| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 4.982278481012658, |
| "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": 6.668051706721059, |
| "learning_rate": 3.2520325203252037e-07, |
| "loss": 1.0701, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.00810126582278481, |
| "grad_norm": 6.689414643891136, |
| "learning_rate": 6.504065040650407e-07, |
| "loss": 1.0755, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.012151898734177215, |
| "grad_norm": 6.322334183060407, |
| "learning_rate": 9.75609756097561e-07, |
| "loss": 1.0379, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01620253164556962, |
| "grad_norm": 6.644170009883363, |
| "learning_rate": 1.3008130081300815e-06, |
| "loss": 1.085, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.020253164556962026, |
| "grad_norm": 6.373962108385111, |
| "learning_rate": 1.6260162601626018e-06, |
| "loss": 1.0763, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.02430379746835443, |
| "grad_norm": 6.169097632630339, |
| "learning_rate": 1.951219512195122e-06, |
| "loss": 1.0515, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.028354430379746835, |
| "grad_norm": 4.833059620265575, |
| "learning_rate": 2.2764227642276426e-06, |
| "loss": 1.0306, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.03240506329113924, |
| "grad_norm": 4.580923776473842, |
| "learning_rate": 2.601626016260163e-06, |
| "loss": 1.0013, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03645569620253165, |
| "grad_norm": 2.479957287059128, |
| "learning_rate": 2.926829268292683e-06, |
| "loss": 0.9356, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.04050632911392405, |
| "grad_norm": 2.3542959047710594, |
| "learning_rate": 3.2520325203252037e-06, |
| "loss": 0.9644, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.044556962025316456, |
| "grad_norm": 2.0475195221208233, |
| "learning_rate": 3.577235772357724e-06, |
| "loss": 0.9079, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.04860759493670886, |
| "grad_norm": 4.331637294543842, |
| "learning_rate": 3.902439024390244e-06, |
| "loss": 0.931, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.052658227848101265, |
| "grad_norm": 4.332222736988816, |
| "learning_rate": 4.227642276422765e-06, |
| "loss": 0.8986, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05670886075949367, |
| "grad_norm": 4.502820782155849, |
| "learning_rate": 4.552845528455285e-06, |
| "loss": 0.9538, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.060759493670886074, |
| "grad_norm": 4.140880031229045, |
| "learning_rate": 4.8780487804878055e-06, |
| "loss": 0.9158, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06481012658227848, |
| "grad_norm": 3.2764702591362056, |
| "learning_rate": 5.203252032520326e-06, |
| "loss": 0.8556, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06886075949367089, |
| "grad_norm": 3.1226953460329128, |
| "learning_rate": 5.528455284552846e-06, |
| "loss": 0.8801, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0729113924050633, |
| "grad_norm": 2.836101900083892, |
| "learning_rate": 5.853658536585366e-06, |
| "loss": 0.8902, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0769620253164557, |
| "grad_norm": 2.4191575894788744, |
| "learning_rate": 6.178861788617887e-06, |
| "loss": 0.8516, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.0810126582278481, |
| "grad_norm": 1.7906263840341736, |
| "learning_rate": 6.504065040650407e-06, |
| "loss": 0.8419, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08506329113924051, |
| "grad_norm": 1.6346903606749945, |
| "learning_rate": 6.829268292682928e-06, |
| "loss": 0.8359, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08911392405063291, |
| "grad_norm": 1.6988202150265592, |
| "learning_rate": 7.154471544715448e-06, |
| "loss": 0.7989, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.09316455696202532, |
| "grad_norm": 1.7901026067803971, |
| "learning_rate": 7.4796747967479676e-06, |
| "loss": 0.7842, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09721518987341772, |
| "grad_norm": 1.6494186976102463, |
| "learning_rate": 7.804878048780489e-06, |
| "loss": 0.7761, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.10126582278481013, |
| "grad_norm": 1.4966926374887763, |
| "learning_rate": 8.130081300813009e-06, |
| "loss": 0.8039, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10531645569620253, |
| "grad_norm": 1.3389294664598348, |
| "learning_rate": 8.45528455284553e-06, |
| "loss": 0.793, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10936708860759493, |
| "grad_norm": 1.4031966330629235, |
| "learning_rate": 8.78048780487805e-06, |
| "loss": 0.7745, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.11341772151898734, |
| "grad_norm": 1.214651673853607, |
| "learning_rate": 9.10569105691057e-06, |
| "loss": 0.7644, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11746835443037974, |
| "grad_norm": 0.9861716503856578, |
| "learning_rate": 9.43089430894309e-06, |
| "loss": 0.7363, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.12151898734177215, |
| "grad_norm": 1.015858958206504, |
| "learning_rate": 9.756097560975611e-06, |
| "loss": 0.7574, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12556962025316457, |
| "grad_norm": 0.9912726165004364, |
| "learning_rate": 1.008130081300813e-05, |
| "loss": 0.7394, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12962025316455697, |
| "grad_norm": 0.8741942068892673, |
| "learning_rate": 1.0406504065040652e-05, |
| "loss": 0.7308, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.13367088607594937, |
| "grad_norm": 0.7147974309748173, |
| "learning_rate": 1.0731707317073172e-05, |
| "loss": 0.765, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13772151898734178, |
| "grad_norm": 0.8717549481055699, |
| "learning_rate": 1.1056910569105692e-05, |
| "loss": 0.7362, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.14177215189873418, |
| "grad_norm": 0.7592235231124019, |
| "learning_rate": 1.1382113821138213e-05, |
| "loss": 0.7528, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.1458227848101266, |
| "grad_norm": 0.7895299778483779, |
| "learning_rate": 1.1707317073170731e-05, |
| "loss": 0.7337, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.149873417721519, |
| "grad_norm": 0.6926525434340636, |
| "learning_rate": 1.2032520325203254e-05, |
| "loss": 0.7274, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.1539240506329114, |
| "grad_norm": 0.7870944336757759, |
| "learning_rate": 1.2357723577235774e-05, |
| "loss": 0.7357, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.1579746835443038, |
| "grad_norm": 0.6877687942577028, |
| "learning_rate": 1.2682926829268294e-05, |
| "loss": 0.7322, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.1620253164556962, |
| "grad_norm": 0.6033005172900113, |
| "learning_rate": 1.3008130081300815e-05, |
| "loss": 0.7217, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1660759493670886, |
| "grad_norm": 0.7352285967106955, |
| "learning_rate": 1.3333333333333333e-05, |
| "loss": 0.7241, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.17012658227848101, |
| "grad_norm": 0.6514873865931173, |
| "learning_rate": 1.3658536585365855e-05, |
| "loss": 0.7171, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.17417721518987342, |
| "grad_norm": 0.5359939032280855, |
| "learning_rate": 1.3983739837398376e-05, |
| "loss": 0.7035, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17822784810126582, |
| "grad_norm": 0.5669464071195061, |
| "learning_rate": 1.4308943089430896e-05, |
| "loss": 0.6804, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.18227848101265823, |
| "grad_norm": 0.6058064317610408, |
| "learning_rate": 1.4634146341463415e-05, |
| "loss": 0.6909, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.18632911392405063, |
| "grad_norm": 0.627047434201408, |
| "learning_rate": 1.4959349593495935e-05, |
| "loss": 0.6988, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.19037974683544304, |
| "grad_norm": 0.6028409281555291, |
| "learning_rate": 1.528455284552846e-05, |
| "loss": 0.6914, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.19443037974683544, |
| "grad_norm": 0.7279297212497261, |
| "learning_rate": 1.5609756097560978e-05, |
| "loss": 0.6932, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19848101265822785, |
| "grad_norm": 0.744370071678772, |
| "learning_rate": 1.5934959349593496e-05, |
| "loss": 0.7017, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.20253164556962025, |
| "grad_norm": 0.5997581768040292, |
| "learning_rate": 1.6260162601626018e-05, |
| "loss": 0.7132, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.20658227848101265, |
| "grad_norm": 0.7872872287161451, |
| "learning_rate": 1.6585365853658537e-05, |
| "loss": 0.6998, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.21063291139240506, |
| "grad_norm": 0.6511428491905927, |
| "learning_rate": 1.691056910569106e-05, |
| "loss": 0.676, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.21468354430379746, |
| "grad_norm": 0.6279292094939638, |
| "learning_rate": 1.7235772357723578e-05, |
| "loss": 0.714, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21873417721518987, |
| "grad_norm": 0.6538421422585304, |
| "learning_rate": 1.75609756097561e-05, |
| "loss": 0.6646, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.22278481012658227, |
| "grad_norm": 0.7237226448957533, |
| "learning_rate": 1.788617886178862e-05, |
| "loss": 0.6818, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.22683544303797468, |
| "grad_norm": 0.839760568433043, |
| "learning_rate": 1.821138211382114e-05, |
| "loss": 0.6858, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.23088607594936708, |
| "grad_norm": 0.7908326255204029, |
| "learning_rate": 1.8536585365853663e-05, |
| "loss": 0.6909, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.23493670886075949, |
| "grad_norm": 0.6618999655981399, |
| "learning_rate": 1.886178861788618e-05, |
| "loss": 0.6877, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.2389873417721519, |
| "grad_norm": 0.7581397195780165, |
| "learning_rate": 1.91869918699187e-05, |
| "loss": 0.6735, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.2430379746835443, |
| "grad_norm": 0.5923914573587943, |
| "learning_rate": 1.9512195121951222e-05, |
| "loss": 0.6821, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2470886075949367, |
| "grad_norm": 0.813459878053388, |
| "learning_rate": 1.983739837398374e-05, |
| "loss": 0.6539, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.25113924050632913, |
| "grad_norm": 0.7131976081753709, |
| "learning_rate": 2.016260162601626e-05, |
| "loss": 0.6977, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.25518987341772154, |
| "grad_norm": 0.7008515722565707, |
| "learning_rate": 2.048780487804878e-05, |
| "loss": 0.6678, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.25924050632911394, |
| "grad_norm": 0.6585822411937272, |
| "learning_rate": 2.0813008130081303e-05, |
| "loss": 0.6381, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.26329113924050634, |
| "grad_norm": 0.648895351523815, |
| "learning_rate": 2.1138211382113822e-05, |
| "loss": 0.6727, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.26734177215189875, |
| "grad_norm": 0.6613222690856796, |
| "learning_rate": 2.1463414634146344e-05, |
| "loss": 0.7051, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.27139240506329115, |
| "grad_norm": 0.7211532184644401, |
| "learning_rate": 2.1788617886178863e-05, |
| "loss": 0.6913, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.27544303797468356, |
| "grad_norm": 0.7084159667911435, |
| "learning_rate": 2.2113821138211385e-05, |
| "loss": 0.6829, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.27949367088607596, |
| "grad_norm": 0.6599499941894094, |
| "learning_rate": 2.2439024390243907e-05, |
| "loss": 0.6771, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.28354430379746837, |
| "grad_norm": 0.6376326528130011, |
| "learning_rate": 2.2764227642276426e-05, |
| "loss": 0.6662, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.28759493670886077, |
| "grad_norm": 0.6711554147500624, |
| "learning_rate": 2.3089430894308948e-05, |
| "loss": 0.6837, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.2916455696202532, |
| "grad_norm": 0.6111430952668357, |
| "learning_rate": 2.3414634146341463e-05, |
| "loss": 0.6544, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.2956962025316456, |
| "grad_norm": 0.7822627037186903, |
| "learning_rate": 2.3739837398373985e-05, |
| "loss": 0.6624, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.299746835443038, |
| "grad_norm": 0.6419459168829424, |
| "learning_rate": 2.4065040650406507e-05, |
| "loss": 0.6766, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.3037974683544304, |
| "grad_norm": 0.7478388389763116, |
| "learning_rate": 2.4390243902439026e-05, |
| "loss": 0.677, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3078481012658228, |
| "grad_norm": 0.756438401190648, |
| "learning_rate": 2.4715447154471548e-05, |
| "loss": 0.6507, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.3118987341772152, |
| "grad_norm": 0.7804711685361412, |
| "learning_rate": 2.5040650406504066e-05, |
| "loss": 0.6843, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.3159493670886076, |
| "grad_norm": 0.7835555162913608, |
| "learning_rate": 2.536585365853659e-05, |
| "loss": 0.6679, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 0.8276855269607756, |
| "learning_rate": 2.569105691056911e-05, |
| "loss": 0.6673, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.3240506329113924, |
| "grad_norm": 0.7369734560549824, |
| "learning_rate": 2.601626016260163e-05, |
| "loss": 0.6406, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3281012658227848, |
| "grad_norm": 1.0274220486406112, |
| "learning_rate": 2.634146341463415e-05, |
| "loss": 0.6737, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3321518987341772, |
| "grad_norm": 0.9194283633650498, |
| "learning_rate": 2.6666666666666667e-05, |
| "loss": 0.6722, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3362025316455696, |
| "grad_norm": 1.0064759352949426, |
| "learning_rate": 2.699186991869919e-05, |
| "loss": 0.6646, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.34025316455696203, |
| "grad_norm": 0.6917258177533991, |
| "learning_rate": 2.731707317073171e-05, |
| "loss": 0.6307, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.34430379746835443, |
| "grad_norm": 0.7873141922284292, |
| "learning_rate": 2.764227642276423e-05, |
| "loss": 0.6634, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.34835443037974684, |
| "grad_norm": 0.8100465528390076, |
| "learning_rate": 2.796747967479675e-05, |
| "loss": 0.6574, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.35240506329113924, |
| "grad_norm": 0.8615807226520026, |
| "learning_rate": 2.829268292682927e-05, |
| "loss": 0.642, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.35645569620253165, |
| "grad_norm": 0.7199907442181335, |
| "learning_rate": 2.8617886178861792e-05, |
| "loss": 0.6776, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.36050632911392405, |
| "grad_norm": 0.8725401140872308, |
| "learning_rate": 2.8943089430894314e-05, |
| "loss": 0.6518, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.36455696202531646, |
| "grad_norm": 0.8783461156652478, |
| "learning_rate": 2.926829268292683e-05, |
| "loss": 0.6434, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.36860759493670886, |
| "grad_norm": 0.76059086571759, |
| "learning_rate": 2.959349593495935e-05, |
| "loss": 0.6399, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.37265822784810126, |
| "grad_norm": 0.7843162660063397, |
| "learning_rate": 2.991869918699187e-05, |
| "loss": 0.6949, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.37670886075949367, |
| "grad_norm": 1.0149978782485007, |
| "learning_rate": 3.0243902439024392e-05, |
| "loss": 0.6253, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3807594936708861, |
| "grad_norm": 0.9742101190158011, |
| "learning_rate": 3.056910569105692e-05, |
| "loss": 0.6357, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.3848101265822785, |
| "grad_norm": 0.6964391864344169, |
| "learning_rate": 3.089430894308943e-05, |
| "loss": 0.6568, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.3888607594936709, |
| "grad_norm": 0.8143182709934148, |
| "learning_rate": 3.1219512195121955e-05, |
| "loss": 0.655, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.3929113924050633, |
| "grad_norm": 0.9159124471919045, |
| "learning_rate": 3.154471544715447e-05, |
| "loss": 0.6821, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.3969620253164557, |
| "grad_norm": 0.9222684191406817, |
| "learning_rate": 3.186991869918699e-05, |
| "loss": 0.6199, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.4010126582278481, |
| "grad_norm": 0.7630251170738187, |
| "learning_rate": 3.2195121951219514e-05, |
| "loss": 0.6511, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.4050632911392405, |
| "grad_norm": 0.8628086651692592, |
| "learning_rate": 3.2520325203252037e-05, |
| "loss": 0.6689, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4091139240506329, |
| "grad_norm": 0.8549069006495394, |
| "learning_rate": 3.284552845528456e-05, |
| "loss": 0.6663, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.4131645569620253, |
| "grad_norm": 0.7784735415797633, |
| "learning_rate": 3.3170731707317074e-05, |
| "loss": 0.6124, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.4172151898734177, |
| "grad_norm": 0.6450225681379378, |
| "learning_rate": 3.3495934959349596e-05, |
| "loss": 0.65, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.4212658227848101, |
| "grad_norm": 0.7487402786267876, |
| "learning_rate": 3.382113821138212e-05, |
| "loss": 0.6716, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4253164556962025, |
| "grad_norm": 0.6826236598431946, |
| "learning_rate": 3.414634146341463e-05, |
| "loss": 0.6253, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4293670886075949, |
| "grad_norm": 0.9933498979975959, |
| "learning_rate": 3.4471544715447155e-05, |
| "loss": 0.6466, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.43341772151898733, |
| "grad_norm": 1.0040527361179448, |
| "learning_rate": 3.479674796747968e-05, |
| "loss": 0.6419, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.43746835443037974, |
| "grad_norm": 0.7058133464555062, |
| "learning_rate": 3.51219512195122e-05, |
| "loss": 0.6324, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.44151898734177214, |
| "grad_norm": 0.7758593728696657, |
| "learning_rate": 3.544715447154472e-05, |
| "loss": 0.6277, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.44556962025316454, |
| "grad_norm": 0.8994332900935913, |
| "learning_rate": 3.577235772357724e-05, |
| "loss": 0.6247, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.44962025316455695, |
| "grad_norm": 1.058795547128779, |
| "learning_rate": 3.609756097560976e-05, |
| "loss": 0.6457, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.45367088607594935, |
| "grad_norm": 0.7455787720999872, |
| "learning_rate": 3.642276422764228e-05, |
| "loss": 0.6441, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.45772151898734176, |
| "grad_norm": 0.7746907714147548, |
| "learning_rate": 3.67479674796748e-05, |
| "loss": 0.6367, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.46177215189873416, |
| "grad_norm": 0.871687322178005, |
| "learning_rate": 3.7073170731707325e-05, |
| "loss": 0.6205, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.46582278481012657, |
| "grad_norm": 0.8854277655325361, |
| "learning_rate": 3.739837398373984e-05, |
| "loss": 0.6531, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.46987341772151897, |
| "grad_norm": 0.9323049162057678, |
| "learning_rate": 3.772357723577236e-05, |
| "loss": 0.644, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.4739240506329114, |
| "grad_norm": 0.8310176490540838, |
| "learning_rate": 3.804878048780488e-05, |
| "loss": 0.6376, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.4779746835443038, |
| "grad_norm": 0.7988957813672862, |
| "learning_rate": 3.83739837398374e-05, |
| "loss": 0.6253, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.4820253164556962, |
| "grad_norm": 1.0125488879304079, |
| "learning_rate": 3.869918699186992e-05, |
| "loss": 0.6253, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.4860759493670886, |
| "grad_norm": 0.7805018670241181, |
| "learning_rate": 3.9024390243902444e-05, |
| "loss": 0.622, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.490126582278481, |
| "grad_norm": 1.3134266485475705, |
| "learning_rate": 3.9349593495934966e-05, |
| "loss": 0.6305, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.4941772151898734, |
| "grad_norm": 0.7790797262556867, |
| "learning_rate": 3.967479674796748e-05, |
| "loss": 0.6199, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4982278481012658, |
| "grad_norm": 1.0679217361725162, |
| "learning_rate": 4e-05, |
| "loss": 0.65, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.5022784810126583, |
| "grad_norm": 1.0154235813084365, |
| "learning_rate": 3.999991946137476e-05, |
| "loss": 0.6159, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.5063291139240507, |
| "grad_norm": 1.0054164620628074, |
| "learning_rate": 3.999967784614766e-05, |
| "loss": 0.6394, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5103797468354431, |
| "grad_norm": 1.0196475225646506, |
| "learning_rate": 3.9999275156264656e-05, |
| "loss": 0.6582, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.5144303797468355, |
| "grad_norm": 0.8562748877695633, |
| "learning_rate": 3.999871139496895e-05, |
| "loss": 0.6459, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.5184810126582279, |
| "grad_norm": 0.9278062987082396, |
| "learning_rate": 3.9997986566800995e-05, |
| "loss": 0.6344, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.5225316455696203, |
| "grad_norm": 1.1962682837644425, |
| "learning_rate": 3.999710067759846e-05, |
| "loss": 0.6312, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.5265822784810127, |
| "grad_norm": 0.8191454119707047, |
| "learning_rate": 3.999605373449617e-05, |
| "loss": 0.6273, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5306329113924051, |
| "grad_norm": 1.2014063890796995, |
| "learning_rate": 3.9994845745926075e-05, |
| "loss": 0.6266, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.5346835443037975, |
| "grad_norm": 0.6904775656409834, |
| "learning_rate": 3.999347672161713e-05, |
| "loss": 0.6421, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5387341772151899, |
| "grad_norm": 1.3054193231050624, |
| "learning_rate": 3.999194667259528e-05, |
| "loss": 0.6286, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5427848101265823, |
| "grad_norm": 0.854846820013133, |
| "learning_rate": 3.999025561118334e-05, |
| "loss": 0.623, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5468354430379747, |
| "grad_norm": 1.0052494099501255, |
| "learning_rate": 3.998840355100086e-05, |
| "loss": 0.6444, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5508860759493671, |
| "grad_norm": 1.009574030465157, |
| "learning_rate": 3.998639050696409e-05, |
| "loss": 0.6364, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5549367088607595, |
| "grad_norm": 0.734218478301677, |
| "learning_rate": 3.998421649528582e-05, |
| "loss": 0.6071, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5589873417721519, |
| "grad_norm": 0.8900782307462244, |
| "learning_rate": 3.9981881533475234e-05, |
| "loss": 0.6355, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.5630379746835443, |
| "grad_norm": 0.7227551633517688, |
| "learning_rate": 3.997938564033779e-05, |
| "loss": 0.6368, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5670886075949367, |
| "grad_norm": 0.6497388672143563, |
| "learning_rate": 3.9976728835975064e-05, |
| "loss": 0.6238, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5711392405063291, |
| "grad_norm": 0.7381768938985102, |
| "learning_rate": 3.9973911141784605e-05, |
| "loss": 0.6095, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.5751898734177215, |
| "grad_norm": 0.7577497252183484, |
| "learning_rate": 3.997093258045973e-05, |
| "loss": 0.6241, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.579240506329114, |
| "grad_norm": 0.6224221747651401, |
| "learning_rate": 3.996779317598936e-05, |
| "loss": 0.6522, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.5832911392405064, |
| "grad_norm": 0.7329422134265952, |
| "learning_rate": 3.996449295365782e-05, |
| "loss": 0.6311, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5873417721518988, |
| "grad_norm": 0.7438904322933009, |
| "learning_rate": 3.996103194004467e-05, |
| "loss": 0.6458, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5913924050632912, |
| "grad_norm": 0.6634205960181782, |
| "learning_rate": 3.995741016302441e-05, |
| "loss": 0.6515, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5954430379746836, |
| "grad_norm": 0.6511990960248553, |
| "learning_rate": 3.9953627651766364e-05, |
| "loss": 0.6204, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.599493670886076, |
| "grad_norm": 0.6550210052624396, |
| "learning_rate": 3.9949684436734325e-05, |
| "loss": 0.6127, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.6035443037974684, |
| "grad_norm": 0.5754424472097128, |
| "learning_rate": 3.994558054968643e-05, |
