0.3405, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.457721518987342, |
| "grad_norm": 0.582354990946231, |
| "learning_rate": 2.397484355008521e-05, |
| "loss": 0.3524, |
| "step": 606 |
| }, |
| { |
| "epoch": 2.4617721518987343, |
| "grad_norm": 0.590077448983923, |
| "learning_rate": 2.3919201176398662e-05, |
| "loss": 0.3483, |
| "step": 607 |
| }, |
| { |
| "epoch": 2.4658227848101264, |
| "grad_norm": 0.5917328982962321, |
| "learning_rate": 2.3863527238004633e-05, |
| "loss": 0.3585, |
| "step": 608 |
| }, |
| { |
| "epoch": 2.469873417721519, |
| "grad_norm": 0.5727056179443517, |
| "learning_rate": 2.380782218329337e-05, |
| "loss": 0.353, |
| "step": 609 |
| }, |
| { |
| "epoch": 2.4739240506329114, |
| "grad_norm": 0.59389310720546, |
| "learning_rate": 2.3752086460905725e-05, |
| "loss": 0.3618, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.477974683544304, |
| "grad_norm": 0.5953266711286123, |
| "learning_rate": 2.3696320519729544e-05, |
| "loss": 0.3543, |
| "step": 611 |
| }, |
| { |
| "epoch": 2.482025316455696, |
| "grad_norm": 0.5855054610768015, |
| "learning_rate": 2.3640524808896045e-05, |
| "loss": 0.3494, |
| "step": 612 |
| }, |
| { |
| "epoch": 2.4860759493670885, |
| "grad_norm": 0.5920223146950461, |
| "learning_rate": 2.3584699777776222e-05, |
| "loss": 0.3473, |
| "step": 613 |
| }, |
| { |
| "epoch": 2.490126582278481, |
| "grad_norm": 0.631242902705408, |
| "learning_rate": 2.3528845875977195e-05, |
| "loss": 0.3593, |
| "step": 614 |
| }, |
| { |
| "epoch": 2.4941772151898736, |
| "grad_norm": 0.5967184535911321, |
| "learning_rate": 2.3472963553338614e-05, |
| "loss": 0.3516, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.4982278481012656, |
| "grad_norm": 0.6166527124494349, |
| "learning_rate": 2.341705325992901e-05, |
| "loss": 0.3435, |
| "step": 616 |
| }, |
| { |
| "epoch": 2.502278481012658, |
| "grad_norm": 0.6019963307747058, |
| "learning_rate": 2.336111544604222e-05, |
| "loss": 0.3494, |
| "step": 617 |
| }, |
| { |
| "epoch": 2.5063291139240507, |
| "grad_norm": 0.6023298083655474, |
| "learning_rate": 2.33051505621937e-05, |
| "loss": 0.3506, |
| "step": 618 |
| }, |
| { |
| "epoch": 2.510379746835443, |
| "grad_norm": 0.6010930472607541, |
| "learning_rate": 2.324915905911693e-05, |
| "loss": 0.3507, |
| "step": 619 |
| }, |
| { |
| "epoch": 2.5144303797468357, |
| "grad_norm": 0.5952386707783621, |
| "learning_rate": 2.319314138775977e-05, |
| "loss": 0.3426, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.5184810126582278, |
| "grad_norm": 0.6125016472709686, |
| "learning_rate": 2.3137097999280856e-05, |
| "loss": 0.3421, |
| "step": 621 |
| }, |
| { |
| "epoch": 2.5225316455696203, |
| "grad_norm": 0.5801630518469453, |
| "learning_rate": 2.308102934504593e-05, |
| "loss": 0.3533, |
| "step": 622 |
| }, |
| { |
| "epoch": 2.526582278481013, |
| "grad_norm": 0.615179605783056, |
| "learning_rate": 2.3024935876624222e-05, |
| "loss": 0.3557, |
| "step": 623 |
| }, |
| { |
| "epoch": 2.530632911392405, |
| "grad_norm": 0.6009672126194706, |
| "learning_rate": 2.2968818045784813e-05, |
| "loss": 0.3313, |
| "step": 624 |
| }, |
| { |
| "epoch": 2.5346835443037974, |
| "grad_norm": 0.6028158093990058, |
| "learning_rate": 2.2912676304493006e-05, |
| "loss": 0.3475, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.53873417721519, |
| "grad_norm": 0.5833492222244325, |
| "learning_rate": 2.2856511104906668e-05, |
| "loss": 0.3562, |
| "step": 626 |
| }, |
| { |
| "epoch": 2.5427848101265824, |
| "grad_norm": 0.5876450855293134, |
| "learning_rate": 2.2800322899372586e-05, |
| "loss": 0.3443, |
| "step": 627 |
| }, |
| { |
| "epoch": 2.546835443037975, |
| "grad_norm": 0.5562011000806907, |
| "learning_rate": 2.2744112140422844e-05, |
| "loss": 0.3381, |
| "step": 628 |
| }, |
| { |
| "epoch": 2.550886075949367, |
| "grad_norm": 0.59835588379094, |
| "learning_rate": 2.2687879280771177e-05, |
| "loss": 0.3509, |
| "step": 629 |
| }, |
| { |
| "epoch": 2.5549367088607595, |
| "grad_norm": 0.5646040712059055, |
| "learning_rate": 2.26316247733093e-05, |
| "loss": 0.3399, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.558987341772152, |
| "grad_norm": 0.5988840554857715, |
| "learning_rate": 2.257534907110328e-05, |
| "loss": 0.3429, |
| "step": 631 |
| }, |
| { |
| "epoch": 2.563037974683544, |
| "grad_norm": 0.6050848720317311, |
| "learning_rate": 2.2519052627389882e-05, |
| "loss": 0.3501, |
| "step": 632 |
| }, |
| { |
| "epoch": 2.5670886075949366, |
| "grad_norm": 0.6217403676571198, |
| "learning_rate": 2.246273589557294e-05, |
| "loss": 0.3425, |
| "step": 633 |
| }, |
| { |
| "epoch": 2.571139240506329, |
| "grad_norm": 0.6098324525030838, |
| "learning_rate": 2.240639932921966e-05, |
| "loss": 0.3563, |
| "step": 634 |
| }, |
| { |
| "epoch": 2.5751898734177217, |
| "grad_norm": 0.592105307623835, |
| "learning_rate": 2.2350043382056995e-05, |
| "loss": 0.3435, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.579240506329114, |
| "grad_norm": 0.6020084047395226, |
| "learning_rate": 2.2293668507968015e-05, |
| "loss": 0.3559, |
| "step": 636 |
| }, |
| { |
| "epoch": 2.5832911392405062, |
| "grad_norm": 0.6204937015151932, |
| "learning_rate": 2.2237275160988186e-05, |
| "loss": 0.3377, |
| "step": 637 |
| }, |
| { |
| "epoch": 2.5873417721518988, |
| "grad_norm": 0.5914665968957243, |
| "learning_rate": 2.2180863795301787e-05, |
| "loss": 0.3384, |
| "step": 638 |
| }, |
| { |
| "epoch": 2.5913924050632913, |
| "grad_norm": 0.6451302117245227, |
| "learning_rate": 2.212443486523819e-05, |
| "loss": 0.3487, |
| "step": 639 |
| }, |
| { |
| "epoch": 2.5954430379746833, |
| "grad_norm": 0.5714193566497999, |
| "learning_rate": 2.2067988825268243e-05, |
| "loss": 0.3481, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.599493670886076, |
| "grad_norm": 0.589825557819433, |
| "learning_rate": 2.2011526130000596e-05, |
| "loss": 0.3466, |
| "step": 641 |
| }, |
| { |
| "epoch": 2.6035443037974684, |
| "grad_norm": 0.5700279778129095, |
| "learning_rate": 2.1955047234178038e-05, |
| "loss": 0.3546, |
| "step": 642 |
| }, |
| { |
| "epoch": 2.607594936708861, |
| "grad_norm": 0.5834881913492758, |
| "learning_rate": 2.1898552592673825e-05, |
| "loss": 0.341, |
| "step": 643 |
| }, |
| { |
| "epoch": 2.6116455696202534, |
| "grad_norm": 0.5735401281978943, |
| "learning_rate": 2.184204266048803e-05, |
| "loss": 0.3456, |
| "step": 644 |
| }, |
| { |
| "epoch": 2.6156962025316455, |
| "grad_norm": 0.581519731324081, |
| "learning_rate": 2.1785517892743887e-05, |
| "loss": 0.3586, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.619746835443038, |
| "grad_norm": 0.5865547844420494, |
| "learning_rate": 2.17289787446841e-05, |
| "loss": 0.3473, |
| "step": 646 |
| }, |
| { |
| "epoch": 2.6237974683544305, |
| "grad_norm": 0.5880979207550592, |
| "learning_rate": 2.1672425671667198e-05, |
| "loss": 0.3402, |
| "step": 647 |
| }, |
| { |
| "epoch": 2.6278481012658226, |
| "grad_norm": 0.5656299904466315, |
| "learning_rate": 2.161585912916385e-05, |
| "loss": 0.358, |
| "step": 648 |
| }, |
| { |
| "epoch": 2.631898734177215, |
| "grad_norm": 0.6060001264725685, |
| "learning_rate": 2.1559279572753214e-05, |
| "loss": 0.3417, |
| "step": 649 |
| }, |
| { |
| "epoch": 2.6359493670886076, |
| "grad_norm": 0.5635480394204186, |
| "learning_rate": 2.1502687458119268e-05, |
| "loss": 0.356, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.64, |
| "grad_norm": 0.5955711660636985, |
| "learning_rate": 2.1446083241047116e-05, |
| "loss": 0.3491, |
| "step": 651 |
| }, |
| { |
| "epoch": 2.6440506329113926, |
| "grad_norm": 0.5776212438019154, |
| "learning_rate": 2.1389467377419333e-05, |
| "loss": 0.345, |
| "step": 652 |
| }, |
| { |
| "epoch": 2.6481012658227847, |
| "grad_norm": 0.5830479821286455, |
| "learning_rate": 2.133284032321232e-05, |
| "loss": 0.3593, |
| "step": 653 |
| }, |
| { |
| "epoch": 2.6521518987341772, |
| "grad_norm": 0.5747412676701164, |
| "learning_rate": 2.1276202534492566e-05, |
| "loss": 0.3488, |
| "step": 654 |
| }, |
| { |
| "epoch": 2.6562025316455697, |
| "grad_norm": 0.5772319337076186, |
| "learning_rate": 2.121955446741306e-05, |
| "loss": 0.3471, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.660253164556962, |
| "grad_norm": 0.5876913762513267, |
| "learning_rate": 2.1162896578209517e-05, |
| "loss": 0.3373, |
| "step": 656 |
| }, |
| { |
| "epoch": 2.6643037974683543, |
| "grad_norm": 0.6067588689193171, |
| "learning_rate": 2.1106229323196813e-05, |
| "loss": 0.3493, |
| "step": 657 |
| }, |
| { |
| "epoch": 2.668354430379747, |
| "grad_norm": 0.5846613715951393, |
| "learning_rate": 2.1049553158765214e-05, |
| "loss": 0.3417, |
| "step": 658 |
| }, |
| { |
| "epoch": 2.6724050632911394, |
| "grad_norm": 0.5784029572860948, |
| "learning_rate": 2.0992868541376764e-05, |
| "loss": 0.3354, |
| "step": 659 |
| }, |
| { |
| "epoch": 2.676455696202532, |
| "grad_norm": 0.6155983470837033, |
| "learning_rate": 2.093617592756158e-05, |
| "loss": 0.3371, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.680506329113924, |
| "grad_norm": 0.5661889560841431, |
| "learning_rate": 2.0879475773914167e-05, |
| "loss": 0.3556, |
| "step": 661 |
| }, |
| { |
| "epoch": 2.6845569620253165, |
| "grad_norm": 0.6055297195407208, |
| "learning_rate": 2.082276853708978e-05, |
| "loss": 0.3594, |
| "step": 662 |
| }, |
| { |
| "epoch": 2.688607594936709, |
| "grad_norm": 0.5877265482942783, |
| "learning_rate": 2.076605467380071e-05, |
| "loss": 0.3473, |
| "step": 663 |
| }, |
| { |
| "epoch": 2.692658227848101, |
| "grad_norm": 0.5980049806072585, |
| "learning_rate": 2.0709334640812613e-05, |
| "loss": 0.3502, |
| "step": 664 |
| }, |
| { |
| "epoch": 2.6967088607594936, |
| "grad_norm": 0.5985315416907165, |
| "learning_rate": 2.0652608894940824e-05, |
| "loss": 0.3556, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.700759493670886, |
| "grad_norm": 0.5596205992051192, |
| "learning_rate": 2.0595877893046722e-05, |
| "loss": 0.3503, |
| "step": 666 |
| }, |
| { |
| "epoch": 2.7048101265822786, |
| "grad_norm": 0.607831895603754, |
| "learning_rate": 2.0539142092033985e-05, |
| "loss": 0.349, |
| "step": 667 |
| }, |
| { |
| "epoch": 2.708860759493671, |
| "grad_norm": 0.5723600926683704, |
| "learning_rate": 2.048240194884496e-05, |
| "loss": 0.3507, |
| "step": 668 |
| }, |
| { |
| "epoch": 2.712911392405063, |
| "grad_norm": 0.5829384874839397, |
| "learning_rate": 2.042565792045695e-05, |
| "loss": 0.3439, |
| "step": 669 |
| }, |
| { |
| "epoch": 2.7169620253164557, |
| "grad_norm": 0.5633834618018123, |
| "learning_rate": 2.036891046387857e-05, |
| "loss": 0.3631, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.721012658227848, |
| "grad_norm": 0.5697024468681174, |
| "learning_rate": 2.0312160036146036e-05, |
| "loss": 0.3374, |
| "step": 671 |
| }, |
| { |
| "epoch": 2.7250632911392403, |
| "grad_norm": 0.5694179445798321, |
| "learning_rate": 2.025540709431948e-05, |
| "loss": 0.349, |
| "step": 672 |
| }, |
| { |
| "epoch": 2.729113924050633, |
| "grad_norm": 0.5811871537663907, |
| "learning_rate": 2.0198652095479298e-05, |
| "loss": 0.3548, |
| "step": 673 |
| }, |
| { |
| "epoch": 2.7331645569620253, |
| "grad_norm": 0.5843057153638428, |
| "learning_rate": 2.014189549672245e-05, |
| "loss": 0.3486, |
| "step": 674 |
| }, |
| { |
| "epoch": 2.737215189873418, |
| "grad_norm": 0.5760990504330805, |
| "learning_rate": 2.0085137755158776e-05, |
| "loss": 0.3497, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.7412658227848103, |
| "grad_norm": 0.5787918195374676, |
| "learning_rate": 2.0028379327907327e-05, |
| "loss": 0.354, |
| "step": 676 |
| }, |
| { |
| "epoch": 2.7453164556962024, |
| "grad_norm": 0.5714974528826536, |
| "learning_rate": 1.9971620672092676e-05, |
| "loss": 0.3525, |
| "step": 677 |
| }, |
| { |
| "epoch": 2.749367088607595, |
| "grad_norm": 0.576295952092862, |
| "learning_rate": 1.991486224484123e-05, |
| "loss": 0.3498, |
| "step": 678 |
| }, |
| { |
| "epoch": 2.7534177215189874, |
| "grad_norm": 0.5825956985295242, |
| "learning_rate": 1.985810450327756e-05, |
| "loss": 0.3472, |
| "step": 679 |
| }, |
| { |
| "epoch": 2.7574683544303795, |
| "grad_norm": 0.5808987506527321, |
| "learning_rate": 1.9801347904520706e-05, |
| "loss": 0.3367, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.761518987341772, |
| "grad_norm": 0.590809720122666, |
| "learning_rate": 1.974459290568053e-05, |
| "loss": 0.3554, |
| "step": 681 |
| }, |
| { |
| "epoch": 2.7655696202531646, |
| "grad_norm": 0.5706788250749534, |
| "learning_rate": 1.968783996385397e-05, |
| "loss": 0.3577, |
| "step": 682 |
| }, |
| { |
| "epoch": 2.769620253164557, |
| "grad_norm": 0.5784531604499272, |
| "learning_rate": 1.963108953612143e-05, |
| "loss": 0.3584, |
| "step": 683 |
| }, |
| { |
