| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.988354430379747, | |
| "eval_steps": 500, | |
| "global_step": 1230, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004050632911392405, | |
| "grad_norm": 9.109492965987231, | |
| "learning_rate": 3.2520325203252037e-07, | |
| "loss": 0.9011, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00810126582278481, | |
| "grad_norm": 9.072326903006184, | |
| "learning_rate": 6.504065040650407e-07, | |
| "loss": 0.9076, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012151898734177215, | |
| "grad_norm": 8.794940962326047, | |
| "learning_rate": 9.75609756097561e-07, | |
| "loss": 0.8885, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01620253164556962, | |
| "grad_norm": 9.05373253333553, | |
| "learning_rate": 1.3008130081300815e-06, | |
| "loss": 0.9066, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.020253164556962026, | |
| "grad_norm": 8.359507853032012, | |
| "learning_rate": 1.6260162601626018e-06, | |
| "loss": 0.8876, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02430379746835443, | |
| "grad_norm": 7.772040353605313, | |
| "learning_rate": 1.951219512195122e-06, | |
| "loss": 0.8529, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028354430379746835, | |
| "grad_norm": 6.463382585723644, | |
| "learning_rate": 2.2764227642276426e-06, | |
| "loss": 0.8326, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03240506329113924, | |
| "grad_norm": 6.019812816511983, | |
| "learning_rate": 2.601626016260163e-06, | |
| "loss": 0.7928, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03645569620253165, | |
| "grad_norm": 3.5328550694792282, | |
| "learning_rate": 2.926829268292683e-06, | |
| "loss": 0.7337, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04050632911392405, | |
| "grad_norm": 3.1589888732974294, | |
| "learning_rate": 3.2520325203252037e-06, | |
| "loss": 0.7358, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044556962025316456, | |
| "grad_norm": 2.9341613473848005, | |
| "learning_rate": 3.577235772357724e-06, | |
| "loss": 0.7383, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04860759493670886, | |
| "grad_norm": 4.287622014906312, | |
| "learning_rate": 3.902439024390244e-06, | |
| "loss": 0.6989, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.052658227848101265, | |
| "grad_norm": 3.8291315840470554, | |
| "learning_rate": 4.227642276422765e-06, | |
| "loss": 0.6923, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05670886075949367, | |
| "grad_norm": 3.786887107595458, | |
| "learning_rate": 4.552845528455285e-06, | |
| "loss": 0.6635, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.060759493670886074, | |
| "grad_norm": 3.2263676465801, | |
| "learning_rate": 4.8780487804878055e-06, | |
| "loss": 0.6285, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06481012658227848, | |
| "grad_norm": 2.861433471605965, | |
| "learning_rate": 5.203252032520326e-06, | |
| "loss": 0.5988, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06886075949367089, | |
| "grad_norm": 2.6564301943018713, | |
| "learning_rate": 5.528455284552846e-06, | |
| "loss": 0.5511, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0729113924050633, | |
| "grad_norm": 2.3081995512582343, | |
| "learning_rate": 5.853658536585366e-06, | |
| "loss": 0.5497, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0769620253164557, | |
| "grad_norm": 2.0637618504455095, | |
| "learning_rate": 6.178861788617887e-06, | |
| "loss": 0.5517, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0810126582278481, | |
| "grad_norm": 1.4082912162506322, | |
| "learning_rate": 6.504065040650407e-06, | |
| "loss": 0.5598, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08506329113924051, | |
| "grad_norm": 1.3269009327409937, | |
| "learning_rate": 6.829268292682928e-06, | |
| "loss": 0.526, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08911392405063291, | |
| "grad_norm": 1.8031896641220637, | |
| "learning_rate": 7.154471544715448e-06, | |
| "loss": 0.5192, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09316455696202532, | |
| "grad_norm": 1.7326448215089711, | |
| "learning_rate": 7.4796747967479676e-06, | |
| "loss": 0.5101, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09721518987341772, | |
| "grad_norm": 1.0237064415420234, | |
| "learning_rate": 7.804878048780489e-06, | |
| "loss": 0.4955, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 1.12604077136826, | |
| "learning_rate": 8.130081300813009e-06, | |
| "loss": 0.5032, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10531645569620253, | |