| "loss": 0.6291, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.6075949367088608, |
| "grad_norm": 0.7982493199839692, |
| "learning_rate": 3.994131602367481e-05, |
| "loss": 0.6434, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6116455696202532, |
| "grad_norm": 0.6716783916125822, |
| "learning_rate": 3.9936890893045376e-05, |
| "loss": 0.6309, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.6156962025316456, |
| "grad_norm": 0.594157383697954, |
| "learning_rate": 3.993230519343752e-05, |
| "loss": 0.6139, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.619746835443038, |
| "grad_norm": 0.6938494201482996, |
| "learning_rate": 3.992755896178383e-05, |
| "loss": 0.6041, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.6237974683544304, |
| "grad_norm": 0.6025315637552626, |
| "learning_rate": 3.992265223630981e-05, |
| "loss": 0.6142, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.6278481012658228, |
| "grad_norm": 0.68510128294434, |
| "learning_rate": 3.991758505653355e-05, |
| "loss": 0.61, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.6318987341772152, |
| "grad_norm": 0.6195304825824063, |
| "learning_rate": 3.991235746326543e-05, |
| "loss": 0.6019, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6359493670886076, |
| "grad_norm": 0.7627697435925584, |
| "learning_rate": 3.9906969498607745e-05, |
| "loss": 0.6229, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 0.713075066194833, |
| "learning_rate": 3.990142120595444e-05, |
| "loss": 0.6081, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.6440506329113924, |
| "grad_norm": 0.6746619531692846, |
| "learning_rate": 3.98957126299907e-05, |
| "loss": 0.6168, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6481012658227848, |
| "grad_norm": 0.742033575936647, |
| "learning_rate": 3.9889843816692596e-05, |
| "loss": 0.6102, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6521518987341772, |
| "grad_norm": 0.6790971096981906, |
| "learning_rate": 3.9883814813326766e-05, |
| "loss": 0.629, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6562025316455696, |
| "grad_norm": 0.7320138310589233, |
| "learning_rate": 3.9877625668449956e-05, |
| "loss": 0.6243, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.660253164556962, |
| "grad_norm": 0.6950259924973899, |
| "learning_rate": 3.98712764319087e-05, |
| "loss": 0.6305, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.6643037974683544, |
| "grad_norm": 0.8809755184806991, |
| "learning_rate": 3.9864767154838864e-05, |
| "loss": 0.5996, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6683544303797468, |
| "grad_norm": 0.6716252280556891, |
| "learning_rate": 3.9858097889665277e-05, |
| "loss": 0.6132, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6724050632911392, |
| "grad_norm": 0.589620948726161, |
| "learning_rate": 3.985126869010129e-05, |
| "loss": 0.6101, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6764556962025317, |
| "grad_norm": 0.5948124824258744, |
| "learning_rate": 3.984427961114833e-05, |
| "loss": 0.6341, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.6805063291139241, |
| "grad_norm": 0.539160344815398, |
| "learning_rate": 3.9837130709095475e-05, |
| "loss": 0.5962, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.6845569620253165, |
| "grad_norm": 0.5962862010789327, |
| "learning_rate": 3.982982204151901e-05, |
| "loss": 0.607, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6886075949367089, |
| "grad_norm": 0.6040285238037283, |
| "learning_rate": 3.982235366728193e-05, |
| "loss": 0.622, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6926582278481013, |
| "grad_norm": 0.5079622310341705, |
| "learning_rate": 3.9814725646533505e-05, |
| "loss": 0.6091, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6967088607594937, |
| "grad_norm": 0.6309500615100253, |
| "learning_rate": 3.9806938040708746e-05, |
| "loss": 0.6215, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.7007594936708861, |
| "grad_norm": 0.53210419854968, |
| "learning_rate": 3.9798990912527976e-05, |
| "loss": 0.6369, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.7048101265822785, |
| "grad_norm": 0.6198896944535545, |
| "learning_rate": 3.979088432599627e-05, |
| "loss": 0.6252, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.7088607594936709, |
| "grad_norm": 0.49519635473425333, |
| "learning_rate": 3.9782618346402964e-05, |
| "loss": 0.6052, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7129113924050633, |
| "grad_norm": 0.6692380901514067, |
| "learning_rate": 3.977419304032111e-05, |
| "loss": 0.5969, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.7169620253164557, |
| "grad_norm": 0.5742338583960168, |
| "learning_rate": 3.976560847560697e-05, |
| "loss": 0.5987, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.7210126582278481, |
| "grad_norm": 0.6183238527387681, |
| "learning_rate": 3.9756864721399456e-05, |
| "loss": 0.6055, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.7250632911392405, |
| "grad_norm": 0.6393582838914289, |
| "learning_rate": 3.974796184811956e-05, |
| "loss": 0.6054, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.7291139240506329, |
| "grad_norm": 0.6573417917508327, |
| "learning_rate": 3.973889992746979e-05, |
| "loss": 0.614, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7331645569620253, |
| "grad_norm": 0.5621430593046283, |
| "learning_rate": 3.972967903243361e-05, |
| "loss": 0.5914, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.7372151898734177, |
| "grad_norm": 0.624586311657193, |
| "learning_rate": 3.972029923727486e-05, |
| "loss": 0.6111, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7412658227848101, |
| "grad_norm": 0.5882101520249411, |
| "learning_rate": 3.971076061753709e-05, |
| "loss": 0.6041, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7453164556962025, |
| "grad_norm": 0.5380162519834443, |
| "learning_rate": 3.9701063250043066e-05, |
| "loss": 0.5938, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7493670886075949, |
| "grad_norm": 0.6932444749861357, |
| "learning_rate": 3.969120721289402e-05, |
| "loss": 0.5786, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.7534177215189873, |
| "grad_norm": 0.5069918081830059, |
| "learning_rate": 3.9681192585469146e-05, |
| "loss": 0.5745, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.7574683544303797, |
| "grad_norm": 0.5726769864861685, |
| "learning_rate": 3.9671019448424865e-05, |
| "loss": 0.6163, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7615189873417721, |
| "grad_norm": 0.5758144868199683, |
| "learning_rate": 3.966068788369422e-05, |
| "loss": 0.6261, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7655696202531646, |
| "grad_norm": 0.5656820887723754, |
| "learning_rate": 3.965019797448622e-05, |
| "loss": 0.6083, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.769620253164557, |
| "grad_norm": 0.5335593944991152, |
| "learning_rate": 3.963954980528515e-05, |
| "loss": 0.605, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.7736708860759494, |
| "grad_norm": 0.5760999413415605, |
| "learning_rate": 3.9628743461849905e-05, |
| "loss": 0.5953, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7777215189873418, |
| "grad_norm": 0.5345244932531948, |
| "learning_rate": 3.961777903121329e-05, |
| "loss": 0.5917, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7817721518987342, |
| "grad_norm": 0.5549047486144421, |
| "learning_rate": 3.960665660168131e-05, |
| "loss": 0.5876, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7858227848101266, |
| "grad_norm": 0.5204984320084429, |
| "learning_rate": 3.9595376262832485e-05, |
| "loss": 0.606, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.789873417721519, |
| "grad_norm": 0.539997057418847, |
| "learning_rate": 3.9583938105517127e-05, |
| "loss": 0.5764, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7939240506329114, |
| "grad_norm": 0.48482825486793957, |
| "learning_rate": 3.957234222185657e-05, |
| "loss": 0.604, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7979746835443038, |
| "grad_norm": 0.4969537590404889, |
| "learning_rate": 3.9560588705242474e-05, |
| "loss": 0.591, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.8020253164556962, |
| "grad_norm": 0.4966990859225105, |
| "learning_rate": 3.954867765033605e-05, |
| "loss": 0.5849, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.8060759493670886, |
| "grad_norm": 0.5953234908564741, |
| "learning_rate": 3.953660915306728e-05, |
| "loss": 0.6136, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.810126582278481, |
| "grad_norm": 0.5385711217008915, |
| "learning_rate": 3.952438331063419e-05, |
| "loss": 0.606, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8141772151898734, |
| "grad_norm": 0.5692702877918502, |
| "learning_rate": 3.951200022150205e-05, |
| "loss": 0.595, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.8182278481012658, |
| "grad_norm": 0.5143319200524892, |
| "learning_rate": 3.949945998540253e-05, |
| "loss": 0.6016, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.8222784810126582, |
| "grad_norm": 0.4916613365595116, |
| "learning_rate": 3.9486762703332993e-05, |
| "loss": 0.5564, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.8263291139240506, |
| "grad_norm": 0.6618302240086578, |
| "learning_rate": 3.947390847755559e-05, |
| "loss": 0.6105, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.830379746835443, |
| "grad_norm": 0.6610147907357069, |
| "learning_rate": 3.946089741159648e-05, |
| "loss": 0.6066, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8344303797468354, |
| "grad_norm": 0.5364235198415442, |
| "learning_rate": 3.944772961024501e-05, |
| "loss": 0.6117, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.8384810126582278, |
| "grad_norm": 0.7704396446161474, |
| "learning_rate": 3.943440517955285e-05, |
| "loss": 0.5972, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.8425316455696202, |
| "grad_norm": 0.5178575016804979, |
| "learning_rate": 3.9420924226833126e-05, |
| "loss": 0.5602, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8465822784810126, |
| "grad_norm": 0.7620792823119523, |
| "learning_rate": 3.9407286860659566e-05, |
| "loss": 0.582, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.850632911392405, |
| "grad_norm": 0.5225736424451888, |
| "learning_rate": 3.9393493190865657e-05, |
| "loss": 0.5972, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8546835443037974, |
| "grad_norm": 0.5487996127984076, |
| "learning_rate": 3.937954332854371e-05, |
| "loss": 0.6073, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8587341772151899, |
| "grad_norm": 0.5454812422991182, |
| "learning_rate": 3.9365437386044016e-05, |
| "loss": 0.5697, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.8627848101265823, |
| "grad_norm": 0.5097934305719052, |
| "learning_rate": 3.935117547697387e-05, |
| "loss": 0.5893, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8668354430379747, |
| "grad_norm": 0.6672091751717397, |
| "learning_rate": 3.933675771619675e-05, |
| "loss": 0.5675, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8708860759493671, |
| "grad_norm": 0.5622426368462822, |
| "learning_rate": 3.932218421983131e-05, |
| "loss": 0.6011, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8749367088607595, |
| "grad_norm": 0.596396187969425, |
| "learning_rate": 3.9307455105250484e-05, |
| "loss": 0.5853, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8789873417721519, |
| "grad_norm": 0.5575403862062193, |
| "learning_rate": 3.929257049108054e-05, |
| "loss": 0.6073, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8830379746835443, |
| "grad_norm": 0.6548601135978452, |
| "learning_rate": 3.927753049720011e-05, |
| "loss": 0.5946, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8870886075949367, |
| "grad_norm": 0.5512370975168387, |
| "learning_rate": 3.9262335244739234e-05, |
| "loss": 0.6018, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8911392405063291, |
| "grad_norm": 0.7242665401753656, |
| "learning_rate": 3.92469848560784e-05, |
| "loss": 0.5938, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8951898734177215, |
| "grad_norm": 0.5514199295253259, |
| "learning_rate": 3.923147945484751e-05, |
| "loss": 0.5865, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8992405063291139, |
| "grad_norm": 0.6993052521129165, |
| "learning_rate": 3.9215819165924956e-05, |
| "loss": 0.5785, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.9032911392405063, |
| "grad_norm": 0.6118919730113623, |
| "learning_rate": 3.920000411543654e-05, |
| "loss": 0.5968, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.9073417721518987, |
| "grad_norm": 0.7788437413379335, |
| "learning_rate": 3.9184034430754495e-05, |
| "loss": 0.584, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.9113924050632911, |
| "grad_norm": 0.6242825753083213, |
| "learning_rate": 3.916791024049648e-05, |
| "loss": 0.6116, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9154430379746835, |
| "grad_norm": 0.8670344794031524, |
| "learning_rate": 3.91516316745245e-05, |
| "loss": 0.6118, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.9194936708860759, |
| "grad_norm": 0.642493728403803, |
| "learning_rate": 3.913519886394389e-05, |
| "loss": 0.5712, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.9235443037974683, |
| "grad_norm": 0.5880870769467297, |
| "learning_rate": 3.911861194110225e-05, |
| "loss": 0.615, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.9275949367088607, |
| "grad_norm": 0.6436297389639383, |
| "learning_rate": 3.910187103958837e-05, |
| "loss": 0.5889, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.9316455696202531, |
| "grad_norm": 0.5415703898305838, |
| "learning_rate": 3.908497629423117e-05, |
| "loss": 0.5832, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9356962025316455, |
| "grad_norm": 0.5930123614114315, |
| "learning_rate": 3.9067927841098614e-05, |
| "loss": 0.6093, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.9397468354430379, |
| "grad_norm": 0.5878088688188932, |
| "learning_rate": 3.9050725817496594e-05, |
| "loss": 0.6046, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.9437974683544303, |
| "grad_norm": 0.5571404937710325, |
| "learning_rate": 3.9033370361967844e-05, |
| "loss": 0.6161, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.9478481012658228, |
| "grad_norm": 0.5732756877915007, |
| "learning_rate": 3.901586161429081e-05, |
| "loss": 0.585, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.9518987341772152, |
| "grad_norm": 0.6844584304326632, |
| "learning_rate": 3.8998199715478545e-05, |
| "loss": 0.5932, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9559493670886076, |
| "grad_norm": 0.4621184864975837, |
| "learning_rate": 3.8980384807777564e-05, |
| "loss": 0.5714, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 0.8164112339993872, |
| "learning_rate": 3.896241703466667e-05, |
| "loss": 0.5895, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9640506329113924, |
| "grad_norm": 0.5299028122621412, |
| "learning_rate": 3.894429654085585e-05, |
| "loss": 0.57, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9681012658227848, |
| "grad_norm": 0.6437516663215663, |
| "learning_rate": 3.892602347228505e-05, |
| "loss": 0.6014, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.9721518987341772, |
| "grad_norm": 0.5901953392779695, |
| "learning_rate": 3.890759797612307e-05, |
| "loss": 0.5976, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9762025316455696, |
| "grad_norm": 0.6132831511227061, |
| "learning_rate": 3.888902020076632e-05, |
| "loss": 0.5984, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.980253164556962, |
| "grad_norm": 0.5958105389564933, |
| "learning_rate": 3.887029029583764e-05, |
| "loss": 0.5591, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9843037974683544, |
| "grad_norm": 0.6236128793376718, |
| "learning_rate": 3.8851408412185125e-05, |
| "loss": 0.5744, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9883544303797468, |
| "grad_norm": 0.5725851134464292, |
| "learning_rate": 3.8832374701880855e-05, |
| "loss": 0.5867, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9924050632911392, |
| "grad_norm": 0.5411041404657524, |
| "learning_rate": 3.881318931821972e-05, |
| "loss": 0.5731, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9964556962025316, |
| "grad_norm": 0.5848361971143093, |
| "learning_rate": 3.879385241571817e-05, |
| "loss": 0.5825, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.000506329113924, |
| "grad_norm": 0.5037709661614345, |
| "learning_rate": 3.8774364150112955e-05, |
| "loss": 0.5546, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.0045569620253165, |
| "grad_norm": 0.6354879477425647, |
| "learning_rate": 3.8754724678359884e-05, |
| "loss": 0.4947, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.0086075949367088, |
| "grad_norm": 0.5064326998220365, |
| "learning_rate": 3.873493415863256e-05, |
| "loss": 0.4985, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.0126582278481013, |
| "grad_norm": 0.6523989871248618, |
| "learning_rate": 3.871499275032111e-05, |
| "loss": 0.4945, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0167088607594936, |
| "grad_norm": 0.5255633531300445, |
| "learning_rate": 3.869490061403091e-05, |
| "loss": 0.4655, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.0207594936708861, |
| "grad_norm": 0.6304192991892732, |
| "learning_rate": 3.867465791158124e-05, |
| "loss": 0.4653, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.0248101265822784, |
| "grad_norm": 0.5727744872193783, |
| "learning_rate": 3.865426480600407e-05, |
| "loss": 0.4776, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.028860759493671, |
| "grad_norm": 0.6461914145454478, |
| "learning_rate": 3.863372146154264e-05, |
| "loss": 0.4776, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.0329113924050632, |
| "grad_norm": 0.5543247748223313, |
| "learning_rate": 3.861302804365024e-05, |
| "loss": 0.5034, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.0369620253164558, |
| "grad_norm": 0.5768583222121332, |
| "learning_rate": 3.85921847189888e-05, |
| "loss": 0.4858, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.041012658227848, |
| "grad_norm": 0.5414556696029865, |
| "learning_rate": 3.85711916554276e-05, |
| "loss": 0.4776, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.0450632911392406, |
| "grad_norm": 0.551066635290613, |
| "learning_rate": 3.85500490220419e-05, |
| "loss": 0.4951, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.0491139240506329, |
| "grad_norm": 0.5489962848076892, |
| "learning_rate": 3.852875698911154e-05, |
| "loss": 0.4844, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.0531645569620254, |
| "grad_norm": 0.6126062537253347, |
| "learning_rate": 3.850731572811963e-05, |
| "loss": 0.4789, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.0572151898734177, |
| "grad_norm": 0.4873408742673492, |
| "learning_rate": 3.848572541175116e-05, |
| "loss": 0.4685, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.0612658227848102, |
| "grad_norm": 0.6693676014730104, |
| "learning_rate": 3.846398621389154e-05, |
| "loss": 0.483, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.0653164556962025, |
| "grad_norm": 0.5220040251935801, |
| "learning_rate": 3.84420983096253e-05, |
| "loss": 0.4927, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.069367088607595, |
| "grad_norm": 0.5648716750630282, |
| "learning_rate": 3.8420061875234606e-05, |
| "loss": 0.4694, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.0734177215189873, |
| "grad_norm": 0.6385670338115153, |
| "learning_rate": 3.839787708819787e-05, |
| "loss": 0.4814, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.0774683544303798, |
| "grad_norm": 0.5220021670730184, |
| "learning_rate": 3.8375544127188325e-05, |
| "loss": 0.465, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.081518987341772, |
| "grad_norm": 0.5675819990596119, |
| "learning_rate": 3.8353063172072564e-05, |
| "loss": 0.4762, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.0855696202531646, |
| "grad_norm": 0.5476666319006838, |
| "learning_rate": 3.8330434403909105e-05, |
| "loss": 0.4578, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.089620253164557, |
| "grad_norm": 0.5121498318363616, |
| "learning_rate": 3.8307658004946934e-05, |
| "loss": 0.4774, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.0936708860759494, |
| "grad_norm": 0.5765900785566105, |
| "learning_rate": 3.8284734158624046e-05, |
| "loss": 0.4674, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.0977215189873417, |
| "grad_norm": 0.5465513503709917, |
| "learning_rate": 3.826166304956594e-05, |
| "loss": 0.4738, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.1017721518987342, |
| "grad_norm": 0.5609285636353945, |
| "learning_rate": 3.8238444863584164e-05, |
| "loss": 0.4662, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.1058227848101265, |
| "grad_norm": 0.523032544203239, |
| "learning_rate": 3.821507978767479e-05, |
| "loss": 0.4588, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.109873417721519, |
| "grad_norm": 0.8017152729800482, |
| "learning_rate": 3.819156801001693e-05, |
| "loss": 0.4873, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.1139240506329113, |
| "grad_norm": 0.6099373316038048, |
| "learning_rate": 3.816790971997121e-05, |
| "loss": 0.4712, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.1179746835443038, |
| "grad_norm": 0.7005169054698475, |
| "learning_rate": 3.8144105108078246e-05, |
| "loss": 0.471, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.1220253164556961, |
| "grad_norm": 0.712803799600362, |
| "learning_rate": 3.81201543660571e-05, |
| "loss": 0.4558, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.1260759493670887, |
| "grad_norm": 0.5719551298162912, |
| "learning_rate": 3.809605768680377e-05, |
| "loss": 0.4708, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.130126582278481, |
| "grad_norm": 0.8084235315302805, |
| "learning_rate": 3.807181526438958e-05, |
| "loss": 0.4863, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.1341772151898735, |
| "grad_norm": 0.6001596477262929, |
| "learning_rate": 3.8047427294059697e-05, |
| "loss": 0.4833, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.1382278481012658, |
| "grad_norm": 0.6924537852304811, |
| "learning_rate": 3.802289397223145e-05, |
| "loss": 0.4773, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.1422784810126583, |
| "grad_norm": 0.7839179331738506, |
| "learning_rate": 3.7998215496492854e-05, |
| "loss": 0.4816, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.1463291139240506, |
| "grad_norm": 0.5571628556364443, |
| "learning_rate": 3.797339206560096e-05, |
| "loss": 0.4892, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.150379746835443, |
| "grad_norm": 0.7678951612467041, |
| "learning_rate": 3.794842387948027e-05, |
| "loss": 0.4884, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.1544303797468354, |
| "grad_norm": 0.5937615478168569, |
| "learning_rate": 3.7923311139221114e-05, |
| "loss": 0.4968, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.158481012658228, |
| "grad_norm": 0.5693886444949976, |
| "learning_rate": 3.7898054047078054e-05, |
| "loss": 0.4933, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.1625316455696202, |
| "grad_norm": 0.662931281054524, |
| "learning_rate": 3.787265280646825e-05, |
| "loss": 0.4937, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.1665822784810127, |
| "grad_norm": 0.5722457173722049, |
| "learning_rate": 3.7847107621969786e-05, |
| "loss": 0.4728, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.170632911392405, |
| "grad_norm": 0.6389522276209824, |
| "learning_rate": 3.7821418699320064e-05, |
| "loss": 0.4906, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.1746835443037975, |
| "grad_norm": 0.5476452406971916, |
| "learning_rate": 3.7795586245414145e-05, |
| "loss": 0.4807, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.1787341772151898, |
| "grad_norm": 0.6056259885165657, |
| "learning_rate": 3.776961046830306e-05, |
| "loss": 0.4818, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.1827848101265823, |
| "grad_norm": 0.5304860567259382, |
| "learning_rate": 3.774349157719215e-05, |
| "loss": 0.5035, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.1868354430379746, |
| "grad_norm": 0.6238180437984675, |
| "learning_rate": 3.7717229782439365e-05, |
| "loss": 0.475, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.1908860759493671, |
| "grad_norm": 0.4960363328625346, |
| "learning_rate": 3.769082529555359e-05, |
| "loss": 0.4742, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.1949367088607594, |
| "grad_norm": 0.6532582689820073, |
| "learning_rate": 3.766427832919294e-05, |
| "loss": 0.4683, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.198987341772152, |
| "grad_norm": 0.49973044788647825, |
| "learning_rate": 3.7637589097163024e-05, |
| "loss": 0.4876, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.2030379746835442, |
| "grad_norm": 0.5519900451662588, |
| "learning_rate": 3.761075781441526e-05, |
| "loss": 0.4896, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.2070886075949367, |
| "grad_norm": 0.49029510004156446, |
| "learning_rate": 3.75837846970451e-05, |
| "loss": 0.4837, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.211139240506329, |
| "grad_norm": 0.5509644920979125, |
| "learning_rate": 3.755666996229032e-05, |