| "epoch": 2.7736708860759496, |
| "grad_norm": 0.5790505561615357, |
| "learning_rate": 1.9574342079543056e-05, |
| "loss": 0.3477, |
| "step": 684 |
| }, |
| { |
| "epoch": 2.7777215189873417, |
| "grad_norm": 0.6000232578400603, |
| "learning_rate": 1.9517598051155046e-05, |
| "loss": 0.3599, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.781772151898734, |
| "grad_norm": 0.597415124669575, |
| "learning_rate": 1.9460857907966025e-05, |
| "loss": 0.3363, |
| "step": 686 |
| }, |
| { |
| "epoch": 2.7858227848101267, |
| "grad_norm": 0.5696041725094757, |
| "learning_rate": 1.9404122106953285e-05, |
| "loss": 0.3476, |
| "step": 687 |
| }, |
| { |
| "epoch": 2.7898734177215188, |
| "grad_norm": 0.5741912903182702, |
| "learning_rate": 1.9347391105059176e-05, |
| "loss": 0.3502, |
| "step": 688 |
| }, |
| { |
| "epoch": 2.7939240506329113, |
| "grad_norm": 0.573576800108603, |
| "learning_rate": 1.92906653591874e-05, |
| "loss": 0.3428, |
| "step": 689 |
| }, |
| { |
| "epoch": 2.797974683544304, |
| "grad_norm": 0.5756646489324376, |
| "learning_rate": 1.9233945326199295e-05, |
| "loss": 0.3436, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.8020253164556963, |
| "grad_norm": 0.580969094671374, |
| "learning_rate": 1.917723146291022e-05, |
| "loss": 0.3548, |
| "step": 691 |
| }, |
| { |
| "epoch": 2.806075949367089, |
| "grad_norm": 0.5754461520617994, |
| "learning_rate": 1.912052422608584e-05, |
| "loss": 0.3327, |
| "step": 692 |
| }, |
| { |
| "epoch": 2.810126582278481, |
| "grad_norm": 0.5985477733719392, |
| "learning_rate": 1.9063824072438428e-05, |
| "loss": 0.3531, |
| "step": 693 |
| }, |
| { |
| "epoch": 2.8141772151898734, |
| "grad_norm": 0.5897348323778643, |
| "learning_rate": 1.9007131458623246e-05, |
| "loss": 0.3426, |
| "step": 694 |
| }, |
| { |
| "epoch": 2.818227848101266, |
| "grad_norm": 0.578765313959326, |
| "learning_rate": 1.895044684123479e-05, |
| "loss": 0.3463, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.822278481012658, |
| "grad_norm": 0.5795689642268431, |
| "learning_rate": 1.8893770676803194e-05, |
| "loss": 0.3485, |
| "step": 696 |
| }, |
| { |
| "epoch": 2.8263291139240505, |
| "grad_norm": 0.6001322239920289, |
| "learning_rate": 1.8837103421790486e-05, |
| "loss": 0.3595, |
| "step": 697 |
| }, |
| { |
| "epoch": 2.830379746835443, |
| "grad_norm": 0.569665781343708, |
| "learning_rate": 1.8780445532586952e-05, |
| "loss": 0.3664, |
| "step": 698 |
| }, |
| { |
| "epoch": 2.8344303797468355, |
| "grad_norm": 0.5864252288365516, |
| "learning_rate": 1.872379746550743e-05, |
| "loss": 0.3405, |
| "step": 699 |
| }, |
| { |
| "epoch": 2.838481012658228, |
| "grad_norm": 0.5621207350742281, |
| "learning_rate": 1.866715967678769e-05, |
| "loss": 0.3431, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.84253164556962, |
| "grad_norm": 0.564408147910918, |
| "learning_rate": 1.861053262258067e-05, |
| "loss": 0.3544, |
| "step": 701 |
| }, |
| { |
| "epoch": 2.8465822784810126, |
| "grad_norm": 0.5890410332793564, |
| "learning_rate": 1.8553916758952897e-05, |
| "loss": 0.3411, |
| "step": 702 |
| }, |
| { |
| "epoch": 2.850632911392405, |
| "grad_norm": 0.5915284950157376, |
| "learning_rate": 1.8497312541880735e-05, |
| "loss": 0.3417, |
| "step": 703 |
| }, |
| { |
| "epoch": 2.8546835443037972, |
| "grad_norm": 0.570632338933207, |
| "learning_rate": 1.8440720427246786e-05, |
| "loss": 0.3546, |
| "step": 704 |
| }, |
| { |
| "epoch": 2.8587341772151897, |
| "grad_norm": 0.5630513323069526, |
| "learning_rate": 1.8384140870836157e-05, |
| "loss": 0.3378, |
| "step": 705 |
| }, |
| { |
| "epoch": 2.8627848101265823, |
| "grad_norm": 0.5738864253404724, |
| "learning_rate": 1.8327574328332806e-05, |
| "loss": 0.3433, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.8668354430379748, |
| "grad_norm": 0.595064508478934, |
| "learning_rate": 1.8271021255315906e-05, |
| "loss": 0.3517, |
| "step": 707 |
| }, |
| { |
| "epoch": 2.8708860759493673, |
| "grad_norm": 0.5497141401897051, |
| "learning_rate": 1.8214482107256117e-05, |
| "loss": 0.3298, |
| "step": 708 |
| }, |
| { |
| "epoch": 2.8749367088607594, |
| "grad_norm": 0.5622924813932927, |
| "learning_rate": 1.8157957339511968e-05, |
| "loss": 0.3496, |
| "step": 709 |
| }, |
| { |
| "epoch": 2.878987341772152, |
| "grad_norm": 0.5895892352853498, |
| "learning_rate": 1.8101447407326182e-05, |
| "loss": 0.3591, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.8830379746835444, |
| "grad_norm": 0.5861827970627633, |
| "learning_rate": 1.8044952765821966e-05, |
| "loss": 0.3432, |
| "step": 711 |
| }, |
| { |
| "epoch": 2.8870886075949365, |
| "grad_norm": 0.5775929745728821, |
| "learning_rate": 1.7988473869999407e-05, |
| "loss": 0.3383, |
| "step": 712 |
| }, |
| { |
| "epoch": 2.891139240506329, |
| "grad_norm": 0.574202454376069, |
| "learning_rate": 1.7932011174731764e-05, |
| "loss": 0.3589, |
| "step": 713 |
| }, |
| { |
| "epoch": 2.8951898734177215, |
| "grad_norm": 0.6157420995346585, |
| "learning_rate": 1.7875565134761817e-05, |
| "loss": 0.3439, |
| "step": 714 |
| }, |
| { |
| "epoch": 2.899240506329114, |
| "grad_norm": 0.571287477212874, |
| "learning_rate": 1.7819136204698226e-05, |
| "loss": 0.3613, |
| "step": 715 |
| }, |
| { |
| "epoch": 2.9032911392405065, |
| "grad_norm": 0.6044574189631108, |
| "learning_rate": 1.776272483901182e-05, |
| "loss": 0.3448, |
| "step": 716 |
| }, |
| { |
| "epoch": 2.9073417721518986, |
| "grad_norm": 0.5672447877552919, |
| "learning_rate": 1.7706331492031995e-05, |
| "loss": 0.348, |
| "step": 717 |
| }, |
| { |
| "epoch": 2.911392405063291, |
| "grad_norm": 0.5551513612710356, |
| "learning_rate": 1.764995661794301e-05, |
| "loss": 0.3471, |
| "step": 718 |
| }, |
| { |
| "epoch": 2.9154430379746836, |
| "grad_norm": 0.5928977788302826, |
| "learning_rate": 1.759360067078035e-05, |
| "loss": 0.3597, |
| "step": 719 |
| }, |
| { |
| "epoch": 2.9194936708860757, |
| "grad_norm": 0.5766393811891416, |
| "learning_rate": 1.7537264104427064e-05, |
| "loss": 0.3561, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.923544303797468, |
| "grad_norm": 0.5770611096831866, |
| "learning_rate": 1.748094737261012e-05, |
| "loss": 0.3433, |
| "step": 721 |
| }, |
| { |
| "epoch": 2.9275949367088607, |
| "grad_norm": 0.5827685533864442, |
| "learning_rate": 1.7424650928896726e-05, |
| "loss": 0.3414, |
| "step": 722 |
| }, |
| { |
| "epoch": 2.9316455696202532, |
| "grad_norm": 0.5851810426378314, |
| "learning_rate": 1.7368375226690712e-05, |
| "loss": 0.3466, |
| "step": 723 |
| }, |
| { |
| "epoch": 2.9356962025316458, |
| "grad_norm": 0.5825143900141643, |
| "learning_rate": 1.731212071922883e-05, |
| "loss": 0.3382, |
| "step": 724 |
| }, |
| { |
| "epoch": 2.939746835443038, |
| "grad_norm": 0.5916346805550766, |
| "learning_rate": 1.7255887859577156e-05, |
| "loss": 0.3421, |
| "step": 725 |
| }, |
| { |
| "epoch": 2.9437974683544303, |
| "grad_norm": 0.5951297043579414, |
| "learning_rate": 1.7199677100627427e-05, |
| "loss": 0.3622, |
| "step": 726 |
| }, |
| { |
| "epoch": 2.947848101265823, |
| "grad_norm": 0.5802844188553519, |
| "learning_rate": 1.7143488895093343e-05, |
| "loss": 0.3494, |
| "step": 727 |
| }, |
| { |
| "epoch": 2.951898734177215, |
| "grad_norm": 0.581943561385102, |
| "learning_rate": 1.7087323695506994e-05, |
| "loss": 0.3367, |
| "step": 728 |
| }, |
| { |
| "epoch": 2.9559493670886074, |
| "grad_norm": 0.5589430178982162, |
| "learning_rate": 1.7031181954215194e-05, |
| "loss": 0.3374, |
| "step": 729 |
| }, |
| { |
| "epoch": 2.96, |
| "grad_norm": 0.560238829306942, |
| "learning_rate": 1.6975064123375788e-05, |
| "loss": 0.3313, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.9640506329113925, |
| "grad_norm": 0.6244985690916002, |
| "learning_rate": 1.6918970654954084e-05, |
| "loss": 0.3479, |
| "step": 731 |
| }, |
| { |
| "epoch": 2.968101265822785, |
| "grad_norm": 0.5777755723855205, |
| "learning_rate": 1.686290200071915e-05, |
| "loss": 0.3416, |
| "step": 732 |
| }, |
| { |
| "epoch": 2.972151898734177, |
| "grad_norm": 0.5629644814325285, |
| "learning_rate": 1.6806858612240234e-05, |
| "loss": 0.3498, |
| "step": 733 |
| }, |
| { |
| "epoch": 2.9762025316455696, |
| "grad_norm": 0.5788135662003859, |
| "learning_rate": 1.6750840940883078e-05, |
| "loss": 0.3365, |
| "step": 734 |
| }, |
| { |
| "epoch": 2.980253164556962, |
| "grad_norm": 0.5593646690154236, |
| "learning_rate": 1.6694849437806305e-05, |
| "loss": 0.3438, |
| "step": 735 |
| }, |
| { |
| "epoch": 2.984303797468354, |
| "grad_norm": 0.5678619888864153, |
| "learning_rate": 1.663888455395778e-05, |
| "loss": 0.3466, |
| "step": 736 |
| }, |
| { |
| "epoch": 2.9883544303797467, |
| "grad_norm": 0.5569263668957953, |
| "learning_rate": 1.6582946740070995e-05, |
| "loss": 0.3484, |
| "step": 737 |
| }, |
| { |
| "epoch": 2.992405063291139, |
| "grad_norm": 0.5758929671994376, |
| "learning_rate": 1.6527036446661396e-05, |
| "loss": 0.3425, |
| "step": 738 |
| }, |
| { |
| "epoch": 2.9964556962025317, |
| "grad_norm": 0.5755822905799529, |
| "learning_rate": 1.6471154124022818e-05, |
| "loss": 0.3392, |
| "step": 739 |
| }, |
| { |
| "epoch": 3.002025316455696, |
| "grad_norm": 0.5790762593261332, |
| "learning_rate": 1.6415300222223788e-05, |
| "loss": 0.259, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0060759493670886, |
| "grad_norm": 0.6643541856720142, |
| "learning_rate": 1.6359475191103958e-05, |
| "loss": 0.1661, |
| "step": 741 |
| }, |
| { |
| "epoch": 3.010126582278481, |
| "grad_norm": 0.5446871515443332, |
| "learning_rate": 1.6303679480270466e-05, |
| "loss": 0.1637, |
| "step": 742 |
| }, |
| { |
| "epoch": 3.0141772151898736, |
| "grad_norm": 0.4775742060630328, |
| "learning_rate": 1.624791353909428e-05, |
| "loss": 0.1628, |
| "step": 743 |
| }, |
| { |
| "epoch": 3.0182278481012657, |
| "grad_norm": 0.5932740247263545, |
| "learning_rate": 1.619217781670663e-05, |
| "loss": 0.1557, |
| "step": 744 |
| }, |
| { |
| "epoch": 3.022278481012658, |
| "grad_norm": 0.7450023650423089, |
| "learning_rate": 1.6136472761995373e-05, |
| "loss": 0.1658, |
| "step": 745 |
| }, |
| { |
| "epoch": 3.0263291139240507, |
| "grad_norm": 0.674674399501268, |
| "learning_rate": 1.608079882360134e-05, |
| "loss": 0.1601, |
| "step": 746 |
| }, |
| { |
| "epoch": 3.030379746835443, |
| "grad_norm": 0.6002552787108303, |
| "learning_rate": 1.60251564499148e-05, |
| "loss": 0.1584, |
| "step": 747 |
| }, |
| { |
| "epoch": 3.0344303797468353, |
| "grad_norm": 0.5145315632919125, |
| "learning_rate": 1.596954608907176e-05, |
| "loss": 0.146, |
| "step": 748 |
| }, |
| { |
| "epoch": 3.038481012658228, |
| "grad_norm": 0.5041393685075924, |
| "learning_rate": 1.591396818895043e-05, |
| "loss": 0.1646, |
| "step": 749 |
| }, |
| { |
| "epoch": 3.0425316455696203, |
| "grad_norm": 0.5135346440680685, |
| "learning_rate": 1.585842319716759e-05, |
| "loss": 0.1562, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.046582278481013, |
| "grad_norm": 0.5168432647424274, |
| "learning_rate": 1.5802911561074944e-05, |
| "loss": 0.1596, |
| "step": 751 |
| }, |
| { |
| "epoch": 3.050632911392405, |
| "grad_norm": 0.49508034483289115, |
| "learning_rate": 1.5747433727755595e-05, |
| "loss": 0.1522, |
| "step": 752 |
| }, |
| { |
| "epoch": 3.0546835443037974, |
| "grad_norm": 0.4894761833660976, |
| "learning_rate": 1.5691990144020376e-05, |
| "loss": 0.1564, |
| "step": 753 |
| }, |
| { |
| "epoch": 3.05873417721519, |
| "grad_norm": 0.5079203004365923, |
| "learning_rate": 1.5636581256404297e-05, |
| "loss": 0.1558, |
| "step": 754 |
| }, |
| { |
| "epoch": 3.0627848101265824, |
| "grad_norm": 0.4982257030211359, |
| "learning_rate": 1.558120751116291e-05, |
| "loss": 0.1403, |
| "step": 755 |
| }, |
| { |
| "epoch": 3.0668354430379745, |
| "grad_norm": 0.5257795259916082, |
| "learning_rate": 1.552586935426876e-05, |
| "loss": 0.1564, |
| "step": 756 |
| }, |
| { |
| "epoch": 3.070886075949367, |
| "grad_norm": 0.5092780407220261, |
| "learning_rate": 1.547056723140774e-05, |
| "loss": 0.1489, |
| "step": 757 |
| }, |
| { |
| "epoch": 3.0749367088607595, |
| "grad_norm": 0.5144981563190801, |
| "learning_rate": 1.5415301587975565e-05, |
| "loss": 0.1595, |
| "step": 758 |
| }, |
| { |
| "epoch": 3.078987341772152, |
| "grad_norm": 0.4946767705649546, |
| "learning_rate": 1.536007286907411e-05, |
| "loss": 0.143, |
| "step": 759 |
| }, |
| { |
| "epoch": 3.083037974683544, |
| "grad_norm": 0.47107364206210844, |
| "learning_rate": 1.5304881519507896e-05, |
| "loss": 0.1422, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.0870886075949366, |
| "grad_norm": 0.46287772957602497, |
| "learning_rate": 1.5249727983780453e-05, |
| "loss": 0.1463, |
| "step": 761 |
| }, |
| { |
| "epoch": 3.091139240506329, |
| "grad_norm": 0.4845774600968683, |