| "grad_norm": 1.2464806084319213, | |
| "learning_rate": 8.45528455284553e-06, | |
| "loss": 0.4972, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10936708860759493, | |
| "grad_norm": 0.9839181745024805, | |
| "learning_rate": 8.78048780487805e-06, | |
| "loss": 0.4605, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11341772151898734, | |
| "grad_norm": 1.0151964627098478, | |
| "learning_rate": 9.10569105691057e-06, | |
| "loss": 0.4854, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11746835443037974, | |
| "grad_norm": 0.8026217038493326, | |
| "learning_rate": 9.43089430894309e-06, | |
| "loss": 0.4713, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12151898734177215, | |
| "grad_norm": 0.8342061836033832, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 0.4753, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12556962025316457, | |
| "grad_norm": 0.9000980659843828, | |
| "learning_rate": 1.008130081300813e-05, | |
| "loss": 0.4762, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12962025316455697, | |
| "grad_norm": 0.7607574964567417, | |
| "learning_rate": 1.0406504065040652e-05, | |
| "loss": 0.4645, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13367088607594937, | |
| "grad_norm": 0.801622592436485, | |
| "learning_rate": 1.0731707317073172e-05, | |
| "loss": 0.4492, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13772151898734178, | |
| "grad_norm": 0.9159762306181388, | |
| "learning_rate": 1.1056910569105692e-05, | |
| "loss": 0.4668, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14177215189873418, | |
| "grad_norm": 0.7037826667289567, | |
| "learning_rate": 1.1382113821138213e-05, | |
| "loss": 0.4542, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1458227848101266, | |
| "grad_norm": 0.7971054334018406, | |
| "learning_rate": 1.1707317073170731e-05, | |
| "loss": 0.4606, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.149873417721519, | |
| "grad_norm": 0.6984017310743663, | |
| "learning_rate": 1.2032520325203254e-05, | |
| "loss": 0.4494, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1539240506329114, | |
| "grad_norm": 0.8493343098715946, | |
| "learning_rate": 1.2357723577235774e-05, | |
| "loss": 0.456, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1579746835443038, | |
| "grad_norm": 0.7355797416821558, | |
| "learning_rate": 1.2682926829268294e-05, | |
| "loss": 0.4517, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1620253164556962, | |
| "grad_norm": 0.7147578783730126, | |
| "learning_rate": 1.3008130081300815e-05, | |
| "loss": 0.4479, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1660759493670886, | |
| "grad_norm": 0.6714946959234974, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.4528, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17012658227848101, | |
| "grad_norm": 0.7816552733944113, | |
| "learning_rate": 1.3658536585365855e-05, | |
| "loss": 0.4387, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17417721518987342, | |
| "grad_norm": 0.649644772548843, | |
| "learning_rate": 1.3983739837398376e-05, | |
| "loss": 0.4443, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17822784810126582, | |
| "grad_norm": 0.7429068573300249, | |
| "learning_rate": 1.4308943089430896e-05, | |
| "loss": 0.4376, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18227848101265823, | |
| "grad_norm": 0.7183868044938194, | |
| "learning_rate": 1.4634146341463415e-05, | |
| "loss": 0.4467, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18632911392405063, | |
| "grad_norm": 0.8922563339336295, | |
| "learning_rate": 1.4959349593495935e-05, | |
| "loss": 0.4406, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.19037974683544304, | |
| "grad_norm": 0.7564804550270525, | |
| "learning_rate": 1.528455284552846e-05, | |
| "loss": 0.4281, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.19443037974683544, | |
| "grad_norm": 0.736901533038964, | |
| "learning_rate": 1.5609756097560978e-05, | |
| "loss": 0.4418, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19848101265822785, | |
| "grad_norm": 0.8909413971491137, | |
| "learning_rate": 1.5934959349593496e-05, | |
| "loss": 0.4348, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 0.7677386212286397, | |
| "learning_rate": 1.6260162601626018e-05, | |
| "loss": 0.4326, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.20658227848101265, | |
| "grad_norm": 0.7350296680307036, | |
| "learning_rate": 1.6585365853658537e-05, | |
| "loss": 0.4202, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21063291139240506, | |
| "grad_norm": 0.6496785156672198, | |