| "loss": 0.4698, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.2151898734177216, |
| "grad_norm": 0.5566539868277111, |
| "learning_rate": 3.752941382852927e-05, |
| "loss": 0.4974, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.2192405063291138, |
| "grad_norm": 0.4704903282111282, |
| "learning_rate": 3.7502016515279115e-05, |
| "loss": 0.4973, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.2232911392405064, |
| "grad_norm": 0.576525890913354, |
| "learning_rate": 3.7474478243194043e-05, |
| "loss": 0.4898, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.2273417721518987, |
| "grad_norm": 0.4706290198822068, |
| "learning_rate": 3.744679923406351e-05, |
| "loss": 0.4868, |
| "step": 303 |
| }, |
| { |
| "epoch": 1.2313924050632912, |
| "grad_norm": 0.487882898038887, |
| "learning_rate": 3.741897971081043e-05, |
| "loss": 0.4779, |
| "step": 304 |
| }, |
| { |
| "epoch": 1.2354430379746835, |
| "grad_norm": 0.4586710431662187, |
| "learning_rate": 3.739101989748946e-05, |
| "loss": 0.4932, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.239493670886076, |
| "grad_norm": 0.48562503285367264, |
| "learning_rate": 3.7362920019285066e-05, |
| "loss": 0.4939, |
| "step": 306 |
| }, |
| { |
| "epoch": 1.2435443037974683, |
| "grad_norm": 0.578154491753082, |
| "learning_rate": 3.73346803025098e-05, |
| "loss": 0.481, |
| "step": 307 |
| }, |
| { |
| "epoch": 1.2475949367088608, |
| "grad_norm": 0.4344444446321055, |
| "learning_rate": 3.730630097460247e-05, |
| "loss": 0.4712, |
| "step": 308 |
| }, |
| { |
| "epoch": 1.251645569620253, |
| "grad_norm": 0.5986136275437102, |
| "learning_rate": 3.727778226412628e-05, |
| "loss": 0.4785, |
| "step": 309 |
| }, |
| { |
| "epoch": 1.2556962025316456, |
| "grad_norm": 0.4771033810456709, |
| "learning_rate": 3.7249124400767006e-05, |
| "loss": 0.4815, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.259746835443038, |
| "grad_norm": 0.4981192574507264, |
| "learning_rate": 3.722032761533114e-05, |
| "loss": 0.4481, |
| "step": 311 |
| }, |
| { |
| "epoch": 1.2637974683544304, |
| "grad_norm": 0.45373097420472924, |
| "learning_rate": 3.719139213974403e-05, |
| "loss": 0.4615, |
| "step": 312 |
| }, |
| { |
| "epoch": 1.2678481012658227, |
| "grad_norm": 0.5681974124097704, |
| "learning_rate": 3.7162318207048006e-05, |
| "loss": 0.4914, |
| "step": 313 |
| }, |
| { |
| "epoch": 1.2718987341772152, |
| "grad_norm": 0.5171832762298352, |
| "learning_rate": 3.713310605140055e-05, |
| "loss": 0.4795, |
| "step": 314 |
| }, |
| { |
| "epoch": 1.2759493670886077, |
| "grad_norm": 0.44772531038441377, |
| "learning_rate": 3.710375590807233e-05, |
| "loss": 0.4766, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.28, |
| "grad_norm": 0.4768245563076455, |
| "learning_rate": 3.7074268013445365e-05, |
| "loss": 0.5081, |
| "step": 316 |
| }, |
| { |
| "epoch": 1.2840506329113923, |
| "grad_norm": 0.4298980460453433, |
| "learning_rate": 3.7044642605011114e-05, |
| "loss": 0.48, |
| "step": 317 |
| }, |
| { |
| "epoch": 1.2881012658227848, |
| "grad_norm": 0.438588526382241, |
| "learning_rate": 3.701487992136854e-05, |
| "loss": 0.4665, |
| "step": 318 |
| }, |
| { |
| "epoch": 1.2921518987341774, |
| "grad_norm": 0.5015503593828158, |
| "learning_rate": 3.69849802022222e-05, |
| "loss": 0.4685, |
| "step": 319 |
| }, |
| { |
| "epoch": 1.2962025316455696, |
| "grad_norm": 0.4492575046289095, |
| "learning_rate": 3.6954943688380334e-05, |
| "loss": 0.4904, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.300253164556962, |
| "grad_norm": 0.5514295056926577, |
| "learning_rate": 3.692477062175289e-05, |
| "loss": 0.4711, |
| "step": 321 |
| }, |
| { |
| "epoch": 1.3043037974683545, |
| "grad_norm": 0.459587589381779, |
| "learning_rate": 3.689446124534958e-05, |
| "loss": 0.4811, |
| "step": 322 |
| }, |
| { |
| "epoch": 1.308354430379747, |
| "grad_norm": 0.5292937910323344, |
| "learning_rate": 3.686401580327799e-05, |
| "loss": 0.4786, |
| "step": 323 |
| }, |
| { |
| "epoch": 1.3124050632911393, |
| "grad_norm": 0.4503115053194189, |
| "learning_rate": 3.683343454074149e-05, |
| "loss": 0.4882, |
| "step": 324 |
| }, |
| { |
| "epoch": 1.3164556962025316, |
| "grad_norm": 0.4905588400008236, |
| "learning_rate": 3.6802717704037386e-05, |
| "loss": 0.463, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.320506329113924, |
| "grad_norm": 0.47136765247413703, |
| "learning_rate": 3.6771865540554855e-05, |
| "loss": 0.4701, |
| "step": 326 |
| }, |
| { |
| "epoch": 1.3245569620253166, |
| "grad_norm": 0.5284825628302522, |
| "learning_rate": 3.674087829877297e-05, |
| "loss": 0.4834, |
| "step": 327 |
| }, |
| { |
| "epoch": 1.3286075949367089, |
| "grad_norm": 0.4256403059750766, |
| "learning_rate": 3.6709756228258735e-05, |
| "loss": 0.4751, |
| "step": 328 |
| }, |
| { |
| "epoch": 1.3326582278481012, |
| "grad_norm": 0.45521913499843164, |
| "learning_rate": 3.667849957966501e-05, |
| "loss": 0.4545, |
| "step": 329 |
| }, |
| { |
| "epoch": 1.3367088607594937, |
| "grad_norm": 0.47819127853050636, |
| "learning_rate": 3.6647108604728546e-05, |
| "loss": 0.4561, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3407594936708862, |
| "grad_norm": 0.49908021465402586, |
| "learning_rate": 3.661558355626795e-05, |
| "loss": 0.4861, |
| "step": 331 |
| }, |
| { |
| "epoch": 1.3448101265822785, |
| "grad_norm": 0.44737588913619475, |
| "learning_rate": 3.658392468818163e-05, |
| "loss": 0.4699, |
| "step": 332 |
| }, |
| { |
| "epoch": 1.3488607594936708, |
| "grad_norm": 0.5503414633926319, |
| "learning_rate": 3.655213225544574e-05, |
| "loss": 0.4879, |
| "step": 333 |
| }, |
| { |
| "epoch": 1.3529113924050633, |
| "grad_norm": 0.4986672619792775, |
| "learning_rate": 3.652020651411218e-05, |
| "loss": 0.4766, |
| "step": 334 |
| }, |
| { |
| "epoch": 1.3569620253164558, |
| "grad_norm": 0.45186204462365503, |
| "learning_rate": 3.6488147721306474e-05, |
| "loss": 0.4684, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.3610126582278481, |
| "grad_norm": 0.5658911961626825, |
| "learning_rate": 3.645595613522574e-05, |
| "loss": 0.4744, |
| "step": 336 |
| }, |
| { |
| "epoch": 1.3650632911392404, |
| "grad_norm": 0.5236360203598153, |
| "learning_rate": 3.642363201513657e-05, |
| "loss": 0.4723, |
| "step": 337 |
| }, |
| { |
| "epoch": 1.369113924050633, |
| "grad_norm": 0.5271425909001062, |
| "learning_rate": 3.6391175621373006e-05, |
| "loss": 0.4697, |
| "step": 338 |
| }, |
| { |
| "epoch": 1.3731645569620254, |
| "grad_norm": 0.4209791093811891, |
| "learning_rate": 3.6358587215334355e-05, |
| "loss": 0.4694, |
| "step": 339 |
| }, |
| { |
| "epoch": 1.3772151898734177, |
| "grad_norm": 0.4719324534521668, |
| "learning_rate": 3.632586705948318e-05, |
| "loss": 0.4884, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.38126582278481, |
| "grad_norm": 0.46720622933215256, |
| "learning_rate": 3.629301541734311e-05, |
| "loss": 0.4884, |
| "step": 341 |
| }, |
| { |
| "epoch": 1.3853164556962025, |
| "grad_norm": 0.5200730183432432, |
| "learning_rate": 3.626003255349676e-05, |
| "loss": 0.469, |
| "step": 342 |
| }, |
| { |
| "epoch": 1.389367088607595, |
| "grad_norm": 0.5411653617602161, |
| "learning_rate": 3.622691873358357e-05, |
| "loss": 0.5047, |
| "step": 343 |
| }, |
| { |
| "epoch": 1.3934177215189874, |
| "grad_norm": 0.43949888548890503, |
| "learning_rate": 3.61936742242977e-05, |
| "loss": 0.474, |
| "step": 344 |
| }, |
| { |
| "epoch": 1.3974683544303796, |
| "grad_norm": 0.47316868049880917, |
| "learning_rate": 3.6160299293385864e-05, |
| "loss": 0.499, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4015189873417722, |
| "grad_norm": 0.4426728578138348, |
| "learning_rate": 3.612679420964516e-05, |
| "loss": 0.4991, |
| "step": 346 |
| }, |
| { |
| "epoch": 1.4055696202531647, |
| "grad_norm": 0.49205190749652766, |
| "learning_rate": 3.609315924292092e-05, |
| "loss": 0.4969, |
| "step": 347 |
| }, |
| { |
| "epoch": 1.409620253164557, |
| "grad_norm": 0.4476647503010313, |
| "learning_rate": 3.6059394664104554e-05, |
| "loss": 0.4841, |
| "step": 348 |
| }, |
| { |
| "epoch": 1.4136708860759493, |
| "grad_norm": 0.48012788731719847, |
| "learning_rate": 3.602550074513133e-05, |
| "loss": 0.4762, |
| "step": 349 |
| }, |
| { |
| "epoch": 1.4177215189873418, |
| "grad_norm": 0.46121241611994784, |
| "learning_rate": 3.599147775897822e-05, |
| "loss": 0.4757, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4217721518987343, |
| "grad_norm": 0.5104356775662501, |
| "learning_rate": 3.595732597966167e-05, |
| "loss": 0.4809, |
| "step": 351 |
| }, |
| { |
| "epoch": 1.4258227848101266, |
| "grad_norm": 0.45453649598149504, |
| "learning_rate": 3.592304568223542e-05, |
| "loss": 0.4761, |
| "step": 352 |
| }, |
| { |
| "epoch": 1.4298734177215189, |
| "grad_norm": 0.46647930560266604, |
| "learning_rate": 3.588863714278826e-05, |
| "loss": 0.4778, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.4339240506329114, |
| "grad_norm": 0.5147868932840715, |
| "learning_rate": 3.585410063844186e-05, |
| "loss": 0.4799, |
| "step": 354 |
| }, |
| { |
| "epoch": 1.437974683544304, |
| "grad_norm": 0.48026260184351705, |
| "learning_rate": 3.581943644734846e-05, |
| "loss": 0.4838, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.4420253164556962, |
| "grad_norm": 0.4948696031548396, |
| "learning_rate": 3.578464484868869e-05, |
| "loss": 0.4541, |
| "step": 356 |
| }, |
| { |
| "epoch": 1.4460759493670885, |
| "grad_norm": 0.4636283074185639, |
| "learning_rate": 3.5749726122669316e-05, |
| "loss": 0.4652, |
| "step": 357 |
| }, |
| { |
| "epoch": 1.450126582278481, |
| "grad_norm": 0.5049853751085159, |
| "learning_rate": 3.5714680550520943e-05, |
| "loss": 0.463, |
| "step": 358 |
| }, |
| { |
| "epoch": 1.4541772151898735, |
| "grad_norm": 0.49736005145816864, |
| "learning_rate": 3.5679508414495794e-05, |
| "loss": 0.4847, |
| "step": 359 |
| }, |
| { |
| "epoch": 1.4582278481012658, |
| "grad_norm": 0.49789967404668833, |
| "learning_rate": 3.564420999786543e-05, |
| "loss": 0.465, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.4622784810126581, |
| "grad_norm": 0.4833594154154734, |
| "learning_rate": 3.560878558491842e-05, |
| "loss": 0.4687, |
| "step": 361 |
| }, |
| { |
| "epoch": 1.4663291139240506, |
| "grad_norm": 0.4378396247223715, |
| "learning_rate": 3.5573235460958145e-05, |
| "loss": 0.4854, |
| "step": 362 |
| }, |
| { |
| "epoch": 1.4703797468354431, |
| "grad_norm": 0.5322724007364603, |
| "learning_rate": 3.553755991230039e-05, |
| "loss": 0.4923, |
| "step": 363 |
| }, |
| { |
| "epoch": 1.4744303797468354, |
| "grad_norm": 0.41469738973350523, |
| "learning_rate": 3.5501759226271144e-05, |
| "loss": 0.4587, |
| "step": 364 |
| }, |
| { |
| "epoch": 1.4784810126582277, |
| "grad_norm": 0.5159458387676435, |
| "learning_rate": 3.546583369120419e-05, |
| "loss": 0.4772, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.4825316455696202, |
| "grad_norm": 0.4087645556391689, |
| "learning_rate": 3.5429783596438864e-05, |
| "loss": 0.4751, |
| "step": 366 |
| }, |
| { |
| "epoch": 1.4865822784810128, |
| "grad_norm": 0.5516978415977648, |
| "learning_rate": 3.539360923231766e-05, |
| "loss": 0.4774, |
| "step": 367 |
| }, |
| { |
| "epoch": 1.490632911392405, |
| "grad_norm": 0.42212406915033507, |
| "learning_rate": 3.535731089018394e-05, |
| "loss": 0.463, |
| "step": 368 |
| }, |
| { |
| "epoch": 1.4946835443037974, |
| "grad_norm": 0.5060721012446758, |
| "learning_rate": 3.532088886237956e-05, |
| "loss": 0.4801, |
| "step": 369 |
| }, |
| { |
| "epoch": 1.4987341772151899, |
| "grad_norm": 0.4753987304009231, |
| "learning_rate": 3.528434344224253e-05, |
| "loss": 0.479, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5027848101265824, |
| "grad_norm": 0.4555774647426061, |
| "learning_rate": 3.524767492410464e-05, |
| "loss": 0.4813, |
| "step": 371 |
| }, |
| { |
| "epoch": 1.5068354430379747, |
| "grad_norm": 0.5236391616375246, |
| "learning_rate": 3.521088360328908e-05, |
| "loss": 0.4806, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.510886075949367, |
| "grad_norm": 0.44878070438511664, |
| "learning_rate": 3.517396977610811e-05, |
| "loss": 0.4844, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.5149367088607595, |
| "grad_norm": 0.5132385053904075, |
| "learning_rate": 3.5136933739860595e-05, |
| "loss": 0.4815, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.518987341772152, |
| "grad_norm": 0.4273141182975724, |
| "learning_rate": 3.509977579282971e-05, |
| "loss": 0.4673, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.5230379746835443, |
| "grad_norm": 0.49236532974722624, |
| "learning_rate": 3.5062496234280424e-05, |
| "loss": 0.4649, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.5270886075949366, |
| "grad_norm": 0.4480852063007958, |
| "learning_rate": 3.502509536445719e-05, |
| "loss": 0.4853, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.531139240506329, |
| "grad_norm": 0.4080457426452025, |
| "learning_rate": 3.498757348458147e-05, |
| "loss": 0.4706, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.5351898734177216, |
| "grad_norm": 0.49977694375681353, |
| "learning_rate": 3.4949930896849324e-05, |
| "loss": 0.4865, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.539240506329114, |
| "grad_norm": 0.4067057901526661, |
| "learning_rate": 3.491216790442899e-05, |
| "loss": 0.4768, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.5432911392405062, |
| "grad_norm": 0.5123170054919219, |
| "learning_rate": 3.487428481145839e-05, |
| "loss": 0.467, |
| "step": 381 |
| }, |
| { |
| "epoch": 1.5473417721518987, |
| "grad_norm": 0.4773485657661029, |
| "learning_rate": 3.483628192304278e-05, |
| "loss": 0.4763, |
| "step": 382 |
| }, |
| { |
| "epoch": 1.5513924050632912, |
| "grad_norm": 0.46808037249216716, |
| "learning_rate": 3.479815954525219e-05, |
| "loss": 0.4595, |
| "step": 383 |
| }, |
| { |
| "epoch": 1.5554430379746835, |
| "grad_norm": 0.4552772195992939, |
| "learning_rate": 3.475991798511899e-05, |
| "loss": 0.4909, |
| "step": 384 |
| }, |
| { |
| "epoch": 1.5594936708860758, |
| "grad_norm": 0.49940506340288926, |
| "learning_rate": 3.4721557550635464e-05, |
| "loss": 0.4805, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.5635443037974683, |
| "grad_norm": 0.5284616649401459, |
| "learning_rate": 3.468307855075128e-05, |
| "loss": 0.4813, |
| "step": 386 |
| }, |
| { |
| "epoch": 1.5675949367088609, |
| "grad_norm": 0.4444856806681571, |
| "learning_rate": 3.4644481295371005e-05, |
| "loss": 0.4628, |
| "step": 387 |
| }, |
| { |
| "epoch": 1.5716455696202531, |
| "grad_norm": 0.44617078419697137, |
| "learning_rate": 3.460576609535163e-05, |
| "loss": 0.4637, |
| "step": 388 |
| }, |
| { |
| "epoch": 1.5756962025316454, |
| "grad_norm": 0.4825285500904799, |
| "learning_rate": 3.456693326250006e-05, |
| "loss": 0.4751, |
| "step": 389 |
| }, |
| { |
| "epoch": 1.579746835443038, |
| "grad_norm": 0.4766485132254008, |
| "learning_rate": 3.452798310957058e-05, |
| "loss": 0.4809, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.5837974683544305, |
| "grad_norm": 0.4464074860882123, |
| "learning_rate": 3.4488915950262386e-05, |
| "loss": 0.4596, |
| "step": 391 |
| }, |
| { |
| "epoch": 1.5878481012658228, |
| "grad_norm": 0.5188653375775143, |
| "learning_rate": 3.4449732099216985e-05, |
| "loss": 0.4793, |
| "step": 392 |
| }, |
| { |
| "epoch": 1.591898734177215, |
| "grad_norm": 0.45075597027240066, |
| "learning_rate": 3.441043187201574e-05, |
| "loss": 0.4456, |
| "step": 393 |
| }, |
| { |
| "epoch": 1.5959493670886076, |
| "grad_norm": 0.5197732131259593, |
| "learning_rate": 3.437101558517728e-05, |
| "loss": 0.486, |
| "step": 394 |
| }, |
| { |
| "epoch": 1.6, |
| "grad_norm": 0.5035254863706539, |
| "learning_rate": 3.433148355615496e-05, |
| "loss": 0.4646, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.6040506329113924, |
| "grad_norm": 0.4959916668912987, |
| "learning_rate": 3.4291836103334294e-05, |
| "loss": 0.4894, |
| "step": 396 |
| }, |
| { |
| "epoch": 1.6081012658227847, |
| "grad_norm": 0.6616258684230307, |
| "learning_rate": 3.425207354603043e-05, |
| "loss": 0.5019, |
| "step": 397 |
| }, |
| { |
| "epoch": 1.6121518987341772, |
| "grad_norm": 0.48565642940309317, |
| "learning_rate": 3.421219620448553e-05, |
| "loss": 0.4688, |
| "step": 398 |
| }, |
| { |
| "epoch": 1.6162025316455697, |
| "grad_norm": 0.6526404461474611, |
| "learning_rate": 3.417220439986623e-05, |
| "loss": 0.4769, |
| "step": 399 |
| }, |
| { |
| "epoch": 1.620253164556962, |
| "grad_norm": 0.5537453250455298, |
| "learning_rate": 3.4132098454261024e-05, |
| "loss": 0.4974, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.6243037974683543, |
| "grad_norm": 0.48592521325373544, |
| "learning_rate": 3.4091878690677676e-05, |
| "loss": 0.4663, |
| "step": 401 |
| }, |
| { |
| "epoch": 1.6283544303797468, |
| "grad_norm": 0.47985356962810694, |
| "learning_rate": 3.405154543304065e-05, |
| "loss": 0.4667, |
| "step": 402 |
| }, |
| { |
| "epoch": 1.6324050632911393, |
| "grad_norm": 0.6112506470885639, |
| "learning_rate": 3.401109900618843e-05, |
| "loss": 0.4635, |
| "step": 403 |
| }, |
| { |
| "epoch": 1.6364556962025316, |
| "grad_norm": 0.4352781541500303, |
| "learning_rate": 3.3970539735870996e-05, |
| "loss": 0.4599, |
| "step": 404 |
| }, |
| { |
| "epoch": 1.640506329113924, |
| "grad_norm": 0.6069828899700545, |
| "learning_rate": 3.392986794874714e-05, |
| "loss": 0.4876, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.6445569620253164, |
| "grad_norm": 0.4916332168876736, |
| "learning_rate": 3.388908397238184e-05, |
| "loss": 0.4773, |
| "step": 406 |
| }, |
| { |
| "epoch": 1.648607594936709, |
| "grad_norm": 0.5192776008010027, |
| "learning_rate": 3.384818813524362e-05, |
| "loss": 0.4828, |
| "step": 407 |
| }, |
| { |
| "epoch": 1.6526582278481012, |
| "grad_norm": 0.523657979144137, |
| "learning_rate": 3.380718076670195e-05, |
| "loss": 0.4689, |
| "step": 408 |
| }, |
| { |
| "epoch": 1.6567088607594935, |
| "grad_norm": 0.4915526918212457, |
| "learning_rate": 3.376606219702454e-05, |
| "loss": 0.4752, |
| "step": 409 |
| }, |
| { |
| "epoch": 1.660759493670886, |
| "grad_norm": 0.51579372102155, |
| "learning_rate": 3.372483275737468e-05, |
| "loss": 0.4795, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.6648101265822786, |
| "grad_norm": 0.4454317848103436, |
| "learning_rate": 3.368349277980861e-05, |
| "loss": 0.4831, |
| "step": 411 |
| }, |
| { |
| "epoch": 1.6688607594936709, |
| "grad_norm": 0.510294531540504, |
| "learning_rate": 3.3642042597272844e-05, |
| "loss": 0.4796, |
| "step": 412 |
| }, |
| { |
| "epoch": 1.6729113924050631, |
| "grad_norm": 0.5076276788374344, |
| "learning_rate": 3.360048254360144e-05, |
| "loss": 0.4778, |
| "step": 413 |
| }, |
| { |
| "epoch": 1.6769620253164557, |
| "grad_norm": 0.5163155653496363, |
| "learning_rate": 3.355881295351336e-05, |
| "loss": 0.473, |
| "step": 414 |
| }, |
| { |
| "epoch": 1.6810126582278482, |
| "grad_norm": 0.4879494702596848, |
| "learning_rate": 3.351703416260975e-05, |
| "loss": 0.4755, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.6850632911392405, |
| "grad_norm": 0.4845518598148072, |
| "learning_rate": 3.347514650737126e-05, |
| "loss": 0.4711, |
| "step": 416 |
| }, |
| { |
| "epoch": 1.6891139240506328, |
| "grad_norm": 0.4815352725426383, |
| "learning_rate": 3.3433150325155295e-05, |
| "loss": 0.4541, |
| "step": 417 |
| }, |
| { |
| "epoch": 1.6931645569620253, |
| "grad_norm": 0.46965594551157586, |
| "learning_rate": 3.339104595419334e-05, |
| "loss": 0.4753, |
| "step": 418 |
| }, |
| { |
| "epoch": 1.6972151898734178, |
| "grad_norm": 0.4584750455315046, |
| "learning_rate": 3.3348833733588204e-05, |
| "loss": 0.4868, |
| "step": 419 |
| }, |
| { |
| "epoch": 1.70126582278481, |
| "grad_norm": 0.471727729184326, |
| "learning_rate": 3.3306514003311305e-05, |
| "loss": 0.4494, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7053164556962024, |
| "grad_norm": 0.46393775729152265, |
| "learning_rate": 3.326408710419996e-05, |
| "loss": 0.4693, |
| "step": 421 |
| }, |
| { |
| "epoch": 1.709367088607595, |
| "grad_norm": 0.4798003140594918, |
| "learning_rate": 3.322155337795454e-05, |
| "loss": 0.4774, |
| "step": 422 |
| }, |
| { |
| "epoch": 1.7134177215189874, |
| "grad_norm": 0.5373763531769086, |
| "learning_rate": 3.317891316713587e-05, |
| "loss": 0.4748, |
| "step": 423 |
| }, |
| { |
| "epoch": 1.7174683544303797, |
| "grad_norm": 0.4881969821578624, |
| "learning_rate": 3.313616681516231e-05, |
| "loss": 0.4868, |
| "step": 424 |
| }, |
| { |
| "epoch": 1.721518987341772, |
| "grad_norm": 0.4967024098908098, |
| "learning_rate": 3.309331466630713e-05, |
| "loss": 0.4647, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.7255696202531645, |
| "grad_norm": 0.4580922117982427, |
| "learning_rate": 3.305035706569563e-05, |
| "loss": 0.4509, |
| "step": 426 |
| }, |
| { |
| "epoch": 1.729620253164557, |
| "grad_norm": 0.5649944308391662, |
| "learning_rate": 3.3007294359302433e-05, |
| "loss": 0.481, |
| "step": 427 |
| }, |
| { |
| "epoch": 1.7336708860759493, |
| "grad_norm": 0.4711601720575769, |
| "learning_rate": 3.296412689394864e-05, |
| "loss": 0.4956, |
| "step": 428 |
| }, |
| { |
| "epoch": 1.7377215189873416, |
| "grad_norm": 0.49405127648476854, |
| "learning_rate": 3.292085501729909e-05, |
| "loss": 0.4554, |
| "step": 429 |
| }, |
| { |
| "epoch": 1.7417721518987341, |
| "grad_norm": 0.42717513914475247, |
| "learning_rate": 3.2877479077859534e-05, |
| "loss": 0.4576, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.7458227848101266, |
| "grad_norm": 0.526133291938095, |
| "learning_rate": 3.283399942497381e-05, |
| "loss": 0.4687, |
| "step": 431 |
| }, |
| { |
| "epoch": 1.749873417721519, |
| "grad_norm": 0.4486166907118644, |
| "learning_rate": 3.279041640882108e-05, |
| "loss": 0.4631, |
| "step": 432 |
| }, |
| { |
| "epoch": 1.7539240506329112, |
| "grad_norm": 0.46513810386811616, |
| "learning_rate": 3.2746730380412964e-05, |
| "loss": 0.4611, |
| "step": 433 |
| }, |
| { |
| "epoch": 1.7579746835443038, |
| "grad_norm": 0.41229810881993384, |
| "learning_rate": 3.2702941691590726e-05, |
| "loss": 0.4666, |
| "step": 434 |
| }, |
| { |
| "epoch": 1.7620253164556963, |
| "grad_norm": 0.45360708284765944, |
| "learning_rate": 3.265905069502244e-05, |
| "loss": 0.4737, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.7660759493670886, |
| "grad_norm": 0.4678099336667204, |
| "learning_rate": 3.261505774420016e-05, |
| "loss": 0.4667, |
| "step": 436 |
| }, |
| { |
| "epoch": 1.7701265822784809, |
| "grad_norm": 0.4565390790057733, |
| "learning_rate": 3.257096319343707e-05, |
| "loss": 0.4651, |
| "step": 437 |
| }, |
| { |
| "epoch": 1.7741772151898734, |
| "grad_norm": 0.445831290818833, |
| "learning_rate": 3.2526767397864614e-05, |
| "loss": 0.4834, |
| "step": 438 |
| }, |
| { |
| "epoch": 1.7782278481012659, |
| "grad_norm": 0.46363626583748874, |
| "learning_rate": 3.248247071342966e-05, |
| "loss": 0.4546, |
| "step": 439 |
| }, |
| { |
| "epoch": 1.7822784810126582, |
| "grad_norm": 0.4373056457777196, |
| "learning_rate": 3.243807349689161e-05, |
| "loss": 0.4753, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.7863291139240505, |
| "grad_norm": 0.51916405175132, |
| "learning_rate": 3.2393576105819544e-05, |
| "loss": 0.4742, |
| "step": 441 |
| }, |
| { |
| "epoch": 1.790379746835443, |
| "grad_norm": 0.41838612022533833, |
| "learning_rate": 3.2348978898589333e-05, |
| "loss": 0.4702, |
| "step": 442 |
| }, |
| { |
| "epoch": 1.7944303797468355, |
| "grad_norm": 0.4711571521304307, |
| "learning_rate": 3.230428223438075e-05, |
| "loss": 0.4603, |
| "step": 443 |
| }, |
| { |
| "epoch": 1.7984810126582278, |
| "grad_norm": 0.40496749061998183, |
| "learning_rate": 3.225948647317459e-05, |
| "loss": 0.4659, |
| "step": 444 |
| }, |
| { |
| "epoch": 1.80253164556962, |
| "grad_norm": 0.44703696753762034, |
| "learning_rate": 3.2214591975749745e-05, |
| "loss": 0.4722, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.8065822784810126, |
| "grad_norm": 0.4356949726576611, |
| "learning_rate": 3.216959910368034e-05, |
| "loss": 0.4783, |
| "step": 446 |
| }, |
| { |
| "epoch": 1.8106329113924051, |
| "grad_norm": 0.4269835886026811, |
| "learning_rate": 3.212450821933277e-05, |
| "loss": 0.4674, |
| "step": 447 |
| }, |
| { |
| "epoch": 1.8146835443037974, |
| "grad_norm": 0.4824224283784243, |
| "learning_rate": 3.207931968586281e-05, |
| "loss": 0.4746, |
| "step": 448 |
| }, |
| { |
| "epoch": 1.81873417721519, |
| "grad_norm": 0.4139220536863564, |
| "learning_rate": 3.203403386721272e-05, |
| "loss": 0.4537, |
| "step": 449 |
| }, |
| { |
| "epoch": 1.8227848101265822, |
| "grad_norm": 0.4622578959077951, |
| "learning_rate": 3.1988651128108245e-05, |
| "loss": 0.4697, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.8268354430379747, |
| "grad_norm": 0.43024272932880664, |
| "learning_rate": 3.194317183405573e-05, |
| "loss": 0.4759, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.8308860759493673, |
| "grad_norm": 0.4319315830751096, |
| "learning_rate": 3.189759635133914e-05, |
| "loss": 0.4949, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.8349367088607595, |
| "grad_norm": 0.42431448930261273, |
| "learning_rate": 3.185192504701718e-05, |
| "loss": 0.4711, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.8389873417721518, |
| "grad_norm": 0.4801882803275491, |
| "learning_rate": 3.1806158288920234e-05, |