| "learning_rate": 1.5194612706090786e-05, |
| "loss": 0.1644, |
| "step": 762 |
| }, |
| { |
| "epoch": 3.0951898734177217, |
| "grad_norm": 0.4700410570939326, |
| "learning_rate": 1.5139536130329771e-05, |
| "loss": 0.1487, |
| "step": 763 |
| }, |
| { |
| "epoch": 3.0992405063291137, |
| "grad_norm": 0.47505336759385564, |
| "learning_rate": 1.508449870007656e-05, |
| "loss": 0.149, |
| "step": 764 |
| }, |
| { |
| "epoch": 3.1032911392405063, |
| "grad_norm": 0.47794569620921223, |
| "learning_rate": 1.5029500858595056e-05, |
| "loss": 0.157, |
| "step": 765 |
| }, |
| { |
| "epoch": 3.1073417721518988, |
| "grad_norm": 0.47027188423266314, |
| "learning_rate": 1.4974543048830328e-05, |
| "loss": 0.1578, |
| "step": 766 |
| }, |
| { |
| "epoch": 3.1113924050632913, |
| "grad_norm": 0.47730983552129475, |
| "learning_rate": 1.4919625713405e-05, |
| "loss": 0.1412, |
| "step": 767 |
| }, |
| { |
| "epoch": 3.1154430379746834, |
| "grad_norm": 0.5012973230344715, |
| "learning_rate": 1.4864749294615756e-05, |
| "loss": 0.1576, |
| "step": 768 |
| }, |
| { |
| "epoch": 3.119493670886076, |
| "grad_norm": 0.48741352365863627, |
| "learning_rate": 1.4809914234429716e-05, |
| "loss": 0.1484, |
| "step": 769 |
| }, |
| { |
| "epoch": 3.1235443037974684, |
| "grad_norm": 0.48627698287469373, |
| "learning_rate": 1.4755120974480923e-05, |
| "loss": 0.155, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.127594936708861, |
| "grad_norm": 0.4895685362253195, |
| "learning_rate": 1.4700369956066771e-05, |
| "loss": 0.1566, |
| "step": 771 |
| }, |
| { |
| "epoch": 3.131645569620253, |
| "grad_norm": 0.4969570244326975, |
| "learning_rate": 1.4645661620144413e-05, |
| "loss": 0.1566, |
| "step": 772 |
| }, |
| { |
| "epoch": 3.1356962025316455, |
| "grad_norm": 0.4613566351014632, |
| "learning_rate": 1.4590996407327284e-05, |
| "loss": 0.1546, |
| "step": 773 |
| }, |
| { |
| "epoch": 3.139746835443038, |
| "grad_norm": 0.4639037701043011, |
| "learning_rate": 1.4536374757881487e-05, |
| "loss": 0.1535, |
| "step": 774 |
| }, |
| { |
| "epoch": 3.1437974683544305, |
| "grad_norm": 0.4537223195480731, |
| "learning_rate": 1.4481797111722271e-05, |
| "loss": 0.1421, |
| "step": 775 |
| }, |
| { |
| "epoch": 3.1478481012658226, |
| "grad_norm": 0.46747530854064645, |
| "learning_rate": 1.4427263908410507e-05, |
| "loss": 0.1541, |
| "step": 776 |
| }, |
| { |
| "epoch": 3.151898734177215, |
| "grad_norm": 0.46824983529511593, |
| "learning_rate": 1.4372775587149108e-05, |
| "loss": 0.1457, |
| "step": 777 |
| }, |
| { |
| "epoch": 3.1559493670886076, |
| "grad_norm": 0.46118089421648745, |
| "learning_rate": 1.4318332586779522e-05, |
| "loss": 0.1439, |
| "step": 778 |
| }, |
| { |
| "epoch": 3.16, |
| "grad_norm": 0.488691781264018, |
| "learning_rate": 1.4263935345778202e-05, |
| "loss": 0.1451, |
| "step": 779 |
| }, |
| { |
| "epoch": 3.164050632911392, |
| "grad_norm": 0.4718969209880653, |
| "learning_rate": 1.420958430225303e-05, |
| "loss": 0.1579, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.1681012658227847, |
| "grad_norm": 0.46182087717817966, |
| "learning_rate": 1.415527989393985e-05, |
| "loss": 0.1495, |
| "step": 781 |
| }, |
| { |
| "epoch": 3.1721518987341772, |
| "grad_norm": 0.5016730949015833, |
| "learning_rate": 1.410102255819891e-05, |
| "loss": 0.1555, |
| "step": 782 |
| }, |
| { |
| "epoch": 3.1762025316455698, |
| "grad_norm": 0.46103130424729916, |
| "learning_rate": 1.404681273201131e-05, |
| "loss": 0.1462, |
| "step": 783 |
| }, |
| { |
| "epoch": 3.180253164556962, |
| "grad_norm": 0.4759579396517062, |
| "learning_rate": 1.399265085197556e-05, |
| "loss": 0.1534, |
| "step": 784 |
| }, |
| { |
| "epoch": 3.1843037974683543, |
| "grad_norm": 0.48113019885632685, |
| "learning_rate": 1.393853735430398e-05, |
| "loss": 0.1519, |
| "step": 785 |
| }, |
| { |
| "epoch": 3.188354430379747, |
| "grad_norm": 0.45463917307401686, |
| "learning_rate": 1.3884472674819246e-05, |
| "loss": 0.1497, |
| "step": 786 |
| }, |
| { |
| "epoch": 3.1924050632911394, |
| "grad_norm": 0.4558706276284719, |
| "learning_rate": 1.3830457248950864e-05, |
| "loss": 0.1496, |
| "step": 787 |
| }, |
| { |
| "epoch": 3.1964556962025314, |
| "grad_norm": 0.45667496640308536, |
| "learning_rate": 1.377649151173163e-05, |
| "loss": 0.1366, |
| "step": 788 |
| }, |
| { |
| "epoch": 3.200506329113924, |
| "grad_norm": 0.4703088863853243, |
| "learning_rate": 1.3722575897794181e-05, |
| "loss": 0.1481, |
| "step": 789 |
| }, |
| { |
| "epoch": 3.2045569620253165, |
| "grad_norm": 0.45585981250899205, |
| "learning_rate": 1.3668710841367472e-05, |
| "loss": 0.14, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.208607594936709, |
| "grad_norm": 0.4728481384192643, |
| "learning_rate": 1.361489677627324e-05, |
| "loss": 0.1433, |
| "step": 791 |
| }, |
| { |
| "epoch": 3.212658227848101, |
| "grad_norm": 0.47190288241491446, |
| "learning_rate": 1.3561134135922585e-05, |
| "loss": 0.1502, |
| "step": 792 |
| }, |
| { |
| "epoch": 3.2167088607594936, |
| "grad_norm": 0.47099024688217994, |
| "learning_rate": 1.350742335331241e-05, |
| "loss": 0.1482, |
| "step": 793 |
| }, |
| { |
| "epoch": 3.220759493670886, |
| "grad_norm": 0.4539812368128493, |
| "learning_rate": 1.345376486102198e-05, |
| "loss": 0.1437, |
| "step": 794 |
| }, |
| { |
| "epoch": 3.2248101265822786, |
| "grad_norm": 0.481746655744389, |
| "learning_rate": 1.3400159091209414e-05, |
| "loss": 0.1518, |
| "step": 795 |
| }, |
| { |
| "epoch": 3.2288607594936707, |
| "grad_norm": 0.475779620895132, |
| "learning_rate": 1.3346606475608216e-05, |
| "loss": 0.1595, |
| "step": 796 |
| }, |
| { |
| "epoch": 3.232911392405063, |
| "grad_norm": 0.4679819348358873, |
| "learning_rate": 1.3293107445523781e-05, |
| "loss": 0.15, |
| "step": 797 |
| }, |
| { |
| "epoch": 3.2369620253164557, |
| "grad_norm": 0.4902653220870392, |
| "learning_rate": 1.3239662431829949e-05, |
| "loss": 0.1499, |
| "step": 798 |
| }, |
| { |
| "epoch": 3.2410126582278482, |
| "grad_norm": 0.48284137093858387, |
| "learning_rate": 1.3186271864965509e-05, |
| "loss": 0.1441, |
| "step": 799 |
| }, |
| { |
| "epoch": 3.2450632911392403, |
| "grad_norm": 0.45576554098906324, |
| "learning_rate": 1.3132936174930756e-05, |
| "loss": 0.1422, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.249113924050633, |
| "grad_norm": 0.484455670366212, |
| "learning_rate": 1.3079655791283995e-05, |
| "loss": 0.1374, |
| "step": 801 |
| }, |
| { |
| "epoch": 3.2531645569620253, |
| "grad_norm": 0.47617612569429246, |
| "learning_rate": 1.3026431143138108e-05, |
| "loss": 0.1449, |
| "step": 802 |
| }, |
| { |
| "epoch": 3.257215189873418, |
| "grad_norm": 0.467161901482994, |
| "learning_rate": 1.2973262659157114e-05, |
| "loss": 0.1468, |
| "step": 803 |
| }, |
| { |
| "epoch": 3.26126582278481, |
| "grad_norm": 0.47469731508540813, |
| "learning_rate": 1.2920150767552651e-05, |
| "loss": 0.1439, |
| "step": 804 |
| }, |
| { |
| "epoch": 3.2653164556962024, |
| "grad_norm": 0.4948911878553713, |
| "learning_rate": 1.2867095896080607e-05, |
| "loss": 0.157, |
| "step": 805 |
| }, |
| { |
| "epoch": 3.269367088607595, |
| "grad_norm": 0.4627683935128917, |
| "learning_rate": 1.2814098472037612e-05, |
| "loss": 0.1589, |
| "step": 806 |
| }, |
| { |
| "epoch": 3.2734177215189875, |
| "grad_norm": 0.47354939246449473, |
| "learning_rate": 1.276115892225764e-05, |
| "loss": 0.152, |
| "step": 807 |
| }, |
| { |
| "epoch": 3.27746835443038, |
| "grad_norm": 0.46999382187162225, |
| "learning_rate": 1.2708277673108555e-05, |
| "loss": 0.1506, |
| "step": 808 |
| }, |
| { |
| "epoch": 3.281518987341772, |
| "grad_norm": 0.4680138962066465, |
| "learning_rate": 1.2655455150488649e-05, |
| "loss": 0.1512, |
| "step": 809 |
| }, |
| { |
| "epoch": 3.2855696202531646, |
| "grad_norm": 0.45639083052076074, |
| "learning_rate": 1.2602691779823272e-05, |
| "loss": 0.143, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.289620253164557, |
| "grad_norm": 0.46833606310133896, |
| "learning_rate": 1.2549987986061355e-05, |
| "loss": 0.1472, |
| "step": 811 |
| }, |
| { |
| "epoch": 3.293670886075949, |
| "grad_norm": 0.46168380097563866, |
| "learning_rate": 1.2497344193672005e-05, |
| "loss": 0.1417, |
| "step": 812 |
| }, |
| { |
| "epoch": 3.2977215189873417, |
| "grad_norm": 0.4855183618071177, |
| "learning_rate": 1.2444760826641092e-05, |
| "loss": 0.1532, |
| "step": 813 |
| }, |
| { |
| "epoch": 3.301772151898734, |
| "grad_norm": 0.4799806305952373, |
| "learning_rate": 1.2392238308467817e-05, |
| "loss": 0.1468, |
| "step": 814 |
| }, |
| { |
| "epoch": 3.3058227848101267, |
| "grad_norm": 0.46663341990656276, |
| "learning_rate": 1.2339777062161326e-05, |
| "loss": 0.1525, |
| "step": 815 |
| }, |
| { |
| "epoch": 3.309873417721519, |
| "grad_norm": 0.45904466690082124, |
| "learning_rate": 1.2287377510237293e-05, |
| "loss": 0.1439, |
| "step": 816 |
| }, |
| { |
| "epoch": 3.3139240506329113, |
| "grad_norm": 0.46186625083680466, |
| "learning_rate": 1.2235040074714488e-05, |
| "loss": 0.143, |
| "step": 817 |
| }, |
| { |
| "epoch": 3.317974683544304, |
| "grad_norm": 0.4620824650054794, |
| "learning_rate": 1.2182765177111434e-05, |
| "loss": 0.1416, |
| "step": 818 |
| }, |
| { |
| "epoch": 3.3220253164556963, |
| "grad_norm": 0.47366212080548054, |
| "learning_rate": 1.213055323844297e-05, |
| "loss": 0.1525, |
| "step": 819 |
| }, |
| { |
| "epoch": 3.3260759493670884, |
| "grad_norm": 0.4453115350053116, |
| "learning_rate": 1.2078404679216864e-05, |
| "loss": 0.1393, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.330126582278481, |
| "grad_norm": 0.459948286718874, |
| "learning_rate": 1.2026319919430458e-05, |
| "loss": 0.1486, |
| "step": 821 |
| }, |
| { |
| "epoch": 3.3341772151898734, |
| "grad_norm": 0.4519870066156903, |
| "learning_rate": 1.1974299378567227e-05, |
| "loss": 0.1344, |
| "step": 822 |
| }, |
| { |
| "epoch": 3.338227848101266, |
| "grad_norm": 0.4743284623893573, |
| "learning_rate": 1.1922343475593462e-05, |
| "loss": 0.1646, |
| "step": 823 |
| }, |
| { |
| "epoch": 3.3422784810126585, |
| "grad_norm": 0.47475587035764516, |
| "learning_rate": 1.187045262895488e-05, |
| "loss": 0.1539, |
| "step": 824 |
| }, |
| { |
| "epoch": 3.3463291139240505, |
| "grad_norm": 0.46946281865051764, |
| "learning_rate": 1.1818627256573203e-05, |
| "loss": 0.1508, |
| "step": 825 |
| }, |
| { |
| "epoch": 3.350379746835443, |
| "grad_norm": 0.4543153132989318, |
| "learning_rate": 1.1766867775842864e-05, |
| "loss": 0.1517, |
| "step": 826 |
| }, |
| { |
| "epoch": 3.3544303797468356, |
| "grad_norm": 0.4668191543711979, |
| "learning_rate": 1.1715174603627615e-05, |
| "loss": 0.1559, |
| "step": 827 |
| }, |
| { |
| "epoch": 3.3584810126582276, |
| "grad_norm": 0.4471649768306851, |
| "learning_rate": 1.1663548156257147e-05, |
| "loss": 0.1438, |
| "step": 828 |
| }, |
| { |
| "epoch": 3.36253164556962, |
| "grad_norm": 0.45850897355540415, |
| "learning_rate": 1.161198884952377e-05, |
| "loss": 0.1454, |
| "step": 829 |
| }, |
| { |
| "epoch": 3.3665822784810127, |
| "grad_norm": 0.4752742474683345, |
| "learning_rate": 1.1560497098679056e-05, |
| "loss": 0.1405, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.370632911392405, |
| "grad_norm": 0.4637370493344981, |
| "learning_rate": 1.1509073318430479e-05, |
| "loss": 0.1471, |
| "step": 831 |
| }, |
| { |
| "epoch": 3.3746835443037977, |
| "grad_norm": 0.4633926302853835, |
| "learning_rate": 1.1457717922938116e-05, |
| "loss": 0.1401, |
| "step": 832 |
| }, |
| { |
| "epoch": 3.3787341772151898, |
| "grad_norm": 0.46612380865216946, |
| "learning_rate": 1.1406431325811233e-05, |
| "loss": 0.1394, |
| "step": 833 |
| }, |
| { |
| "epoch": 3.3827848101265823, |
| "grad_norm": 0.4717082885780641, |
| "learning_rate": 1.135521394010506e-05, |
| "loss": 0.1477, |
| "step": 834 |
| }, |
| { |
| "epoch": 3.386835443037975, |
| "grad_norm": 0.4581973231688191, |
| "learning_rate": 1.1304066178317367e-05, |
| "loss": 0.1353, |
| "step": 835 |
| }, |
| { |
| "epoch": 3.390886075949367, |
| "grad_norm": 0.46062807864154054, |
| "learning_rate": 1.1252988452385199e-05, |
| "loss": 0.1388, |
| "step": 836 |
| }, |
| { |
| "epoch": 3.3949367088607594, |
| "grad_norm": 0.45618778079224875, |
| "learning_rate": 1.1201981173681536e-05, |
| "loss": 0.1365, |
| "step": 837 |
| }, |
| { |
| "epoch": 3.398987341772152, |
| "grad_norm": 0.4663059087546015, |
| "learning_rate": 1.1151044753011991e-05, |
| "loss": 0.1582, |
| "step": 838 |
| }, |
| { |
| "epoch": 3.4030379746835444, |
| "grad_norm": 0.46158367029572966, |
| "learning_rate": 1.1100179600611491e-05, |
| "loss": 0.1412, |
| "step": 839 |
| }, |
| { |
| "epoch": 3.407088607594937, |
| "grad_norm": 0.4474678593149604, |
| "learning_rate": 1.1049386126140985e-05, |