| "learning_rate": 1.691056910569106e-05, | |
| "loss": 0.415, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21468354430379746, | |
| "grad_norm": 0.7005244246480724, | |
| "learning_rate": 1.7235772357723578e-05, | |
| "loss": 0.4355, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21873417721518987, | |
| "grad_norm": 0.6795226522508696, | |
| "learning_rate": 1.75609756097561e-05, | |
| "loss": 0.4291, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22278481012658227, | |
| "grad_norm": 0.6318573325717359, | |
| "learning_rate": 1.788617886178862e-05, | |
| "loss": 0.4422, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22683544303797468, | |
| "grad_norm": 0.8592320447255203, | |
| "learning_rate": 1.821138211382114e-05, | |
| "loss": 0.3966, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23088607594936708, | |
| "grad_norm": 0.7694198729719192, | |
| "learning_rate": 1.8536585365853663e-05, | |
| "loss": 0.4228, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23493670886075949, | |
| "grad_norm": 0.6669329408131133, | |
| "learning_rate": 1.886178861788618e-05, | |
| "loss": 0.4188, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2389873417721519, | |
| "grad_norm": 0.669690501563595, | |
| "learning_rate": 1.91869918699187e-05, | |
| "loss": 0.4165, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2430379746835443, | |
| "grad_norm": 0.7071314795025191, | |
| "learning_rate": 1.9512195121951222e-05, | |
| "loss": 0.4166, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2470886075949367, | |
| "grad_norm": 0.6291224644608354, | |
| "learning_rate": 1.983739837398374e-05, | |
| "loss": 0.4218, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25113924050632913, | |
| "grad_norm": 0.6513900502802937, | |
| "learning_rate": 2.016260162601626e-05, | |
| "loss": 0.436, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.25518987341772154, | |
| "grad_norm": 0.6748875980416997, | |
| "learning_rate": 2.048780487804878e-05, | |
| "loss": 0.4134, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.25924050632911394, | |
| "grad_norm": 0.6295876902765336, | |
| "learning_rate": 2.0813008130081303e-05, | |
| "loss": 0.4033, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.26329113924050634, | |
| "grad_norm": 0.7075915399020716, | |
| "learning_rate": 2.1138211382113822e-05, | |
| "loss": 0.399, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26734177215189875, | |
| "grad_norm": 0.6519358844543466, | |
| "learning_rate": 2.1463414634146344e-05, | |
| "loss": 0.4235, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27139240506329115, | |
| "grad_norm": 0.7979270134824519, | |
| "learning_rate": 2.1788617886178863e-05, | |
| "loss": 0.4069, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.27544303797468356, | |
| "grad_norm": 0.7694220986883973, | |
| "learning_rate": 2.2113821138211385e-05, | |
| "loss": 0.3934, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.27949367088607596, | |
| "grad_norm": 0.6102149696309562, | |
| "learning_rate": 2.2439024390243907e-05, | |
| "loss": 0.4143, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.28354430379746837, | |
| "grad_norm": 0.7511021995991899, | |
| "learning_rate": 2.2764227642276426e-05, | |
| "loss": 0.4116, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.28759493670886077, | |
| "grad_norm": 0.624615418325311, | |
| "learning_rate": 2.3089430894308948e-05, | |
| "loss": 0.4201, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2916455696202532, | |
| "grad_norm": 0.773943377020293, | |
| "learning_rate": 2.3414634146341463e-05, | |
| "loss": 0.4137, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2956962025316456, | |
| "grad_norm": 0.7291378572197738, | |
| "learning_rate": 2.3739837398373985e-05, | |
| "loss": 0.4287, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.299746835443038, | |
| "grad_norm": 0.7653173181960037, | |
| "learning_rate": 2.4065040650406507e-05, | |
| "loss": 0.3938, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 0.8644079877484884, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "loss": 0.4328, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3078481012658228, | |
| "grad_norm": 0.7024591936510975, | |
| "learning_rate": 2.4715447154471548e-05, | |
| "loss": 0.3945, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3118987341772152, | |
| "grad_norm": 0.8616560002776101, | |
| "learning_rate": 2.5040650406504066e-05, | |
| "loss": 0.3963, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3159493670886076, | |
| "grad_norm": 1.005603843194374, | |
| "learning_rate": 2.536585365853659e-05, | |