| "loss": 0.4707, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.8430379746835444, |
| "grad_norm": 0.4532742418018245, |
| "learning_rate": 3.1760296445647477e-05, |
| "loss": 0.47, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.8470886075949369, |
| "grad_norm": 0.4163820273806894, |
| "learning_rate": 3.1714339886563896e-05, |
| "loss": 0.4729, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.8511392405063292, |
| "grad_norm": 0.39878392800104656, |
| "learning_rate": 3.166828898179731e-05, |
| "loss": 0.4514, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.8551898734177215, |
| "grad_norm": 0.4758062084026422, |
| "learning_rate": 3.162214410223536e-05, |
| "loss": 0.4947, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.859240506329114, |
| "grad_norm": 0.4323261669026794, |
| "learning_rate": 3.157590561952257e-05, |
| "loss": 0.4834, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.8632911392405065, |
| "grad_norm": 0.41765945147201483, |
| "learning_rate": 3.152957390605732e-05, |
| "loss": 0.4764, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.8673417721518988, |
| "grad_norm": 0.44745856939386236, |
| "learning_rate": 3.148314933498886e-05, |
| "loss": 0.4763, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.871392405063291, |
| "grad_norm": 0.41273605652582784, |
| "learning_rate": 3.143663228021431e-05, |
| "loss": 0.4646, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.8754430379746836, |
| "grad_norm": 0.39584392383285244, |
| "learning_rate": 3.1390023116375624e-05, |
| "loss": 0.4424, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.879493670886076, |
| "grad_norm": 0.43916488359210626, |
| "learning_rate": 3.134332221885661e-05, |
| "loss": 0.4961, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.8835443037974684, |
| "grad_norm": 0.4530017093124715, |
| "learning_rate": 3.129652996377987e-05, |
| "loss": 0.4769, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.8875949367088607, |
| "grad_norm": 0.4323587572723995, |
| "learning_rate": 3.12496467280038e-05, |
| "loss": 0.4885, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.8916455696202532, |
| "grad_norm": 0.4333971858007138, |
| "learning_rate": 3.120267288911952e-05, |
| "loss": 0.4612, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.8956962025316457, |
| "grad_norm": 0.46235723178337224, |
| "learning_rate": 3.11556088254479e-05, |
| "loss": 0.4952, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.899746835443038, |
| "grad_norm": 0.44909118044352847, |
| "learning_rate": 3.11084549160364e-05, |
| "loss": 0.4808, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.9037974683544303, |
| "grad_norm": 0.49178574724723156, |
| "learning_rate": 3.106121154065615e-05, |
| "loss": 0.4477, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.9078481012658228, |
| "grad_norm": 0.42611563498275146, |
| "learning_rate": 3.1013879079798805e-05, |
| "loss": 0.4745, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.9118987341772153, |
| "grad_norm": 0.4732969808040858, |
| "learning_rate": 3.096645791467348e-05, |
| "loss": 0.4634, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.9159493670886076, |
| "grad_norm": 0.41592382367242436, |
| "learning_rate": 3.091894842720373e-05, |
| "loss": 0.472, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.92, |
| "grad_norm": 0.4344571575532055, |
| "learning_rate": 3.0871351000024425e-05, |
| "loss": 0.4662, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.9240506329113924, |
| "grad_norm": 0.4203853655750657, |
| "learning_rate": 3.0823666016478716e-05, |
| "loss": 0.4758, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.928101265822785, |
| "grad_norm": 0.4145408374305995, |
| "learning_rate": 3.0775893860614896e-05, |
| "loss": 0.459, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.9321518987341773, |
| "grad_norm": 0.4259726639026889, |
| "learning_rate": 3.0728034917183336e-05, |
| "loss": 0.4839, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.9362025316455695, |
| "grad_norm": 0.4154097475405872, |
| "learning_rate": 3.06800895716334e-05, |
| "loss": 0.4708, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.940253164556962, |
| "grad_norm": 0.4106037994784759, |
| "learning_rate": 3.063205821011029e-05, |
| "loss": 0.4781, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.9443037974683546, |
| "grad_norm": 0.4195444292965047, |
| "learning_rate": 3.0583941219452016e-05, |
| "loss": 0.4786, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.9483544303797469, |
| "grad_norm": 0.4105098026451314, |
| "learning_rate": 3.053573898718618e-05, |
| "loss": 0.4738, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.9524050632911392, |
| "grad_norm": 0.4901549889751573, |
| "learning_rate": 3.0487451901526956e-05, |
| "loss": 0.4891, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.9564556962025317, |
| "grad_norm": 0.4080220529698759, |
| "learning_rate": 3.0439080351371875e-05, |
| "loss": 0.4505, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.9605063291139242, |
| "grad_norm": 0.4886203488643362, |
| "learning_rate": 3.0390624726298764e-05, |
| "loss": 0.4696, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.9645569620253165, |
| "grad_norm": 0.4715090299038299, |
| "learning_rate": 3.034208541656255e-05, |
| "loss": 0.4719, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.9686075949367088, |
| "grad_norm": 0.481615864140709, |
| "learning_rate": 3.029346281309218e-05, |
| "loss": 0.4736, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.9726582278481013, |
| "grad_norm": 0.4763002847097705, |
| "learning_rate": 3.0244757307487415e-05, |
| "loss": 0.4635, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.9767088607594938, |
| "grad_norm": 0.45839404832720193, |
| "learning_rate": 3.019596929201569e-05, |
| "loss": 0.4477, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.980759493670886, |
| "grad_norm": 0.5127366671862449, |
| "learning_rate": 3.0147099159608985e-05, |
| "loss": 0.4697, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.9848101265822784, |
| "grad_norm": 0.43416600738548305, |
| "learning_rate": 3.0098147303860616e-05, |
| "loss": 0.4605, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.988860759493671, |
| "grad_norm": 0.43703069001465666, |
| "learning_rate": 3.0049114119022117e-05, |
| "loss": 0.4321, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.9929113924050634, |
| "grad_norm": 0.4215342701858305, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.4597, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.9969620253164557, |
| "grad_norm": 0.41238268760402197, |
| "learning_rate": 2.995080534235264e-05, |
| "loss": 0.4744, |
| "step": 493 |
| }, |
| { |
| "epoch": 2.001012658227848, |
| "grad_norm": 0.42214094056655277, |
| "learning_rate": 2.9901530542287044e-05, |
| "loss": 0.4423, |
| "step": 494 |
| }, |
| { |
| "epoch": 2.0050632911392405, |
| "grad_norm": 0.549484172657868, |
| "learning_rate": 2.9852175996655676e-05, |
| "loss": 0.3541, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.009113924050633, |
| "grad_norm": 0.4907049853165995, |
| "learning_rate": 2.980274210295326e-05, |
| "loss": 0.326, |
| "step": 496 |
| }, |
| { |
| "epoch": 2.013164556962025, |
| "grad_norm": 0.7520682577592449, |
| "learning_rate": 2.9753229259313578e-05, |
| "loss": 0.3511, |
| "step": 497 |
| }, |
| { |
| "epoch": 2.0172151898734176, |
| "grad_norm": 0.5211553769265185, |
| "learning_rate": 2.9703637864506274e-05, |
| "loss": 0.3533, |
| "step": 498 |
| }, |
| { |
| "epoch": 2.02126582278481, |
| "grad_norm": 0.5224666291210901, |
| "learning_rate": 2.965396831793362e-05, |
| "loss": 0.3342, |
| "step": 499 |
| }, |
| { |
| "epoch": 2.0253164556962027, |
| "grad_norm": 0.45654148882687096, |
| "learning_rate": 2.9604221019627316e-05, |
| "loss": 0.3375, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0293670886075947, |
| "grad_norm": 0.46302110804743657, |
| "learning_rate": 2.955439637024526e-05, |
| "loss": 0.3478, |
| "step": 501 |
| }, |
| { |
| "epoch": 2.0334177215189873, |
| "grad_norm": 0.48264347385411555, |
| "learning_rate": 2.9504494771068334e-05, |
| "loss": 0.3492, |
| "step": 502 |
| }, |
| { |
| "epoch": 2.0374683544303798, |
| "grad_norm": 0.4682572599148682, |
| "learning_rate": 2.9454516623997156e-05, |
| "loss": 0.3352, |
| "step": 503 |
| }, |
| { |
| "epoch": 2.0415189873417723, |
| "grad_norm": 0.4582038031936385, |
| "learning_rate": 2.9404462331548847e-05, |
| "loss": 0.3554, |
| "step": 504 |
| }, |
| { |
| "epoch": 2.0455696202531644, |
| "grad_norm": 0.43294886405489724, |
| "learning_rate": 2.93543322968538e-05, |
| "loss": 0.3412, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.049620253164557, |
| "grad_norm": 0.4966857328928292, |
| "learning_rate": 2.9304126923652428e-05, |
| "loss": 0.3112, |
| "step": 506 |
| }, |
| { |
| "epoch": 2.0536708860759494, |
| "grad_norm": 0.4076193518676165, |
| "learning_rate": 2.9253846616291896e-05, |
| "loss": 0.3316, |
| "step": 507 |
| }, |
| { |
| "epoch": 2.057721518987342, |
| "grad_norm": 0.5068817977074173, |
| "learning_rate": 2.9203491779722896e-05, |
| "loss": 0.3343, |
| "step": 508 |
| }, |
| { |
| "epoch": 2.061772151898734, |
| "grad_norm": 0.4815752763201104, |
| "learning_rate": 2.9153062819496357e-05, |
| "loss": 0.3352, |
| "step": 509 |
| }, |
| { |
| "epoch": 2.0658227848101265, |
| "grad_norm": 0.441988522812003, |
| "learning_rate": 2.9102560141760178e-05, |
| "loss": 0.3509, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.069873417721519, |
| "grad_norm": 0.4560715321516646, |
| "learning_rate": 2.9051984153256004e-05, |
| "loss": 0.3323, |
| "step": 511 |
| }, |
| { |
| "epoch": 2.0739240506329115, |
| "grad_norm": 0.47010032255771783, |
| "learning_rate": 2.900133526131588e-05, |
| "loss": 0.3331, |
| "step": 512 |
| }, |
| { |
| "epoch": 2.0779746835443036, |
| "grad_norm": 0.4050617790975188, |
| "learning_rate": 2.8950613873859025e-05, |
| "loss": 0.3356, |
| "step": 513 |
| }, |
| { |
| "epoch": 2.082025316455696, |
| "grad_norm": 0.39973710679684216, |
| "learning_rate": 2.8899820399388515e-05, |
| "loss": 0.3413, |
| "step": 514 |
| }, |
| { |
| "epoch": 2.0860759493670886, |
| "grad_norm": 0.43292651923392106, |
| "learning_rate": 2.8848955246988012e-05, |
| "loss": 0.3427, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.090126582278481, |
| "grad_norm": 0.3838652159176441, |
| "learning_rate": 2.879801882631847e-05, |
| "loss": 0.3314, |
| "step": 516 |
| }, |
| { |
| "epoch": 2.094177215189873, |
| "grad_norm": 0.4077172316391536, |
| "learning_rate": 2.8747011547614808e-05, |
| "loss": 0.3581, |
| "step": 517 |
| }, |
| { |
| "epoch": 2.0982278481012657, |
| "grad_norm": 0.402921127435316, |
| "learning_rate": 2.8695933821682635e-05, |
| "loss": 0.3355, |
| "step": 518 |
| }, |
| { |
| "epoch": 2.1022784810126582, |
| "grad_norm": 0.40504150492126845, |
| "learning_rate": 2.864478605989494e-05, |
| "loss": 0.3405, |
| "step": 519 |
| }, |
| { |
| "epoch": 2.1063291139240508, |
| "grad_norm": 0.4046599443308935, |
| "learning_rate": 2.8593568674188765e-05, |
| "loss": 0.342, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.110379746835443, |
| "grad_norm": 0.4011088802876108, |
| "learning_rate": 2.8542282077061892e-05, |
| "loss": 0.333, |
| "step": 521 |
| }, |
| { |
| "epoch": 2.1144303797468353, |
| "grad_norm": 0.4007122760701765, |
| "learning_rate": 2.8490926681569523e-05, |
| "loss": 0.3451, |
| "step": 522 |
| }, |
| { |
| "epoch": 2.118481012658228, |
| "grad_norm": 0.40841570649186254, |
| "learning_rate": 2.8439502901320956e-05, |
| "loss": 0.3509, |
| "step": 523 |
| }, |
| { |
| "epoch": 2.1225316455696204, |
| "grad_norm": 0.43275187422986844, |
| "learning_rate": 2.8388011150476237e-05, |
| "loss": 0.3189, |
| "step": 524 |
| }, |
| { |
| "epoch": 2.1265822784810124, |
| "grad_norm": 0.4185681763651052, |
| "learning_rate": 2.8336451843742866e-05, |
| "loss": 0.3495, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.130632911392405, |
| "grad_norm": 0.4468746861938347, |
| "learning_rate": 2.8284825396372387e-05, |
| "loss": 0.3305, |
| "step": 526 |
| }, |
| { |
| "epoch": 2.1346835443037975, |
| "grad_norm": 0.4301096471319844, |
| "learning_rate": 2.8233132224157132e-05, |
| "loss": 0.3398, |
| "step": 527 |
| }, |
| { |
| "epoch": 2.13873417721519, |
| "grad_norm": 0.4458175531339097, |
| "learning_rate": 2.8181372743426805e-05, |
| "loss": 0.3431, |
| "step": 528 |
| }, |
| { |
| "epoch": 2.1427848101265825, |
| "grad_norm": 0.43418481539374504, |
| "learning_rate": 2.8129547371045128e-05, |
| "loss": 0.3391, |
| "step": 529 |
| }, |
| { |
| "epoch": 2.1468354430379746, |
| "grad_norm": 0.41217115639583757, |
| "learning_rate": 2.8077656524406534e-05, |
| "loss": 0.3291, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.150886075949367, |
| "grad_norm": 0.4686563996591434, |
| "learning_rate": 2.802570062143278e-05, |
| "loss": 0.3257, |
| "step": 531 |
| }, |
| { |
| "epoch": 2.1549367088607596, |
| "grad_norm": 0.39055897084239, |
| "learning_rate": 2.7973680080569555e-05, |
| "loss": 0.3439, |
| "step": 532 |
| }, |
| { |
| "epoch": 2.1589873417721517, |
| "grad_norm": 0.41737306800766344, |
| "learning_rate": 2.792159532078314e-05, |
| "loss": 0.3488, |
| "step": 533 |
| }, |
| { |
| "epoch": 2.163037974683544, |
| "grad_norm": 0.41145172350643666, |
| "learning_rate": 2.7869446761557033e-05, |
| "loss": 0.3348, |
| "step": 534 |
| }, |
| { |
| "epoch": 2.1670886075949367, |
| "grad_norm": 0.42292309870658645, |
| "learning_rate": 2.781723482288857e-05, |
| "loss": 0.34, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.1711392405063292, |
| "grad_norm": 0.4171104976780024, |
| "learning_rate": 2.7764959925285517e-05, |
| "loss": 0.3447, |
| "step": 536 |
| }, |
| { |
| "epoch": 2.1751898734177217, |
| "grad_norm": 0.4024571196464202, |
| "learning_rate": 2.771262248976272e-05, |
| "loss": 0.335, |
| "step": 537 |
| }, |
| { |
| "epoch": 2.179240506329114, |
| "grad_norm": 0.4345395242948398, |
| "learning_rate": 2.7660222937838677e-05, |
| "loss": 0.3191, |
| "step": 538 |
| }, |
| { |
| "epoch": 2.1832911392405063, |
| "grad_norm": 0.38608679725949574, |
| "learning_rate": 2.7607761691532186e-05, |
| "loss": 0.3367, |
| "step": 539 |
| }, |
| { |
| "epoch": 2.187341772151899, |
| "grad_norm": 0.38546063366699324, |
| "learning_rate": 2.7555239173358916e-05, |
| "loss": 0.3346, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.191392405063291, |
| "grad_norm": 0.4336134671349893, |
| "learning_rate": 2.7502655806328e-05, |
| "loss": 0.341, |
| "step": 541 |
| }, |
| { |
| "epoch": 2.1954430379746834, |
| "grad_norm": 0.40181279009307463, |
| "learning_rate": 2.7450012013938648e-05, |
| "loss": 0.3449, |
| "step": 542 |
| }, |
| { |
| "epoch": 2.199493670886076, |
| "grad_norm": 0.39760429663827374, |
| "learning_rate": 2.739730822017673e-05, |
| "loss": 0.3305, |
| "step": 543 |
| }, |
| { |
| "epoch": 2.2035443037974685, |
| "grad_norm": 0.44069075402066515, |
| "learning_rate": 2.7344544849511355e-05, |
| "loss": 0.3355, |
| "step": 544 |
| }, |
| { |
| "epoch": 2.207594936708861, |
| "grad_norm": 0.39917835649423933, |
| "learning_rate": 2.7291722326891456e-05, |
| "loss": 0.3428, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.211645569620253, |
| "grad_norm": 0.38998614734890114, |
| "learning_rate": 2.723884107774236e-05, |
| "loss": 0.3447, |
| "step": 546 |
| }, |
| { |
| "epoch": 2.2156962025316456, |
| "grad_norm": 0.3777694025298611, |
| "learning_rate": 2.718590152796239e-05, |
| "loss": 0.3445, |
| "step": 547 |
| }, |
| { |
| "epoch": 2.219746835443038, |
| "grad_norm": 0.39858025352516696, |
| "learning_rate": 2.71329041039194e-05, |
| "loss": 0.3363, |
| "step": 548 |
| }, |
| { |
| "epoch": 2.22379746835443, |
| "grad_norm": 0.39157121660115113, |
| "learning_rate": 2.7079849232447357e-05, |
| "loss": 0.3517, |
| "step": 549 |
| }, |
| { |
| "epoch": 2.2278481012658227, |
| "grad_norm": 0.39056269719809134, |
| "learning_rate": 2.7026737340842895e-05, |
| "loss": 0.3411, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.231898734177215, |
| "grad_norm": 0.47170660761954086, |
| "learning_rate": 2.697356885686189e-05, |
| "loss": 0.3408, |
| "step": 551 |
| }, |
| { |
| "epoch": 2.2359493670886077, |
| "grad_norm": 0.37481719245814166, |
| "learning_rate": 2.6920344208716014e-05, |
| "loss": 0.3129, |
| "step": 552 |
| }, |
| { |
| "epoch": 2.24, |
| "grad_norm": 0.4389385897225364, |
| "learning_rate": 2.6867063825069252e-05, |
| "loss": 0.3434, |
| "step": 553 |
| }, |
| { |
| "epoch": 2.2440506329113923, |
| "grad_norm": 0.48365280365624486, |
| "learning_rate": 2.6813728135034494e-05, |
| "loss": 0.3391, |
| "step": 554 |
| }, |
| { |
| "epoch": 2.248101265822785, |
| "grad_norm": 0.39155884567401084, |
| "learning_rate": 2.6760337568170056e-05, |
| "loss": 0.3606, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.2521518987341773, |
| "grad_norm": 0.43719693847153923, |
| "learning_rate": 2.6706892554476226e-05, |
| "loss": 0.3274, |
| "step": 556 |
| }, |
| { |
| "epoch": 2.2562025316455694, |
| "grad_norm": 0.431091720563098, |
| "learning_rate": 2.6653393524391795e-05, |
| "loss": 0.3439, |
| "step": 557 |
| }, |
| { |
| "epoch": 2.260253164556962, |
| "grad_norm": 0.4350664392892626, |
| "learning_rate": 2.6599840908790592e-05, |
| "loss": 0.3475, |
| "step": 558 |
| }, |
| { |
| "epoch": 2.2643037974683544, |
| "grad_norm": 0.4703442530353007, |
| "learning_rate": 2.6546235138978028e-05, |
| "loss": 0.3564, |
| "step": 559 |
| }, |
| { |
| "epoch": 2.268354430379747, |
| "grad_norm": 0.4808205012991322, |
| "learning_rate": 2.6492576646687597e-05, |
| "loss": 0.3288, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.2724050632911394, |
| "grad_norm": 0.4371691232950181, |
| "learning_rate": 2.6438865864077425e-05, |
| "loss": 0.3355, |
| "step": 561 |
| }, |
| { |
| "epoch": 2.2764556962025315, |
| "grad_norm": 0.4237622440154571, |
| "learning_rate": 2.6385103223726766e-05, |
| "loss": 0.3538, |
| "step": 562 |
| }, |
| { |
| "epoch": 2.280506329113924, |
| "grad_norm": 0.4186113972751265, |
| "learning_rate": 2.6331289158632537e-05, |
| "loss": 0.3558, |
| "step": 563 |
| }, |
| { |
| "epoch": 2.2845569620253166, |
| "grad_norm": 0.44030891021049995, |
| "learning_rate": 2.6277424102205817e-05, |
| "loss": 0.3345, |
| "step": 564 |
| }, |
| { |
| "epoch": 2.2886075949367086, |
| "grad_norm": 0.39212994134082896, |
| "learning_rate": 2.6223508488268374e-05, |
| "loss": 0.3468, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.292658227848101, |
| "grad_norm": 0.40282128210133955, |
| "learning_rate": 2.6169542751049148e-05, |
| "loss": 0.3447, |
| "step": 566 |
| }, |
| { |
| "epoch": 2.2967088607594937, |
| "grad_norm": 0.41008174104074724, |
| "learning_rate": 2.6115527325180754e-05, |
| "loss": 0.3392, |
| "step": 567 |
| }, |
| { |
| "epoch": 2.300759493670886, |
| "grad_norm": 0.41312984666077796, |
| "learning_rate": 2.606146264569603e-05, |
| "loss": 0.3518, |
| "step": 568 |
| }, |
| { |
| "epoch": 2.3048101265822787, |
| "grad_norm": 0.44013475765865645, |
| "learning_rate": 2.6007349148024447e-05, |
| "loss": 0.3495, |
| "step": 569 |
| }, |
| { |
| "epoch": 2.3088607594936708, |
| "grad_norm": 0.40599207058910863, |
| "learning_rate": 2.5953187267988694e-05, |
| "loss": 0.327, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.3129113924050633, |
| "grad_norm": 0.4398348228572484, |
| "learning_rate": 2.5898977441801097e-05, |
| "loss": 0.3306, |
| "step": 571 |
| }, |
| { |
| "epoch": 2.316962025316456, |
| "grad_norm": 0.41934547845118686, |
| "learning_rate": 2.584472010606015e-05, |
| "loss": 0.3475, |
| "step": 572 |
| }, |
| { |
| "epoch": 2.321012658227848, |
| "grad_norm": 0.4100501614070898, |
| "learning_rate": 2.5790415697746976e-05, |
| "loss": 0.3408, |
| "step": 573 |
| }, |
| { |
| "epoch": 2.3250632911392404, |
| "grad_norm": 0.39128576084508426, |
| "learning_rate": 2.5736064654221808e-05, |
| "loss": 0.339, |
| "step": 574 |
| }, |
| { |
| "epoch": 2.329113924050633, |
| "grad_norm": 0.43813790274584125, |
| "learning_rate": 2.568166741322048e-05, |
| "loss": 0.3563, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.3331645569620254, |
| "grad_norm": 0.380803535240516, |
| "learning_rate": 2.56272244128509e-05, |
| "loss": 0.3286, |
| "step": 576 |
| }, |
| { |
| "epoch": 2.337215189873418, |
| "grad_norm": 0.45073455453655853, |
| "learning_rate": 2.55727360915895e-05, |
| "loss": 0.3385, |
| "step": 577 |
| }, |
| { |
| "epoch": 2.34126582278481, |
| "grad_norm": 0.4085623611034995, |
| "learning_rate": 2.5518202888277734e-05, |
| "loss": 0.3373, |
| "step": 578 |
| }, |
| { |
| "epoch": 2.3453164556962025, |
| "grad_norm": 0.40453110432224704, |
| "learning_rate": 2.5463625242118523e-05, |
| "loss": 0.3384, |
| "step": 579 |
| }, |
| { |
| "epoch": 2.349367088607595, |
| "grad_norm": 0.3857888906141202, |
| "learning_rate": 2.5409003592672723e-05, |
| "loss": 0.3503, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.353417721518987, |
| "grad_norm": 0.40165274394356887, |
| "learning_rate": 2.535433837985559e-05, |
| "loss": 0.3333, |
| "step": 581 |
| }, |
| { |
| "epoch": 2.3574683544303796, |
| "grad_norm": 0.4161740059306817, |
| "learning_rate": 2.529963004393324e-05, |
| "loss": 0.341, |
| "step": 582 |
| }, |
| { |
| "epoch": 2.361518987341772, |
| "grad_norm": 0.4018780768093404, |
| "learning_rate": 2.524487902551908e-05, |
| "loss": 0.3439, |
| "step": 583 |
| }, |
| { |
| "epoch": 2.3655696202531646, |
| "grad_norm": 0.4189164916276957, |
| "learning_rate": 2.519008576557029e-05, |
| "loss": 0.3388, |
| "step": 584 |
| }, |
| { |
| "epoch": 2.369620253164557, |
| "grad_norm": 0.4108658751975085, |
| "learning_rate": 2.5135250705384254e-05, |
| "loss": 0.3411, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.3736708860759492, |
| "grad_norm": 0.43531449098882663, |
| "learning_rate": 2.5080374286595007e-05, |
| "loss": 0.3447, |
| "step": 586 |
| }, |
| { |
| "epoch": 2.3777215189873417, |
| "grad_norm": 0.39362822706723855, |
| "learning_rate": 2.5025456951169677e-05, |
| "loss": 0.336, |
| "step": 587 |
| }, |
| { |
| "epoch": 2.3817721518987343, |
| "grad_norm": 0.4280403321716144, |
| "learning_rate": 2.4970499141404942e-05, |
| "loss": 0.3412, |
| "step": 588 |
| }, |
| { |
| "epoch": 2.3858227848101268, |
| "grad_norm": 0.40419006669193436, |
| "learning_rate": 2.491550129992345e-05, |
| "loss": 0.3516, |
| "step": 589 |
| }, |
| { |
| "epoch": 2.389873417721519, |
| "grad_norm": 0.39015374818988424, |
| "learning_rate": 2.486046386967024e-05, |
| "loss": 0.3351, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.3939240506329114, |
| "grad_norm": 0.37862468567023366, |
| "learning_rate": 2.4805387293909214e-05, |
| "loss": 0.3369, |
| "step": 591 |
| }, |
| { |
| "epoch": 2.397974683544304, |
| "grad_norm": 0.3746769778028936, |
| "learning_rate": 2.4750272016219552e-05, |
| "loss": 0.3489, |
| "step": 592 |
| }, |
| { |
| "epoch": 2.4020253164556964, |
| "grad_norm": 0.3978581194927632, |
| "learning_rate": 2.4695118480492114e-05, |
| "loss": 0.3486, |
| "step": 593 |
| }, |
| { |
| "epoch": 2.4060759493670885, |
| "grad_norm": 0.38454781035219093, |
| "learning_rate": 2.4639927130925898e-05, |
| "loss": 0.3456, |
| "step": 594 |
| }, |
| { |
| "epoch": 2.410126582278481, |
| "grad_norm": 0.45437302919851846, |
| "learning_rate": 2.458469841202444e-05, |
| "loss": 0.3523, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.4141772151898735, |
| "grad_norm": 0.39352405699904897, |
| "learning_rate": 2.452943276859226e-05, |
| "loss": 0.3318, |
| "step": 596 |
| }, |
| { |
| "epoch": 2.418227848101266, |
| "grad_norm": 0.4201986883982845, |
| "learning_rate": 2.447413064573125e-05, |
| "loss": 0.3461, |
| "step": 597 |
| }, |
| { |
| "epoch": 2.422278481012658, |
| "grad_norm": 0.3967942929358372, |
| "learning_rate": 2.4418792488837095e-05, |
| "loss": 0.3336, |
| "step": 598 |
| }, |
| { |
| "epoch": 2.4263291139240506, |
| "grad_norm": 0.42327982878665554, |
| "learning_rate": 2.4363418743595713e-05, |
| "loss": 0.3226, |
| "step": 599 |
| }, |
| { |
| "epoch": 2.430379746835443, |
| "grad_norm": 0.3977034526977596, |
| "learning_rate": 2.430800985597963e-05, |
| "loss": 0.3537, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4344303797468356, |
| "grad_norm": 0.4020554412959414, |
| "learning_rate": 2.4252566272244415e-05, |
| "loss": 0.337, |
| "step": 601 |
| }, |
| { |
| "epoch": 2.4384810126582277, |
| "grad_norm": 0.3773320119773115, |
| "learning_rate": 2.4197088438925063e-05, |
| "loss": 0.3391, |
| "step": 602 |
| }, |
| { |
| "epoch": 2.44253164556962, |
| "grad_norm": 0.36880955448455754, |
| "learning_rate": 2.4141576802832417e-05, |
| "loss": 0.3342, |
| "step": 603 |
| }, |
| { |
| "epoch": 2.4465822784810127, |
| "grad_norm": 0.40093757334060226, |
| "learning_rate": 2.408603181104957e-05, |
| "loss": 0.3376, |
| "step": 604 |
| }, |
| { |
| "epoch": 2.4506329113924052, |
| "grad_norm": 0.3908714538446605, |
| "learning_rate": 2.4030453910928245e-05, |
| "loss": 0.3486, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.4546835443037973, |
| "grad_norm": 0.37901789012124065, |
| "learning_rate": 2.397484355008521e-05, |
| "loss": 0.3325, |
| "step": 606 |
| }, |
| { |
| "epoch": 2.45873417721519, |
| "grad_norm": 0.38030636390582684, |
| "learning_rate": 2.3919201176398662e-05, |
| "loss": 0.3604, |
| "step": 607 |
| }, |
| { |
| "epoch": 2.4627848101265823, |
| "grad_norm": 0.41596144406665636, |