| "loss": 0.1401, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.411139240506329, |
| "grad_norm": 0.45193580780642706, |
| "learning_rate": 1.0998664738684128e-05, |
| "loss": 0.1494, |
| "step": 841 |
| }, |
| { |
| "epoch": 3.4151898734177215, |
| "grad_norm": 0.4494867536532663, |
| "learning_rate": 1.0948015846744e-05, |
| "loss": 0.1462, |
| "step": 842 |
| }, |
| { |
| "epoch": 3.419240506329114, |
| "grad_norm": 0.4573568854616399, |
| "learning_rate": 1.0897439858239832e-05, |
| "loss": 0.1434, |
| "step": 843 |
| }, |
| { |
| "epoch": 3.423291139240506, |
| "grad_norm": 0.4546557326231091, |
| "learning_rate": 1.0846937180503652e-05, |
| "loss": 0.1414, |
| "step": 844 |
| }, |
| { |
| "epoch": 3.4273417721518986, |
| "grad_norm": 0.46404663541693403, |
| "learning_rate": 1.0796508220277117e-05, |
| "loss": 0.1391, |
| "step": 845 |
| }, |
| { |
| "epoch": 3.431392405063291, |
| "grad_norm": 0.4612461123712428, |
| "learning_rate": 1.0746153383708107e-05, |
| "loss": 0.1404, |
| "step": 846 |
| }, |
| { |
| "epoch": 3.4354430379746836, |
| "grad_norm": 0.47345198058975885, |
| "learning_rate": 1.0695873076347579e-05, |
| "loss": 0.1524, |
| "step": 847 |
| }, |
| { |
| "epoch": 3.439493670886076, |
| "grad_norm": 0.4684808816179674, |
| "learning_rate": 1.0645667703146205e-05, |
| "loss": 0.1482, |
| "step": 848 |
| }, |
| { |
| "epoch": 3.4435443037974682, |
| "grad_norm": 0.45219280088770153, |
| "learning_rate": 1.0595537668451161e-05, |
| "loss": 0.1418, |
| "step": 849 |
| }, |
| { |
| "epoch": 3.4475949367088607, |
| "grad_norm": 0.46336765651609746, |
| "learning_rate": 1.0545483376002854e-05, |
| "loss": 0.1476, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.4516455696202533, |
| "grad_norm": 0.45379890993066757, |
| "learning_rate": 1.0495505228931676e-05, |
| "loss": 0.146, |
| "step": 851 |
| }, |
| { |
| "epoch": 3.4556962025316453, |
| "grad_norm": 0.4727323314750382, |
| "learning_rate": 1.044560362975474e-05, |
| "loss": 0.1511, |
| "step": 852 |
| }, |
| { |
| "epoch": 3.459746835443038, |
| "grad_norm": 0.46072094553719173, |
| "learning_rate": 1.0395778980372695e-05, |
| "loss": 0.1504, |
| "step": 853 |
| }, |
| { |
| "epoch": 3.4637974683544304, |
| "grad_norm": 0.4531227736074988, |
| "learning_rate": 1.0346031682066381e-05, |
| "loss": 0.1403, |
| "step": 854 |
| }, |
| { |
| "epoch": 3.467848101265823, |
| "grad_norm": 0.46701291210702395, |
| "learning_rate": 1.0296362135493724e-05, |
| "loss": 0.1483, |
| "step": 855 |
| }, |
| { |
| "epoch": 3.4718987341772154, |
| "grad_norm": 0.45308855593562325, |
| "learning_rate": 1.0246770740686422e-05, |
| "loss": 0.1387, |
| "step": 856 |
| }, |
| { |
| "epoch": 3.4759493670886075, |
| "grad_norm": 0.44402701628067043, |
| "learning_rate": 1.0197257897046743e-05, |
| "loss": 0.1444, |
| "step": 857 |
| }, |
| { |
| "epoch": 3.48, |
| "grad_norm": 0.4802588331487622, |
| "learning_rate": 1.014782400334433e-05, |
| "loss": 0.1549, |
| "step": 858 |
| }, |
| { |
| "epoch": 3.4840506329113925, |
| "grad_norm": 0.4729540713212242, |
| "learning_rate": 1.009846945771296e-05, |
| "loss": 0.1517, |
| "step": 859 |
| }, |
| { |
| "epoch": 3.4881012658227846, |
| "grad_norm": 0.4689959506942431, |
| "learning_rate": 1.0049194657647363e-05, |
| "loss": 0.1509, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.492151898734177, |
| "grad_norm": 0.470355663857068, |
| "learning_rate": 1.0000000000000006e-05, |
| "loss": 0.1441, |
| "step": 861 |
| }, |
| { |
| "epoch": 3.4962025316455696, |
| "grad_norm": 0.4801017632274188, |
| "learning_rate": 9.950885880977891e-06, |
| "loss": 0.1494, |
| "step": 862 |
| }, |
| { |
| "epoch": 3.500253164556962, |
| "grad_norm": 0.4586490546968498, |
| "learning_rate": 9.901852696139382e-06, |
| "loss": 0.1347, |
| "step": 863 |
| }, |
| { |
| "epoch": 3.5043037974683546, |
| "grad_norm": 0.44720613740970344, |
| "learning_rate": 9.852900840391027e-06, |
| "loss": 0.148, |
| "step": 864 |
| }, |
| { |
| "epoch": 3.5083544303797467, |
| "grad_norm": 0.4929322395026425, |
| "learning_rate": 9.804030707984313e-06, |
| "loss": 0.1543, |
| "step": 865 |
| }, |
| { |
| "epoch": 3.512405063291139, |
| "grad_norm": 0.4708319719815674, |
| "learning_rate": 9.755242692512599e-06, |
| "loss": 0.1493, |
| "step": 866 |
| }, |
| { |
| "epoch": 3.5164556962025317, |
| "grad_norm": 0.4696663725335733, |
| "learning_rate": 9.70653718690782e-06, |
| "loss": 0.1555, |
| "step": 867 |
| }, |
| { |
| "epoch": 3.520506329113924, |
| "grad_norm": 0.4694626102305619, |
| "learning_rate": 9.657914583437454e-06, |
| "loss": 0.1541, |
| "step": 868 |
| }, |
| { |
| "epoch": 3.5245569620253163, |
| "grad_norm": 0.454184149481797, |
| "learning_rate": 9.609375273701246e-06, |
| "loss": 0.147, |
| "step": 869 |
| }, |
| { |
| "epoch": 3.528607594936709, |
| "grad_norm": 0.4544412602103682, |
| "learning_rate": 9.560919648628133e-06, |
| "loss": 0.1373, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.5326582278481014, |
| "grad_norm": 0.4606160290221402, |
| "learning_rate": 9.512548098473047e-06, |
| "loss": 0.1433, |
| "step": 871 |
| }, |
| { |
| "epoch": 3.536708860759494, |
| "grad_norm": 0.45677365522004526, |
| "learning_rate": 9.464261012813825e-06, |
| "loss": 0.1456, |
| "step": 872 |
| }, |
| { |
| "epoch": 3.540759493670886, |
| "grad_norm": 0.4504134973397786, |
| "learning_rate": 9.416058780547987e-06, |
| "loss": 0.1407, |
| "step": 873 |
| }, |
| { |
| "epoch": 3.5448101265822785, |
| "grad_norm": 0.4654006996970872, |
| "learning_rate": 9.367941789889714e-06, |
| "loss": 0.1509, |
| "step": 874 |
| }, |
| { |
| "epoch": 3.548860759493671, |
| "grad_norm": 0.47864125610152997, |
| "learning_rate": 9.319910428366607e-06, |
| "loss": 0.1474, |
| "step": 875 |
| }, |
| { |
| "epoch": 3.552911392405063, |
| "grad_norm": 0.4641929509738862, |
| "learning_rate": 9.271965082816667e-06, |
| "loss": 0.1491, |
| "step": 876 |
| }, |
| { |
| "epoch": 3.5569620253164556, |
| "grad_norm": 0.4577188429450541, |
| "learning_rate": 9.224106139385111e-06, |
| "loss": 0.1396, |
| "step": 877 |
| }, |
| { |
| "epoch": 3.561012658227848, |
| "grad_norm": 0.4795220867122461, |
| "learning_rate": 9.176333983521291e-06, |
| "loss": 0.1497, |
| "step": 878 |
| }, |
| { |
| "epoch": 3.5650632911392406, |
| "grad_norm": 0.45090794825749136, |
| "learning_rate": 9.12864899997558e-06, |
| "loss": 0.1465, |
| "step": 879 |
| }, |
| { |
| "epoch": 3.569113924050633, |
| "grad_norm": 0.4662767984501957, |
| "learning_rate": 9.08105157279628e-06, |
| "loss": 0.1475, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.573164556962025, |
| "grad_norm": 0.45136509460748186, |
| "learning_rate": 9.03354208532653e-06, |
| "loss": 0.1509, |
| "step": 881 |
| }, |
| { |
| "epoch": 3.5772151898734177, |
| "grad_norm": 0.45169900123471435, |
| "learning_rate": 8.986120920201205e-06, |
| "loss": 0.1543, |
| "step": 882 |
| }, |
| { |
| "epoch": 3.58126582278481, |
| "grad_norm": 0.45482026406395487, |
| "learning_rate": 8.938788459343852e-06, |
| "loss": 0.1458, |
| "step": 883 |
| }, |
| { |
| "epoch": 3.5853164556962023, |
| "grad_norm": 0.4521284572049662, |
| "learning_rate": 8.8915450839636e-06, |
| "loss": 0.1368, |
| "step": 884 |
| }, |
| { |
| "epoch": 3.589367088607595, |
| "grad_norm": 0.45104667294019807, |
| "learning_rate": 8.844391174552116e-06, |
| "loss": 0.145, |
| "step": 885 |
| }, |
| { |
| "epoch": 3.5934177215189873, |
| "grad_norm": 0.44630798008081896, |
| "learning_rate": 8.797327110880479e-06, |
| "loss": 0.1294, |
| "step": 886 |
| }, |
| { |
| "epoch": 3.59746835443038, |
| "grad_norm": 0.4596179927404309, |
| "learning_rate": 8.750353271996206e-06, |
| "loss": 0.146, |
| "step": 887 |
| }, |
| { |
| "epoch": 3.6015189873417723, |
| "grad_norm": 0.45020722133972646, |
| "learning_rate": 8.703470036220132e-06, |
| "loss": 0.1335, |
| "step": 888 |
| }, |
| { |
| "epoch": 3.6055696202531644, |
| "grad_norm": 0.44366387597306417, |
| "learning_rate": 8.656677781143394e-06, |
| "loss": 0.1424, |
| "step": 889 |
| }, |
| { |
| "epoch": 3.609620253164557, |
| "grad_norm": 0.4660524327959872, |
| "learning_rate": 8.609976883624377e-06, |
| "loss": 0.1425, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.6136708860759494, |
| "grad_norm": 0.45083080793044844, |
| "learning_rate": 8.563367719785698e-06, |
| "loss": 0.1358, |
| "step": 891 |
| }, |
| { |
| "epoch": 3.6177215189873415, |
| "grad_norm": 0.4678511611456318, |
| "learning_rate": 8.516850665011138e-06, |
| "loss": 0.1444, |
| "step": 892 |
| }, |
| { |
| "epoch": 3.621772151898734, |
| "grad_norm": 0.4517342710943577, |
| "learning_rate": 8.47042609394269e-06, |
| "loss": 0.1534, |
| "step": 893 |
| }, |
| { |
| "epoch": 3.6258227848101265, |
| "grad_norm": 0.4640643572055444, |
| "learning_rate": 8.424094380477432e-06, |
| "loss": 0.1525, |
| "step": 894 |
| }, |
| { |
| "epoch": 3.629873417721519, |
| "grad_norm": 0.4556586488537981, |
| "learning_rate": 8.37785589776465e-06, |
| "loss": 0.1366, |
| "step": 895 |
| }, |
| { |
| "epoch": 3.6339240506329116, |
| "grad_norm": 0.45417851572307294, |
| "learning_rate": 8.331711018202694e-06, |
| "loss": 0.1462, |
| "step": 896 |
| }, |
| { |
| "epoch": 3.6379746835443036, |
| "grad_norm": 0.45119248340372076, |
| "learning_rate": 8.285660113436104e-06, |
| "loss": 0.1467, |
| "step": 897 |
| }, |
| { |
| "epoch": 3.642025316455696, |
| "grad_norm": 0.45667537814287934, |
| "learning_rate": 8.239703554352527e-06, |
| "loss": 0.1469, |
| "step": 898 |
| }, |
| { |
| "epoch": 3.6460759493670887, |
| "grad_norm": 0.47151760858537717, |
| "learning_rate": 8.193841711079775e-06, |
| "loss": 0.1494, |
| "step": 899 |
| }, |
| { |
| "epoch": 3.6501265822784807, |
| "grad_norm": 0.4608188740878505, |
| "learning_rate": 8.148074952982828e-06, |
| "loss": 0.1427, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.6541772151898733, |
| "grad_norm": 0.4609529644790624, |
| "learning_rate": 8.102403648660859e-06, |
| "loss": 0.1499, |
| "step": 901 |
| }, |
| { |
| "epoch": 3.6582278481012658, |
| "grad_norm": 0.4653027826581259, |
| "learning_rate": 8.056828165944282e-06, |
| "loss": 0.1488, |
| "step": 902 |
| }, |
| { |
| "epoch": 3.6622784810126583, |
| "grad_norm": 0.4571951737757143, |
| "learning_rate": 8.011348871891762e-06, |
| "loss": 0.1429, |
| "step": 903 |
| }, |
| { |
| "epoch": 3.666329113924051, |
| "grad_norm": 0.4505414935953383, |
| "learning_rate": 7.965966132787287e-06, |
| "loss": 0.1364, |
| "step": 904 |
| }, |
| { |
| "epoch": 3.670379746835443, |
| "grad_norm": 0.4512197703771189, |
| "learning_rate": 7.920680314137189e-06, |
| "loss": 0.1446, |
| "step": 905 |
| }, |
| { |
| "epoch": 3.6744303797468354, |
| "grad_norm": 0.46399003059878313, |
| "learning_rate": 7.875491780667246e-06, |
| "loss": 0.1434, |
| "step": 906 |
| }, |
| { |
| "epoch": 3.678481012658228, |
| "grad_norm": 0.4549899608699432, |
| "learning_rate": 7.830400896319667e-06, |
| "loss": 0.1421, |
| "step": 907 |
| }, |
| { |
| "epoch": 3.68253164556962, |
| "grad_norm": 0.4576757835003457, |
| "learning_rate": 7.785408024250259e-06, |
| "loss": 0.1442, |
| "step": 908 |
| }, |
| { |
| "epoch": 3.6865822784810125, |
| "grad_norm": 0.45386701668542606, |
| "learning_rate": 7.74051352682542e-06, |
| "loss": 0.1479, |
| "step": 909 |
| }, |
| { |
| "epoch": 3.690632911392405, |
| "grad_norm": 0.45306038197016607, |
| "learning_rate": 7.695717765619257e-06, |
| "loss": 0.1446, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.6946835443037975, |
| "grad_norm": 0.46530890530945046, |
| "learning_rate": 7.651021101410673e-06, |
| "loss": 0.1469, |
| "step": 911 |
| }, |
| { |
| "epoch": 3.69873417721519, |
| "grad_norm": 0.45793561253964093, |
| "learning_rate": 7.606423894180464e-06, |
| "loss": 0.1404, |
| "step": 912 |
| }, |
| { |
| "epoch": 3.702784810126582, |
| "grad_norm": 0.4524526659828809, |
| "learning_rate": 7.56192650310839e-06, |
| "loss": 0.1498, |
| "step": 913 |
| }, |
| { |
| "epoch": 3.7068354430379746, |
| "grad_norm": 0.4708439998674182, |
| "learning_rate": 7.517529286570349e-06, |
| "loss": 0.1498, |
| "step": 914 |
| }, |
| { |
| "epoch": 3.710886075949367, |
| "grad_norm": 0.4408032092120141, |
| "learning_rate": 7.473232602135387e-06, |
| "loss": 0.1349, |
| "step": 915 |
| }, |
| { |
| "epoch": 3.714936708860759, |
| "grad_norm": 0.4430413356131997, |
| "learning_rate": 7.429036806562935e-06, |
| "loss": 0.1425, |
| "step": 916 |
| }, |
| { |
| "epoch": 3.7189873417721517, |
| "grad_norm": 0.4511598290752781, |
| "learning_rate": 7.3849422557998455e-06, |
| "loss": 0.1453, |
| "step": 917 |
| }, |
| { |
| "epoch": 3.7230379746835442, |
| "grad_norm": 0.4462515114115929, |
| "learning_rate": 7.340949304977567e-06, |
| "loss": 0.1362, |
| "step": 918 |
| }, |
| { |
| "epoch": 3.7270886075949368, |
| "grad_norm": 0.4478320884345968, |
| "learning_rate": 7.297058308409282e-06, |