| "loss": 0.397, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1.182178898269826, | |
| "learning_rate": 2.569105691056911e-05, | |
| "loss": 0.4042, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3240506329113924, | |
| "grad_norm": 0.732609278937735, | |
| "learning_rate": 2.601626016260163e-05, | |
| "loss": 0.3978, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3281012658227848, | |
| "grad_norm": 0.9946834640057717, | |
| "learning_rate": 2.634146341463415e-05, | |
| "loss": 0.4151, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3321518987341772, | |
| "grad_norm": 0.9588762329804992, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.4035, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3362025316455696, | |
| "grad_norm": 1.0892011063861522, | |
| "learning_rate": 2.699186991869919e-05, | |
| "loss": 0.4022, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.34025316455696203, | |
| "grad_norm": 1.0471283551168618, | |
| "learning_rate": 2.731707317073171e-05, | |
| "loss": 0.4025, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.34430379746835443, | |
| "grad_norm": 0.7110498677446814, | |
| "learning_rate": 2.764227642276423e-05, | |
| "loss": 0.3947, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.34835443037974684, | |
| "grad_norm": 0.8960187224119011, | |
| "learning_rate": 2.796747967479675e-05, | |
| "loss": 0.4128, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35240506329113924, | |
| "grad_norm": 1.129981560885454, | |
| "learning_rate": 2.829268292682927e-05, | |
| "loss": 0.3885, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.35645569620253165, | |
| "grad_norm": 0.749510462435202, | |
| "learning_rate": 2.8617886178861792e-05, | |
| "loss": 0.4082, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.36050632911392405, | |
| "grad_norm": 0.8284907311161714, | |
| "learning_rate": 2.8943089430894314e-05, | |
| "loss": 0.4049, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36455696202531646, | |
| "grad_norm": 0.5915945309208293, | |
| "learning_rate": 2.926829268292683e-05, | |
| "loss": 0.401, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.36860759493670886, | |
| "grad_norm": 0.8322985022912504, | |
| "learning_rate": 2.959349593495935e-05, | |
| "loss": 0.3875, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37265822784810126, | |
| "grad_norm": 0.7063608048568366, | |
| "learning_rate": 2.991869918699187e-05, | |
| "loss": 0.4005, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.37670886075949367, | |
| "grad_norm": 0.8550785768522986, | |
| "learning_rate": 3.0243902439024392e-05, | |
| "loss": 0.3892, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3807594936708861, | |
| "grad_norm": 0.8079247431142896, | |
| "learning_rate": 3.056910569105692e-05, | |
| "loss": 0.4068, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3848101265822785, | |
| "grad_norm": 0.8723343912069774, | |
| "learning_rate": 3.089430894308943e-05, | |
| "loss": 0.404, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3888607594936709, | |
| "grad_norm": 0.7843073501613207, | |
| "learning_rate": 3.1219512195121955e-05, | |
| "loss": 0.405, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3929113924050633, | |
| "grad_norm": 0.7888073075156519, | |
| "learning_rate": 3.154471544715447e-05, | |
| "loss": 0.4164, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3969620253164557, | |
| "grad_norm": 0.7656115976307064, | |
| "learning_rate": 3.186991869918699e-05, | |
| "loss": 0.3959, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4010126582278481, | |
| "grad_norm": 0.8161976206671736, | |
| "learning_rate": 3.2195121951219514e-05, | |
| "loss": 0.4038, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 0.8250255567576916, | |
| "learning_rate": 3.2520325203252037e-05, | |
| "loss": 0.3917, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4091139240506329, | |
| "grad_norm": 0.8866491093716544, | |
| "learning_rate": 3.284552845528456e-05, | |
| "loss": 0.421, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4131645569620253, | |
| "grad_norm": 0.7483581060201607, | |
| "learning_rate": 3.3170731707317074e-05, | |
| "loss": 0.4163, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4172151898734177, | |
| "grad_norm": 0.7919525972421952, | |
| "learning_rate": 3.3495934959349596e-05, | |
| "loss": 0.4005, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4212658227848101, | |
| "grad_norm": 0.713598768210866, | |
| "learning_rate": 3.382113821138212e-05, | |
| "loss": 0.406, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4253164556962025, | |