| "learning_rate": 2.3863527238004633e-05, |
| "loss": 0.3451, |
| "step": 608 |
| }, |
| { |
| "epoch": 2.466835443037975, |
| "grad_norm": 0.40789590313588525, |
| "learning_rate": 2.380782218329337e-05, |
| "loss": 0.3397, |
| "step": 609 |
| }, |
| { |
| "epoch": 2.470886075949367, |
| "grad_norm": 0.4261662605216207, |
| "learning_rate": 2.3752086460905725e-05, |
| "loss": 0.3475, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.4749367088607594, |
| "grad_norm": 0.4091835365200391, |
| "learning_rate": 2.3696320519729544e-05, |
| "loss": 0.3421, |
| "step": 611 |
| }, |
| { |
| "epoch": 2.478987341772152, |
| "grad_norm": 0.41166142166480124, |
| "learning_rate": 2.3640524808896045e-05, |
| "loss": 0.3551, |
| "step": 612 |
| }, |
| { |
| "epoch": 2.4830379746835445, |
| "grad_norm": 0.38132702937622376, |
| "learning_rate": 2.3584699777776222e-05, |
| "loss": 0.3481, |
| "step": 613 |
| }, |
| { |
| "epoch": 2.4870886075949366, |
| "grad_norm": 0.4582273828158824, |
| "learning_rate": 2.3528845875977195e-05, |
| "loss": 0.3428, |
| "step": 614 |
| }, |
| { |
| "epoch": 2.491139240506329, |
| "grad_norm": 0.4189136878522859, |
| "learning_rate": 2.3472963553338614e-05, |
| "loss": 0.3485, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.4951898734177216, |
| "grad_norm": 0.38157229358583883, |
| "learning_rate": 2.341705325992901e-05, |
| "loss": 0.3532, |
| "step": 616 |
| }, |
| { |
| "epoch": 2.499240506329114, |
| "grad_norm": 0.3921625257075792, |
| "learning_rate": 2.336111544604222e-05, |
| "loss": 0.331, |
| "step": 617 |
| }, |
| { |
| "epoch": 2.503291139240506, |
| "grad_norm": 0.38848410468871447, |
| "learning_rate": 2.33051505621937e-05, |
| "loss": 0.3523, |
| "step": 618 |
| }, |
| { |
| "epoch": 2.5073417721518987, |
| "grad_norm": 0.4055022585677755, |
| "learning_rate": 2.324915905911693e-05, |
| "loss": 0.3339, |
| "step": 619 |
| }, |
| { |
| "epoch": 2.511392405063291, |
| "grad_norm": 0.39016424621253637, |
| "learning_rate": 2.319314138775977e-05, |
| "loss": 0.3227, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.5154430379746833, |
| "grad_norm": 0.38942678406244824, |
| "learning_rate": 2.3137097999280856e-05, |
| "loss": 0.3521, |
| "step": 621 |
| }, |
| { |
| "epoch": 2.519493670886076, |
| "grad_norm": 0.3830316218381617, |
| "learning_rate": 2.308102934504593e-05, |
| "loss": 0.3493, |
| "step": 622 |
| }, |
| { |
| "epoch": 2.5235443037974683, |
| "grad_norm": 0.40959344646529927, |
| "learning_rate": 2.3024935876624222e-05, |
| "loss": 0.3109, |
| "step": 623 |
| }, |
| { |
| "epoch": 2.527594936708861, |
| "grad_norm": 0.3674484558088014, |
| "learning_rate": 2.2968818045784813e-05, |
| "loss": 0.342, |
| "step": 624 |
| }, |
| { |
| "epoch": 2.5316455696202533, |
| "grad_norm": 0.36394300660879714, |
| "learning_rate": 2.2912676304493006e-05, |
| "loss": 0.3368, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.5356962025316454, |
| "grad_norm": 0.39628935901706225, |
| "learning_rate": 2.2856511104906668e-05, |
| "loss": 0.3479, |
| "step": 626 |
| }, |
| { |
| "epoch": 2.539746835443038, |
| "grad_norm": 0.37135624867667694, |
| "learning_rate": 2.2800322899372586e-05, |
| "loss": 0.3307, |
| "step": 627 |
| }, |
| { |
| "epoch": 2.5437974683544304, |
| "grad_norm": 0.3572821230624279, |
| "learning_rate": 2.2744112140422844e-05, |
| "loss": 0.349, |
| "step": 628 |
| }, |
| { |
| "epoch": 2.547848101265823, |
| "grad_norm": 0.3941431166382233, |
| "learning_rate": 2.2687879280771177e-05, |
| "loss": 0.3393, |
| "step": 629 |
| }, |
| { |
| "epoch": 2.5518987341772155, |
| "grad_norm": 0.35258037288239147, |
| "learning_rate": 2.26316247733093e-05, |
| "loss": 0.3357, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.5559493670886075, |
| "grad_norm": 0.39392467192752406, |
| "learning_rate": 2.257534907110328e-05, |
| "loss": 0.3451, |
| "step": 631 |
| }, |
| { |
| "epoch": 2.56, |
| "grad_norm": 0.4168007101967973, |
| "learning_rate": 2.2519052627389882e-05, |
| "loss": 0.3459, |
| "step": 632 |
| }, |
| { |
| "epoch": 2.5640506329113926, |
| "grad_norm": 0.4148172503884062, |
| "learning_rate": 2.246273589557294e-05, |
| "loss": 0.3265, |
| "step": 633 |
| }, |
| { |
| "epoch": 2.5681012658227846, |
| "grad_norm": 0.3949677171082751, |
| "learning_rate": 2.240639932921966e-05, |
| "loss": 0.3322, |
| "step": 634 |
| }, |
| { |
| "epoch": 2.572151898734177, |
| "grad_norm": 0.4564028958751634, |
| "learning_rate": 2.2350043382056995e-05, |
| "loss": 0.3498, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.5762025316455697, |
| "grad_norm": 0.40697901428756716, |
| "learning_rate": 2.2293668507968015e-05, |
| "loss": 0.3263, |
| "step": 636 |
| }, |
| { |
| "epoch": 2.580253164556962, |
| "grad_norm": 0.48896575747946563, |
| "learning_rate": 2.2237275160988186e-05, |
| "loss": 0.3464, |
| "step": 637 |
| }, |
| { |
| "epoch": 2.5843037974683547, |
| "grad_norm": 0.4262315940164504, |
| "learning_rate": 2.2180863795301787e-05, |
| "loss": 0.3431, |
| "step": 638 |
| }, |
| { |
| "epoch": 2.5883544303797468, |
| "grad_norm": 0.402781586069299, |
| "learning_rate": 2.212443486523819e-05, |
| "loss": 0.3488, |
| "step": 639 |
| }, |
| { |
| "epoch": 2.5924050632911393, |
| "grad_norm": 0.4380183914742864, |
| "learning_rate": 2.2067988825268243e-05, |
| "loss": 0.3626, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.596455696202532, |
| "grad_norm": 0.38795894474690934, |
| "learning_rate": 2.2011526130000596e-05, |
| "loss": 0.3343, |
| "step": 641 |
| }, |
| { |
| "epoch": 2.600506329113924, |
| "grad_norm": 0.4164707018565618, |
| "learning_rate": 2.1955047234178038e-05, |
| "loss": 0.3233, |
| "step": 642 |
| }, |
| { |
| "epoch": 2.6045569620253164, |
| "grad_norm": 0.3484070346588608, |
| "learning_rate": 2.1898552592673825e-05, |
| "loss": 0.3482, |
| "step": 643 |
| }, |
| { |
| "epoch": 2.608607594936709, |
| "grad_norm": 0.42607341688085754, |
| "learning_rate": 2.184204266048803e-05, |
| "loss": 0.3471, |
| "step": 644 |
| }, |
| { |
| "epoch": 2.6126582278481014, |
| "grad_norm": 0.3554847625436213, |
| "learning_rate": 2.1785517892743887e-05, |
| "loss": 0.3543, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.616708860759494, |
| "grad_norm": 0.390620885167572, |
| "learning_rate": 2.17289787446841e-05, |
| "loss": 0.3384, |
| "step": 646 |
| }, |
| { |
| "epoch": 2.620759493670886, |
| "grad_norm": 0.43049920950313225, |
| "learning_rate": 2.1672425671667198e-05, |
| "loss": 0.3405, |
| "step": 647 |
| }, |
| { |
| "epoch": 2.6248101265822785, |
| "grad_norm": 0.3779985585268527, |
| "learning_rate": 2.161585912916385e-05, |
| "loss": 0.3405, |
| "step": 648 |
| }, |
| { |
| "epoch": 2.628860759493671, |
| "grad_norm": 0.3921412276604478, |
| "learning_rate": 2.1559279572753214e-05, |
| "loss": 0.3497, |
| "step": 649 |
| }, |
| { |
| "epoch": 2.632911392405063, |
| "grad_norm": 0.3565401261757519, |
| "learning_rate": 2.1502687458119268e-05, |
| "loss": 0.341, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.6369620253164556, |
| "grad_norm": 0.4197856604797413, |
| "learning_rate": 2.1446083241047116e-05, |
| "loss": 0.3386, |
| "step": 651 |
| }, |
| { |
| "epoch": 2.641012658227848, |
| "grad_norm": 0.37336954675867967, |
| "learning_rate": 2.1389467377419333e-05, |
| "loss": 0.3314, |
| "step": 652 |
| }, |
| { |
| "epoch": 2.6450632911392407, |
| "grad_norm": 0.36103070360821227, |
| "learning_rate": 2.133284032321232e-05, |
| "loss": 0.3308, |
| "step": 653 |
| }, |
| { |
| "epoch": 2.649113924050633, |
| "grad_norm": 0.4028062620270954, |
| "learning_rate": 2.1276202534492566e-05, |
| "loss": 0.3209, |
| "step": 654 |
| }, |
| { |
| "epoch": 2.6531645569620252, |
| "grad_norm": 0.3790262001986638, |
| "learning_rate": 2.121955446741306e-05, |
| "loss": 0.3399, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.6572151898734178, |
| "grad_norm": 0.39340641714335095, |
| "learning_rate": 2.1162896578209517e-05, |
| "loss": 0.3395, |
| "step": 656 |
| }, |
| { |
| "epoch": 2.6612658227848103, |
| "grad_norm": 0.37725030852920877, |
| "learning_rate": 2.1106229323196813e-05, |
| "loss": 0.3452, |
| "step": 657 |
| }, |
| { |
| "epoch": 2.6653164556962023, |
| "grad_norm": 0.39807892096171954, |
| "learning_rate": 2.1049553158765214e-05, |
| "loss": 0.3039, |
| "step": 658 |
| }, |
| { |
| "epoch": 2.669367088607595, |
| "grad_norm": 0.41337253272157104, |
| "learning_rate": 2.0992868541376764e-05, |
| "loss": 0.3454, |
| "step": 659 |
| }, |
| { |
| "epoch": 2.6734177215189874, |
| "grad_norm": 0.37532099558478466, |
| "learning_rate": 2.093617592756158e-05, |
| "loss": 0.3422, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.67746835443038, |
| "grad_norm": 0.40376960865656075, |
| "learning_rate": 2.0879475773914167e-05, |
| "loss": 0.3469, |
| "step": 661 |
| }, |
| { |
| "epoch": 2.6815189873417724, |
| "grad_norm": 0.3544164440446397, |
| "learning_rate": 2.082276853708978e-05, |
| "loss": 0.3327, |
| "step": 662 |
| }, |
| { |
| "epoch": 2.6855696202531645, |
| "grad_norm": 0.40517600161675615, |
| "learning_rate": 2.076605467380071e-05, |
| "loss": 0.3363, |
| "step": 663 |
| }, |
| { |
| "epoch": 2.689620253164557, |
| "grad_norm": 0.36903753419369967, |
| "learning_rate": 2.0709334640812613e-05, |
| "loss": 0.3344, |
| "step": 664 |
| }, |
| { |
| "epoch": 2.6936708860759495, |
| "grad_norm": 0.41716013545429104, |
| "learning_rate": 2.0652608894940824e-05, |
| "loss": 0.3395, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.6977215189873416, |
| "grad_norm": 0.368704209168687, |
| "learning_rate": 2.0595877893046722e-05, |
| "loss": 0.3327, |
| "step": 666 |
| }, |
| { |
| "epoch": 2.701772151898734, |
| "grad_norm": 0.385114890005095, |
| "learning_rate": 2.0539142092033985e-05, |
| "loss": 0.3386, |
| "step": 667 |
| }, |
| { |
| "epoch": 2.7058227848101266, |
| "grad_norm": 0.3618221634577607, |
| "learning_rate": 2.048240194884496e-05, |
| "loss": 0.323, |
| "step": 668 |
| }, |
| { |
| "epoch": 2.709873417721519, |
| "grad_norm": 0.3529259345423914, |
| "learning_rate": 2.042565792045695e-05, |
| "loss": 0.3239, |
| "step": 669 |
| }, |
| { |
| "epoch": 2.7139240506329116, |
| "grad_norm": 0.3781609375643004, |
| "learning_rate": 2.036891046387857e-05, |
| "loss": 0.3282, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.7179746835443037, |
| "grad_norm": 0.3853209212924312, |
| "learning_rate": 2.0312160036146036e-05, |
| "loss": 0.3696, |
| "step": 671 |
| }, |
| { |
| "epoch": 2.7220253164556962, |
| "grad_norm": 0.37803571790981827, |
| "learning_rate": 2.025540709431948e-05, |
| "loss": 0.3413, |
| "step": 672 |
| }, |
| { |
| "epoch": 2.7260759493670887, |
| "grad_norm": 0.3745187077446651, |
| "learning_rate": 2.0198652095479298e-05, |
| "loss": 0.3478, |
| "step": 673 |
| }, |
| { |
| "epoch": 2.730126582278481, |
| "grad_norm": 0.4058673009163546, |
| "learning_rate": 2.014189549672245e-05, |
| "loss": 0.343, |
| "step": 674 |
| }, |
| { |
| "epoch": 2.7341772151898733, |
| "grad_norm": 0.37164874089122135, |
| "learning_rate": 2.0085137755158776e-05, |
| "loss": 0.3351, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.738227848101266, |
| "grad_norm": 0.38167558425253034, |
| "learning_rate": 2.0028379327907327e-05, |
| "loss": 0.3313, |
| "step": 676 |
| }, |
| { |
| "epoch": 2.7422784810126584, |
| "grad_norm": 0.41050699910234, |
| "learning_rate": 1.9971620672092676e-05, |
| "loss": 0.3408, |
| "step": 677 |
| }, |
| { |
| "epoch": 2.746329113924051, |
| "grad_norm": 0.36222371472895654, |
| "learning_rate": 1.991486224484123e-05, |
| "loss": 0.3442, |
| "step": 678 |
| }, |
| { |
| "epoch": 2.750379746835443, |
| "grad_norm": 0.36273607504756344, |
| "learning_rate": 1.985810450327756e-05, |
| "loss": 0.3365, |
| "step": 679 |
| }, |
| { |
| "epoch": 2.7544303797468355, |
| "grad_norm": 0.3735462980523313, |
| "learning_rate": 1.9801347904520706e-05, |
| "loss": 0.354, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.758481012658228, |
| "grad_norm": 0.36316464653684793, |
| "learning_rate": 1.974459290568053e-05, |
| "loss": 0.3399, |
| "step": 681 |
| }, |
| { |
| "epoch": 2.76253164556962, |
| "grad_norm": 0.39573551837419124, |
| "learning_rate": 1.968783996385397e-05, |
| "loss": 0.3713, |
| "step": 682 |
| }, |
| { |
| "epoch": 2.7665822784810126, |
| "grad_norm": 0.39293430120128897, |
| "learning_rate": 1.963108953612143e-05, |
| "loss": 0.3281, |
| "step": 683 |
| }, |
| { |
| "epoch": 2.770632911392405, |
| "grad_norm": 0.37638047081437903, |
| "learning_rate": 1.9574342079543056e-05, |
| "loss": 0.3453, |
| "step": 684 |
| }, |
| { |
| "epoch": 2.7746835443037976, |
| "grad_norm": 0.45367045259868255, |
| "learning_rate": 1.9517598051155046e-05, |
| "loss": 0.3449, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.77873417721519, |
| "grad_norm": 0.40136308601954457, |
| "learning_rate": 1.9460857907966025e-05, |
| "loss": 0.3234, |
| "step": 686 |
| }, |
| { |
| "epoch": 2.782784810126582, |
| "grad_norm": 0.4223215702056556, |
| "learning_rate": 1.9404122106953285e-05, |
| "loss": 0.3416, |
| "step": 687 |
| }, |
| { |
| "epoch": 2.7868354430379747, |
| "grad_norm": 0.4285569256003927, |
| "learning_rate": 1.9347391105059176e-05, |
| "loss": 0.3294, |
| "step": 688 |
| }, |
| { |
| "epoch": 2.790886075949367, |
| "grad_norm": 0.3656122006796164, |
| "learning_rate": 1.92906653591874e-05, |
| "loss": 0.3409, |
| "step": 689 |
| }, |
| { |
| "epoch": 2.7949367088607593, |
| "grad_norm": 0.3986824064259397, |
| "learning_rate": 1.9233945326199295e-05, |
| "loss": 0.3661, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.798987341772152, |
| "grad_norm": 0.40591471408867597, |
| "learning_rate": 1.917723146291022e-05, |
| "loss": 0.3548, |
| "step": 691 |
| }, |
| { |
| "epoch": 2.8030379746835443, |
| "grad_norm": 0.3738119266413488, |
| "learning_rate": 1.912052422608584e-05, |
| "loss": 0.3344, |
| "step": 692 |
| }, |
| { |
| "epoch": 2.807088607594937, |
| "grad_norm": 0.4055437325850107, |
| "learning_rate": 1.9063824072438428e-05, |
| "loss": 0.3525, |
| "step": 693 |
| }, |
| { |
| "epoch": 2.8111392405063294, |
| "grad_norm": 0.4114351939302134, |
| "learning_rate": 1.9007131458623246e-05, |
| "loss": 0.3352, |
| "step": 694 |
| }, |
| { |
| "epoch": 2.8151898734177214, |
| "grad_norm": 0.38442929428935146, |
| "learning_rate": 1.895044684123479e-05, |
| "loss": 0.3448, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.819240506329114, |
| "grad_norm": 0.3805036706884265, |
| "learning_rate": 1.8893770676803194e-05, |
| "loss": 0.3375, |
| "step": 696 |
| }, |
| { |
| "epoch": 2.8232911392405065, |
| "grad_norm": 0.37402485852165546, |
| "learning_rate": 1.8837103421790486e-05, |
| "loss": 0.3397, |
| "step": 697 |
| }, |
| { |
| "epoch": 2.8273417721518985, |
| "grad_norm": 0.37631462462086723, |
| "learning_rate": 1.8780445532586952e-05, |
| "loss": 0.345, |
| "step": 698 |
| }, |
| { |
| "epoch": 2.831392405063291, |
| "grad_norm": 0.36137180188740575, |
| "learning_rate": 1.872379746550743e-05, |
| "loss": 0.3272, |
| "step": 699 |
| }, |
| { |
| "epoch": 2.8354430379746836, |
| "grad_norm": 0.39988521378052383, |
| "learning_rate": 1.866715967678769e-05, |
| "loss": 0.3296, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.839493670886076, |
| "grad_norm": 0.39519625321056384, |
| "learning_rate": 1.861053262258067e-05, |
| "loss": 0.3388, |
| "step": 701 |
| }, |
| { |
| "epoch": 2.8435443037974686, |
| "grad_norm": 0.4038056334040143, |
| "learning_rate": 1.8553916758952897e-05, |
| "loss": 0.3434, |
| "step": 702 |
| }, |
| { |
| "epoch": 2.8475949367088607, |
| "grad_norm": 0.3745692801845034, |
| "learning_rate": 1.8497312541880735e-05, |
| "loss": 0.3221, |
| "step": 703 |
| }, |
| { |
| "epoch": 2.851645569620253, |
| "grad_norm": 0.37528612409797, |
| "learning_rate": 1.8440720427246786e-05, |
| "loss": 0.3419, |
| "step": 704 |
| }, |
| { |
| "epoch": 2.8556962025316457, |
| "grad_norm": 0.3784025357601745, |
| "learning_rate": 1.8384140870836157e-05, |
| "loss": 0.3366, |
| "step": 705 |
| }, |
| { |
| "epoch": 2.8597468354430378, |
| "grad_norm": 0.375743754196614, |
| "learning_rate": 1.8327574328332806e-05, |
| "loss": 0.3325, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.8637974683544303, |
| "grad_norm": 0.37252992951527647, |
| "learning_rate": 1.8271021255315906e-05, |
| "loss": 0.3426, |
| "step": 707 |
| }, |
| { |
| "epoch": 2.867848101265823, |
| "grad_norm": 0.3828969576045123, |
| "learning_rate": 1.8214482107256117e-05, |
| "loss": 0.3434, |
| "step": 708 |
| }, |
| { |
| "epoch": 2.8718987341772153, |
| "grad_norm": 0.35415766037984503, |
| "learning_rate": 1.8157957339511968e-05, |
| "loss": 0.3509, |
| "step": 709 |
| }, |
| { |
| "epoch": 2.875949367088608, |
| "grad_norm": 0.3909679487046648, |
| "learning_rate": 1.8101447407326182e-05, |
| "loss": 0.3136, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.88, |
| "grad_norm": 0.36233820301991504, |
| "learning_rate": 1.8044952765821966e-05, |
| "loss": 0.3204, |
| "step": 711 |
| }, |
| { |
| "epoch": 2.8840506329113924, |
| "grad_norm": 0.3753105866899036, |
| "learning_rate": 1.7988473869999407e-05, |
| "loss": 0.3326, |
| "step": 712 |
| }, |
| { |
| "epoch": 2.888101265822785, |
| "grad_norm": 0.3870877384124527, |
| "learning_rate": 1.7932011174731764e-05, |
| "loss": 0.3263, |
| "step": 713 |
| }, |
| { |
| "epoch": 2.892151898734177, |
| "grad_norm": 0.34181594470916243, |
| "learning_rate": 1.7875565134761817e-05, |
| "loss": 0.3425, |
| "step": 714 |
| }, |
| { |
| "epoch": 2.8962025316455695, |
| "grad_norm": 0.35825809356821087, |
| "learning_rate": 1.7819136204698226e-05, |
| "loss": 0.3395, |
| "step": 715 |
| }, |
| { |
| "epoch": 2.900253164556962, |
| "grad_norm": 0.3857580855304102, |
| "learning_rate": 1.776272483901182e-05, |
| "loss": 0.3316, |
| "step": 716 |
| }, |
| { |
| "epoch": 2.9043037974683545, |
| "grad_norm": 0.3712213523777954, |
| "learning_rate": 1.7706331492031995e-05, |
| "loss": 0.3401, |
| "step": 717 |
| }, |
| { |
| "epoch": 2.908354430379747, |
| "grad_norm": 0.37888276861921644, |
| "learning_rate": 1.764995661794301e-05, |
| "loss": 0.3237, |
| "step": 718 |
| }, |
| { |
| "epoch": 2.912405063291139, |
| "grad_norm": 0.38605349982904225, |
| "learning_rate": 1.759360067078035e-05, |
| "loss": 0.3463, |
| "step": 719 |
| }, |
| { |
| "epoch": 2.9164556962025316, |
| "grad_norm": 0.37274825066397815, |
| "learning_rate": 1.7537264104427064e-05, |
| "loss": 0.3345, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.920506329113924, |
| "grad_norm": 0.40621618277123606, |
| "learning_rate": 1.748094737261012e-05, |
| "loss": 0.335, |
| "step": 721 |
| }, |
| { |
| "epoch": 2.9245569620253162, |
| "grad_norm": 0.3794675600607056, |
| "learning_rate": 1.7424650928896726e-05, |
| "loss": 0.3446, |
| "step": 722 |
| }, |
| { |
| "epoch": 2.9286075949367087, |
| "grad_norm": 0.34905996786517873, |
| "learning_rate": 1.7368375226690712e-05, |
| "loss": 0.3428, |
| "step": 723 |
| }, |
| { |
| "epoch": 2.9326582278481013, |
| "grad_norm": 0.37594009359888514, |
| "learning_rate": 1.731212071922883e-05, |
| "loss": 0.3266, |
| "step": 724 |
| }, |
| { |
| "epoch": 2.9367088607594938, |
| "grad_norm": 0.3631844405314312, |
| "learning_rate": 1.7255887859577156e-05, |
| "loss": 0.3394, |
| "step": 725 |
| }, |
| { |
| "epoch": 2.9407594936708863, |
| "grad_norm": 0.3825799063124562, |
| "learning_rate": 1.7199677100627427e-05, |
| "loss": 0.3373, |
| "step": 726 |
| }, |
| { |
| "epoch": 2.9448101265822784, |
| "grad_norm": 0.3933700059147295, |
| "learning_rate": 1.7143488895093343e-05, |
| "loss": 0.3223, |
| "step": 727 |
| }, |
| { |
| "epoch": 2.948860759493671, |
| "grad_norm": 0.36948281962698015, |
| "learning_rate": 1.7087323695506994e-05, |
| "loss": 0.3282, |
| "step": 728 |
| }, |
| { |
| "epoch": 2.9529113924050634, |
| "grad_norm": 0.36216008051658977, |
| "learning_rate": 1.7031181954215194e-05, |
| "loss": 0.3372, |
| "step": 729 |
| }, |
| { |
| "epoch": 2.9569620253164555, |
| "grad_norm": 0.3478288152326165, |
| "learning_rate": 1.6975064123375788e-05, |
| "loss": 0.3498, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.961012658227848, |
| "grad_norm": 0.35523682350386865, |
| "learning_rate": 1.6918970654954084e-05, |
| "loss": 0.3317, |
| "step": 731 |
| }, |
| { |
| "epoch": 2.9650632911392405, |
| "grad_norm": 0.36219939035122, |
| "learning_rate": 1.686290200071915e-05, |
| "loss": 0.338, |
| "step": 732 |
| }, |
| { |
| "epoch": 2.969113924050633, |
| "grad_norm": 0.3342489437547706, |
| "learning_rate": 1.6806858612240234e-05, |
| "loss": 0.3332, |
| "step": 733 |
| }, |
| { |
| "epoch": 2.9731645569620255, |
| "grad_norm": 0.34725618476340153, |
| "learning_rate": 1.6750840940883078e-05, |
| "loss": 0.3306, |
| "step": 734 |
| }, |
| { |
| "epoch": 2.9772151898734176, |
| "grad_norm": 0.3695142740546524, |
| "learning_rate": 1.6694849437806305e-05, |
| "loss": 0.3574, |
| "step": 735 |
| }, |
| { |
| "epoch": 2.98126582278481, |
| "grad_norm": 0.3392409151684075, |
| "learning_rate": 1.663888455395778e-05, |
| "loss": 0.3252, |
| "step": 736 |
| }, |
| { |
| "epoch": 2.9853164556962026, |
| "grad_norm": 0.3658965142533706, |
| "learning_rate": 1.6582946740070995e-05, |
| "loss": 0.3426, |
| "step": 737 |
| }, |
| { |
| "epoch": 2.9893670886075947, |
| "grad_norm": 0.3631891411944074, |
| "learning_rate": 1.6527036446661396e-05, |
| "loss": 0.3201, |
| "step": 738 |
| }, |
| { |
| "epoch": 2.993417721518987, |
| "grad_norm": 0.34477779134434816, |
| "learning_rate": 1.6471154124022818e-05, |
| "loss": 0.3385, |
| "step": 739 |
| }, |
| { |
| "epoch": 2.9974683544303797, |
| "grad_norm": 0.3524880939165382, |
| "learning_rate": 1.6415300222223788e-05, |
| "loss": 0.3396, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0015189873417722, |
| "grad_norm": 0.3766333836299262, |
| "learning_rate": 1.6359475191103958e-05, |
| "loss": 0.3126, |
| "step": 741 |
| }, |
| { |
| "epoch": 3.0055696202531648, |
| "grad_norm": 0.5175900638681576, |
| "learning_rate": 1.6303679480270466e-05, |
| "loss": 0.2367, |
| "step": 742 |
| }, |
| { |
| "epoch": 3.009620253164557, |
| "grad_norm": 0.34198044550431994, |
| "learning_rate": 1.624791353909428e-05, |
| "loss": 0.2283, |
| "step": 743 |
| }, |
| { |
| "epoch": 3.0136708860759494, |
| "grad_norm": 0.8114858190356592, |
| "learning_rate": 1.619217781670663e-05, |
| "loss": 0.2448, |
| "step": 744 |
| }, |
| { |
| "epoch": 3.017721518987342, |
| "grad_norm": 0.4698108550565904, |
| "learning_rate": 1.6136472761995373e-05, |
| "loss": 0.2138, |
| "step": 745 |
| }, |
| { |
| "epoch": 3.0217721518987344, |
| "grad_norm": 0.43861184061592073, |
| "learning_rate": 1.608079882360134e-05, |
| "loss": 0.2357, |
| "step": 746 |
| }, |
| { |
| "epoch": 3.0258227848101265, |
| "grad_norm": 0.538209440025781, |
| "learning_rate": 1.60251564499148e-05, |
| "loss": 0.2289, |
| "step": 747 |
| }, |
| { |
| "epoch": 3.029873417721519, |
| "grad_norm": 0.4526631588984132, |
| "learning_rate": 1.596954608907176e-05, |
| "loss": 0.2187, |
| "step": 748 |
| }, |
| { |
| "epoch": 3.0339240506329115, |
| "grad_norm": 0.40294837100212194, |
| "learning_rate": 1.591396818895043e-05, |
| "loss": 0.2346, |
| "step": 749 |
| }, |
| { |
| "epoch": 3.037974683544304, |
| "grad_norm": 0.4577002559906199, |
| "learning_rate": 1.585842319716759e-05, |
| "loss": 0.2332, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.042025316455696, |
| "grad_norm": 0.46697408884834923, |
| "learning_rate": 1.5802911561074944e-05, |
| "loss": 0.2343, |
| "step": 751 |
| }, |
| { |
| "epoch": 3.0460759493670886, |
| "grad_norm": 0.3886649730756613, |
| "learning_rate": 1.5747433727755595e-05, |
| "loss": 0.23, |
| "step": 752 |
| }, |
| { |
| "epoch": 3.050126582278481, |
| "grad_norm": 0.3991058739622296, |
| "learning_rate": 1.5691990144020376e-05, |
| "loss": 0.2166, |
| "step": 753 |
| }, |
| { |
| "epoch": 3.0541772151898736, |
| "grad_norm": 0.3804929293597602, |
| "learning_rate": 1.5636581256404297e-05, |
| "loss": 0.2361, |
| "step": 754 |
| }, |
| { |
| "epoch": 3.0582278481012657, |
| "grad_norm": 0.3777064005384867, |
| "learning_rate": 1.558120751116291e-05, |
| "loss": 0.2317, |
| "step": 755 |
| }, |
| { |
| "epoch": 3.062278481012658, |
| "grad_norm": 0.38527719278402417, |
| "learning_rate": 1.552586935426876e-05, |
| "loss": 0.2445, |
| "step": 756 |
| }, |
| { |
| "epoch": 3.0663291139240507, |
| "grad_norm": 0.3461478057932524, |
| "learning_rate": 1.547056723140774e-05, |
| "loss": 0.2115, |
| "step": 757 |
| }, |
| { |
| "epoch": 3.0703797468354432, |
| "grad_norm": 0.3584026332750357, |
| "learning_rate": 1.5415301587975565e-05, |
| "loss": 0.2209, |
| "step": 758 |
| }, |
| { |
| "epoch": 3.0744303797468353, |
| "grad_norm": 0.40733117397117863, |
| "learning_rate": 1.536007286907411e-05, |
| "loss": 0.2504, |
| "step": 759 |
| }, |
| { |
| "epoch": 3.078481012658228, |
| "grad_norm": 0.37570828938747647, |
| "learning_rate": 1.5304881519507896e-05, |
| "loss": 0.2572, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.0825316455696203, |
| "grad_norm": 0.35506651072587103, |
| "learning_rate": 1.5249727983780453e-05, |
| "loss": 0.2217, |
| "step": 761 |
| }, |
| { |
| "epoch": 3.086582278481013, |
| "grad_norm": 0.35435768484290436, |
| "learning_rate": 1.5194612706090786e-05, |
| "loss": 0.232, |
| "step": 762 |
| }, |
| { |
| "epoch": 3.090632911392405, |
| "grad_norm": 0.3366051608775223, |
| "learning_rate": 1.5139536130329771e-05, |
| "loss": 0.2518, |
| "step": 763 |
| }, |
| { |