| "loss": 0.1375, |
| "step": 919 |
| }, |
| { |
| "epoch": 3.7311392405063293, |
| "grad_norm": 0.44494978888477404, |
| "learning_rate": 7.25326961958704e-06, |
| "loss": 0.1371, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.7351898734177214, |
| "grad_norm": 0.4649832758917675, |
| "learning_rate": 7.209583591178921e-06, |
| "loss": 0.1577, |
| "step": 921 |
| }, |
| { |
| "epoch": 3.739240506329114, |
| "grad_norm": 0.4519442593591131, |
| "learning_rate": 7.1660005750261925e-06, |
| "loss": 0.1468, |
| "step": 922 |
| }, |
| { |
| "epoch": 3.7432911392405064, |
| "grad_norm": 0.45686989364043956, |
| "learning_rate": 7.1225209221404765e-06, |
| "loss": 0.1558, |
| "step": 923 |
| }, |
| { |
| "epoch": 3.747341772151899, |
| "grad_norm": 0.4492702777132558, |
| "learning_rate": 7.079144982700909e-06, |
| "loss": 0.1373, |
| "step": 924 |
| }, |
| { |
| "epoch": 3.7513924050632914, |
| "grad_norm": 0.45264598093077096, |
| "learning_rate": 7.0358731060513695e-06, |
| "loss": 0.1498, |
| "step": 925 |
| }, |
| { |
| "epoch": 3.7554430379746835, |
| "grad_norm": 0.44787123350879404, |
| "learning_rate": 6.99270564069757e-06, |
| "loss": 0.1349, |
| "step": 926 |
| }, |
| { |
| "epoch": 3.759493670886076, |
| "grad_norm": 0.44473434016990576, |
| "learning_rate": 6.949642934304375e-06, |
| "loss": 0.1366, |
| "step": 927 |
| }, |
| { |
| "epoch": 3.7635443037974685, |
| "grad_norm": 0.4510335466942928, |
| "learning_rate": 6.906685333692871e-06, |
| "loss": 0.1434, |
| "step": 928 |
| }, |
| { |
| "epoch": 3.7675949367088606, |
| "grad_norm": 0.45328138150321684, |
| "learning_rate": 6.86383318483769e-06, |
| "loss": 0.142, |
| "step": 929 |
| }, |
| { |
| "epoch": 3.771645569620253, |
| "grad_norm": 0.45794059238654394, |
| "learning_rate": 6.821086832864139e-06, |
| "loss": 0.1514, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.7756962025316456, |
| "grad_norm": 0.44302689100865866, |
| "learning_rate": 6.77844662204546e-06, |
| "loss": 0.1389, |
| "step": 931 |
| }, |
| { |
| "epoch": 3.779746835443038, |
| "grad_norm": 0.44427430035020443, |
| "learning_rate": 6.7359128958000455e-06, |
| "loss": 0.1394, |
| "step": 932 |
| }, |
| { |
| "epoch": 3.7837974683544306, |
| "grad_norm": 0.4402454176106253, |
| "learning_rate": 6.693485996688695e-06, |
| "loss": 0.1466, |
| "step": 933 |
| }, |
| { |
| "epoch": 3.7878481012658227, |
| "grad_norm": 0.4573271956144137, |
| "learning_rate": 6.651166266411801e-06, |
| "loss": 0.1529, |
| "step": 934 |
| }, |
| { |
| "epoch": 3.7918987341772152, |
| "grad_norm": 0.45146250139252553, |
| "learning_rate": 6.6089540458066725e-06, |
| "loss": 0.1346, |
| "step": 935 |
| }, |
| { |
| "epoch": 3.7959493670886078, |
| "grad_norm": 0.4478104969577133, |
| "learning_rate": 6.566849674844711e-06, |
| "loss": 0.1407, |
| "step": 936 |
| }, |
| { |
| "epoch": 3.8, |
| "grad_norm": 0.4489179638360717, |
| "learning_rate": 6.524853492628747e-06, |
| "loss": 0.1355, |
| "step": 937 |
| }, |
| { |
| "epoch": 3.8040506329113923, |
| "grad_norm": 0.4446173239885877, |
| "learning_rate": 6.4829658373902536e-06, |
| "loss": 0.1365, |
| "step": 938 |
| }, |
| { |
| "epoch": 3.808101265822785, |
| "grad_norm": 0.45368459526014737, |
| "learning_rate": 6.441187046486648e-06, |
| "loss": 0.1469, |
| "step": 939 |
| }, |
| { |
| "epoch": 3.8121518987341774, |
| "grad_norm": 0.4585275205757105, |
| "learning_rate": 6.399517456398567e-06, |
| "loss": 0.1441, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.81620253164557, |
| "grad_norm": 0.4478396772443359, |
| "learning_rate": 6.357957402727164e-06, |
| "loss": 0.1391, |
| "step": 941 |
| }, |
| { |
| "epoch": 3.820253164556962, |
| "grad_norm": 0.4536896143262987, |
| "learning_rate": 6.316507220191395e-06, |
| "loss": 0.1442, |
| "step": 942 |
| }, |
| { |
| "epoch": 3.8243037974683545, |
| "grad_norm": 0.45187915537651946, |
| "learning_rate": 6.275167242625331e-06, |
| "loss": 0.1494, |
| "step": 943 |
| }, |
| { |
| "epoch": 3.828354430379747, |
| "grad_norm": 0.45413295567180595, |
| "learning_rate": 6.233937802975471e-06, |
| "loss": 0.1414, |
| "step": 944 |
| }, |
| { |
| "epoch": 3.832405063291139, |
| "grad_norm": 0.44565125036680925, |
| "learning_rate": 6.192819233298046e-06, |
| "loss": 0.1355, |
| "step": 945 |
| }, |
| { |
| "epoch": 3.8364556962025316, |
| "grad_norm": 0.44065471076993323, |
| "learning_rate": 6.151811864756383e-06, |
| "loss": 0.1334, |
| "step": 946 |
| }, |
| { |
| "epoch": 3.840506329113924, |
| "grad_norm": 0.44483660800180636, |
| "learning_rate": 6.1109160276181655e-06, |
| "loss": 0.1309, |
| "step": 947 |
| }, |
| { |
| "epoch": 3.8445569620253166, |
| "grad_norm": 0.4543864487648866, |
| "learning_rate": 6.070132051252868e-06, |
| "loss": 0.1456, |
| "step": 948 |
| }, |
| { |
| "epoch": 3.848607594936709, |
| "grad_norm": 0.4437239723652923, |
| "learning_rate": 6.0294602641290034e-06, |
| "loss": 0.149, |
| "step": 949 |
| }, |
| { |
| "epoch": 3.852658227848101, |
| "grad_norm": 0.4562857907395769, |
| "learning_rate": 5.988900993811575e-06, |
| "loss": 0.1446, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.8567088607594937, |
| "grad_norm": 0.4395548230722485, |
| "learning_rate": 5.948454566959363e-06, |
| "loss": 0.1437, |
| "step": 951 |
| }, |
| { |
| "epoch": 3.8607594936708862, |
| "grad_norm": 0.4479270974264699, |
| "learning_rate": 5.908121309322328e-06, |
| "loss": 0.1367, |
| "step": 952 |
| }, |
| { |
| "epoch": 3.8648101265822783, |
| "grad_norm": 0.4497969864808348, |
| "learning_rate": 5.867901545738976e-06, |
| "loss": 0.141, |
| "step": 953 |
| }, |
| { |
| "epoch": 3.868860759493671, |
| "grad_norm": 0.4537517813757227, |
| "learning_rate": 5.827795600133774e-06, |
| "loss": 0.1386, |
| "step": 954 |
| }, |
| { |
| "epoch": 3.8729113924050633, |
| "grad_norm": 0.44629280520636344, |
| "learning_rate": 5.787803795514466e-06, |
| "loss": 0.1368, |
| "step": 955 |
| }, |
| { |
| "epoch": 3.876962025316456, |
| "grad_norm": 0.45227139744987827, |
| "learning_rate": 5.747926453969576e-06, |
| "loss": 0.1452, |
| "step": 956 |
| }, |
| { |
| "epoch": 3.8810126582278484, |
| "grad_norm": 0.43960281339542706, |
| "learning_rate": 5.708163896665708e-06, |
| "loss": 0.1333, |
| "step": 957 |
| }, |
| { |
| "epoch": 3.8850632911392404, |
| "grad_norm": 0.4543731073352413, |
| "learning_rate": 5.668516443845047e-06, |
| "loss": 0.1521, |
| "step": 958 |
| }, |
| { |
| "epoch": 3.889113924050633, |
| "grad_norm": 0.4430700392923175, |
| "learning_rate": 5.6289844148227225e-06, |
| "loss": 0.1461, |
| "step": 959 |
| }, |
| { |
| "epoch": 3.8931645569620255, |
| "grad_norm": 0.4538292940625044, |
| "learning_rate": 5.5895681279842615e-06, |
| "loss": 0.1435, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.8972151898734175, |
| "grad_norm": 0.43034668509067714, |
| "learning_rate": 5.550267900783019e-06, |
| "loss": 0.1358, |
| "step": 961 |
| }, |
| { |
| "epoch": 3.90126582278481, |
| "grad_norm": 0.444932777369951, |
| "learning_rate": 5.511084049737623e-06, |
| "loss": 0.1464, |
| "step": 962 |
| }, |
| { |
| "epoch": 3.9053164556962026, |
| "grad_norm": 0.45455617433007567, |
| "learning_rate": 5.4720168904294215e-06, |
| "loss": 0.1481, |
| "step": 963 |
| }, |
| { |
| "epoch": 3.909367088607595, |
| "grad_norm": 0.4357502061104318, |
| "learning_rate": 5.433066737499948e-06, |
| "loss": 0.1393, |
| "step": 964 |
| }, |
| { |
| "epoch": 3.9134177215189876, |
| "grad_norm": 0.44085714771447526, |
| "learning_rate": 5.394233904648376e-06, |
| "loss": 0.1411, |
| "step": 965 |
| }, |
| { |
| "epoch": 3.9174683544303797, |
| "grad_norm": 0.4463162383403825, |
| "learning_rate": 5.355518704628997e-06, |
| "loss": 0.1505, |
| "step": 966 |
| }, |
| { |
| "epoch": 3.921518987341772, |
| "grad_norm": 0.44744286176019993, |
| "learning_rate": 5.316921449248731e-06, |
| "loss": 0.147, |
| "step": 967 |
| }, |
| { |
| "epoch": 3.9255696202531647, |
| "grad_norm": 0.4365968922871633, |
| "learning_rate": 5.278442449364538e-06, |
| "loss": 0.1367, |
| "step": 968 |
| }, |
| { |
| "epoch": 3.9296202531645568, |
| "grad_norm": 0.44156624966977986, |
| "learning_rate": 5.240082014881016e-06, |
| "loss": 0.1333, |
| "step": 969 |
| }, |
| { |
| "epoch": 3.9336708860759493, |
| "grad_norm": 0.44894278010688005, |
| "learning_rate": 5.201840454747822e-06, |
| "loss": 0.1421, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.937721518987342, |
| "grad_norm": 0.4471808549166584, |
| "learning_rate": 5.163718076957223e-06, |
| "loss": 0.1369, |
| "step": 971 |
| }, |
| { |
| "epoch": 3.9417721518987343, |
| "grad_norm": 0.4405099899087764, |
| "learning_rate": 5.125715188541609e-06, |
| "loss": 0.1389, |
| "step": 972 |
| }, |
| { |
| "epoch": 3.945822784810127, |
| "grad_norm": 0.4541021469688468, |
| "learning_rate": 5.087832095571021e-06, |
| "loss": 0.1507, |
| "step": 973 |
| }, |
| { |
| "epoch": 3.949873417721519, |
| "grad_norm": 0.4319602528914583, |
| "learning_rate": 5.0500691031506766e-06, |
| "loss": 0.1367, |
| "step": 974 |
| }, |
| { |
| "epoch": 3.9539240506329114, |
| "grad_norm": 0.43450362727795255, |
| "learning_rate": 5.01242651541854e-06, |
| "loss": 0.1392, |
| "step": 975 |
| }, |
| { |
| "epoch": 3.957974683544304, |
| "grad_norm": 0.44155314608473695, |
| "learning_rate": 4.974904635542815e-06, |
| "loss": 0.1434, |
| "step": 976 |
| }, |
| { |
| "epoch": 3.962025316455696, |
| "grad_norm": 0.4453793700429282, |
| "learning_rate": 4.937503765719582e-06, |
| "loss": 0.1371, |
| "step": 977 |
| }, |
| { |
| "epoch": 3.9660759493670885, |
| "grad_norm": 0.45607005471905526, |
| "learning_rate": 4.900224207170299e-06, |
| "loss": 0.1339, |
| "step": 978 |
| }, |
| { |
| "epoch": 3.970126582278481, |
| "grad_norm": 0.4531585432250915, |
| "learning_rate": 4.8630662601394065e-06, |
| "loss": 0.1414, |
| "step": 979 |
| }, |
| { |
| "epoch": 3.9741772151898735, |
| "grad_norm": 0.4390249031917463, |
| "learning_rate": 4.8260302238918995e-06, |
| "loss": 0.1388, |
| "step": 980 |
| }, |
| { |
| "epoch": 3.978227848101266, |
| "grad_norm": 0.4489307305754667, |
| "learning_rate": 4.789116396710924e-06, |
| "loss": 0.142, |
| "step": 981 |
| }, |
| { |
| "epoch": 3.982278481012658, |
| "grad_norm": 0.4426288390968658, |
| "learning_rate": 4.752325075895368e-06, |
| "loss": 0.1434, |
| "step": 982 |
| }, |
| { |
| "epoch": 3.9863291139240506, |
| "grad_norm": 0.4573099932672487, |
| "learning_rate": 4.715656557757473e-06, |
| "loss": 0.1455, |
| "step": 983 |
| }, |
| { |
| "epoch": 3.990379746835443, |
| "grad_norm": 0.42999243501221757, |
| "learning_rate": 4.679111137620442e-06, |
| "loss": 0.1385, |
| "step": 984 |
| }, |
| { |
| "epoch": 3.9944303797468352, |
| "grad_norm": 0.4380572577583787, |
| "learning_rate": 4.6426891098160585e-06, |
| "loss": 0.1307, |
| "step": 985 |
| }, |
| { |
| "epoch": 3.9984810126582278, |
| "grad_norm": 0.4401949415250603, |
| "learning_rate": 4.6063907676823474e-06, |
| "loss": 0.1444, |
| "step": 986 |
| }, |
| { |
| "epoch": 4.004050632911392, |
| "grad_norm": 0.43352756130848125, |
| "learning_rate": 4.570216403561141e-06, |
| "loss": 0.0701, |
| "step": 987 |
| }, |
| { |
| "epoch": 4.008101265822785, |
| "grad_norm": 0.41167595045804145, |
| "learning_rate": 4.534166308795815e-06, |
| "loss": 0.0681, |
| "step": 988 |
| }, |
| { |
| "epoch": 4.012151898734177, |
| "grad_norm": 0.38553533674730495, |
| "learning_rate": 4.498240773728859e-06, |
| "loss": 0.0682, |
| "step": 989 |
| }, |
| { |
| "epoch": 4.01620253164557, |
| "grad_norm": 0.3641510669145834, |
| "learning_rate": 4.462440087699609e-06, |
| "loss": 0.0728, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.020253164556962, |
| "grad_norm": 0.32406401609862234, |
| "learning_rate": 4.426764539041861e-06, |
| "loss": 0.0669, |
| "step": 991 |
| }, |
| { |
| "epoch": 4.024303797468354, |
| "grad_norm": 0.31130926447589913, |
| "learning_rate": 4.391214415081582e-06, |
| "loss": 0.0662, |
| "step": 992 |
| }, |
| { |
| "epoch": 4.028354430379747, |
| "grad_norm": 0.2882930509898079, |
| "learning_rate": 4.355790002134579e-06, |
| "loss": 0.06, |
| "step": 993 |
| }, |
| { |
| "epoch": 4.032405063291139, |
| "grad_norm": 0.306797848307614, |
| "learning_rate": 4.320491585504207e-06, |
| "loss": 0.0581, |
| "step": 994 |
| }, |
| { |
| "epoch": 4.036455696202531, |
| "grad_norm": 0.3618422264245487, |
| "learning_rate": 4.2853194494790615e-06, |
| "loss": 0.0743, |
| "step": 995 |
| }, |
| { |
| "epoch": 4.040506329113924, |
| "grad_norm": 0.36129130546155586, |
| "learning_rate": 4.250273877330691e-06, |
| "loss": 0.0624, |
| "step": 996 |
| }, |
| { |
| "epoch": 4.044556962025316, |
| "grad_norm": 0.383867890270933, |
| "learning_rate": 4.215355151311313e-06, |
| "loss": 0.066, |
| "step": 997 |
| }, |
| { |
| "epoch": 4.048607594936709, |
| "grad_norm": 0.3739719558787052, |