| "grad_norm": 0.6681620145703522, | |
| "learning_rate": 3.414634146341463e-05, | |
| "loss": 0.3877, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4293670886075949, | |
| "grad_norm": 0.7483442975956845, | |
| "learning_rate": 3.4471544715447155e-05, | |
| "loss": 0.4091, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.43341772151898733, | |
| "grad_norm": 0.621426649431444, | |
| "learning_rate": 3.479674796747968e-05, | |
| "loss": 0.3962, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43746835443037974, | |
| "grad_norm": 0.9214795856952341, | |
| "learning_rate": 3.51219512195122e-05, | |
| "loss": 0.3961, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44151898734177214, | |
| "grad_norm": 0.8289568973445475, | |
| "learning_rate": 3.544715447154472e-05, | |
| "loss": 0.3997, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44556962025316454, | |
| "grad_norm": 0.8165673750165096, | |
| "learning_rate": 3.577235772357724e-05, | |
| "loss": 0.3955, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44962025316455695, | |
| "grad_norm": 0.6229681378183765, | |
| "learning_rate": 3.609756097560976e-05, | |
| "loss": 0.4003, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.45367088607594935, | |
| "grad_norm": 0.8181221831649735, | |
| "learning_rate": 3.642276422764228e-05, | |
| "loss": 0.3893, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.45772151898734176, | |
| "grad_norm": 0.7641537060590827, | |
| "learning_rate": 3.67479674796748e-05, | |
| "loss": 0.3844, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.46177215189873416, | |
| "grad_norm": 0.7642177526817286, | |
| "learning_rate": 3.7073170731707325e-05, | |
| "loss": 0.3931, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46582278481012657, | |
| "grad_norm": 0.8935668159511603, | |
| "learning_rate": 3.739837398373984e-05, | |
| "loss": 0.403, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.46987341772151897, | |
| "grad_norm": 0.7258185703605878, | |
| "learning_rate": 3.772357723577236e-05, | |
| "loss": 0.3976, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4739240506329114, | |
| "grad_norm": 0.7234895966845565, | |
| "learning_rate": 3.804878048780488e-05, | |
| "loss": 0.4003, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4779746835443038, | |
| "grad_norm": 0.9259277635368078, | |
| "learning_rate": 3.83739837398374e-05, | |
| "loss": 0.4045, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4820253164556962, | |
| "grad_norm": 0.9148561647476231, | |
| "learning_rate": 3.869918699186992e-05, | |
| "loss": 0.4174, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4860759493670886, | |
| "grad_norm": 0.6038207327660995, | |
| "learning_rate": 3.9024390243902444e-05, | |
| "loss": 0.4012, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.490126582278481, | |
| "grad_norm": 0.7858405630709782, | |
| "learning_rate": 3.9349593495934966e-05, | |
| "loss": 0.3949, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4941772151898734, | |
| "grad_norm": 0.694675420788813, | |
| "learning_rate": 3.967479674796748e-05, | |
| "loss": 0.412, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.4982278481012658, | |
| "grad_norm": 0.6982284127565913, | |
| "learning_rate": 4e-05, | |
| "loss": 0.3961, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5022784810126583, | |
| "grad_norm": 0.766394710258729, | |
| "learning_rate": 3.999991946137476e-05, | |
| "loss": 0.4282, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 0.7465187021859281, | |
| "learning_rate": 3.999967784614766e-05, | |
| "loss": 0.404, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5103797468354431, | |
| "grad_norm": 0.6783849522840777, | |
| "learning_rate": 3.9999275156264656e-05, | |
| "loss": 0.389, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5144303797468355, | |
| "grad_norm": 0.7696825908946998, | |
| "learning_rate": 3.999871139496895e-05, | |
| "loss": 0.3939, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5184810126582279, | |
| "grad_norm": 0.7984656584012791, | |
| "learning_rate": 3.9997986566800995e-05, | |
| "loss": 0.4149, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5225316455696203, | |
| "grad_norm": 0.6172573428335758, | |
| "learning_rate": 3.999710067759846e-05, | |
| "loss": 0.4025, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5265822784810127, | |
| "grad_norm": 0.651742831873365, | |
| "learning_rate": 3.999605373449617e-05, | |
| "loss": 0.4137, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5306329113924051, | |
| "grad_norm": 0.6955933143620815, | |
| "learning_rate": 3.9994845745926075e-05, | |