| "epoch": 3.0946835443037974, |
| "grad_norm": 0.34109262613233643, |
| "learning_rate": 1.508449870007656e-05, |
| "loss": 0.2419, |
| "step": 764 |
| }, |
| { |
| "epoch": 3.09873417721519, |
| "grad_norm": 0.33604835317855336, |
| "learning_rate": 1.5029500858595056e-05, |
| "loss": 0.2165, |
| "step": 765 |
| }, |
| { |
| "epoch": 3.1027848101265825, |
| "grad_norm": 0.3328211251563782, |
| "learning_rate": 1.4974543048830328e-05, |
| "loss": 0.231, |
| "step": 766 |
| }, |
| { |
| "epoch": 3.1068354430379745, |
| "grad_norm": 0.3420956553591052, |
| "learning_rate": 1.4919625713405e-05, |
| "loss": 0.2484, |
| "step": 767 |
| }, |
| { |
| "epoch": 3.110886075949367, |
| "grad_norm": 0.3376285690013058, |
| "learning_rate": 1.4864749294615756e-05, |
| "loss": 0.2217, |
| "step": 768 |
| }, |
| { |
| "epoch": 3.1149367088607596, |
| "grad_norm": 0.34454585059986487, |
| "learning_rate": 1.4809914234429716e-05, |
| "loss": 0.2354, |
| "step": 769 |
| }, |
| { |
| "epoch": 3.118987341772152, |
| "grad_norm": 0.3448339992699649, |
| "learning_rate": 1.4755120974480923e-05, |
| "loss": 0.2238, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.123037974683544, |
| "grad_norm": 0.3344585974762778, |
| "learning_rate": 1.4700369956066771e-05, |
| "loss": 0.2181, |
| "step": 771 |
| }, |
| { |
| "epoch": 3.1270886075949367, |
| "grad_norm": 0.3838524298451627, |
| "learning_rate": 1.4645661620144413e-05, |
| "loss": 0.2321, |
| "step": 772 |
| }, |
| { |
| "epoch": 3.131139240506329, |
| "grad_norm": 0.34076956152677557, |
| "learning_rate": 1.4590996407327284e-05, |
| "loss": 0.2266, |
| "step": 773 |
| }, |
| { |
| "epoch": 3.1351898734177217, |
| "grad_norm": 0.3364422781051453, |
| "learning_rate": 1.4536374757881487e-05, |
| "loss": 0.2305, |
| "step": 774 |
| }, |
| { |
| "epoch": 3.1392405063291138, |
| "grad_norm": 0.3908636616690709, |
| "learning_rate": 1.4481797111722271e-05, |
| "loss": 0.2383, |
| "step": 775 |
| }, |
| { |
| "epoch": 3.1432911392405063, |
| "grad_norm": 0.3314457612544679, |
| "learning_rate": 1.4427263908410507e-05, |
| "loss": 0.2344, |
| "step": 776 |
| }, |
| { |
| "epoch": 3.147341772151899, |
| "grad_norm": 0.3283337928965765, |
| "learning_rate": 1.4372775587149108e-05, |
| "loss": 0.2362, |
| "step": 777 |
| }, |
| { |
| "epoch": 3.1513924050632913, |
| "grad_norm": 0.33618333668604977, |
| "learning_rate": 1.4318332586779522e-05, |
| "loss": 0.2369, |
| "step": 778 |
| }, |
| { |
| "epoch": 3.1554430379746834, |
| "grad_norm": 0.31017478433787976, |
| "learning_rate": 1.4263935345778202e-05, |
| "loss": 0.2314, |
| "step": 779 |
| }, |
| { |
| "epoch": 3.159493670886076, |
| "grad_norm": 0.3214506584893053, |
| "learning_rate": 1.420958430225303e-05, |
| "loss": 0.24, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.1635443037974684, |
| "grad_norm": 0.3359190669633716, |
| "learning_rate": 1.415527989393985e-05, |
| "loss": 0.2375, |
| "step": 781 |
| }, |
| { |
| "epoch": 3.167594936708861, |
| "grad_norm": 0.3246039595813393, |
| "learning_rate": 1.410102255819891e-05, |
| "loss": 0.2472, |
| "step": 782 |
| }, |
| { |
| "epoch": 3.171645569620253, |
| "grad_norm": 0.3308867691164173, |
| "learning_rate": 1.404681273201131e-05, |
| "loss": 0.225, |
| "step": 783 |
| }, |
| { |
| "epoch": 3.1756962025316455, |
| "grad_norm": 0.3185026809984867, |
| "learning_rate": 1.399265085197556e-05, |
| "loss": 0.2448, |
| "step": 784 |
| }, |
| { |
| "epoch": 3.179746835443038, |
| "grad_norm": 0.3398217443655176, |
| "learning_rate": 1.393853735430398e-05, |
| "loss": 0.2382, |
| "step": 785 |
| }, |
| { |
| "epoch": 3.1837974683544306, |
| "grad_norm": 0.35905875420492084, |
| "learning_rate": 1.3884472674819246e-05, |
| "loss": 0.2237, |
| "step": 786 |
| }, |
| { |
| "epoch": 3.1878481012658226, |
| "grad_norm": 0.33466049214427324, |
| "learning_rate": 1.3830457248950864e-05, |
| "loss": 0.2387, |
| "step": 787 |
| }, |
| { |
| "epoch": 3.191898734177215, |
| "grad_norm": 0.32494851599396224, |
| "learning_rate": 1.377649151173163e-05, |
| "loss": 0.226, |
| "step": 788 |
| }, |
| { |
| "epoch": 3.1959493670886077, |
| "grad_norm": 0.34342444818890616, |
| "learning_rate": 1.3722575897794181e-05, |
| "loss": 0.2261, |
| "step": 789 |
| }, |
| { |
| "epoch": 3.2, |
| "grad_norm": 0.3336656402992833, |
| "learning_rate": 1.3668710841367472e-05, |
| "loss": 0.2029, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.2040506329113922, |
| "grad_norm": 0.3221658216404002, |
| "learning_rate": 1.361489677627324e-05, |
| "loss": 0.2247, |
| "step": 791 |
| }, |
| { |
| "epoch": 3.2081012658227848, |
| "grad_norm": 0.34162390105353635, |
| "learning_rate": 1.3561134135922585e-05, |
| "loss": 0.2312, |
| "step": 792 |
| }, |
| { |
| "epoch": 3.2121518987341773, |
| "grad_norm": 0.3259219946782709, |
| "learning_rate": 1.350742335331241e-05, |
| "loss": 0.2226, |
| "step": 793 |
| }, |
| { |
| "epoch": 3.21620253164557, |
| "grad_norm": 0.3334648860476423, |
| "learning_rate": 1.345376486102198e-05, |
| "loss": 0.2313, |
| "step": 794 |
| }, |
| { |
| "epoch": 3.220253164556962, |
| "grad_norm": 0.32437839008267055, |
| "learning_rate": 1.3400159091209414e-05, |
| "loss": 0.2206, |
| "step": 795 |
| }, |
| { |
| "epoch": 3.2243037974683544, |
| "grad_norm": 0.31913986173209824, |
| "learning_rate": 1.3346606475608216e-05, |
| "loss": 0.2272, |
| "step": 796 |
| }, |
| { |
| "epoch": 3.228354430379747, |
| "grad_norm": 0.34607007494610814, |
| "learning_rate": 1.3293107445523781e-05, |
| "loss": 0.2147, |
| "step": 797 |
| }, |
| { |
| "epoch": 3.2324050632911394, |
| "grad_norm": 0.3424114445991121, |
| "learning_rate": 1.3239662431829949e-05, |
| "loss": 0.2212, |
| "step": 798 |
| }, |
| { |
| "epoch": 3.2364556962025315, |
| "grad_norm": 0.32721210753166585, |
| "learning_rate": 1.3186271864965509e-05, |
| "loss": 0.2311, |
| "step": 799 |
| }, |
| { |
| "epoch": 3.240506329113924, |
| "grad_norm": 0.33882724770088657, |
| "learning_rate": 1.3132936174930756e-05, |
| "loss": 0.2361, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.2445569620253165, |
| "grad_norm": 0.3160595796309291, |
| "learning_rate": 1.3079655791283995e-05, |
| "loss": 0.2521, |
| "step": 801 |
| }, |
| { |
| "epoch": 3.248607594936709, |
| "grad_norm": 0.35862229988459915, |
| "learning_rate": 1.3026431143138108e-05, |
| "loss": 0.2514, |
| "step": 802 |
| }, |
| { |
| "epoch": 3.252658227848101, |
| "grad_norm": 0.3163333458601768, |
| "learning_rate": 1.2973262659157114e-05, |
| "loss": 0.2133, |
| "step": 803 |
| }, |
| { |
| "epoch": 3.2567088607594936, |
| "grad_norm": 0.3180021998026855, |
| "learning_rate": 1.2920150767552651e-05, |
| "loss": 0.2185, |
| "step": 804 |
| }, |
| { |
| "epoch": 3.260759493670886, |
| "grad_norm": 0.33385831876209454, |
| "learning_rate": 1.2867095896080607e-05, |
| "loss": 0.252, |
| "step": 805 |
| }, |
| { |
| "epoch": 3.2648101265822786, |
| "grad_norm": 0.3294623263116459, |
| "learning_rate": 1.2814098472037612e-05, |
| "loss": 0.2193, |
| "step": 806 |
| }, |
| { |
| "epoch": 3.2688607594936707, |
| "grad_norm": 0.32100322104483286, |
| "learning_rate": 1.276115892225764e-05, |
| "loss": 0.2457, |
| "step": 807 |
| }, |
| { |
| "epoch": 3.2729113924050632, |
| "grad_norm": 0.32845309405223955, |
| "learning_rate": 1.2708277673108555e-05, |
| "loss": 0.2341, |
| "step": 808 |
| }, |
| { |
| "epoch": 3.2769620253164558, |
| "grad_norm": 0.3427945579585441, |
| "learning_rate": 1.2655455150488649e-05, |
| "loss": 0.2197, |
| "step": 809 |
| }, |
| { |
| "epoch": 3.2810126582278483, |
| "grad_norm": 0.2955438220450271, |
| "learning_rate": 1.2602691779823272e-05, |
| "loss": 0.234, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.2850632911392403, |
| "grad_norm": 0.31763599658631025, |
| "learning_rate": 1.2549987986061355e-05, |
| "loss": 0.2317, |
| "step": 811 |
| }, |
| { |
| "epoch": 3.289113924050633, |
| "grad_norm": 0.34037855439778625, |
| "learning_rate": 1.2497344193672005e-05, |
| "loss": 0.2235, |
| "step": 812 |
| }, |
| { |
| "epoch": 3.2931645569620254, |
| "grad_norm": 0.31545195641544527, |
| "learning_rate": 1.2444760826641092e-05, |
| "loss": 0.2323, |
| "step": 813 |
| }, |
| { |
| "epoch": 3.297215189873418, |
| "grad_norm": 0.33409228266254626, |
| "learning_rate": 1.2392238308467817e-05, |
| "loss": 0.2092, |
| "step": 814 |
| }, |
| { |
| "epoch": 3.30126582278481, |
| "grad_norm": 0.314363794937502, |
| "learning_rate": 1.2339777062161326e-05, |
| "loss": 0.2489, |
| "step": 815 |
| }, |
| { |
| "epoch": 3.3053164556962025, |
| "grad_norm": 0.3197020346086439, |
| "learning_rate": 1.2287377510237293e-05, |
| "loss": 0.2122, |
| "step": 816 |
| }, |
| { |
| "epoch": 3.309367088607595, |
| "grad_norm": 0.32157774007948414, |
| "learning_rate": 1.2235040074714488e-05, |
| "loss": 0.2366, |
| "step": 817 |
| }, |
| { |
| "epoch": 3.3134177215189875, |
| "grad_norm": 0.30361574599792895, |
| "learning_rate": 1.2182765177111434e-05, |
| "loss": 0.2422, |
| "step": 818 |
| }, |
| { |
| "epoch": 3.3174683544303796, |
| "grad_norm": 0.315891741682938, |
| "learning_rate": 1.213055323844297e-05, |
| "loss": 0.2325, |
| "step": 819 |
| }, |
| { |
| "epoch": 3.321518987341772, |
| "grad_norm": 0.3170579928342589, |
| "learning_rate": 1.2078404679216864e-05, |
| "loss": 0.2235, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.3255696202531646, |
| "grad_norm": 0.33079438039025316, |
| "learning_rate": 1.2026319919430458e-05, |
| "loss": 0.2334, |
| "step": 821 |
| }, |
| { |
| "epoch": 3.329620253164557, |
| "grad_norm": 0.33189710497748204, |
| "learning_rate": 1.1974299378567227e-05, |
| "loss": 0.2365, |
| "step": 822 |
| }, |
| { |
| "epoch": 3.333670886075949, |
| "grad_norm": 0.31115956118830673, |
| "learning_rate": 1.1922343475593462e-05, |
| "loss": 0.2325, |
| "step": 823 |
| }, |
| { |
| "epoch": 3.3377215189873417, |
| "grad_norm": 0.3295708215496758, |
| "learning_rate": 1.187045262895488e-05, |
| "loss": 0.237, |
| "step": 824 |
| }, |
| { |
| "epoch": 3.3417721518987342, |
| "grad_norm": 0.3363822431062975, |
| "learning_rate": 1.1818627256573203e-05, |
| "loss": 0.2403, |
| "step": 825 |
| }, |
| { |
| "epoch": 3.3458227848101267, |
| "grad_norm": 0.3186010542724897, |
| "learning_rate": 1.1766867775842864e-05, |
| "loss": 0.2306, |
| "step": 826 |
| }, |
| { |
| "epoch": 3.349873417721519, |
| "grad_norm": 0.3231505207544207, |
| "learning_rate": 1.1715174603627615e-05, |
| "loss": 0.2393, |
| "step": 827 |
| }, |
| { |
| "epoch": 3.3539240506329113, |
| "grad_norm": 0.3217702764159803, |
| "learning_rate": 1.1663548156257147e-05, |
| "loss": 0.2239, |
| "step": 828 |
| }, |
| { |
| "epoch": 3.357974683544304, |
| "grad_norm": 0.31727626361164696, |
| "learning_rate": 1.161198884952377e-05, |
| "loss": 0.2257, |
| "step": 829 |
| }, |
| { |
| "epoch": 3.3620253164556964, |
| "grad_norm": 0.32285489714598115, |
| "learning_rate": 1.1560497098679056e-05, |
| "loss": 0.2377, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.3660759493670884, |
| "grad_norm": 0.32566769695769143, |
| "learning_rate": 1.1509073318430479e-05, |
| "loss": 0.2375, |
| "step": 831 |
| }, |
| { |
| "epoch": 3.370126582278481, |
| "grad_norm": 0.31124127520175787, |
| "learning_rate": 1.1457717922938116e-05, |
| "loss": 0.2379, |
| "step": 832 |
| }, |
| { |
| "epoch": 3.3741772151898735, |
| "grad_norm": 0.33079423234426386, |
| "learning_rate": 1.1406431325811233e-05, |
| "loss": 0.2223, |
| "step": 833 |
| }, |
| { |
| "epoch": 3.378227848101266, |
| "grad_norm": 0.3218933542061997, |
| "learning_rate": 1.135521394010506e-05, |
| "loss": 0.2198, |
| "step": 834 |
| }, |
| { |
| "epoch": 3.382278481012658, |
| "grad_norm": 0.3366831313189657, |
| "learning_rate": 1.1304066178317367e-05, |
| "loss": 0.2332, |
| "step": 835 |
| }, |
| { |
| "epoch": 3.3863291139240506, |
| "grad_norm": 0.34648066051050286, |
| "learning_rate": 1.1252988452385199e-05, |
| "loss": 0.2311, |
| "step": 836 |
| }, |
| { |
| "epoch": 3.390379746835443, |
| "grad_norm": 0.3136709102533596, |
| "learning_rate": 1.1201981173681536e-05, |
| "loss": 0.2394, |
| "step": 837 |
| }, |
| { |
| "epoch": 3.3944303797468356, |
| "grad_norm": 0.322583033012318, |
| "learning_rate": 1.1151044753011991e-05, |
| "loss": 0.231, |
| "step": 838 |
| }, |
| { |
| "epoch": 3.3984810126582277, |
| "grad_norm": 0.33597658150373677, |
| "learning_rate": 1.1100179600611491e-05, |
| "loss": 0.2145, |
| "step": 839 |
| }, |
| { |
| "epoch": 3.40253164556962, |
| "grad_norm": 0.32371605330404596, |
| "learning_rate": 1.1049386126140985e-05, |
| "loss": 0.2204, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.4065822784810127, |
| "grad_norm": 0.33747677126838543, |
| "learning_rate": 1.0998664738684128e-05, |
| "loss": 0.2344, |
| "step": 841 |
| }, |
| { |
| "epoch": 3.410632911392405, |
| "grad_norm": 0.3207163404026187, |
| "learning_rate": 1.0948015846744e-05, |
| "loss": 0.2255, |
| "step": 842 |
| }, |
| { |
| "epoch": 3.4146835443037973, |
| "grad_norm": 0.32493290007971215, |
| "learning_rate": 1.0897439858239832e-05, |
| "loss": 0.2326, |
| "step": 843 |
| }, |
| { |
| "epoch": 3.41873417721519, |
| "grad_norm": 0.3082780901192136, |
| "learning_rate": 1.0846937180503652e-05, |
| "loss": 0.2285, |
| "step": 844 |
| }, |
| { |
| "epoch": 3.4227848101265823, |
| "grad_norm": 0.29833058408867147, |
| "learning_rate": 1.0796508220277117e-05, |
| "loss": 0.23, |
| "step": 845 |
| }, |
| { |
| "epoch": 3.426835443037975, |
| "grad_norm": 0.31756345958690846, |
| "learning_rate": 1.0746153383708107e-05, |
| "loss": 0.2426, |
| "step": 846 |
| }, |
| { |
| "epoch": 3.430886075949367, |
| "grad_norm": 0.31060532893970705, |
| "learning_rate": 1.0695873076347579e-05, |
| "loss": 0.2433, |
| "step": 847 |
| }, |
| { |
| "epoch": 3.4349367088607594, |
| "grad_norm": 0.31287469597468215, |
| "learning_rate": 1.0645667703146205e-05, |
| "loss": 0.2522, |
| "step": 848 |
| }, |
| { |
| "epoch": 3.438987341772152, |
| "grad_norm": 0.32700087918845944, |
| "learning_rate": 1.0595537668451161e-05, |
| "loss": 0.2357, |
| "step": 849 |
| }, |
| { |
| "epoch": 3.4430379746835444, |
| "grad_norm": 0.32982645842990493, |
| "learning_rate": 1.0545483376002854e-05, |
| "loss": 0.2378, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.4470886075949365, |
| "grad_norm": 0.313247023703437, |
| "learning_rate": 1.0495505228931676e-05, |
| "loss": 0.2387, |
| "step": 851 |
| }, |
| { |
| "epoch": 3.451139240506329, |
| "grad_norm": 0.30461617545287567, |
| "learning_rate": 1.044560362975474e-05, |
| "loss": 0.2355, |
| "step": 852 |
| }, |
| { |
| "epoch": 3.4551898734177215, |
| "grad_norm": 0.3020560879065176, |
| "learning_rate": 1.0395778980372695e-05, |
| "loss": 0.226, |
| "step": 853 |
| }, |
| { |
| "epoch": 3.459240506329114, |
| "grad_norm": 0.3247630277397608, |
| "learning_rate": 1.0346031682066381e-05, |
| "loss": 0.2183, |
| "step": 854 |
| }, |
| { |
| "epoch": 3.463291139240506, |
| "grad_norm": 0.33258172256389984, |
| "learning_rate": 1.0296362135493724e-05, |
| "loss": 0.2309, |
| "step": 855 |
| }, |
| { |
| "epoch": 3.4673417721518986, |
| "grad_norm": 0.32567477544932205, |
| "learning_rate": 1.0246770740686422e-05, |
| "loss": 0.229, |
| "step": 856 |
| }, |
| { |
| "epoch": 3.471392405063291, |
| "grad_norm": 0.31076925704406344, |
| "learning_rate": 1.0197257897046743e-05, |
| "loss": 0.2136, |
| "step": 857 |
| }, |
| { |
| "epoch": 3.4754430379746837, |
| "grad_norm": 0.3184200469209035, |
| "learning_rate": 1.014782400334433e-05, |
| "loss": 0.242, |
| "step": 858 |
| }, |
| { |
| "epoch": 3.479493670886076, |
| "grad_norm": 0.32172226127680875, |
| "learning_rate": 1.009846945771296e-05, |
| "loss": 0.2429, |
| "step": 859 |
| }, |
| { |
| "epoch": 3.4835443037974683, |
| "grad_norm": 0.30760804086303556, |
| "learning_rate": 1.0049194657647363e-05, |
| "loss": 0.2427, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.487594936708861, |
| "grad_norm": 0.3205531005872481, |
| "learning_rate": 1.0000000000000006e-05, |
| "loss": 0.23, |
| "step": 861 |
| }, |
| { |
| "epoch": 3.4916455696202533, |
| "grad_norm": 0.32056747755761716, |
| "learning_rate": 9.950885880977891e-06, |
| "loss": 0.2475, |
| "step": 862 |
| }, |
| { |
| "epoch": 3.4956962025316454, |
| "grad_norm": 0.32116299031559875, |
| "learning_rate": 9.901852696139382e-06, |
| "loss": 0.2329, |
| "step": 863 |
| }, |
| { |
| "epoch": 3.499746835443038, |
| "grad_norm": 0.29769368834294135, |
| "learning_rate": 9.852900840391027e-06, |
| "loss": 0.2535, |
| "step": 864 |
| }, |
| { |
| "epoch": 3.5037974683544304, |
| "grad_norm": 0.30973799781276246, |
| "learning_rate": 9.804030707984313e-06, |
| "loss": 0.2336, |
| "step": 865 |
| }, |
| { |
| "epoch": 3.507848101265823, |
| "grad_norm": 0.3271472729974498, |
| "learning_rate": 9.755242692512599e-06, |
| "loss": 0.2369, |
| "step": 866 |
| }, |
| { |
| "epoch": 3.5118987341772154, |
| "grad_norm": 0.3131787821334698, |
| "learning_rate": 9.70653718690782e-06, |
| "loss": 0.2227, |
| "step": 867 |
| }, |
| { |
| "epoch": 3.5159493670886075, |
| "grad_norm": 0.3291091414719241, |
| "learning_rate": 9.657914583437454e-06, |
| "loss": 0.2207, |
| "step": 868 |
| }, |
| { |
| "epoch": 3.52, |
| "grad_norm": 0.33329319352209935, |
| "learning_rate": 9.609375273701246e-06, |
| "loss": 0.2113, |
| "step": 869 |
| }, |
| { |
| "epoch": 3.5240506329113925, |
| "grad_norm": 0.3284368819854331, |
| "learning_rate": 9.560919648628133e-06, |
| "loss": 0.2341, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.5281012658227846, |
| "grad_norm": 0.31141188866604047, |
| "learning_rate": 9.512548098473047e-06, |
| "loss": 0.2159, |
| "step": 871 |
| }, |
| { |
| "epoch": 3.532151898734177, |
| "grad_norm": 0.3225711070048204, |
| "learning_rate": 9.464261012813825e-06, |
| "loss": 0.2207, |
| "step": 872 |
| }, |
| { |
| "epoch": 3.5362025316455696, |
| "grad_norm": 0.32516018865999935, |
| "learning_rate": 9.416058780547987e-06, |
| "loss": 0.2458, |
| "step": 873 |
| }, |
| { |
| "epoch": 3.540253164556962, |
| "grad_norm": 0.3133027525149711, |
| "learning_rate": 9.367941789889714e-06, |
| "loss": 0.2271, |
| "step": 874 |
| }, |
| { |
| "epoch": 3.5443037974683547, |
| "grad_norm": 0.3124853570469816, |
| "learning_rate": 9.319910428366607e-06, |
| "loss": 0.2414, |
| "step": 875 |
| }, |
| { |
| "epoch": 3.5483544303797467, |
| "grad_norm": 0.3071656745471624, |
| "learning_rate": 9.271965082816667e-06, |
| "loss": 0.2281, |
| "step": 876 |
| }, |
| { |
| "epoch": 3.5524050632911393, |
| "grad_norm": 0.3268248271918991, |
| "learning_rate": 9.224106139385111e-06, |
| "loss": 0.2278, |
| "step": 877 |
| }, |
| { |
| "epoch": 3.5564556962025318, |
| "grad_norm": 0.3188581261889334, |
| "learning_rate": 9.176333983521291e-06, |
| "loss": 0.2283, |
| "step": 878 |
| }, |
| { |
| "epoch": 3.560506329113924, |
| "grad_norm": 0.320321082701254, |
| "learning_rate": 9.12864899997558e-06, |
| "loss": 0.2314, |
| "step": 879 |
| }, |
| { |
| "epoch": 3.5645569620253164, |
| "grad_norm": 0.31149314832528024, |
| "learning_rate": 9.08105157279628e-06, |
| "loss": 0.2366, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.568607594936709, |
| "grad_norm": 0.31329392736006795, |
| "learning_rate": 9.03354208532653e-06, |
| "loss": 0.2229, |
| "step": 881 |
| }, |
| { |
| "epoch": 3.5726582278481014, |
| "grad_norm": 0.3285117798495477, |
| "learning_rate": 8.986120920201205e-06, |
| "loss": 0.2406, |
| "step": 882 |
| }, |
| { |
| "epoch": 3.576708860759494, |
| "grad_norm": 0.32206867076732615, |
| "learning_rate": 8.938788459343852e-06, |
| "loss": 0.214, |
| "step": 883 |
| }, |
| { |
| "epoch": 3.580759493670886, |
| "grad_norm": 0.32128145747722786, |
| "learning_rate": 8.8915450839636e-06, |
| "loss": 0.2253, |
| "step": 884 |
| }, |
| { |
| "epoch": 3.5848101265822785, |
| "grad_norm": 0.3212449321103388, |
| "learning_rate": 8.844391174552116e-06, |
| "loss": 0.2462, |
| "step": 885 |
| }, |
| { |
| "epoch": 3.588860759493671, |
| "grad_norm": 0.30678915791411, |
| "learning_rate": 8.797327110880479e-06, |
| "loss": 0.2316, |
| "step": 886 |
| }, |
| { |
| "epoch": 3.592911392405063, |
| "grad_norm": 0.3125293762334648, |
| "learning_rate": 8.750353271996206e-06, |
| "loss": 0.2176, |
| "step": 887 |
| }, |
| { |
| "epoch": 3.5969620253164556, |
| "grad_norm": 0.3162369046156705, |
| "learning_rate": 8.703470036220132e-06, |
| "loss": 0.2367, |
| "step": 888 |
| }, |
| { |
| "epoch": 3.601012658227848, |
| "grad_norm": 0.3014562310298116, |
| "learning_rate": 8.656677781143394e-06, |
| "loss": 0.228, |
| "step": 889 |
| }, |
| { |
| "epoch": 3.6050632911392406, |
| "grad_norm": 0.3114526211569808, |
| "learning_rate": 8.609976883624377e-06, |
| "loss": 0.2308, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.609113924050633, |
| "grad_norm": 0.30866959200025845, |
| "learning_rate": 8.563367719785698e-06, |
| "loss": 0.2295, |
| "step": 891 |
| }, |
| { |
| "epoch": 3.613164556962025, |
| "grad_norm": 0.3231469136630656, |
| "learning_rate": 8.516850665011138e-06, |
| "loss": 0.2081, |
| "step": 892 |
| }, |
| { |
| "epoch": 3.6172151898734177, |
| "grad_norm": 0.31088949299455176, |
| "learning_rate": 8.47042609394269e-06, |
| "loss": 0.2468, |
| "step": 893 |
| }, |
| { |
| "epoch": 3.6212658227848102, |
| "grad_norm": 0.30881298652121864, |
| "learning_rate": 8.424094380477432e-06, |
| "loss": 0.2132, |
| "step": 894 |
| }, |
| { |
| "epoch": 3.6253164556962023, |
| "grad_norm": 0.31072353239248524, |
| "learning_rate": 8.37785589776465e-06, |
| "loss": 0.2319, |
| "step": 895 |
| }, |
| { |
| "epoch": 3.629367088607595, |
| "grad_norm": 0.31050712592190954, |
| "learning_rate": 8.331711018202694e-06, |
| "loss": 0.201, |
| "step": 896 |
| }, |
| { |
| "epoch": 3.6334177215189873, |
| "grad_norm": 0.3280088523401034, |
| "learning_rate": 8.285660113436104e-06, |
| "loss": 0.2405, |
| "step": 897 |
| }, |
| { |
| "epoch": 3.63746835443038, |
| "grad_norm": 0.32063930653808026, |
| "learning_rate": 8.239703554352527e-06, |
| "loss": 0.2371, |
| "step": 898 |
| }, |
| { |
| "epoch": 3.6415189873417724, |
| "grad_norm": 0.31429570441116567, |
| "learning_rate": 8.193841711079775e-06, |
| "loss": 0.2178, |
| "step": 899 |
| }, |
| { |
| "epoch": 3.6455696202531644, |
| "grad_norm": 0.32982559563103486, |
| "learning_rate": 8.148074952982828e-06, |
| "loss": 0.234, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.649620253164557, |
| "grad_norm": 0.32847186400152406, |
| "learning_rate": 8.102403648660859e-06, |
| "loss": 0.2281, |
| "step": 901 |
| }, |
| { |
| "epoch": 3.6536708860759495, |
| "grad_norm": 0.3282650001428093, |
| "learning_rate": 8.056828165944282e-06, |
| "loss": 0.239, |
| "step": 902 |
| }, |
| { |
| "epoch": 3.6577215189873415, |
| "grad_norm": 0.3132286537270214, |
| "learning_rate": 8.011348871891762e-06, |
| "loss": 0.236, |
| "step": 903 |
| }, |
| { |
| "epoch": 3.661772151898734, |
| "grad_norm": 0.3269658199317102, |
| "learning_rate": 7.965966132787287e-06, |
| "loss": 0.2149, |
| "step": 904 |
| }, |
| { |
| "epoch": 3.6658227848101266, |
| "grad_norm": 0.32259781054329023, |
| "learning_rate": 7.920680314137189e-06, |
| "loss": 0.238, |
| "step": 905 |
| }, |
| { |
| "epoch": 3.669873417721519, |
| "grad_norm": 0.3210744499870592, |
| "learning_rate": 7.875491780667246e-06, |
| "loss": 0.2395, |
| "step": 906 |
| }, |
| { |
| "epoch": 3.6739240506329116, |
| "grad_norm": 0.3086090866792456, |
| "learning_rate": 7.830400896319667e-06, |
| "loss": 0.2347, |
| "step": 907 |
| }, |
| { |
| "epoch": 3.6779746835443037, |
| "grad_norm": 0.31846018031217127, |
| "learning_rate": 7.785408024250259e-06, |
| "loss": 0.2159, |
| "step": 908 |
| }, |
| { |
| "epoch": 3.682025316455696, |
| "grad_norm": 0.33926003364785284, |
| "learning_rate": 7.74051352682542e-06, |
| "loss": 0.228, |
| "step": 909 |
| }, |
| { |
| "epoch": 3.6860759493670887, |
| "grad_norm": 0.3100259656067674, |
| "learning_rate": 7.695717765619257e-06, |
| "loss": 0.2366, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.690126582278481, |
| "grad_norm": 0.32042372547502895, |
| "learning_rate": 7.651021101410673e-06, |
| "loss": 0.2293, |
| "step": 911 |
| }, |
| { |
| "epoch": 3.6941772151898733, |
| "grad_norm": 0.3144052139145267, |
| "learning_rate": 7.606423894180464e-06, |
| "loss": 0.2289, |
| "step": 912 |
| }, |
| { |
| "epoch": 3.698227848101266, |
| "grad_norm": 0.31447603364309196, |
| "learning_rate": 7.56192650310839e-06, |
| "loss": 0.2369, |
| "step": 913 |
| }, |
| { |
| "epoch": 3.7022784810126583, |
| "grad_norm": 0.30366812408055993, |
| "learning_rate": 7.517529286570349e-06, |
| "loss": 0.2261, |
| "step": 914 |
| }, |
| { |
| "epoch": 3.706329113924051, |
| "grad_norm": 0.32629018723595693, |
| "learning_rate": 7.473232602135387e-06, |
| "loss": 0.2315, |
| "step": 915 |
| }, |
| { |
| "epoch": 3.710379746835443, |
| "grad_norm": 0.3062597206553022, |
| "learning_rate": 7.429036806562935e-06, |
| "loss": 0.2192, |
| "step": 916 |
| }, |
| { |
| "epoch": 3.7144303797468354, |
| "grad_norm": 0.30423137489117913, |
| "learning_rate": 7.3849422557998455e-06, |
| "loss": 0.2399, |
| "step": 917 |
| }, |
| { |
| "epoch": 3.718481012658228, |
| "grad_norm": 0.3009247638848356, |
| "learning_rate": 7.340949304977567e-06, |
| "loss": 0.214, |
| "step": 918 |
| }, |
| { |
| "epoch": 3.72253164556962, |
| "grad_norm": 0.30934184994478625, |
| "learning_rate": 7.297058308409282e-06, |
| "loss": 0.2451, |
| "step": 919 |
| }, |
| { |
| "epoch": 3.7265822784810125, |
| "grad_norm": 0.31371598879835993, |
| "learning_rate": 7.25326961958704e-06, |