| "learning_rate": 4.180563552651542e-06, |
| "loss": 0.0632, |
| "step": 998 |
| }, |
| { |
| "epoch": 4.052658227848101, |
| "grad_norm": 0.3805245273382703, |
| "learning_rate": 4.145899361558147e-06, |
| "loss": 0.0627, |
| "step": 999 |
| }, |
| { |
| "epoch": 4.056708860759493, |
| "grad_norm": 0.3789486027916783, |
| "learning_rate": 4.111362857211738e-06, |
| "loss": 0.0647, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.060759493670886, |
| "grad_norm": 0.35324954859182073, |
| "learning_rate": 4.076954317764592e-06, |
| "loss": 0.0651, |
| "step": 1001 |
| }, |
| { |
| "epoch": 4.0648101265822785, |
| "grad_norm": 0.3193994523745831, |
| "learning_rate": 4.042674020338335e-06, |
| "loss": 0.0621, |
| "step": 1002 |
| }, |
| { |
| "epoch": 4.0688607594936705, |
| "grad_norm": 0.310463567766949, |
| "learning_rate": 4.0085222410217835e-06, |
| "loss": 0.0583, |
| "step": 1003 |
| }, |
| { |
| "epoch": 4.0729113924050635, |
| "grad_norm": 0.31283769171545805, |
| "learning_rate": 3.974499254868674e-06, |
| "loss": 0.0664, |
| "step": 1004 |
| }, |
| { |
| "epoch": 4.076962025316456, |
| "grad_norm": 0.30664258052771315, |
| "learning_rate": 3.940605335895451e-06, |
| "loss": 0.0648, |
| "step": 1005 |
| }, |
| { |
| "epoch": 4.0810126582278485, |
| "grad_norm": 0.2968394696570242, |
| "learning_rate": 3.90684075707908e-06, |
| "loss": 0.0634, |
| "step": 1006 |
| }, |
| { |
| "epoch": 4.085063291139241, |
| "grad_norm": 0.3014504021529523, |
| "learning_rate": 3.8732057903548505e-06, |
| "loss": 0.0634, |
| "step": 1007 |
| }, |
| { |
| "epoch": 4.089113924050633, |
| "grad_norm": 0.28538988860016623, |
| "learning_rate": 3.8397007066141375e-06, |
| "loss": 0.0563, |
| "step": 1008 |
| }, |
| { |
| "epoch": 4.093164556962026, |
| "grad_norm": 0.30005486715352664, |
| "learning_rate": 3.806325775702304e-06, |
| "loss": 0.0619, |
| "step": 1009 |
| }, |
| { |
| "epoch": 4.097215189873418, |
| "grad_norm": 0.3038867334470333, |
| "learning_rate": 3.773081266416434e-06, |
| "loss": 0.0701, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.10126582278481, |
| "grad_norm": 0.3041830610814425, |
| "learning_rate": 3.739967446503245e-06, |
| "loss": 0.0704, |
| "step": 1011 |
| }, |
| { |
| "epoch": 4.105316455696203, |
| "grad_norm": 0.29817272538079814, |
| "learning_rate": 3.706984582656894e-06, |
| "loss": 0.0606, |
| "step": 1012 |
| }, |
| { |
| "epoch": 4.109367088607595, |
| "grad_norm": 0.2911356671464797, |
| "learning_rate": 3.6741329405168237e-06, |
| "loss": 0.059, |
| "step": 1013 |
| }, |
| { |
| "epoch": 4.113417721518988, |
| "grad_norm": 0.29737454091319304, |
| "learning_rate": 3.641412784665648e-06, |
| "loss": 0.0634, |
| "step": 1014 |
| }, |
| { |
| "epoch": 4.11746835443038, |
| "grad_norm": 0.29522633247232405, |
| "learning_rate": 3.608824378627005e-06, |
| "loss": 0.0635, |
| "step": 1015 |
| }, |
| { |
| "epoch": 4.121518987341772, |
| "grad_norm": 0.30472936588452626, |
| "learning_rate": 3.5763679848634337e-06, |
| "loss": 0.0662, |
| "step": 1016 |
| }, |
| { |
| "epoch": 4.125569620253165, |
| "grad_norm": 0.2981486071090898, |
| "learning_rate": 3.544043864774269e-06, |
| "loss": 0.0654, |
| "step": 1017 |
| }, |
| { |
| "epoch": 4.129620253164557, |
| "grad_norm": 0.3133592185129186, |
| "learning_rate": 3.5118522786935282e-06, |
| "loss": 0.0657, |
| "step": 1018 |
| }, |
| { |
| "epoch": 4.133670886075949, |
| "grad_norm": 0.32100111840429474, |
| "learning_rate": 3.479793485887819e-06, |
| "loss": 0.0642, |
| "step": 1019 |
| }, |
| { |
| "epoch": 4.137721518987342, |
| "grad_norm": 0.29538770133360803, |
| "learning_rate": 3.4478677445542653e-06, |
| "loss": 0.0561, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.141772151898734, |
| "grad_norm": 0.3185971455075416, |
| "learning_rate": 3.4160753118183767e-06, |
| "loss": 0.0622, |
| "step": 1021 |
| }, |
| { |
| "epoch": 4.145822784810127, |
| "grad_norm": 0.3205117468248187, |
| "learning_rate": 3.3844164437320527e-06, |
| "loss": 0.0662, |
| "step": 1022 |
| }, |
| { |
| "epoch": 4.149873417721519, |
| "grad_norm": 0.31265287212807247, |
| "learning_rate": 3.3528913952714558e-06, |
| "loss": 0.0645, |
| "step": 1023 |
| }, |
| { |
| "epoch": 4.153924050632911, |
| "grad_norm": 0.3002874392732933, |
| "learning_rate": 3.321500420335e-06, |
| "loss": 0.0609, |
| "step": 1024 |
| }, |
| { |
| "epoch": 4.157974683544304, |
| "grad_norm": 0.31668059100825796, |
| "learning_rate": 3.290243771741275e-06, |
| "loss": 0.069, |
| "step": 1025 |
| }, |
| { |
| "epoch": 4.162025316455696, |
| "grad_norm": 0.2979955247007704, |
| "learning_rate": 3.2591217012270325e-06, |
| "loss": 0.0599, |
| "step": 1026 |
| }, |
| { |
| "epoch": 4.166075949367088, |
| "grad_norm": 0.2936973457137949, |
| "learning_rate": 3.228134459445149e-06, |
| "loss": 0.0564, |
| "step": 1027 |
| }, |
| { |
| "epoch": 4.170126582278481, |
| "grad_norm": 0.30664080806004707, |
| "learning_rate": 3.1972822959626205e-06, |
| "loss": 0.0624, |
| "step": 1028 |
| }, |
| { |
| "epoch": 4.174177215189873, |
| "grad_norm": 0.3068656592901503, |
| "learning_rate": 3.166565459258513e-06, |
| "loss": 0.0648, |
| "step": 1029 |
| }, |
| { |
| "epoch": 4.178227848101266, |
| "grad_norm": 0.32011953170293866, |
| "learning_rate": 3.1359841967220193e-06, |
| "loss": 0.0687, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.182278481012658, |
| "grad_norm": 0.29545865181996767, |
| "learning_rate": 3.105538754650419e-06, |
| "loss": 0.0557, |
| "step": 1031 |
| }, |
| { |
| "epoch": 4.18632911392405, |
| "grad_norm": 0.30728246578891993, |
| "learning_rate": 3.07522937824712e-06, |
| "loss": 0.0666, |
| "step": 1032 |
| }, |
| { |
| "epoch": 4.190379746835443, |
| "grad_norm": 0.3147930520389128, |
| "learning_rate": 3.0450563116196697e-06, |
| "loss": 0.0718, |
| "step": 1033 |
| }, |
| { |
| "epoch": 4.194430379746835, |
| "grad_norm": 0.3073288814517406, |
| "learning_rate": 3.0150197977778008e-06, |
| "loss": 0.0606, |
| "step": 1034 |
| }, |
| { |
| "epoch": 4.1984810126582275, |
| "grad_norm": 0.29114305525927203, |
| "learning_rate": 2.985120078631465e-06, |
| "loss": 0.0547, |
| "step": 1035 |
| }, |
| { |
| "epoch": 4.2025316455696204, |
| "grad_norm": 0.3009779345560235, |
| "learning_rate": 2.9553573949888893e-06, |
| "loss": 0.0632, |
| "step": 1036 |
| }, |
| { |
| "epoch": 4.2065822784810125, |
| "grad_norm": 0.3016457706961184, |
| "learning_rate": 2.9257319865546384e-06, |
| "loss": 0.0647, |
| "step": 1037 |
| }, |
| { |
| "epoch": 4.2106329113924055, |
| "grad_norm": 0.2923466437988415, |
| "learning_rate": 2.896244091927678e-06, |
| "loss": 0.059, |
| "step": 1038 |
| }, |
| { |
| "epoch": 4.2146835443037975, |
| "grad_norm": 0.304418265279203, |
| "learning_rate": 2.8668939485994584e-06, |
| "loss": 0.0647, |
| "step": 1039 |
| }, |
| { |
| "epoch": 4.21873417721519, |
| "grad_norm": 0.30384524128246454, |
| "learning_rate": 2.837681792951994e-06, |
| "loss": 0.0622, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.222784810126583, |
| "grad_norm": 0.3217627373949058, |
| "learning_rate": 2.808607860255981e-06, |
| "loss": 0.0742, |
| "step": 1041 |
| }, |
| { |
| "epoch": 4.226835443037975, |
| "grad_norm": 0.30472410496512414, |
| "learning_rate": 2.7796723846688634e-06, |
| "loss": 0.0628, |
| "step": 1042 |
| }, |
| { |
| "epoch": 4.230886075949367, |
| "grad_norm": 0.31453461326941073, |
| "learning_rate": 2.7508755992329937e-06, |
| "loss": 0.0627, |
| "step": 1043 |
| }, |
| { |
| "epoch": 4.23493670886076, |
| "grad_norm": 0.315348337830132, |
| "learning_rate": 2.722217735873718e-06, |
| "loss": 0.072, |
| "step": 1044 |
| }, |
| { |
| "epoch": 4.238987341772152, |
| "grad_norm": 0.29891281440214584, |
| "learning_rate": 2.6936990253975315e-06, |
| "loss": 0.0632, |
| "step": 1045 |
| }, |
| { |
| "epoch": 4.243037974683545, |
| "grad_norm": 0.3003075617720225, |
| "learning_rate": 2.665319697490205e-06, |
| "loss": 0.0609, |
| "step": 1046 |
| }, |
| { |
| "epoch": 4.247088607594937, |
| "grad_norm": 0.31224860745678584, |
| "learning_rate": 2.637079980714945e-06, |
| "loss": 0.068, |
| "step": 1047 |
| }, |
| { |
| "epoch": 4.251139240506329, |
| "grad_norm": 0.30343128768769584, |
| "learning_rate": 2.6089801025105453e-06, |
| "loss": 0.0628, |
| "step": 1048 |
| }, |
| { |
| "epoch": 4.255189873417722, |
| "grad_norm": 0.3036106149543612, |
| "learning_rate": 2.581020289189571e-06, |
| "loss": 0.0584, |
| "step": 1049 |
| }, |
| { |
| "epoch": 4.259240506329114, |
| "grad_norm": 0.30750017360579474, |
| "learning_rate": 2.553200765936501e-06, |
| "loss": 0.0592, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.263291139240506, |
| "grad_norm": 0.3101895425289598, |
| "learning_rate": 2.525521756805962e-06, |
| "loss": 0.0685, |
| "step": 1051 |
| }, |
| { |
| "epoch": 4.267341772151899, |
| "grad_norm": 0.29173548914027786, |
| "learning_rate": 2.497983484720885e-06, |
| "loss": 0.0592, |
| "step": 1052 |
| }, |
| { |
| "epoch": 4.271392405063291, |
| "grad_norm": 0.30872453009373463, |
| "learning_rate": 2.470586171470728e-06, |
| "loss": 0.0654, |
| "step": 1053 |
| }, |
| { |
| "epoch": 4.275443037974684, |
| "grad_norm": 0.3021412699571817, |
| "learning_rate": 2.4433300377096836e-06, |
| "loss": 0.0624, |
| "step": 1054 |
| }, |
| { |
| "epoch": 4.279493670886076, |
| "grad_norm": 0.303060003945744, |
| "learning_rate": 2.4162153029549073e-06, |
| "loss": 0.0651, |
| "step": 1055 |
| }, |
| { |
| "epoch": 4.283544303797468, |
| "grad_norm": 0.29900831358909, |
| "learning_rate": 2.3892421855847458e-06, |
| "loss": 0.0596, |
| "step": 1056 |
| }, |
| { |
| "epoch": 4.287594936708861, |
| "grad_norm": 0.3077441952732954, |
| "learning_rate": 2.362410902836978e-06, |
| "loss": 0.0671, |
| "step": 1057 |
| }, |
| { |
| "epoch": 4.291645569620253, |
| "grad_norm": 0.3165591748586177, |
| "learning_rate": 2.3357216708070653e-06, |
| "loss": 0.0622, |
| "step": 1058 |
| }, |
| { |
| "epoch": 4.295696202531645, |
| "grad_norm": 0.30575257386517823, |
| "learning_rate": 2.309174704446411e-06, |
| "loss": 0.0645, |
| "step": 1059 |
| }, |
| { |
| "epoch": 4.299746835443038, |
| "grad_norm": 0.297470666057587, |
| "learning_rate": 2.2827702175606437e-06, |
| "loss": 0.0597, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.30379746835443, |
| "grad_norm": 0.30670686254265206, |
| "learning_rate": 2.256508422807855e-06, |
| "loss": 0.0668, |
| "step": 1061 |
| }, |
| { |
| "epoch": 4.307848101265823, |
| "grad_norm": 0.3013753904737074, |
| "learning_rate": 2.230389531696946e-06, |
| "loss": 0.0616, |
| "step": 1062 |
| }, |
| { |
| "epoch": 4.311898734177215, |
| "grad_norm": 0.2972050068229906, |
| "learning_rate": 2.204413754585857e-06, |
| "loss": 0.0629, |
| "step": 1063 |
| }, |
| { |
| "epoch": 4.315949367088607, |
| "grad_norm": 0.30241420345490305, |
| "learning_rate": 2.1785813006799406e-06, |
| "loss": 0.0624, |
| "step": 1064 |
| }, |
| { |
| "epoch": 4.32, |
| "grad_norm": 0.3078070495276007, |
| "learning_rate": 2.1528923780302224e-06, |
| "loss": 0.0624, |
| "step": 1065 |
| }, |
| { |
| "epoch": 4.324050632911392, |
| "grad_norm": 0.30194343673209684, |
| "learning_rate": 2.127347193531757e-06, |
| "loss": 0.0677, |
| "step": 1066 |
| }, |
| { |
| "epoch": 4.328101265822784, |
| "grad_norm": 0.2948674642992075, |
| "learning_rate": 2.101945952921942e-06, |
| "loss": 0.0594, |
| "step": 1067 |
| }, |
| { |
| "epoch": 4.332151898734177, |
| "grad_norm": 0.29616321603693013, |
| "learning_rate": 2.0766888607788906e-06, |
| "loss": 0.0626, |
| "step": 1068 |
| }, |
| { |
| "epoch": 4.3362025316455695, |
| "grad_norm": 0.3017424481986944, |
| "learning_rate": 2.0515761205197337e-06, |
| "loss": 0.0597, |
| "step": 1069 |
| }, |
| { |
| "epoch": 4.340253164556962, |
| "grad_norm": 0.30199573939585755, |
| "learning_rate": 2.0266079343990453e-06, |
| "loss": 0.0648, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.3443037974683545, |
| "grad_norm": 0.29418071381287647, |
| "learning_rate": 2.0017845035071494e-06, |
| "loss": 0.0561, |
| "step": 1071 |
| }, |
| { |
| "epoch": 4.348354430379747, |
| "grad_norm": 0.2857826877409487, |
| "learning_rate": 1.9771060277685537e-06, |
| "loss": 0.0606, |
| "step": 1072 |
| }, |
| { |
| "epoch": 4.3524050632911395, |
| "grad_norm": 0.302701012520005, |
| "learning_rate": 1.95257270594031e-06, |
| "loss": 0.06, |
| "step": 1073 |
| }, |
| { |
| "epoch": 4.356455696202532, |
| "grad_norm": 0.3064482728261592, |
| "learning_rate": 1.9281847356104188e-06, |
| "loss": 0.0616, |
| "step": 1074 |
| }, |
| { |
| "epoch": 4.360506329113924, |
| "grad_norm": 0.30066553040340366, |
| "learning_rate": 1.9039423131962365e-06, |
| "loss": 0.0635, |
| "step": 1075 |
| }, |
| { |
| "epoch": 4.364556962025317, |
| "grad_norm": 0.30303324432984047, |
| "learning_rate": 1.8798456339429027e-06, |