| "loss": 0.4141, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5346835443037975, | |
| "grad_norm": 0.7533811843897135, | |
| "learning_rate": 3.999347672161713e-05, | |
| "loss": 0.4195, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5387341772151899, | |
| "grad_norm": 0.602370100398438, | |
| "learning_rate": 3.999194667259528e-05, | |
| "loss": 0.4063, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5427848101265823, | |
| "grad_norm": 0.5532654134940315, | |
| "learning_rate": 3.999025561118334e-05, | |
| "loss": 0.4122, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5468354430379747, | |
| "grad_norm": 0.6561523356300223, | |
| "learning_rate": 3.998840355100086e-05, | |
| "loss": 0.3976, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5508860759493671, | |
| "grad_norm": 0.6048301913755542, | |
| "learning_rate": 3.998639050696409e-05, | |
| "loss": 0.4084, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5549367088607595, | |
| "grad_norm": 0.5977305599570522, | |
| "learning_rate": 3.998421649528582e-05, | |
| "loss": 0.4075, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5589873417721519, | |
| "grad_norm": 0.5049015386978117, | |
| "learning_rate": 3.9981881533475234e-05, | |
| "loss": 0.3782, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5630379746835443, | |
| "grad_norm": 0.5733912800979764, | |
| "learning_rate": 3.997938564033779e-05, | |
| "loss": 0.3802, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5670886075949367, | |
| "grad_norm": 0.5862448069339645, | |
| "learning_rate": 3.9976728835975064e-05, | |
| "loss": 0.3912, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5711392405063291, | |
| "grad_norm": 0.6823927404953157, | |
| "learning_rate": 3.9973911141784605e-05, | |
| "loss": 0.407, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5751898734177215, | |
| "grad_norm": 0.5560924494458189, | |
| "learning_rate": 3.997093258045973e-05, | |
| "loss": 0.3908, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.579240506329114, | |
| "grad_norm": 0.6003443009595604, | |
| "learning_rate": 3.996779317598936e-05, | |
| "loss": 0.3881, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5832911392405064, | |
| "grad_norm": 0.7829473726398428, | |
| "learning_rate": 3.996449295365782e-05, | |
| "loss": 0.3948, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5873417721518988, | |
| "grad_norm": 0.6837096958661966, | |
| "learning_rate": 3.996103194004467e-05, | |
| "loss": 0.4075, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5913924050632912, | |
| "grad_norm": 0.735869137091558, | |
| "learning_rate": 3.995741016302441e-05, | |
| "loss": 0.402, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5954430379746836, | |
| "grad_norm": 0.5401232370872774, | |
| "learning_rate": 3.9953627651766364e-05, | |
| "loss": 0.4065, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.599493670886076, | |
| "grad_norm": 0.7820019661213403, | |
| "learning_rate": 3.9949684436734325e-05, | |
| "loss": 0.3856, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6035443037974684, | |
| "grad_norm": 0.7118302667820393, | |
| "learning_rate": 3.994558054968643e-05, | |
| "loss": 0.4126, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 0.6699903654065317, | |
| "learning_rate": 3.994131602367481e-05, | |
| "loss": 0.3876, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6116455696202532, | |
| "grad_norm": 0.640040387049364, | |
| "learning_rate": 3.9936890893045376e-05, | |
| "loss": 0.3894, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6156962025316456, | |
| "grad_norm": 0.5441478312201011, | |
| "learning_rate": 3.993230519343752e-05, | |
| "loss": 0.3893, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.619746835443038, | |
| "grad_norm": 0.7051785510552822, | |
| "learning_rate": 3.992755896178383e-05, | |
| "loss": 0.3968, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6237974683544304, | |
| "grad_norm": 0.5731926149532757, | |
| "learning_rate": 3.992265223630981e-05, | |
| "loss": 0.4033, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6278481012658228, | |
| "grad_norm": 0.6336167907225031, | |
| "learning_rate": 3.991758505653355e-05, | |
| "loss": 0.3962, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6318987341772152, | |
| "grad_norm": 0.608315362311643, | |
| "learning_rate": 3.991235746326543e-05, | |
| "loss": 0.3985, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6359493670886076, | |
| "grad_norm": 0.6327124295909283, | |
| "learning_rate": 3.9906969498607745e-05, | |