| "loss": 0.2079, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.730632911392405, |
| "grad_norm": 0.31187801570548557, |
| "learning_rate": 7.209583591178921e-06, |
| "loss": 0.2141, |
| "step": 921 |
| }, |
| { |
| "epoch": 3.7346835443037976, |
| "grad_norm": 0.2981939225526193, |
| "learning_rate": 7.1660005750261925e-06, |
| "loss": 0.2366, |
| "step": 922 |
| }, |
| { |
| "epoch": 3.73873417721519, |
| "grad_norm": 0.31790182441337306, |
| "learning_rate": 7.1225209221404765e-06, |
| "loss": 0.2333, |
| "step": 923 |
| }, |
| { |
| "epoch": 3.742784810126582, |
| "grad_norm": 0.299393939853749, |
| "learning_rate": 7.079144982700909e-06, |
| "loss": 0.2472, |
| "step": 924 |
| }, |
| { |
| "epoch": 3.7468354430379747, |
| "grad_norm": 0.3046105785856575, |
| "learning_rate": 7.0358731060513695e-06, |
| "loss": 0.247, |
| "step": 925 |
| }, |
| { |
| "epoch": 3.750886075949367, |
| "grad_norm": 0.29773547576759407, |
| "learning_rate": 6.99270564069757e-06, |
| "loss": 0.2276, |
| "step": 926 |
| }, |
| { |
| "epoch": 3.7549367088607593, |
| "grad_norm": 0.31063694257543834, |
| "learning_rate": 6.949642934304375e-06, |
| "loss": 0.239, |
| "step": 927 |
| }, |
| { |
| "epoch": 3.7589873417721518, |
| "grad_norm": 0.3182805098707621, |
| "learning_rate": 6.906685333692871e-06, |
| "loss": 0.215, |
| "step": 928 |
| }, |
| { |
| "epoch": 3.7630379746835443, |
| "grad_norm": 0.30309015225398306, |
| "learning_rate": 6.86383318483769e-06, |
| "loss": 0.234, |
| "step": 929 |
| }, |
| { |
| "epoch": 3.767088607594937, |
| "grad_norm": 0.30932620957025725, |
| "learning_rate": 6.821086832864139e-06, |
| "loss": 0.208, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.7711392405063293, |
| "grad_norm": 0.31526772415361615, |
| "learning_rate": 6.77844662204546e-06, |
| "loss": 0.2221, |
| "step": 931 |
| }, |
| { |
| "epoch": 3.7751898734177214, |
| "grad_norm": 0.31037721473316215, |
| "learning_rate": 6.7359128958000455e-06, |
| "loss": 0.2165, |
| "step": 932 |
| }, |
| { |
| "epoch": 3.779240506329114, |
| "grad_norm": 0.30370408881356226, |
| "learning_rate": 6.693485996688695e-06, |
| "loss": 0.2287, |
| "step": 933 |
| }, |
| { |
| "epoch": 3.7832911392405064, |
| "grad_norm": 0.2932290141498711, |
| "learning_rate": 6.651166266411801e-06, |
| "loss": 0.2419, |
| "step": 934 |
| }, |
| { |
| "epoch": 3.7873417721518985, |
| "grad_norm": 0.3207252370913895, |
| "learning_rate": 6.6089540458066725e-06, |
| "loss": 0.224, |
| "step": 935 |
| }, |
| { |
| "epoch": 3.791392405063291, |
| "grad_norm": 0.3264016128793882, |
| "learning_rate": 6.566849674844711e-06, |
| "loss": 0.2303, |
| "step": 936 |
| }, |
| { |
| "epoch": 3.7954430379746835, |
| "grad_norm": 0.29995320356253075, |
| "learning_rate": 6.524853492628747e-06, |
| "loss": 0.2464, |
| "step": 937 |
| }, |
| { |
| "epoch": 3.799493670886076, |
| "grad_norm": 0.301454390908644, |
| "learning_rate": 6.4829658373902536e-06, |
| "loss": 0.2319, |
| "step": 938 |
| }, |
| { |
| "epoch": 3.8035443037974686, |
| "grad_norm": 0.3112903321950993, |
| "learning_rate": 6.441187046486648e-06, |
| "loss": 0.2244, |
| "step": 939 |
| }, |
| { |
| "epoch": 3.8075949367088606, |
| "grad_norm": 0.30931355577394287, |
| "learning_rate": 6.399517456398567e-06, |
| "loss": 0.224, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.811645569620253, |
| "grad_norm": 0.31881942395476004, |
| "learning_rate": 6.357957402727164e-06, |
| "loss": 0.1994, |
| "step": 941 |
| }, |
| { |
| "epoch": 3.8156962025316457, |
| "grad_norm": 0.3087564224426162, |
| "learning_rate": 6.316507220191395e-06, |
| "loss": 0.2334, |
| "step": 942 |
| }, |
| { |
| "epoch": 3.8197468354430377, |
| "grad_norm": 0.30832878696231836, |
| "learning_rate": 6.275167242625331e-06, |
| "loss": 0.2222, |
| "step": 943 |
| }, |
| { |
| "epoch": 3.8237974683544302, |
| "grad_norm": 0.31919742550199276, |
| "learning_rate": 6.233937802975471e-06, |
| "loss": 0.2531, |
| "step": 944 |
| }, |
| { |
| "epoch": 3.8278481012658228, |
| "grad_norm": 0.3123852396250325, |
| "learning_rate": 6.192819233298046e-06, |
| "loss": 0.2325, |
| "step": 945 |
| }, |
| { |
| "epoch": 3.8318987341772153, |
| "grad_norm": 0.3216968636574637, |
| "learning_rate": 6.151811864756383e-06, |
| "loss": 0.2482, |
| "step": 946 |
| }, |
| { |
| "epoch": 3.835949367088608, |
| "grad_norm": 0.32127091551782727, |
| "learning_rate": 6.1109160276181655e-06, |
| "loss": 0.2075, |
| "step": 947 |
| }, |
| { |
| "epoch": 3.84, |
| "grad_norm": 0.31149148472882454, |
| "learning_rate": 6.070132051252868e-06, |
| "loss": 0.215, |
| "step": 948 |
| }, |
| { |
| "epoch": 3.8440506329113924, |
| "grad_norm": 0.30693527183638303, |
| "learning_rate": 6.0294602641290034e-06, |
| "loss": 0.208, |
| "step": 949 |
| }, |
| { |
| "epoch": 3.848101265822785, |
| "grad_norm": 0.3131412212786968, |
| "learning_rate": 5.988900993811575e-06, |
| "loss": 0.228, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.852151898734177, |
| "grad_norm": 0.32389137478715246, |
| "learning_rate": 5.948454566959363e-06, |
| "loss": 0.2507, |
| "step": 951 |
| }, |
| { |
| "epoch": 3.8562025316455695, |
| "grad_norm": 0.33263730681869436, |
| "learning_rate": 5.908121309322328e-06, |
| "loss": 0.2379, |
| "step": 952 |
| }, |
| { |
| "epoch": 3.860253164556962, |
| "grad_norm": 0.2957092235772375, |
| "learning_rate": 5.867901545738976e-06, |
| "loss": 0.2375, |
| "step": 953 |
| }, |
| { |
| "epoch": 3.8643037974683545, |
| "grad_norm": 0.3058630150606596, |
| "learning_rate": 5.827795600133774e-06, |
| "loss": 0.2235, |
| "step": 954 |
| }, |
| { |
| "epoch": 3.868354430379747, |
| "grad_norm": 0.30207016220192623, |
| "learning_rate": 5.787803795514466e-06, |
| "loss": 0.2107, |
| "step": 955 |
| }, |
| { |
| "epoch": 3.872405063291139, |
| "grad_norm": 0.31037907373203516, |
| "learning_rate": 5.747926453969576e-06, |
| "loss": 0.2171, |
| "step": 956 |
| }, |
| { |
| "epoch": 3.8764556962025316, |
| "grad_norm": 0.3344209014690423, |
| "learning_rate": 5.708163896665708e-06, |
| "loss": 0.2257, |
| "step": 957 |
| }, |
| { |
| "epoch": 3.880506329113924, |
| "grad_norm": 0.5308641061654318, |
| "learning_rate": 5.668516443845047e-06, |
| "loss": 0.2112, |
| "step": 958 |
| }, |
| { |
| "epoch": 3.884556962025316, |
| "grad_norm": 0.2985112216188278, |
| "learning_rate": 5.6289844148227225e-06, |
| "loss": 0.2439, |
| "step": 959 |
| }, |
| { |
| "epoch": 3.8886075949367087, |
| "grad_norm": 0.3172767074177479, |
| "learning_rate": 5.5895681279842615e-06, |
| "loss": 0.2379, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.8926582278481012, |
| "grad_norm": 0.3055568715648384, |
| "learning_rate": 5.550267900783019e-06, |
| "loss": 0.1966, |
| "step": 961 |
| }, |
| { |
| "epoch": 3.8967088607594937, |
| "grad_norm": 0.3058615657204689, |
| "learning_rate": 5.511084049737623e-06, |
| "loss": 0.2257, |
| "step": 962 |
| }, |
| { |
| "epoch": 3.9007594936708863, |
| "grad_norm": 0.301052744612958, |
| "learning_rate": 5.4720168904294215e-06, |
| "loss": 0.2269, |
| "step": 963 |
| }, |
| { |
| "epoch": 3.9048101265822783, |
| "grad_norm": 0.3019050613917395, |
| "learning_rate": 5.433066737499948e-06, |
| "loss": 0.2395, |
| "step": 964 |
| }, |
| { |
| "epoch": 3.908860759493671, |
| "grad_norm": 0.3022196892857648, |
| "learning_rate": 5.394233904648376e-06, |
| "loss": 0.2105, |
| "step": 965 |
| }, |
| { |
| "epoch": 3.9129113924050634, |
| "grad_norm": 0.29512642197526723, |
| "learning_rate": 5.355518704628997e-06, |
| "loss": 0.2204, |
| "step": 966 |
| }, |
| { |
| "epoch": 3.9169620253164554, |
| "grad_norm": 0.3332168104499327, |
| "learning_rate": 5.316921449248731e-06, |
| "loss": 0.2293, |
| "step": 967 |
| }, |
| { |
| "epoch": 3.921012658227848, |
| "grad_norm": 0.29884942001990333, |
| "learning_rate": 5.278442449364538e-06, |
| "loss": 0.2354, |
| "step": 968 |
| }, |
| { |
| "epoch": 3.9250632911392405, |
| "grad_norm": 0.3131691590717457, |
| "learning_rate": 5.240082014881016e-06, |
| "loss": 0.2396, |
| "step": 969 |
| }, |
| { |
| "epoch": 3.929113924050633, |
| "grad_norm": 0.2949165830351333, |
| "learning_rate": 5.201840454747822e-06, |
| "loss": 0.211, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.9331645569620255, |
| "grad_norm": 0.2917470707381296, |
| "learning_rate": 5.163718076957223e-06, |
| "loss": 0.2378, |
| "step": 971 |
| }, |
| { |
| "epoch": 3.9372151898734176, |
| "grad_norm": 0.2941469223513319, |
| "learning_rate": 5.125715188541609e-06, |
| "loss": 0.2296, |
| "step": 972 |
| }, |
| { |
| "epoch": 3.94126582278481, |
| "grad_norm": 0.3124724043130332, |
| "learning_rate": 5.087832095571021e-06, |
| "loss": 0.2243, |
| "step": 973 |
| }, |
| { |
| "epoch": 3.9453164556962026, |
| "grad_norm": 0.29512196526839585, |
| "learning_rate": 5.0500691031506766e-06, |
| "loss": 0.2174, |
| "step": 974 |
| }, |
| { |
| "epoch": 3.9493670886075947, |
| "grad_norm": 0.30273315287973473, |
| "learning_rate": 5.01242651541854e-06, |
| "loss": 0.2188, |
| "step": 975 |
| }, |
| { |
| "epoch": 3.953417721518987, |
| "grad_norm": 0.31590772884014295, |
| "learning_rate": 4.974904635542815e-06, |
| "loss": 0.2251, |
| "step": 976 |
| }, |
| { |
| "epoch": 3.9574683544303797, |
| "grad_norm": 0.31460474200253186, |
| "learning_rate": 4.937503765719582e-06, |
| "loss": 0.1876, |
| "step": 977 |
| }, |
| { |
| "epoch": 3.961518987341772, |
| "grad_norm": 0.313299576075186, |
| "learning_rate": 4.900224207170299e-06, |
| "loss": 0.2277, |
| "step": 978 |
| }, |
| { |
| "epoch": 3.9655696202531647, |
| "grad_norm": 0.3028663691662953, |
| "learning_rate": 4.8630662601394065e-06, |
| "loss": 0.2262, |
| "step": 979 |
| }, |
| { |
| "epoch": 3.969620253164557, |
| "grad_norm": 0.2878220840632792, |
| "learning_rate": 4.8260302238918995e-06, |
| "loss": 0.2025, |
| "step": 980 |
| }, |
| { |
| "epoch": 3.9736708860759493, |
| "grad_norm": 0.30323642384904126, |
| "learning_rate": 4.789116396710924e-06, |
| "loss": 0.2325, |
| "step": 981 |
| }, |
| { |
| "epoch": 3.977721518987342, |
| "grad_norm": 0.2935762683351535, |
| "learning_rate": 4.752325075895368e-06, |
| "loss": 0.2439, |
| "step": 982 |
| }, |
| { |
| "epoch": 3.981772151898734, |
| "grad_norm": 0.3038791065984245, |
| "learning_rate": 4.715656557757473e-06, |
| "loss": 0.2061, |
| "step": 983 |
| }, |
| { |
| "epoch": 3.9858227848101264, |
| "grad_norm": 0.30546509364061286, |
| "learning_rate": 4.679111137620442e-06, |
| "loss": 0.2202, |
| "step": 984 |
| }, |
| { |
| "epoch": 3.989873417721519, |
| "grad_norm": 0.3177757124046917, |
| "learning_rate": 4.6426891098160585e-06, |
| "loss": 0.2258, |
| "step": 985 |
| }, |
| { |
| "epoch": 3.9939240506329114, |
| "grad_norm": 0.3044523401712507, |
| "learning_rate": 4.6063907676823474e-06, |
| "loss": 0.2087, |
| "step": 986 |
| }, |
| { |
| "epoch": 3.997974683544304, |
| "grad_norm": 0.30514698418828246, |
| "learning_rate": 4.570216403561141e-06, |
| "loss": 0.2127, |
| "step": 987 |
| }, |
| { |
| "epoch": 4.002025316455696, |
| "grad_norm": 0.33576671645691264, |
| "learning_rate": 4.534166308795815e-06, |
| "loss": 0.1977, |
| "step": 988 |
| }, |
| { |
| "epoch": 4.006075949367089, |
| "grad_norm": 0.4240498523038077, |
| "learning_rate": 4.498240773728859e-06, |
| "loss": 0.1511, |
| "step": 989 |
| }, |
| { |
| "epoch": 4.010126582278481, |
| "grad_norm": 0.35225900037049745, |
| "learning_rate": 4.462440087699609e-06, |
| "loss": 0.1648, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.014177215189873, |
| "grad_norm": 0.2605711144286495, |
| "learning_rate": 4.426764539041861e-06, |
| "loss": 0.1855, |
| "step": 991 |
| }, |
| { |
| "epoch": 4.018227848101266, |
| "grad_norm": 0.3010725890981119, |
| "learning_rate": 4.391214415081582e-06, |
| "loss": 0.162, |
| "step": 992 |
| }, |
| { |
| "epoch": 4.022278481012658, |
| "grad_norm": 0.48824464455643496, |
| "learning_rate": 4.355790002134579e-06, |
| "loss": 0.1766, |
| "step": 993 |
| }, |
| { |
| "epoch": 4.02632911392405, |
| "grad_norm": 0.41214932071459043, |
| "learning_rate": 4.320491585504207e-06, |
| "loss": 0.1678, |
| "step": 994 |
| }, |
| { |
| "epoch": 4.030379746835443, |
| "grad_norm": 0.36598672900893714, |
| "learning_rate": 4.2853194494790615e-06, |
| "loss": 0.1622, |
| "step": 995 |
| }, |
| { |
| "epoch": 4.034430379746835, |
| "grad_norm": 0.30767810422712893, |
| "learning_rate": 4.250273877330691e-06, |
| "loss": 0.1615, |
| "step": 996 |
| }, |
| { |
| "epoch": 4.038481012658228, |
| "grad_norm": 0.3072074425692325, |
| "learning_rate": 4.215355151311313e-06, |
| "loss": 0.1681, |
| "step": 997 |
| }, |
| { |
| "epoch": 4.04253164556962, |
| "grad_norm": 0.35671769256279934, |
| "learning_rate": 4.180563552651542e-06, |
| "loss": 0.1846, |
| "step": 998 |
| }, |
| { |
| "epoch": 4.046582278481012, |
| "grad_norm": 0.3318558956435703, |
| "learning_rate": 4.145899361558147e-06, |
| "loss": 0.1734, |
| "step": 999 |
| }, |
| { |
| "epoch": 4.050632911392405, |
| "grad_norm": 0.3188727040342821, |
| "learning_rate": 4.111362857211738e-06, |
| "loss": 0.1879, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.054683544303797, |
| "grad_norm": 0.276514705554858, |
| "learning_rate": 4.076954317764592e-06, |
| "loss": 0.1615, |
| "step": 1001 |
| }, |
| { |
| "epoch": 4.0587341772151895, |
| "grad_norm": 0.2946537962294623, |
| "learning_rate": 4.042674020338335e-06, |
| "loss": 0.1713, |
| "step": 1002 |
| }, |
| { |
| "epoch": 4.062784810126582, |
| "grad_norm": 0.30727761934474446, |
| "learning_rate": 4.0085222410217835e-06, |
| "loss": 0.1487, |
| "step": 1003 |
| }, |
| { |
| "epoch": 4.0668354430379745, |
| "grad_norm": 0.3516145060040981, |
| "learning_rate": 3.974499254868674e-06, |
| "loss": 0.185, |
| "step": 1004 |
| }, |
| { |
| "epoch": 4.0708860759493675, |
| "grad_norm": 0.28948804995507266, |
| "learning_rate": 3.940605335895451e-06, |
| "loss": 0.1613, |
| "step": 1005 |
| }, |
| { |
| "epoch": 4.0749367088607595, |
| "grad_norm": 0.30522334368104226, |
| "learning_rate": 3.90684075707908e-06, |
| "loss": 0.1936, |
| "step": 1006 |
| }, |
| { |
| "epoch": 4.078987341772152, |
| "grad_norm": 0.2701075420324479, |
| "learning_rate": 3.8732057903548505e-06, |
| "loss": 0.1771, |
| "step": 1007 |
| }, |
| { |
| "epoch": 4.083037974683545, |
| "grad_norm": 0.29382244177359723, |
| "learning_rate": 3.8397007066141375e-06, |
| "loss": 0.1693, |
| "step": 1008 |
| }, |
| { |
| "epoch": 4.087088607594937, |
| "grad_norm": 0.285592867048655, |
| "learning_rate": 3.806325775702304e-06, |
| "loss": 0.1778, |
| "step": 1009 |
| }, |
| { |
| "epoch": 4.091139240506329, |
| "grad_norm": 0.27894024507681436, |
| "learning_rate": 3.773081266416434e-06, |
| "loss": 0.1711, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.095189873417722, |
| "grad_norm": 0.27284369156814076, |
| "learning_rate": 3.739967446503245e-06, |
| "loss": 0.1725, |
| "step": 1011 |
| }, |
| { |
| "epoch": 4.099240506329114, |
| "grad_norm": 0.264309063144552, |
| "learning_rate": 3.706984582656894e-06, |
| "loss": 0.1692, |
| "step": 1012 |
| }, |
| { |
| "epoch": 4.103291139240507, |
| "grad_norm": 0.24795301586215296, |
| "learning_rate": 3.6741329405168237e-06, |
| "loss": 0.1592, |
| "step": 1013 |
| }, |
| { |
| "epoch": 4.107341772151899, |
| "grad_norm": 0.26073988566609235, |
| "learning_rate": 3.641412784665648e-06, |
| "loss": 0.1685, |
| "step": 1014 |
| }, |
| { |
| "epoch": 4.111392405063291, |
| "grad_norm": 0.2624231345631046, |
| "learning_rate": 3.608824378627005e-06, |
| "loss": 0.1882, |
| "step": 1015 |
| }, |
| { |
| "epoch": 4.115443037974684, |
| "grad_norm": 0.2569006506678885, |
| "learning_rate": 3.5763679848634337e-06, |
| "loss": 0.1684, |
| "step": 1016 |
| }, |
| { |
| "epoch": 4.119493670886076, |
| "grad_norm": 0.26955862439080114, |
| "learning_rate": 3.544043864774269e-06, |
| "loss": 0.1791, |
| "step": 1017 |
| }, |
| { |
| "epoch": 4.123544303797468, |
| "grad_norm": 0.26530046526896184, |
| "learning_rate": 3.5118522786935282e-06, |
| "loss": 0.1651, |
| "step": 1018 |
| }, |
| { |
| "epoch": 4.127594936708861, |
| "grad_norm": 0.2645608958858635, |
| "learning_rate": 3.479793485887819e-06, |
| "loss": 0.1494, |
| "step": 1019 |
| }, |
| { |
| "epoch": 4.131645569620253, |
| "grad_norm": 0.25156319529483606, |
| "learning_rate": 3.4478677445542653e-06, |
| "loss": 0.1637, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.135696202531646, |
| "grad_norm": 0.2573199948902076, |
| "learning_rate": 3.4160753118183767e-06, |
| "loss": 0.1493, |
| "step": 1021 |
| }, |
| { |
| "epoch": 4.139746835443038, |
| "grad_norm": 0.2702492233154174, |
| "learning_rate": 3.3844164437320527e-06, |
| "loss": 0.1615, |
| "step": 1022 |
| }, |
| { |
| "epoch": 4.14379746835443, |
| "grad_norm": 0.2570535105237843, |
| "learning_rate": 3.3528913952714558e-06, |
| "loss": 0.1733, |
| "step": 1023 |
| }, |
| { |
| "epoch": 4.147848101265823, |
| "grad_norm": 0.25888162325308633, |
| "learning_rate": 3.321500420335e-06, |
| "loss": 0.1856, |
| "step": 1024 |
| }, |
| { |
| "epoch": 4.151898734177215, |
| "grad_norm": 0.2624635499783615, |
| "learning_rate": 3.290243771741275e-06, |
| "loss": 0.1914, |
| "step": 1025 |
| }, |
| { |
| "epoch": 4.155949367088607, |
| "grad_norm": 0.2465101956887112, |
| "learning_rate": 3.2591217012270325e-06, |
| "loss": 0.1704, |
| "step": 1026 |
| }, |
| { |
| "epoch": 4.16, |
| "grad_norm": 0.25283710489475403, |
| "learning_rate": 3.228134459445149e-06, |
| "loss": 0.1683, |
| "step": 1027 |
| }, |
| { |
| "epoch": 4.164050632911392, |
| "grad_norm": 0.24592830285040448, |
| "learning_rate": 3.1972822959626205e-06, |
| "loss": 0.1557, |
| "step": 1028 |
| }, |
| { |
| "epoch": 4.168101265822785, |
| "grad_norm": 0.25476625025609767, |
| "learning_rate": 3.166565459258513e-06, |
| "loss": 0.1719, |
| "step": 1029 |
| }, |
| { |
| "epoch": 4.172151898734177, |
| "grad_norm": 0.2561761180679288, |
| "learning_rate": 3.1359841967220193e-06, |
| "loss": 0.1704, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.176202531645569, |
| "grad_norm": 0.2522783180022918, |
| "learning_rate": 3.105538754650419e-06, |
| "loss": 0.1418, |
| "step": 1031 |
| }, |
| { |
| "epoch": 4.180253164556962, |
| "grad_norm": 0.24380115488202916, |
| "learning_rate": 3.07522937824712e-06, |
| "loss": 0.1865, |
| "step": 1032 |
| }, |
| { |
| "epoch": 4.184303797468354, |
| "grad_norm": 0.24351228641664377, |
| "learning_rate": 3.0450563116196697e-06, |
| "loss": 0.1814, |
| "step": 1033 |
| }, |
| { |
| "epoch": 4.188354430379746, |
| "grad_norm": 0.25145163178932134, |
| "learning_rate": 3.0150197977778008e-06, |
| "loss": 0.1861, |
| "step": 1034 |
| }, |
| { |
| "epoch": 4.192405063291139, |
| "grad_norm": 0.24436514448333707, |
| "learning_rate": 2.985120078631465e-06, |
| "loss": 0.1793, |
| "step": 1035 |
| }, |
| { |
| "epoch": 4.1964556962025314, |
| "grad_norm": 0.2649452241895016, |
| "learning_rate": 2.9553573949888893e-06, |
| "loss": 0.1537, |
| "step": 1036 |
| }, |
| { |
| "epoch": 4.200506329113924, |
| "grad_norm": 0.2561330001627475, |
| "learning_rate": 2.9257319865546384e-06, |
| "loss": 0.1771, |
| "step": 1037 |
| }, |
| { |
| "epoch": 4.2045569620253165, |
| "grad_norm": 0.2531969940730145, |
| "learning_rate": 2.896244091927678e-06, |
| "loss": 0.1664, |
| "step": 1038 |
| }, |
| { |
| "epoch": 4.2086075949367086, |
| "grad_norm": 0.2623322768530426, |
| "learning_rate": 2.8668939485994584e-06, |
| "loss": 0.1501, |
| "step": 1039 |
| }, |
| { |
| "epoch": 4.2126582278481015, |
| "grad_norm": 0.2557763753624446, |
| "learning_rate": 2.837681792951994e-06, |
| "loss": 0.1897, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.216708860759494, |
| "grad_norm": 0.2701721246140197, |
| "learning_rate": 2.808607860255981e-06, |
| "loss": 0.1731, |
| "step": 1041 |
| }, |
| { |
| "epoch": 4.220759493670886, |
| "grad_norm": 0.26657212833788185, |
| "learning_rate": 2.7796723846688634e-06, |
| "loss": 0.1644, |
| "step": 1042 |
| }, |
| { |
| "epoch": 4.224810126582279, |
| "grad_norm": 0.24429690712036994, |
| "learning_rate": 2.7508755992329937e-06, |
| "loss": 0.1884, |
| "step": 1043 |
| }, |
| { |
| "epoch": 4.228860759493671, |
| "grad_norm": 0.251823842122866, |
| "learning_rate": 2.722217735873718e-06, |
| "loss": 0.1763, |
| "step": 1044 |
| }, |
| { |
| "epoch": 4.232911392405064, |
| "grad_norm": 0.2524527943995315, |
| "learning_rate": 2.6936990253975315e-06, |
| "loss": 0.1666, |
| "step": 1045 |
| }, |
| { |
| "epoch": 4.236962025316456, |
| "grad_norm": 0.2672884633700076, |
| "learning_rate": 2.665319697490205e-06, |
| "loss": 0.1542, |
| "step": 1046 |
| }, |
| { |
| "epoch": 4.241012658227848, |
| "grad_norm": 0.2594523143422785, |
| "learning_rate": 2.637079980714945e-06, |
| "loss": 0.1678, |
| "step": 1047 |
| }, |
| { |
| "epoch": 4.245063291139241, |
| "grad_norm": 0.2552708705957773, |
| "learning_rate": 2.6089801025105453e-06, |
| "loss": 0.1682, |
| "step": 1048 |
| }, |
| { |
| "epoch": 4.249113924050633, |
| "grad_norm": 0.2530446835932571, |
| "learning_rate": 2.581020289189571e-06, |
| "loss": 0.1636, |
| "step": 1049 |
| }, |
| { |
| "epoch": 4.253164556962025, |
| "grad_norm": 0.26021693723284617, |
| "learning_rate": 2.553200765936501e-06, |
| "loss": 0.188, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.257215189873418, |
| "grad_norm": 0.25273115199267077, |
| "learning_rate": 2.525521756805962e-06, |
| "loss": 0.1791, |
| "step": 1051 |
| }, |
| { |
| "epoch": 4.26126582278481, |
| "grad_norm": 0.2623237209330316, |
| "learning_rate": 2.497983484720885e-06, |
| "loss": 0.1641, |
| "step": 1052 |
| }, |
| { |
| "epoch": 4.265316455696203, |
| "grad_norm": 0.25626451639057896, |
| "learning_rate": 2.470586171470728e-06, |
| "loss": 0.1641, |
| "step": 1053 |
| }, |
| { |
| "epoch": 4.269367088607595, |
| "grad_norm": 0.25133596955081156, |
| "learning_rate": 2.4433300377096836e-06, |
| "loss": 0.1603, |
| "step": 1054 |
| }, |
| { |
| "epoch": 4.273417721518987, |
| "grad_norm": 0.24992610545777982, |
| "learning_rate": 2.4162153029549073e-06, |
| "loss": 0.1751, |
| "step": 1055 |
| }, |
| { |
| "epoch": 4.27746835443038, |
| "grad_norm": 0.24198491919073517, |
| "learning_rate": 2.3892421855847458e-06, |
| "loss": 0.1793, |
| "step": 1056 |
| }, |
| { |
| "epoch": 4.281518987341772, |
| "grad_norm": 0.24940763550756753, |
| "learning_rate": 2.362410902836978e-06, |
| "loss": 0.1924, |
| "step": 1057 |
| }, |
| { |
| "epoch": 4.285569620253165, |
| "grad_norm": 0.2475624740433846, |
| "learning_rate": 2.3357216708070653e-06, |
| "loss": 0.1825, |
| "step": 1058 |
| }, |
| { |
| "epoch": 4.289620253164557, |
| "grad_norm": 0.25408126384476065, |
| "learning_rate": 2.309174704446411e-06, |
| "loss": 0.1719, |
| "step": 1059 |
| }, |
| { |
| "epoch": 4.293670886075949, |
| "grad_norm": 0.2549557050071026, |
| "learning_rate": 2.2827702175606437e-06, |
| "loss": 0.165, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.297721518987342, |
| "grad_norm": 0.2559497056470194, |
| "learning_rate": 2.256508422807855e-06, |
| "loss": 0.1727, |
| "step": 1061 |
| }, |
| { |
| "epoch": 4.301772151898734, |
| "grad_norm": 0.2522030908796696, |
| "learning_rate": 2.230389531696946e-06, |
| "loss": 0.1523, |
| "step": 1062 |
| }, |
| { |
| "epoch": 4.305822784810126, |
| "grad_norm": 0.26249376575738975, |
| "learning_rate": 2.204413754585857e-06, |
| "loss": 0.1671, |
| "step": 1063 |
| }, |
| { |
| "epoch": 4.309873417721519, |
| "grad_norm": 0.24877118262950834, |
| "learning_rate": 2.1785813006799406e-06, |
| "loss": 0.1775, |
| "step": 1064 |
| }, |
| { |
| "epoch": 4.313924050632911, |
| "grad_norm": 0.2561813992239209, |
| "learning_rate": 2.1528923780302224e-06, |
| "loss": 0.1576, |
| "step": 1065 |
| }, |
| { |
| "epoch": 4.317974683544303, |
| "grad_norm": 0.250810846700604, |
| "learning_rate": 2.127347193531757e-06, |
| "loss": 0.1589, |
| "step": 1066 |
| }, |
| { |
| "epoch": 4.322025316455696, |
| "grad_norm": 0.24811723118048135, |
| "learning_rate": 2.101945952921942e-06, |
| "loss": 0.182, |
| "step": 1067 |
| }, |
| { |
| "epoch": 4.326075949367088, |
| "grad_norm": 0.24868816529543178, |
| "learning_rate": 2.0766888607788906e-06, |
| "loss": 0.1656, |
| "step": 1068 |
| }, |
| { |
| "epoch": 4.330126582278481, |
| "grad_norm": 0.2481625326532568, |
| "learning_rate": 2.0515761205197337e-06, |
| "loss": 0.178, |
| "step": 1069 |
| }, |
| { |
| "epoch": 4.334177215189873, |
| "grad_norm": 0.2505144860950905, |
| "learning_rate": 2.0266079343990453e-06, |
| "loss": 0.157, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.3382278481012655, |
| "grad_norm": 0.2539896781311021, |
| "learning_rate": 2.0017845035071494e-06, |
| "loss": 0.1471, |
| "step": 1071 |
| }, |
| { |
| "epoch": 4.3422784810126585, |
| "grad_norm": 0.25690087260347705, |
| "learning_rate": 1.9771060277685537e-06, |
| "loss": 0.174, |
| "step": 1072 |
| }, |
| { |
| "epoch": 4.3463291139240505, |
| "grad_norm": 0.25686483751226485, |
| "learning_rate": 1.95257270594031e-06, |
| "loss": 0.1721, |
| "step": 1073 |
| }, |
| { |
| "epoch": 4.3503797468354435, |
| "grad_norm": 0.24604982486662838, |
| "learning_rate": 1.9281847356104188e-06, |
| "loss": 0.1592, |
| "step": 1074 |
| }, |
| { |
| "epoch": 4.3544303797468356, |
| "grad_norm": 0.24893517675281732, |
| "learning_rate": 1.9039423131962365e-06, |
| "loss": 0.1787, |
| "step": 1075 |
| }, |
| { |
| "epoch": 4.358481012658228, |
| "grad_norm": 0.24890486335541467, |
| "learning_rate": 1.8798456339429027e-06, |
| "loss": 0.1695, |
| "step": 1076 |
| }, |
| { |