| "loss": 0.0616, |
| "step": 1076 |
| }, |
| { |
| "epoch": 4.368607594936709, |
| "grad_norm": 0.3156872368392941, |
| "learning_rate": 1.8558948919217612e-06, |
| "loss": 0.0708, |
| "step": 1077 |
| }, |
| { |
| "epoch": 4.372658227848102, |
| "grad_norm": 0.2986548943969268, |
| "learning_rate": 1.8320902800287954e-06, |
| "loss": 0.0624, |
| "step": 1078 |
| }, |
| { |
| "epoch": 4.376708860759494, |
| "grad_norm": 0.29426441646322266, |
| "learning_rate": 1.8084319899830726e-06, |
| "loss": 0.0574, |
| "step": 1079 |
| }, |
| { |
| "epoch": 4.380759493670886, |
| "grad_norm": 0.2925116038675086, |
| "learning_rate": 1.7849202123252097e-06, |
| "loss": 0.0598, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.384810126582279, |
| "grad_norm": 0.30654453797534503, |
| "learning_rate": 1.7615551364158401e-06, |
| "loss": 0.0647, |
| "step": 1081 |
| }, |
| { |
| "epoch": 4.388860759493671, |
| "grad_norm": 0.29041688840289387, |
| "learning_rate": 1.738336950434061e-06, |
| "loss": 0.0561, |
| "step": 1082 |
| }, |
| { |
| "epoch": 4.392911392405063, |
| "grad_norm": 0.2969422853044283, |
| "learning_rate": 1.715265841375957e-06, |
| "loss": 0.0579, |
| "step": 1083 |
| }, |
| { |
| "epoch": 4.396962025316456, |
| "grad_norm": 0.30300414458836195, |
| "learning_rate": 1.6923419950530684e-06, |
| "loss": 0.0626, |
| "step": 1084 |
| }, |
| { |
| "epoch": 4.401012658227848, |
| "grad_norm": 0.3107773822970189, |
| "learning_rate": 1.6695655960909008e-06, |
| "loss": 0.0651, |
| "step": 1085 |
| }, |
| { |
| "epoch": 4.405063291139241, |
| "grad_norm": 0.2861664859491347, |
| "learning_rate": 1.646936827927441e-06, |
| "loss": 0.0597, |
| "step": 1086 |
| }, |
| { |
| "epoch": 4.409113924050633, |
| "grad_norm": 0.3056627609353436, |
| "learning_rate": 1.6244558728116766e-06, |
| "loss": 0.0619, |
| "step": 1087 |
| }, |
| { |
| "epoch": 4.413164556962025, |
| "grad_norm": 0.3109556222150808, |
| "learning_rate": 1.6021229118021265e-06, |
| "loss": 0.0625, |
| "step": 1088 |
| }, |
| { |
| "epoch": 4.417215189873418, |
| "grad_norm": 0.2885419249143678, |
| "learning_rate": 1.5799381247653967e-06, |
| "loss": 0.0617, |
| "step": 1089 |
| }, |
| { |
| "epoch": 4.42126582278481, |
| "grad_norm": 0.2892629819928812, |
| "learning_rate": 1.5579016903747013e-06, |
| "loss": 0.0569, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.425316455696202, |
| "grad_norm": 0.31836496469328457, |
| "learning_rate": 1.5360137861084656e-06, |
| "loss": 0.0657, |
| "step": 1091 |
| }, |
| { |
| "epoch": 4.429367088607595, |
| "grad_norm": 0.28798223555291214, |
| "learning_rate": 1.5142745882488475e-06, |
| "loss": 0.0567, |
| "step": 1092 |
| }, |
| { |
| "epoch": 4.433417721518987, |
| "grad_norm": 0.3115489911357705, |
| "learning_rate": 1.4926842718803691e-06, |
| "loss": 0.0636, |
| "step": 1093 |
| }, |
| { |
| "epoch": 4.43746835443038, |
| "grad_norm": 0.29556325483355866, |
| "learning_rate": 1.4712430108884657e-06, |
| "loss": 0.0597, |
| "step": 1094 |
| }, |
| { |
| "epoch": 4.441518987341772, |
| "grad_norm": 0.3106605277945633, |
| "learning_rate": 1.4499509779581078e-06, |
| "loss": 0.0591, |
| "step": 1095 |
| }, |
| { |
| "epoch": 4.445569620253164, |
| "grad_norm": 0.30721491832929715, |
| "learning_rate": 1.4288083445723988e-06, |
| "loss": 0.064, |
| "step": 1096 |
| }, |
| { |
| "epoch": 4.449620253164557, |
| "grad_norm": 0.29908894301518635, |
| "learning_rate": 1.4078152810112045e-06, |
| "loss": 0.0605, |
| "step": 1097 |
| }, |
| { |
| "epoch": 4.453670886075949, |
| "grad_norm": 0.3001980707755454, |
| "learning_rate": 1.3869719563497697e-06, |
| "loss": 0.0581, |
| "step": 1098 |
| }, |
| { |
| "epoch": 4.457721518987341, |
| "grad_norm": 0.3011777841361766, |
| "learning_rate": 1.3662785384573663e-06, |
| "loss": 0.0643, |
| "step": 1099 |
| }, |
| { |
| "epoch": 4.461772151898734, |
| "grad_norm": 0.31191013624492064, |
| "learning_rate": 1.3457351939959383e-06, |
| "loss": 0.071, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.465822784810126, |
| "grad_norm": 0.3093696934240766, |
| "learning_rate": 1.3253420884187551e-06, |
| "loss": 0.0653, |
| "step": 1101 |
| }, |
| { |
| "epoch": 4.469873417721519, |
| "grad_norm": 0.2924153379111255, |
| "learning_rate": 1.3050993859690953e-06, |
| "loss": 0.06, |
| "step": 1102 |
| }, |
| { |
| "epoch": 4.473924050632911, |
| "grad_norm": 0.2960965616297582, |
| "learning_rate": 1.2850072496788869e-06, |
| "loss": 0.0596, |
| "step": 1103 |
| }, |
| { |
| "epoch": 4.4779746835443035, |
| "grad_norm": 0.30245448174625633, |
| "learning_rate": 1.2650658413674434e-06, |
| "loss": 0.0584, |
| "step": 1104 |
| }, |
| { |
| "epoch": 4.4820253164556965, |
| "grad_norm": 0.29032883397219916, |
| "learning_rate": 1.2452753216401226e-06, |
| "loss": 0.0577, |
| "step": 1105 |
| }, |
| { |
| "epoch": 4.4860759493670885, |
| "grad_norm": 0.2990379134861573, |
| "learning_rate": 1.2256358498870503e-06, |
| "loss": 0.0631, |
| "step": 1106 |
| }, |
| { |
| "epoch": 4.490126582278481, |
| "grad_norm": 0.30811722532177493, |
| "learning_rate": 1.2061475842818337e-06, |
| "loss": 0.0682, |
| "step": 1107 |
| }, |
| { |
| "epoch": 4.494177215189874, |
| "grad_norm": 0.2899843059637487, |
| "learning_rate": 1.1868106817802816e-06, |
| "loss": 0.0542, |
| "step": 1108 |
| }, |
| { |
| "epoch": 4.498227848101266, |
| "grad_norm": 0.2903846950550272, |
| "learning_rate": 1.1676252981191482e-06, |
| "loss": 0.0565, |
| "step": 1109 |
| }, |
| { |
| "epoch": 4.502278481012659, |
| "grad_norm": 0.2909487394746265, |
| "learning_rate": 1.1485915878148823e-06, |
| "loss": 0.0553, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.506329113924051, |
| "grad_norm": 0.2838628146569836, |
| "learning_rate": 1.1297097041623584e-06, |
| "loss": 0.06, |
| "step": 1111 |
| }, |
| { |
| "epoch": 4.510379746835443, |
| "grad_norm": 0.30545036804227704, |
| "learning_rate": 1.1109797992336847e-06, |
| "loss": 0.0608, |
| "step": 1112 |
| }, |
| { |
| "epoch": 4.514430379746836, |
| "grad_norm": 0.30363094934093116, |
| "learning_rate": 1.092402023876933e-06, |
| "loss": 0.0614, |
| "step": 1113 |
| }, |
| { |
| "epoch": 4.518481012658228, |
| "grad_norm": 0.3024373184556324, |
| "learning_rate": 1.0739765277149527e-06, |
| "loss": 0.0675, |
| "step": 1114 |
| }, |
| { |
| "epoch": 4.52253164556962, |
| "grad_norm": 0.2934465431355414, |
| "learning_rate": 1.0557034591441596e-06, |
| "loss": 0.0563, |
| "step": 1115 |
| }, |
| { |
| "epoch": 4.526582278481013, |
| "grad_norm": 0.31162440656811363, |
| "learning_rate": 1.0375829653333324e-06, |
| "loss": 0.0671, |
| "step": 1116 |
| }, |
| { |
| "epoch": 4.530632911392405, |
| "grad_norm": 0.3068651996191024, |
| "learning_rate": 1.0196151922224385e-06, |
| "loss": 0.065, |
| "step": 1117 |
| }, |
| { |
| "epoch": 4.534683544303798, |
| "grad_norm": 0.2978740249192706, |
| "learning_rate": 1.0018002845214526e-06, |
| "loss": 0.0586, |
| "step": 1118 |
| }, |
| { |
| "epoch": 4.53873417721519, |
| "grad_norm": 0.29301930882021404, |
| "learning_rate": 9.841383857091947e-07, |
| "loss": 0.0597, |
| "step": 1119 |
| }, |
| { |
| "epoch": 4.542784810126582, |
| "grad_norm": 0.296471417942431, |
| "learning_rate": 9.666296380321616e-07, |
| "loss": 0.0608, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.546835443037975, |
| "grad_norm": 0.2986617921762481, |
| "learning_rate": 9.492741825034124e-07, |
| "loss": 0.0595, |
| "step": 1121 |
| }, |
| { |
| "epoch": 4.550886075949367, |
| "grad_norm": 0.3015136926060872, |
| "learning_rate": 9.320721589013892e-07, |
| "loss": 0.063, |
| "step": 1122 |
| }, |
| { |
| "epoch": 4.55493670886076, |
| "grad_norm": 0.30424304448084805, |
| "learning_rate": 9.150237057688339e-07, |
| "loss": 0.0611, |
| "step": 1123 |
| }, |
| { |
| "epoch": 4.558987341772152, |
| "grad_norm": 0.2886311301736603, |
| "learning_rate": 8.981289604116328e-07, |
| "loss": 0.0578, |
| "step": 1124 |
| }, |
| { |
| "epoch": 4.563037974683544, |
| "grad_norm": 0.2982866542844636, |
| "learning_rate": 8.813880588977542e-07, |
| "loss": 0.0613, |
| "step": 1125 |
| }, |
| { |
| "epoch": 4.567088607594937, |
| "grad_norm": 0.29565998261414683, |
| "learning_rate": 8.648011360561126e-07, |
| "loss": 0.0618, |
| "step": 1126 |
| }, |
| { |
| "epoch": 4.571139240506329, |
| "grad_norm": 0.30987594374694916, |
| "learning_rate": 8.483683254755037e-07, |
| "loss": 0.0669, |
| "step": 1127 |
| }, |
| { |
| "epoch": 4.575189873417721, |
| "grad_norm": 0.30069973158796043, |
| "learning_rate": 8.320897595035227e-07, |
| "loss": 0.0596, |
| "step": 1128 |
| }, |
| { |
| "epoch": 4.579240506329114, |
| "grad_norm": 0.2961418166208025, |
| "learning_rate": 8.159655692455093e-07, |
| "loss": 0.0664, |
| "step": 1129 |
| }, |
| { |
| "epoch": 4.583291139240506, |
| "grad_norm": 0.30590700951489763, |
| "learning_rate": 7.999958845634648e-07, |
| "loss": 0.0621, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.587341772151898, |
| "grad_norm": 0.299181570614845, |
| "learning_rate": 7.841808340750478e-07, |
| "loss": 0.0576, |
| "step": 1131 |
| }, |
| { |
| "epoch": 4.591392405063291, |
| "grad_norm": 0.2886449871010561, |
| "learning_rate": 7.685205451524869e-07, |
| "loss": 0.0576, |
| "step": 1132 |
| }, |
| { |
| "epoch": 4.595443037974683, |
| "grad_norm": 0.30465085606372816, |
| "learning_rate": 7.530151439216027e-07, |
| "loss": 0.0574, |
| "step": 1133 |
| }, |
| { |
| "epoch": 4.599493670886076, |
| "grad_norm": 0.2939191714871278, |
| "learning_rate": 7.376647552607675e-07, |
| "loss": 0.0606, |
| "step": 1134 |
| }, |
| { |
| "epoch": 4.603544303797468, |
| "grad_norm": 0.30343483093737644, |
| "learning_rate": 7.224695027998963e-07, |
| "loss": 0.0625, |
| "step": 1135 |
| }, |
| { |
| "epoch": 4.6075949367088604, |
| "grad_norm": 0.3051640352374809, |
| "learning_rate": 7.07429508919466e-07, |
| "loss": 0.0658, |
| "step": 1136 |
| }, |
| { |
| "epoch": 4.611645569620253, |
| "grad_norm": 0.29680590352502567, |
| "learning_rate": 6.925448947495206e-07, |
| "loss": 0.0593, |
| "step": 1137 |
| }, |
| { |
| "epoch": 4.6156962025316455, |
| "grad_norm": 0.29489545404561, |
| "learning_rate": 6.778157801686936e-07, |
| "loss": 0.0597, |
| "step": 1138 |
| }, |
| { |
| "epoch": 4.619746835443038, |
| "grad_norm": 0.2907781695683926, |
| "learning_rate": 6.632422838032515e-07, |
| "loss": 0.0553, |
| "step": 1139 |
| }, |
| { |
| "epoch": 4.6237974683544305, |
| "grad_norm": 0.27566035569085523, |
| "learning_rate": 6.488245230261281e-07, |
| "loss": 0.0515, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.627848101265823, |
| "grad_norm": 0.2941833520844426, |
| "learning_rate": 6.345626139559868e-07, |
| "loss": 0.0621, |
| "step": 1141 |
| }, |
| { |
| "epoch": 4.6318987341772155, |
| "grad_norm": 0.3019034350628998, |
| "learning_rate": 6.204566714562866e-07, |
| "loss": 0.0616, |
| "step": 1142 |
| }, |
| { |
| "epoch": 4.635949367088608, |
| "grad_norm": 0.2996877883735102, |
| "learning_rate": 6.06506809134344e-07, |
| "loss": 0.0646, |
| "step": 1143 |
| }, |
| { |
| "epoch": 4.64, |
| "grad_norm": 0.30456348871229894, |
| "learning_rate": 5.927131393404373e-07, |
| "loss": 0.0663, |
| "step": 1144 |
| }, |
| { |
| "epoch": 4.644050632911393, |
| "grad_norm": 0.2942967816864042, |
| "learning_rate": 5.790757731668817e-07, |
| "loss": 0.0582, |
| "step": 1145 |
| }, |
| { |
| "epoch": 4.648101265822785, |
| "grad_norm": 0.30864111900566604, |
| "learning_rate": 5.655948204471507e-07, |
| "loss": 0.0646, |
| "step": 1146 |
| }, |
| { |
| "epoch": 4.652151898734177, |
| "grad_norm": 0.2921654911037382, |
| "learning_rate": 5.522703897549875e-07, |
| "loss": 0.0608, |
| "step": 1147 |
| }, |
| { |
| "epoch": 4.65620253164557, |
| "grad_norm": 0.30749586676726903, |
| "learning_rate": 5.391025884035239e-07, |
| "loss": 0.0646, |
| "step": 1148 |
| }, |
| { |
| "epoch": 4.660253164556962, |
| "grad_norm": 0.28525357395189427, |
| "learning_rate": 5.260915224444207e-07, |
| "loss": 0.0569, |
| "step": 1149 |
| }, |
| { |
| "epoch": 4.664303797468355, |
| "grad_norm": 0.29668413600995025, |
| "learning_rate": 5.132372966670129e-07, |
| "loss": 0.0561, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.668354430379747, |
| "grad_norm": 0.29322279677625634, |
| "learning_rate": 5.005400145974704e-07, |
| "loss": 0.0584, |
| "step": 1151 |
| }, |
| { |
| "epoch": 4.672405063291139, |
| "grad_norm": 0.3000193817935681, |
| "learning_rate": 4.879997784979562e-07, |
| "loss": 0.0579, |
| "step": 1152 |
| }, |
| { |
| "epoch": 4.676455696202532, |
| "grad_norm": 0.3041021717708857, |
| "learning_rate": 4.7561668936580984e-07, |
| "loss": 0.061, |
| "step": 1153 |
| }, |
| { |
| "epoch": 4.680506329113924, |
| "grad_norm": 0.29775762930979566, |
| "learning_rate": 4.6339084693272306e-07, |
| "loss": 0.0637, |
| "step": 1154 |
| }, |