| "loss": 0.4056, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.7137645214297197, | |
| "learning_rate": 3.990142120595444e-05, | |
| "loss": 0.4194, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6440506329113924, | |
| "grad_norm": 0.570295655009781, | |
| "learning_rate": 3.98957126299907e-05, | |
| "loss": 0.3886, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6481012658227848, | |
| "grad_norm": 0.6454735339604967, | |
| "learning_rate": 3.9889843816692596e-05, | |
| "loss": 0.3921, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6521518987341772, | |
| "grad_norm": 0.6100700718455481, | |
| "learning_rate": 3.9883814813326766e-05, | |
| "loss": 0.3926, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6562025316455696, | |
| "grad_norm": 0.5711987111866352, | |
| "learning_rate": 3.9877625668449956e-05, | |
| "loss": 0.3959, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.660253164556962, | |
| "grad_norm": 0.4971495355484315, | |
| "learning_rate": 3.98712764319087e-05, | |
| "loss": 0.4024, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6643037974683544, | |
| "grad_norm": 0.5825568282400481, | |
| "learning_rate": 3.9864767154838864e-05, | |
| "loss": 0.3869, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6683544303797468, | |
| "grad_norm": 0.6372933057624928, | |
| "learning_rate": 3.9858097889665277e-05, | |
| "loss": 0.4187, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6724050632911392, | |
| "grad_norm": 0.5755132948554388, | |
| "learning_rate": 3.985126869010129e-05, | |
| "loss": 0.3935, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6764556962025317, | |
| "grad_norm": 0.5018487299052393, | |
| "learning_rate": 3.984427961114833e-05, | |
| "loss": 0.3931, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6805063291139241, | |
| "grad_norm": 0.6120622348006484, | |
| "learning_rate": 3.9837130709095475e-05, | |
| "loss": 0.4178, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6845569620253165, | |
| "grad_norm": 0.5738076561782685, | |
| "learning_rate": 3.982982204151901e-05, | |
| "loss": 0.3808, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6886075949367089, | |
| "grad_norm": 0.5360660835784177, | |
| "learning_rate": 3.982235366728193e-05, | |
| "loss": 0.3915, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6926582278481013, | |
| "grad_norm": 0.5290295240536526, | |
| "learning_rate": 3.9814725646533505e-05, | |
| "loss": 0.3983, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6967088607594937, | |
| "grad_norm": 0.5613967380786643, | |
| "learning_rate": 3.9806938040708746e-05, | |
| "loss": 0.4028, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7007594936708861, | |
| "grad_norm": 0.621994140094819, | |
| "learning_rate": 3.9798990912527976e-05, | |
| "loss": 0.3934, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7048101265822785, | |
| "grad_norm": 0.5293982732994768, | |
| "learning_rate": 3.979088432599627e-05, | |
| "loss": 0.3962, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 0.5391722873185413, | |
| "learning_rate": 3.9782618346402964e-05, | |
| "loss": 0.3857, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7129113924050633, | |
| "grad_norm": 0.5458549291084576, | |
| "learning_rate": 3.977419304032111e-05, | |
| "loss": 0.39, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7169620253164557, | |
| "grad_norm": 0.5751004015840897, | |
| "learning_rate": 3.976560847560697e-05, | |
| "loss": 0.3977, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7210126582278481, | |
| "grad_norm": 0.5079531707902396, | |
| "learning_rate": 3.9756864721399456e-05, | |
| "loss": 0.4, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7250632911392405, | |
| "grad_norm": 0.5854183190605096, | |
| "learning_rate": 3.974796184811956e-05, | |
| "loss": 0.3907, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7291139240506329, | |
| "grad_norm": 0.5613212859489137, | |
| "learning_rate": 3.973889992746979e-05, | |
| "loss": 0.3885, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7331645569620253, | |
| "grad_norm": 0.5201534478018532, | |
| "learning_rate": 3.972967903243361e-05, | |
| "loss": 0.3917, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7372151898734177, | |
| "grad_norm": 0.5146037846451228, | |
| "learning_rate": 3.972029923727486e-05, | |
| "loss": 0.376, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7412658227848101, | |
| "grad_norm": 0.5587789152321736, | |
| "learning_rate": 3.971076061753709e-05, | |
| "loss": 0.3829, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7453164556962025, | |
| "grad_norm": 0.6328337575935045, | |