| "epoch": 4.362531645569621, |
| "grad_norm": 0.2534502538132449, |
| "learning_rate": 1.8558948919217612e-06, |
| "loss": 0.1749, |
| "step": 1077 |
| }, |
| { |
| "epoch": 4.366582278481013, |
| "grad_norm": 0.2513409937696382, |
| "learning_rate": 1.8320902800287954e-06, |
| "loss": 0.1497, |
| "step": 1078 |
| }, |
| { |
| "epoch": 4.370632911392405, |
| "grad_norm": 0.24681415845044039, |
| "learning_rate": 1.8084319899830726e-06, |
| "loss": 0.1885, |
| "step": 1079 |
| }, |
| { |
| "epoch": 4.374683544303798, |
| "grad_norm": 0.25927704155990117, |
| "learning_rate": 1.7849202123252097e-06, |
| "loss": 0.1718, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.37873417721519, |
| "grad_norm": 0.24796797942692653, |
| "learning_rate": 1.7615551364158401e-06, |
| "loss": 0.1778, |
| "step": 1081 |
| }, |
| { |
| "epoch": 4.382784810126582, |
| "grad_norm": 0.24805045152792465, |
| "learning_rate": 1.738336950434061e-06, |
| "loss": 0.157, |
| "step": 1082 |
| }, |
| { |
| "epoch": 4.386835443037975, |
| "grad_norm": 0.25819426721172345, |
| "learning_rate": 1.715265841375957e-06, |
| "loss": 0.1697, |
| "step": 1083 |
| }, |
| { |
| "epoch": 4.390886075949367, |
| "grad_norm": 0.2507857577166902, |
| "learning_rate": 1.6923419950530684e-06, |
| "loss": 0.1795, |
| "step": 1084 |
| }, |
| { |
| "epoch": 4.39493670886076, |
| "grad_norm": 0.25959624723276464, |
| "learning_rate": 1.6695655960909008e-06, |
| "loss": 0.1778, |
| "step": 1085 |
| }, |
| { |
| "epoch": 4.398987341772152, |
| "grad_norm": 0.24613444937985585, |
| "learning_rate": 1.646936827927441e-06, |
| "loss": 0.1662, |
| "step": 1086 |
| }, |
| { |
| "epoch": 4.403037974683544, |
| "grad_norm": 0.24767515502814458, |
| "learning_rate": 1.6244558728116766e-06, |
| "loss": 0.162, |
| "step": 1087 |
| }, |
| { |
| "epoch": 4.407088607594937, |
| "grad_norm": 0.2457153641824694, |
| "learning_rate": 1.6021229118021265e-06, |
| "loss": 0.1617, |
| "step": 1088 |
| }, |
| { |
| "epoch": 4.411139240506329, |
| "grad_norm": 0.2519173192430547, |
| "learning_rate": 1.5799381247653967e-06, |
| "loss": 0.1665, |
| "step": 1089 |
| }, |
| { |
| "epoch": 4.415189873417722, |
| "grad_norm": 0.24385384132529042, |
| "learning_rate": 1.5579016903747013e-06, |
| "loss": 0.1722, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.419240506329114, |
| "grad_norm": 0.25430196344886236, |
| "learning_rate": 1.5360137861084656e-06, |
| "loss": 0.1841, |
| "step": 1091 |
| }, |
| { |
| "epoch": 4.423291139240506, |
| "grad_norm": 0.2539011308218622, |
| "learning_rate": 1.5142745882488475e-06, |
| "loss": 0.1599, |
| "step": 1092 |
| }, |
| { |
| "epoch": 4.427341772151899, |
| "grad_norm": 0.2451330126419005, |
| "learning_rate": 1.4926842718803691e-06, |
| "loss": 0.1486, |
| "step": 1093 |
| }, |
| { |
| "epoch": 4.431392405063291, |
| "grad_norm": 0.2472707183231816, |
| "learning_rate": 1.4712430108884657e-06, |
| "loss": 0.1765, |
| "step": 1094 |
| }, |
| { |
| "epoch": 4.435443037974683, |
| "grad_norm": 0.2611696161747097, |
| "learning_rate": 1.4499509779581078e-06, |
| "loss": 0.1947, |
| "step": 1095 |
| }, |
| { |
| "epoch": 4.439493670886076, |
| "grad_norm": 0.23621519045925327, |
| "learning_rate": 1.4288083445723988e-06, |
| "loss": 0.1437, |
| "step": 1096 |
| }, |
| { |
| "epoch": 4.443544303797468, |
| "grad_norm": 0.2555968862015179, |
| "learning_rate": 1.4078152810112045e-06, |
| "loss": 0.1812, |
| "step": 1097 |
| }, |
| { |
| "epoch": 4.44759493670886, |
| "grad_norm": 0.2539006315145, |
| "learning_rate": 1.3869719563497697e-06, |
| "loss": 0.1671, |
| "step": 1098 |
| }, |
| { |
| "epoch": 4.451645569620253, |
| "grad_norm": 0.258588428734053, |
| "learning_rate": 1.3662785384573663e-06, |
| "loss": 0.1779, |
| "step": 1099 |
| }, |
| { |
| "epoch": 4.455696202531645, |
| "grad_norm": 0.25452006013363476, |
| "learning_rate": 1.3457351939959383e-06, |
| "loss": 0.1698, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.459746835443038, |
| "grad_norm": 0.2409197638356423, |
| "learning_rate": 1.3253420884187551e-06, |
| "loss": 0.1649, |
| "step": 1101 |
| }, |
| { |
| "epoch": 4.46379746835443, |
| "grad_norm": 0.24758117613665695, |
| "learning_rate": 1.3050993859690953e-06, |
| "loss": 0.1503, |
| "step": 1102 |
| }, |
| { |
| "epoch": 4.467848101265822, |
| "grad_norm": 0.26292217269369145, |
| "learning_rate": 1.2850072496788869e-06, |
| "loss": 0.1769, |
| "step": 1103 |
| }, |
| { |
| "epoch": 4.471898734177215, |
| "grad_norm": 0.2400494502082909, |
| "learning_rate": 1.2650658413674434e-06, |
| "loss": 0.1635, |
| "step": 1104 |
| }, |
| { |
| "epoch": 4.4759493670886075, |
| "grad_norm": 0.24856802658207408, |
| "learning_rate": 1.2452753216401226e-06, |
| "loss": 0.1762, |
| "step": 1105 |
| }, |
| { |
| "epoch": 4.48, |
| "grad_norm": 0.24307213649420842, |
| "learning_rate": 1.2256358498870503e-06, |
| "loss": 0.1759, |
| "step": 1106 |
| }, |
| { |
| "epoch": 4.4840506329113925, |
| "grad_norm": 0.2504835024928859, |
| "learning_rate": 1.2061475842818337e-06, |
| "loss": 0.1637, |
| "step": 1107 |
| }, |
| { |
| "epoch": 4.488101265822785, |
| "grad_norm": 0.254576826220558, |
| "learning_rate": 1.1868106817802816e-06, |
| "loss": 0.173, |
| "step": 1108 |
| }, |
| { |
| "epoch": 4.4921518987341775, |
| "grad_norm": 0.30259134353666095, |
| "learning_rate": 1.1676252981191482e-06, |
| "loss": 0.1848, |
| "step": 1109 |
| }, |
| { |
| "epoch": 4.49620253164557, |
| "grad_norm": 0.25025102876031385, |
| "learning_rate": 1.1485915878148823e-06, |
| "loss": 0.1657, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.500253164556962, |
| "grad_norm": 0.25800741407754696, |
| "learning_rate": 1.1297097041623584e-06, |
| "loss": 0.169, |
| "step": 1111 |
| }, |
| { |
| "epoch": 4.504303797468355, |
| "grad_norm": 0.24557871760981728, |
| "learning_rate": 1.1109797992336847e-06, |
| "loss": 0.1625, |
| "step": 1112 |
| }, |
| { |
| "epoch": 4.508354430379747, |
| "grad_norm": 0.25917066627740293, |
| "learning_rate": 1.092402023876933e-06, |
| "loss": 0.1793, |
| "step": 1113 |
| }, |
| { |
| "epoch": 4.512405063291139, |
| "grad_norm": 0.7544193089585305, |
| "learning_rate": 1.0739765277149527e-06, |
| "loss": 0.162, |
| "step": 1114 |
| }, |
| { |
| "epoch": 4.516455696202532, |
| "grad_norm": 0.2601148240000392, |
| "learning_rate": 1.0557034591441596e-06, |
| "loss": 0.17, |
| "step": 1115 |
| }, |
| { |
| "epoch": 4.520506329113924, |
| "grad_norm": 0.24462461699421123, |
| "learning_rate": 1.0375829653333324e-06, |
| "loss": 0.1506, |
| "step": 1116 |
| }, |
| { |
| "epoch": 4.524556962025317, |
| "grad_norm": 0.24035445718159879, |
| "learning_rate": 1.0196151922224385e-06, |
| "loss": 0.161, |
| "step": 1117 |
| }, |
| { |
| "epoch": 4.528607594936709, |
| "grad_norm": 0.2506869600204385, |
| "learning_rate": 1.0018002845214526e-06, |
| "loss": 0.1476, |
| "step": 1118 |
| }, |
| { |
| "epoch": 4.532658227848101, |
| "grad_norm": 0.2579141371781443, |
| "learning_rate": 9.841383857091947e-07, |
| "loss": 0.1523, |
| "step": 1119 |
| }, |
| { |
| "epoch": 4.536708860759494, |
| "grad_norm": 0.24870836688642142, |
| "learning_rate": 9.666296380321616e-07, |
| "loss": 0.1657, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.540759493670886, |
| "grad_norm": 0.25086824354121046, |
| "learning_rate": 9.492741825034124e-07, |
| "loss": 0.1722, |
| "step": 1121 |
| }, |
| { |
| "epoch": 4.544810126582279, |
| "grad_norm": 0.2526287453397886, |
| "learning_rate": 9.320721589013892e-07, |
| "loss": 0.1616, |
| "step": 1122 |
| }, |
| { |
| "epoch": 4.548860759493671, |
| "grad_norm": 0.2522210632231441, |
| "learning_rate": 9.150237057688339e-07, |
| "loss": 0.1792, |
| "step": 1123 |
| }, |
| { |
| "epoch": 4.552911392405063, |
| "grad_norm": 0.24598287376168385, |
| "learning_rate": 8.981289604116328e-07, |
| "loss": 0.163, |
| "step": 1124 |
| }, |
| { |
| "epoch": 4.556962025316456, |
| "grad_norm": 0.2395731295337235, |
| "learning_rate": 8.813880588977542e-07, |
| "loss": 0.1719, |
| "step": 1125 |
| }, |
| { |
| "epoch": 4.561012658227848, |
| "grad_norm": 0.25861650895507116, |
| "learning_rate": 8.648011360561126e-07, |
| "loss": 0.1891, |
| "step": 1126 |
| }, |
| { |
| "epoch": 4.56506329113924, |
| "grad_norm": 0.24820004670483423, |
| "learning_rate": 8.483683254755037e-07, |
| "loss": 0.177, |
| "step": 1127 |
| }, |
| { |
| "epoch": 4.569113924050633, |
| "grad_norm": 0.2614299443743974, |
| "learning_rate": 8.320897595035227e-07, |
| "loss": 0.1398, |
| "step": 1128 |
| }, |
| { |
| "epoch": 4.573164556962025, |
| "grad_norm": 0.24514860990313472, |
| "learning_rate": 8.159655692455093e-07, |
| "loss": 0.1886, |
| "step": 1129 |
| }, |
| { |
| "epoch": 4.577215189873417, |
| "grad_norm": 0.25477295006172757, |
| "learning_rate": 7.999958845634648e-07, |
| "loss": 0.1617, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.58126582278481, |
| "grad_norm": 0.25373176785844287, |
| "learning_rate": 7.841808340750478e-07, |
| "loss": 0.1629, |
| "step": 1131 |
| }, |
| { |
| "epoch": 4.585316455696202, |
| "grad_norm": 0.24946094175907213, |
| "learning_rate": 7.685205451524869e-07, |
| "loss": 0.1586, |
| "step": 1132 |
| }, |
| { |
| "epoch": 4.589367088607595, |
| "grad_norm": 0.25164322648448, |
| "learning_rate": 7.530151439216027e-07, |
| "loss": 0.1714, |
| "step": 1133 |
| }, |
| { |
| "epoch": 4.593417721518987, |
| "grad_norm": 0.2482187083912238, |
| "learning_rate": 7.376647552607675e-07, |
| "loss": 0.1605, |
| "step": 1134 |
| }, |
| { |
| "epoch": 4.597468354430379, |
| "grad_norm": 0.254681832291675, |
| "learning_rate": 7.224695027998963e-07, |
| "loss": 0.1612, |
| "step": 1135 |
| }, |
| { |
| "epoch": 4.601518987341772, |
| "grad_norm": 0.24701343583994967, |
| "learning_rate": 7.07429508919466e-07, |
| "loss": 0.1697, |
| "step": 1136 |
| }, |
| { |
| "epoch": 4.605569620253164, |
| "grad_norm": 0.250214270556023, |
| "learning_rate": 6.925448947495206e-07, |
| "loss": 0.1684, |
| "step": 1137 |
| }, |
| { |
| "epoch": 4.609620253164557, |
| "grad_norm": 0.24606017084159842, |
| "learning_rate": 6.778157801686936e-07, |
| "loss": 0.1616, |
| "step": 1138 |
| }, |
| { |
| "epoch": 4.613670886075949, |
| "grad_norm": 0.24649683121693114, |
| "learning_rate": 6.632422838032515e-07, |
| "loss": 0.157, |
| "step": 1139 |
| }, |
| { |
| "epoch": 4.6177215189873415, |
| "grad_norm": 0.25213227354523965, |
| "learning_rate": 6.488245230261281e-07, |
| "loss": 0.1795, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.6217721518987345, |
| "grad_norm": 0.24552196368775253, |
| "learning_rate": 6.345626139559868e-07, |
| "loss": 0.1655, |
| "step": 1141 |
| }, |
| { |
| "epoch": 4.6258227848101265, |
| "grad_norm": 0.24919014075877305, |
| "learning_rate": 6.204566714562866e-07, |
| "loss": 0.1605, |
| "step": 1142 |
| }, |
| { |
| "epoch": 4.629873417721519, |
| "grad_norm": 0.25750416192078346, |
| "learning_rate": 6.06506809134344e-07, |
| "loss": 0.16, |
| "step": 1143 |
| }, |
| { |
| "epoch": 4.633924050632912, |
| "grad_norm": 0.23945213990950456, |
| "learning_rate": 5.927131393404373e-07, |
| "loss": 0.1867, |
| "step": 1144 |
| }, |
| { |
| "epoch": 4.637974683544304, |
| "grad_norm": 0.25060688571127837, |
| "learning_rate": 5.790757731668817e-07, |
| "loss": 0.1677, |
| "step": 1145 |
| }, |
| { |
| "epoch": 4.642025316455696, |
| "grad_norm": 0.2554903379485748, |
| "learning_rate": 5.655948204471507e-07, |
| "loss": 0.1483, |
| "step": 1146 |
| }, |
| { |
| "epoch": 4.646075949367089, |
| "grad_norm": 0.2457861288465182, |
| "learning_rate": 5.522703897549875e-07, |
| "loss": 0.1568, |
| "step": 1147 |
| }, |
| { |
| "epoch": 4.650126582278481, |
| "grad_norm": 0.24955469145295922, |
| "learning_rate": 5.391025884035239e-07, |
| "loss": 0.1788, |
| "step": 1148 |
| }, |
| { |
| "epoch": 4.654177215189874, |
| "grad_norm": 0.245281998612079, |
| "learning_rate": 5.260915224444207e-07, |
| "loss": 0.1667, |
| "step": 1149 |
| }, |
| { |
| "epoch": 4.658227848101266, |
| "grad_norm": 0.25665943427912546, |
| "learning_rate": 5.132372966670129e-07, |
| "loss": 0.1741, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.662278481012658, |
| "grad_norm": 0.2592509771896657, |
| "learning_rate": 5.005400145974704e-07, |
| "loss": 0.1786, |
| "step": 1151 |
| }, |
| { |
| "epoch": 4.666329113924051, |
| "grad_norm": 0.2379174437638, |
| "learning_rate": 4.879997784979562e-07, |
| "loss": 0.1556, |
| "step": 1152 |
| }, |
| { |
| "epoch": 4.670379746835443, |
| "grad_norm": 0.24899676219865396, |
| "learning_rate": 4.7561668936580984e-07, |
| "loss": 0.1503, |
| "step": 1153 |
| }, |
| { |
| "epoch": 4.674430379746836, |
| "grad_norm": 0.24997520985957647, |
| "learning_rate": 4.6339084693272306e-07, |
| "loss": 0.1861, |
| "step": 1154 |
| }, |
| { |
| "epoch": 4.678481012658228, |
| "grad_norm": 0.24749599732115252, |
| "learning_rate": 4.5132234966395847e-07, |
| "loss": 0.1581, |
| "step": 1155 |
| }, |
| { |
| "epoch": 4.68253164556962, |
| "grad_norm": 0.23948123643752908, |
| "learning_rate": 4.3941129475752795e-07, |
| "loss": 0.1668, |
| "step": 1156 |
| }, |
| { |
| "epoch": 4.686582278481013, |
| "grad_norm": 0.2523326789525837, |
| "learning_rate": 4.27657778143431e-07, |
| "loss": 0.1669, |
| "step": 1157 |
| }, |
| { |
| "epoch": 4.690632911392405, |
| "grad_norm": 0.24413930826300798, |
| "learning_rate": 4.1606189448287757e-07, |
| "loss": 0.1846, |
| "step": 1158 |
| }, |
| { |
| "epoch": 4.694683544303797, |
| "grad_norm": 0.2502736687818893, |
| "learning_rate": 4.046237371675177e-07, |
| "loss": 0.179, |
| "step": 1159 |
| }, |
| { |
| "epoch": 4.69873417721519, |
| "grad_norm": 0.25003022784624535, |
| "learning_rate": 3.9334339831869963e-07, |
| "loss": 0.1734, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.702784810126582, |
| "grad_norm": 0.2630714783635645, |
| "learning_rate": 3.8222096878671955e-07, |
| "loss": 0.1427, |
| "step": 1161 |
| }, |
| { |
| "epoch": 4.706835443037974, |
| "grad_norm": 0.2468601822291086, |
| "learning_rate": 3.7125653815009545e-07, |
| "loss": 0.167, |
| "step": 1162 |
| }, |
| { |
| "epoch": 4.710886075949367, |
| "grad_norm": 0.2509630322622235, |
| "learning_rate": 3.6045019471484974e-07, |
| "loss": 0.1671, |
| "step": 1163 |
| }, |
| { |
| "epoch": 4.714936708860759, |
| "grad_norm": 0.25926661259863687, |
| "learning_rate": 3.498020255137813e-07, |
| "loss": 0.1649, |
| "step": 1164 |
| }, |
| { |
| "epoch": 4.718987341772152, |
| "grad_norm": 0.24737963313275468, |
| "learning_rate": 3.393121163057811e-07, |
| "loss": 0.1588, |
| "step": 1165 |
| }, |
| { |
| "epoch": 4.723037974683544, |
| "grad_norm": 0.2473387349198756, |
| "learning_rate": 3.289805515751399e-07, |
| "loss": 0.1658, |
| "step": 1166 |
| }, |
| { |
| "epoch": 4.727088607594936, |
| "grad_norm": 0.24856694790034461, |
| "learning_rate": 3.188074145308573e-07, |
| "loss": 0.1773, |
| "step": 1167 |
| }, |
| { |
| "epoch": 4.731139240506329, |
| "grad_norm": 0.25065162108131533, |
| "learning_rate": 3.087927871059804e-07, |
| "loss": 0.1759, |
| "step": 1168 |
| }, |
| { |
| "epoch": 4.735189873417721, |
| "grad_norm": 0.23929826629527337, |
| "learning_rate": 2.989367499569418e-07, |
| "loss": 0.1763, |
| "step": 1169 |
| }, |
| { |
| "epoch": 4.739240506329114, |
| "grad_norm": 0.2522319374337004, |
| "learning_rate": 2.8923938246290917e-07, |
| "loss": 0.1637, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.743291139240506, |
| "grad_norm": 0.24651645910247258, |
| "learning_rate": 2.7970076272514804e-07, |
| "loss": 0.1663, |
| "step": 1171 |
| }, |
| { |
| "epoch": 4.7473417721518985, |
| "grad_norm": 0.2444971199245313, |
| "learning_rate": 2.703209675663887e-07, |
| "loss": 0.1751, |
| "step": 1172 |
| }, |
| { |
| "epoch": 4.751392405063291, |
| "grad_norm": 0.2586070837589992, |
| "learning_rate": 2.6110007253021374e-07, |
| "loss": 0.1721, |
| "step": 1173 |
| }, |
| { |
| "epoch": 4.7554430379746835, |
| "grad_norm": 0.2465185352170262, |
| "learning_rate": 2.520381518804471e-07, |
| "loss": 0.1804, |
| "step": 1174 |
| }, |
| { |
| "epoch": 4.759493670886076, |
| "grad_norm": 0.24386437668703717, |
| "learning_rate": 2.4313527860054585e-07, |
| "loss": 0.1587, |
| "step": 1175 |
| }, |
| { |
| "epoch": 4.7635443037974685, |
| "grad_norm": 0.24928924089435972, |
| "learning_rate": 2.343915243930317e-07, |
| "loss": 0.1762, |
| "step": 1176 |
| }, |
| { |
| "epoch": 4.767594936708861, |
| "grad_norm": 0.24173069339567727, |
| "learning_rate": 2.2580695967889367e-07, |
| "loss": 0.1356, |
| "step": 1177 |
| }, |
| { |
| "epoch": 4.7716455696202535, |
| "grad_norm": 0.24379855804148468, |
| "learning_rate": 2.1738165359704189e-07, |
| "loss": 0.1661, |
| "step": 1178 |
| }, |
| { |
| "epoch": 4.775696202531646, |
| "grad_norm": 0.25630903967165347, |
| "learning_rate": 2.0911567400373257e-07, |
| "loss": 0.1888, |
| "step": 1179 |
| }, |
| { |
| "epoch": 4.779746835443038, |
| "grad_norm": 0.24891219588488542, |
| "learning_rate": 2.0100908747202607e-07, |
| "loss": 0.1721, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.783797468354431, |
| "grad_norm": 0.2461967745823798, |
| "learning_rate": 1.9306195929125638e-07, |
| "loss": 0.182, |
| "step": 1181 |
| }, |
| { |
| "epoch": 4.787848101265823, |
| "grad_norm": 0.24897960340893166, |
| "learning_rate": 1.8527435346650247e-07, |
| "loss": 0.1318, |
| "step": 1182 |
| }, |
| { |
| "epoch": 4.791898734177215, |
| "grad_norm": 0.24435245409717682, |
| "learning_rate": 1.7764633271807108e-07, |
| "loss": 0.1672, |
| "step": 1183 |
| }, |
| { |
| "epoch": 4.795949367088608, |
| "grad_norm": 0.24731628008235412, |
| "learning_rate": 1.7017795848099262e-07, |
| "loss": 0.1594, |
| "step": 1184 |
| }, |
| { |
| "epoch": 4.8, |
| "grad_norm": 0.2450467692065222, |
| "learning_rate": 1.6286929090452596e-07, |
| "loss": 0.1625, |
| "step": 1185 |
| }, |
| { |
| "epoch": 4.804050632911393, |
| "grad_norm": 0.24955190351919856, |
| "learning_rate": 1.557203888516745e-07, |
| "loss": 0.1659, |
| "step": 1186 |
| }, |
| { |
| "epoch": 4.808101265822785, |
| "grad_norm": 0.2605114443049789, |
| "learning_rate": 1.487313098987131e-07, |
| "loss": 0.1622, |
| "step": 1187 |
| }, |
| { |
| "epoch": 4.812151898734177, |
| "grad_norm": 0.24479021330914083, |
| "learning_rate": 1.4190211033472402e-07, |
| "loss": 0.1727, |
| "step": 1188 |
| }, |
| { |
| "epoch": 4.81620253164557, |
| "grad_norm": 0.2517877420881239, |
| "learning_rate": 1.3523284516113955e-07, |
| "loss": 0.1747, |
| "step": 1189 |
| }, |
| { |
| "epoch": 4.820253164556962, |
| "grad_norm": 0.25528618187563434, |
| "learning_rate": 1.2872356809130682e-07, |
| "loss": 0.1736, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.824303797468355, |
| "grad_norm": 0.24601699538547153, |
| "learning_rate": 1.2237433155004807e-07, |
| "loss": 0.1489, |
| "step": 1191 |
| }, |
| { |
| "epoch": 4.828354430379747, |
| "grad_norm": 0.2553322363094143, |
| "learning_rate": 1.1618518667323886e-07, |
| "loss": 0.1812, |
| "step": 1192 |
| }, |
| { |
| "epoch": 4.832405063291139, |
| "grad_norm": 0.2507365864822915, |
| "learning_rate": 1.1015618330740385e-07, |
| "loss": 0.1806, |
| "step": 1193 |
| }, |
| { |
| "epoch": 4.836455696202532, |
| "grad_norm": 0.24908693060376746, |
| "learning_rate": 1.042873700093061e-07, |
| "loss": 0.1702, |
| "step": 1194 |
| }, |
| { |
| "epoch": 4.840506329113924, |
| "grad_norm": 0.2506830493606635, |
| "learning_rate": 9.857879404556291e-08, |
| "loss": 0.1866, |
| "step": 1195 |
| }, |
| { |
| "epoch": 4.844556962025316, |
| "grad_norm": 0.2614829497899598, |
| "learning_rate": 9.303050139225722e-08, |
| "loss": 0.1639, |
| "step": 1196 |
| }, |
| { |
| "epoch": 4.848607594936709, |
| "grad_norm": 0.24948539756474658, |
| "learning_rate": 8.76425367345779e-08, |
| "loss": 0.1654, |
| "step": 1197 |
| }, |
| { |
| "epoch": 4.852658227848101, |
| "grad_norm": 0.252597951532225, |
| "learning_rate": 8.241494346644897e-08, |
| "loss": 0.1479, |
| "step": 1198 |
| }, |
| { |
| "epoch": 4.856708860759493, |
| "grad_norm": 0.24703234844543911, |
| "learning_rate": 7.734776369019204e-08, |
| "loss": 0.1536, |
| "step": 1199 |
| }, |
| { |
| "epoch": 4.860759493670886, |
| "grad_norm": 0.2471488915764657, |
| "learning_rate": 7.244103821617332e-08, |
| "loss": 0.164, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.864810126582278, |
| "grad_norm": 0.2485139014142052, |
| "learning_rate": 6.769480656248606e-08, |
| "loss": 0.1698, |
| "step": 1201 |
| }, |
| { |
| "epoch": 4.868860759493671, |
| "grad_norm": 0.23885647489818349, |
| "learning_rate": 6.310910695462635e-08, |
| "loss": 0.1633, |
| "step": 1202 |
| }, |
| { |
| "epoch": 4.872911392405063, |
| "grad_norm": 0.2600562108195837, |
| "learning_rate": 5.8683976325191185e-08, |
| "loss": 0.164, |
| "step": 1203 |
| }, |
| { |
| "epoch": 4.876962025316455, |
| "grad_norm": 0.24773973700947138, |
| "learning_rate": 5.4419450313571984e-08, |
| "loss": 0.1607, |
| "step": 1204 |
| }, |
| { |
| "epoch": 4.881012658227848, |
| "grad_norm": 0.24187138407273887, |
| "learning_rate": 5.031556326567488e-08, |
| "loss": 0.1969, |
| "step": 1205 |
| }, |
| { |
| "epoch": 4.88506329113924, |
| "grad_norm": 0.25116418518274686, |
| "learning_rate": 4.637234823364312e-08, |
| "loss": 0.1735, |
| "step": 1206 |
| }, |
| { |
| "epoch": 4.889113924050633, |
| "grad_norm": 0.24951370408949555, |
| "learning_rate": 4.258983697558838e-08, |
| "loss": 0.1653, |
| "step": 1207 |
| }, |
| { |
| "epoch": 4.8931645569620255, |
| "grad_norm": 0.2576640013511211, |
| "learning_rate": 3.896805995533548e-08, |
| "loss": 0.1674, |
| "step": 1208 |
| }, |
| { |
| "epoch": 4.8972151898734175, |
| "grad_norm": 0.24794653806830305, |
| "learning_rate": 3.550704634218028e-08, |
| "loss": 0.1574, |
| "step": 1209 |
| }, |
| { |
| "epoch": 4.9012658227848105, |
| "grad_norm": 0.25934738001516044, |
| "learning_rate": 3.2206824010647676e-08, |
| "loss": 0.163, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.905316455696203, |
| "grad_norm": 0.251197821825792, |
| "learning_rate": 2.9067419540278476e-08, |
| "loss": 0.1625, |
| "step": 1211 |
| }, |
| { |
| "epoch": 4.909367088607595, |
| "grad_norm": 0.2449394658188603, |
| "learning_rate": 2.6088858215400638e-08, |
| "loss": 0.1606, |
| "step": 1212 |
| }, |
| { |
| "epoch": 4.913417721518988, |
| "grad_norm": 0.24959789329299378, |
| "learning_rate": 2.3271164024940564e-08, |
| "loss": 0.1761, |
| "step": 1213 |
| }, |
| { |
| "epoch": 4.91746835443038, |
| "grad_norm": 0.2494405478067043, |
| "learning_rate": 2.061435966221881e-08, |
| "loss": 0.185, |
| "step": 1214 |
| }, |
| { |
| "epoch": 4.921518987341772, |
| "grad_norm": 0.2438368779393808, |
| "learning_rate": 1.811846652477245e-08, |
| "loss": 0.177, |
| "step": 1215 |
| }, |
| { |
| "epoch": 4.925569620253165, |
| "grad_norm": 0.23569528814449206, |
| "learning_rate": 1.5783504714184106e-08, |
| "loss": 0.1887, |
| "step": 1216 |
| }, |
| { |
| "epoch": 4.929620253164557, |
| "grad_norm": 0.25783342034879503, |
| "learning_rate": 1.360949303591097e-08, |
| "loss": 0.1721, |
| "step": 1217 |
| }, |
| { |
| "epoch": 4.93367088607595, |
| "grad_norm": 0.2490821570976755, |
| "learning_rate": 1.1596448999144916e-08, |
| "loss": 0.1854, |
| "step": 1218 |
| }, |
| { |
| "epoch": 4.937721518987342, |
| "grad_norm": 0.25742991347645655, |
| "learning_rate": 9.744388816668172e-09, |
| "loss": 0.1492, |
| "step": 1219 |
| }, |
| { |
| "epoch": 4.941772151898734, |
| "grad_norm": 0.24463004074875397, |
| "learning_rate": 8.05332740472009e-09, |
| "loss": 0.1658, |
| "step": 1220 |
| }, |
| { |
| "epoch": 4.945822784810127, |
| "grad_norm": 0.2426031643484562, |
| "learning_rate": 6.523278382872811e-09, |
| "loss": 0.1732, |
| "step": 1221 |
| }, |
| { |
| "epoch": 4.949873417721519, |
| "grad_norm": 0.2534269520588756, |
| "learning_rate": 5.15425407393133e-09, |
| "loss": 0.1924, |
| "step": 1222 |
| }, |
| { |
| "epoch": 4.953924050632912, |
| "grad_norm": 0.24658232971598232, |
| "learning_rate": 3.94626550383137e-09, |
| "loss": 0.1768, |
| "step": 1223 |
| }, |
| { |
| "epoch": 4.957974683544304, |
| "grad_norm": 0.24169534983026078, |
| "learning_rate": 2.899322401546112e-09, |
| "loss": 0.1685, |
| "step": 1224 |
| }, |
| { |
| "epoch": 4.962025316455696, |
| "grad_norm": 0.2480416152486689, |
| "learning_rate": 2.013433199010706e-09, |
| "loss": 0.1786, |
| "step": 1225 |
| }, |
| { |
| "epoch": 4.966075949367089, |
| "grad_norm": 0.24661193404651735, |
| "learning_rate": 1.2886050310556563e-09, |
| "loss": 0.1657, |
| "step": 1226 |
| }, |
| { |
| "epoch": 4.970126582278481, |
| "grad_norm": 0.2395416892211065, |
| "learning_rate": 7.248437353468695e-10, |
| "loss": 0.1726, |
| "step": 1227 |
| }, |
| { |
| "epoch": 4.974177215189873, |
| "grad_norm": 0.2476747733982238, |
| "learning_rate": 3.221538523412449e-10, |
| "loss": 0.1764, |
| "step": 1228 |
| }, |
| { |
| "epoch": 4.978227848101266, |
| "grad_norm": 0.23991938409564686, |
| "learning_rate": 8.053862524670663e-11, |
| "loss": 0.1812, |
| "step": 1229 |
| }, |
| { |
| "epoch": 4.982278481012658, |
| "grad_norm": 0.24423711225003375, |
| "learning_rate": 0.0, |
| "loss": 0.1733, |
| "step": 1230 |
| }, |
| { |
| "epoch": 4.982278481012658, |
| "step": 1230, |
| "total_flos": 2.502462916635525e+18, |
| "train_loss": 0.376850108160236, |
| "train_runtime": 39530.5983, |
| "train_samples_per_second": 3.997, |
| "train_steps_per_second": 0.031 |
| } |
| ], |
| "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": 2.502462916635525e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