| { |
| "epoch": 4.684556962025317, |
| "grad_norm": 0.3119940041978111, |
| "learning_rate": 4.5132234966395847e-07, |
| "loss": 0.062, |
| "step": 1155 |
| }, |
| { |
| "epoch": 4.688607594936709, |
| "grad_norm": 0.31181760313651424, |
| "learning_rate": 4.3941129475752795e-07, |
| "loss": 0.0627, |
| "step": 1156 |
| }, |
| { |
| "epoch": 4.692658227848101, |
| "grad_norm": 0.28984500501689303, |
| "learning_rate": 4.27657778143431e-07, |
| "loss": 0.0585, |
| "step": 1157 |
| }, |
| { |
| "epoch": 4.696708860759494, |
| "grad_norm": 0.3116795812343656, |
| "learning_rate": 4.1606189448287757e-07, |
| "loss": 0.0692, |
| "step": 1158 |
| }, |
| { |
| "epoch": 4.700759493670886, |
| "grad_norm": 0.2979249986687452, |
| "learning_rate": 4.046237371675177e-07, |
| "loss": 0.0601, |
| "step": 1159 |
| }, |
| { |
| "epoch": 4.704810126582278, |
| "grad_norm": 0.298576022168448, |
| "learning_rate": 3.9334339831869963e-07, |
| "loss": 0.06, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.708860759493671, |
| "grad_norm": 0.29485775192564545, |
| "learning_rate": 3.8222096878671955e-07, |
| "loss": 0.0623, |
| "step": 1161 |
| }, |
| { |
| "epoch": 4.712911392405063, |
| "grad_norm": 0.3052157691494499, |
| "learning_rate": 3.7125653815009545e-07, |
| "loss": 0.0605, |
| "step": 1162 |
| }, |
| { |
| "epoch": 4.716962025316455, |
| "grad_norm": 0.33148825252403236, |
| "learning_rate": 3.6045019471484974e-07, |
| "loss": 0.0685, |
| "step": 1163 |
| }, |
| { |
| "epoch": 4.721012658227848, |
| "grad_norm": 0.2942623268027151, |
| "learning_rate": 3.498020255137813e-07, |
| "loss": 0.0573, |
| "step": 1164 |
| }, |
| { |
| "epoch": 4.72506329113924, |
| "grad_norm": 0.29506167249311893, |
| "learning_rate": 3.393121163057811e-07, |
| "loss": 0.0598, |
| "step": 1165 |
| }, |
| { |
| "epoch": 4.729113924050633, |
| "grad_norm": 0.3027855869828897, |
| "learning_rate": 3.289805515751399e-07, |
| "loss": 0.0632, |
| "step": 1166 |
| }, |
| { |
| "epoch": 4.733164556962025, |
| "grad_norm": 0.3078171593102951, |
| "learning_rate": 3.188074145308573e-07, |
| "loss": 0.066, |
| "step": 1167 |
| }, |
| { |
| "epoch": 4.737215189873417, |
| "grad_norm": 0.2817894203857008, |
| "learning_rate": 3.087927871059804e-07, |
| "loss": 0.0547, |
| "step": 1168 |
| }, |
| { |
| "epoch": 4.74126582278481, |
| "grad_norm": 0.28958327366948133, |
| "learning_rate": 2.989367499569418e-07, |
| "loss": 0.0568, |
| "step": 1169 |
| }, |
| { |
| "epoch": 4.745316455696202, |
| "grad_norm": 0.29826892116115866, |
| "learning_rate": 2.8923938246290917e-07, |
| "loss": 0.0618, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.749367088607595, |
| "grad_norm": 0.3138302302400074, |
| "learning_rate": 2.7970076272514804e-07, |
| "loss": 0.07, |
| "step": 1171 |
| }, |
| { |
| "epoch": 4.7534177215189874, |
| "grad_norm": 0.29883452636584, |
| "learning_rate": 2.703209675663887e-07, |
| "loss": 0.0621, |
| "step": 1172 |
| }, |
| { |
| "epoch": 4.7574683544303795, |
| "grad_norm": 0.2925407655698031, |
| "learning_rate": 2.6110007253021374e-07, |
| "loss": 0.0545, |
| "step": 1173 |
| }, |
| { |
| "epoch": 4.7615189873417725, |
| "grad_norm": 0.29391536263182694, |
| "learning_rate": 2.520381518804471e-07, |
| "loss": 0.0559, |
| "step": 1174 |
| }, |
| { |
| "epoch": 4.7655696202531646, |
| "grad_norm": 0.3079717622625679, |
| "learning_rate": 2.4313527860054585e-07, |
| "loss": 0.0719, |
| "step": 1175 |
| }, |
| { |
| "epoch": 4.769620253164557, |
| "grad_norm": 0.29652131021808376, |
| "learning_rate": 2.343915243930317e-07, |
| "loss": 0.061, |
| "step": 1176 |
| }, |
| { |
| "epoch": 4.77367088607595, |
| "grad_norm": 0.2955517244302827, |
| "learning_rate": 2.2580695967889367e-07, |
| "loss": 0.0638, |
| "step": 1177 |
| }, |
| { |
| "epoch": 4.777721518987342, |
| "grad_norm": 0.3120836278659896, |
| "learning_rate": 2.1738165359704189e-07, |
| "loss": 0.0649, |
| "step": 1178 |
| }, |
| { |
| "epoch": 4.781772151898734, |
| "grad_norm": 0.2875429680404458, |
| "learning_rate": 2.0911567400373257e-07, |
| "loss": 0.0577, |
| "step": 1179 |
| }, |
| { |
| "epoch": 4.785822784810127, |
| "grad_norm": 0.2978376073389291, |
| "learning_rate": 2.0100908747202607e-07, |
| "loss": 0.0622, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.789873417721519, |
| "grad_norm": 0.29944650274596685, |
| "learning_rate": 1.9306195929125638e-07, |
| "loss": 0.0609, |
| "step": 1181 |
| }, |
| { |
| "epoch": 4.793924050632912, |
| "grad_norm": 0.29372886319114555, |
| "learning_rate": 1.8527435346650247e-07, |
| "loss": 0.0606, |
| "step": 1182 |
| }, |
| { |
| "epoch": 4.797974683544304, |
| "grad_norm": 0.2982277832256123, |
| "learning_rate": 1.7764633271807108e-07, |
| "loss": 0.061, |
| "step": 1183 |
| }, |
| { |
| "epoch": 4.802025316455696, |
| "grad_norm": 0.28992347019402875, |
| "learning_rate": 1.7017795848099262e-07, |
| "loss": 0.0554, |
| "step": 1184 |
| }, |
| { |
| "epoch": 4.806075949367089, |
| "grad_norm": 0.2930457012143097, |
| "learning_rate": 1.6286929090452596e-07, |
| "loss": 0.0586, |
| "step": 1185 |
| }, |
| { |
| "epoch": 4.810126582278481, |
| "grad_norm": 0.29759092207695725, |
| "learning_rate": 1.557203888516745e-07, |
| "loss": 0.0632, |
| "step": 1186 |
| }, |
| { |
| "epoch": 4.814177215189874, |
| "grad_norm": 0.2924144559360316, |
| "learning_rate": 1.487313098987131e-07, |
| "loss": 0.0602, |
| "step": 1187 |
| }, |
| { |
| "epoch": 4.818227848101266, |
| "grad_norm": 0.29935629013284576, |
| "learning_rate": 1.4190211033472402e-07, |
| "loss": 0.0608, |
| "step": 1188 |
| }, |
| { |
| "epoch": 4.822278481012658, |
| "grad_norm": 0.30509180313508694, |
| "learning_rate": 1.3523284516113955e-07, |
| "loss": 0.0632, |
| "step": 1189 |
| }, |
| { |
| "epoch": 4.826329113924051, |
| "grad_norm": 0.30158133358686817, |
| "learning_rate": 1.2872356809130682e-07, |
| "loss": 0.0594, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.830379746835443, |
| "grad_norm": 0.30585863031876065, |
| "learning_rate": 1.2237433155004807e-07, |
| "loss": 0.0649, |
| "step": 1191 |
| }, |
| { |
| "epoch": 4.834430379746835, |
| "grad_norm": 0.31122303143032004, |
| "learning_rate": 1.1618518667323886e-07, |
| "loss": 0.0638, |
| "step": 1192 |
| }, |
| { |
| "epoch": 4.838481012658228, |
| "grad_norm": 0.30503994547173485, |
| "learning_rate": 1.1015618330740385e-07, |
| "loss": 0.0669, |
| "step": 1193 |
| }, |
| { |
| "epoch": 4.84253164556962, |
| "grad_norm": 0.2932332153394253, |
| "learning_rate": 1.042873700093061e-07, |
| "loss": 0.0577, |
| "step": 1194 |
| }, |
| { |
| "epoch": 4.846582278481012, |
| "grad_norm": 0.3042256657024357, |
| "learning_rate": 9.857879404556291e-08, |
| "loss": 0.068, |
| "step": 1195 |
| }, |
| { |
| "epoch": 4.850632911392405, |
| "grad_norm": 0.30243572168670546, |
| "learning_rate": 9.303050139225722e-08, |
| "loss": 0.0671, |
| "step": 1196 |
| }, |
| { |
| "epoch": 4.854683544303797, |
| "grad_norm": 0.2919639368326357, |
| "learning_rate": 8.76425367345779e-08, |
| "loss": 0.0599, |
| "step": 1197 |
| }, |
| { |
| "epoch": 4.85873417721519, |
| "grad_norm": 0.2965550631038998, |
| "learning_rate": 8.241494346644897e-08, |
| "loss": 0.0623, |
| "step": 1198 |
| }, |
| { |
| "epoch": 4.862784810126582, |
| "grad_norm": 0.2940298331912529, |
| "learning_rate": 7.734776369019204e-08, |
| "loss": 0.062, |
| "step": 1199 |
| }, |
| { |
| "epoch": 4.866835443037974, |
| "grad_norm": 0.2988974140740747, |
| "learning_rate": 7.244103821617332e-08, |
| "loss": 0.0696, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.870886075949367, |
| "grad_norm": 0.29835993433449204, |
| "learning_rate": 6.769480656248606e-08, |
| "loss": 0.0616, |
| "step": 1201 |
| }, |
| { |
| "epoch": 4.874936708860759, |
| "grad_norm": 0.29271440657501424, |
| "learning_rate": 6.310910695462635e-08, |
| "loss": 0.0616, |
| "step": 1202 |
| }, |
| { |
| "epoch": 4.878987341772152, |
| "grad_norm": 0.2948968306966981, |
| "learning_rate": 5.8683976325191185e-08, |
| "loss": 0.0612, |
| "step": 1203 |
| }, |
| { |
| "epoch": 4.883037974683544, |
| "grad_norm": 0.3171537679420532, |
| "learning_rate": 5.4419450313571984e-08, |
| "loss": 0.0674, |
| "step": 1204 |
| }, |
| { |
| "epoch": 4.8870886075949365, |
| "grad_norm": 0.2886940401949461, |
| "learning_rate": 5.031556326567488e-08, |
| "loss": 0.0576, |
| "step": 1205 |
| }, |
| { |
| "epoch": 4.891139240506329, |
| "grad_norm": 0.28911299557882514, |
| "learning_rate": 4.637234823364312e-08, |
| "loss": 0.054, |
| "step": 1206 |
| }, |
| { |
| "epoch": 4.8951898734177215, |
| "grad_norm": 0.3072450047808017, |
| "learning_rate": 4.258983697558838e-08, |
| "loss": 0.0678, |
| "step": 1207 |
| }, |
| { |
| "epoch": 4.899240506329114, |
| "grad_norm": 0.3023730266358617, |
| "learning_rate": 3.896805995533548e-08, |
| "loss": 0.0591, |
| "step": 1208 |
| }, |
| { |
| "epoch": 4.9032911392405065, |
| "grad_norm": 0.2913979716014068, |
| "learning_rate": 3.550704634218028e-08, |
| "loss": 0.0597, |
| "step": 1209 |
| }, |
| { |
| "epoch": 4.907341772151899, |
| "grad_norm": 0.28894373144608926, |
| "learning_rate": 3.2206824010647676e-08, |
| "loss": 0.0568, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.911392405063291, |
| "grad_norm": 0.291228808796867, |
| "learning_rate": 2.9067419540278476e-08, |
| "loss": 0.0652, |
| "step": 1211 |
| }, |
| { |
| "epoch": 4.915443037974684, |
| "grad_norm": 0.30290835442927067, |
| "learning_rate": 2.6088858215400638e-08, |
| "loss": 0.0688, |
| "step": 1212 |
| }, |
| { |
| "epoch": 4.919493670886076, |
| "grad_norm": 0.31071320458062845, |
| "learning_rate": 2.3271164024940564e-08, |
| "loss": 0.0672, |
| "step": 1213 |
| }, |
| { |
| "epoch": 4.923544303797469, |
| "grad_norm": 0.29292032921796785, |
| "learning_rate": 2.061435966221881e-08, |
| "loss": 0.059, |
| "step": 1214 |
| }, |
| { |
| "epoch": 4.927594936708861, |
| "grad_norm": 0.29334719074614746, |
| "learning_rate": 1.811846652477245e-08, |
| "loss": 0.0607, |
| "step": 1215 |
| }, |
| { |
| "epoch": 4.931645569620253, |
| "grad_norm": 0.3018437066575217, |
| "learning_rate": 1.5783504714184106e-08, |
| "loss": 0.0603, |
| "step": 1216 |
| }, |
| { |
| "epoch": 4.935696202531646, |
| "grad_norm": 0.2999373305046984, |
| "learning_rate": 1.360949303591097e-08, |
| "loss": 0.0644, |
| "step": 1217 |
| }, |
| { |
| "epoch": 4.939746835443038, |
| "grad_norm": 0.2942885135818735, |
| "learning_rate": 1.1596448999144916e-08, |
| "loss": 0.0582, |
| "step": 1218 |
| }, |
| { |
| "epoch": 4.943797468354431, |
| "grad_norm": 0.29996231264156015, |
| "learning_rate": 9.744388816668172e-09, |
| "loss": 0.0629, |
| "step": 1219 |
| }, |
| { |
| "epoch": 4.947848101265823, |
| "grad_norm": 0.30103990732725827, |
| "learning_rate": 8.05332740472009e-09, |
| "loss": 0.0668, |
| "step": 1220 |
| }, |
| { |
| "epoch": 4.951898734177215, |
| "grad_norm": 0.2961951660988688, |
| "learning_rate": 6.523278382872811e-09, |
| "loss": 0.058, |
| "step": 1221 |
| }, |
| { |
| "epoch": 4.955949367088608, |
| "grad_norm": 0.2932796829654494, |
| "learning_rate": 5.15425407393133e-09, |
| "loss": 0.061, |
| "step": 1222 |
| }, |
| { |
| "epoch": 4.96, |
| "grad_norm": 0.2929700683032455, |
| "learning_rate": 3.94626550383137e-09, |
| "loss": 0.0589, |
| "step": 1223 |
| }, |
| { |
| "epoch": 4.964050632911392, |
| "grad_norm": 0.3108212626988706, |
| "learning_rate": 2.899322401546112e-09, |
| "loss": 0.0622, |
| "step": 1224 |
| }, |
| { |
| "epoch": 4.968101265822785, |
| "grad_norm": 0.3014327179239255, |
| "learning_rate": 2.013433199010706e-09, |
| "loss": 0.0576, |
| "step": 1225 |
| }, |
| { |
| "epoch": 4.972151898734177, |
| "grad_norm": 0.2919425555022161, |
| "learning_rate": 1.2886050310556563e-09, |
| "loss": 0.0619, |
| "step": 1226 |
| }, |
| { |
| "epoch": 4.976202531645569, |
| "grad_norm": 0.288586268844552, |
| "learning_rate": 7.248437353468695e-10, |
| "loss": 0.0583, |
| "step": 1227 |
| }, |
| { |
| "epoch": 4.980253164556962, |
| "grad_norm": 0.2997923219421742, |
| "learning_rate": 3.221538523412449e-10, |
| "loss": 0.0622, |
| "step": 1228 |
| }, |
| { |
| "epoch": 4.984303797468354, |
| "grad_norm": 0.29097569939136586, |
| "learning_rate": 8.053862524670663e-11, |
| "loss": 0.0571, |
| "step": 1229 |
| }, |
| { |
| "epoch": 4.988354430379747, |
| "grad_norm": 0.2862940065356984, |
| "learning_rate": 0.0, |
| "loss": 0.0582, |
| "step": 1230 |
| }, |
| { |
| "epoch": 4.988354430379747, |
| "step": 1230, |
| "total_flos": 8.253949447294157e+17, |
| "train_loss": 0.4173717436418543, |
| "train_runtime": 21703.0763, |
| "train_samples_per_second": 7.28, |
| "train_steps_per_second": 0.057 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1230, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 8.253949447294157e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