| "learning_rate": 3.9701063250043066e-05, | |
| "loss": 0.3804, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7493670886075949, | |
| "grad_norm": 0.4924849666371829, | |
| "learning_rate": 3.969120721289402e-05, | |
| "loss": 0.3713, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7534177215189873, | |
| "grad_norm": 0.7359688721157082, | |
| "learning_rate": 3.9681192585469146e-05, | |
| "loss": 0.4002, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7574683544303797, | |
| "grad_norm": 0.6467249502520852, | |
| "learning_rate": 3.9671019448424865e-05, | |
| "loss": 0.3821, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7615189873417721, | |
| "grad_norm": 0.6532333138478211, | |
| "learning_rate": 3.966068788369422e-05, | |
| "loss": 0.4132, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7655696202531646, | |
| "grad_norm": 0.5250249750536133, | |
| "learning_rate": 3.965019797448622e-05, | |
| "loss": 0.3777, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.769620253164557, | |
| "grad_norm": 0.6264164107575475, | |
| "learning_rate": 3.963954980528515e-05, | |
| "loss": 0.3909, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7736708860759494, | |
| "grad_norm": 0.5332671324575613, | |
| "learning_rate": 3.9628743461849905e-05, | |
| "loss": 0.3866, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7777215189873418, | |
| "grad_norm": 0.5490481863553106, | |
| "learning_rate": 3.961777903121329e-05, | |
| "loss": 0.3961, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7817721518987342, | |
| "grad_norm": 0.6501474988072067, | |
| "learning_rate": 3.960665660168131e-05, | |
| "loss": 0.3863, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7858227848101266, | |
| "grad_norm": 0.5798102076557473, | |
| "learning_rate": 3.9595376262832485e-05, | |
| "loss": 0.3845, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.789873417721519, | |
| "grad_norm": 0.6155895520096855, | |
| "learning_rate": 3.9583938105517127e-05, | |
| "loss": 0.4107, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7939240506329114, | |
| "grad_norm": 0.6136003153905183, | |
| "learning_rate": 3.957234222185657e-05, | |
| "loss": 0.3938, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7979746835443038, | |
| "grad_norm": 0.5336139996475284, | |
| "learning_rate": 3.9560588705242474e-05, | |
| "loss": 0.3816, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8020253164556962, | |
| "grad_norm": 0.6450305389244427, | |
| "learning_rate": 3.954867765033605e-05, | |
| "loss": 0.3877, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8060759493670886, | |
| "grad_norm": 0.47242089991087033, | |
| "learning_rate": 3.953660915306728e-05, | |
| "loss": 0.3717, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 0.4993058452370321, | |
| "learning_rate": 3.952438331063419e-05, | |
| "loss": 0.3846, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8141772151898734, | |
| "grad_norm": 0.4838017266391022, | |
| "learning_rate": 3.951200022150205e-05, | |
| "loss": 0.3765, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8182278481012658, | |
| "grad_norm": 0.46918173117982054, | |
| "learning_rate": 3.949945998540253e-05, | |
| "loss": 0.3785, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8222784810126582, | |
| "grad_norm": 0.6730651836127816, | |
| "learning_rate": 3.9486762703332993e-05, | |
| "loss": 0.3882, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8263291139240506, | |
| "grad_norm": 0.5413975356122073, | |
| "learning_rate": 3.947390847755559e-05, | |
| "loss": 0.38, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.830379746835443, | |
| "grad_norm": 0.5893999376891482, | |
| "learning_rate": 3.946089741159648e-05, | |
| "loss": 0.3803, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8344303797468354, | |
| "grad_norm": 0.6646081219756133, | |
| "learning_rate": 3.944772961024501e-05, | |
| "loss": 0.3881, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8384810126582278, | |
| "grad_norm": 0.5448417044040943, | |
| "learning_rate": 3.943440517955285e-05, | |
| "loss": 0.3886, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8425316455696202, | |
| "grad_norm": 0.5396977992485523, | |
| "learning_rate": 3.9420924226833126e-05, | |
| "loss": 0.3764, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8465822784810126, | |
| "grad_norm": 0.5635439494399355, | |
| "learning_rate": 3.9407286860659566e-05, | |
| "loss": 0.3956, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.850632911392405, | |
| "grad_norm": 0.6303943465107783, | |
| "learning_rate": 3.9393493190865657e-05, | |
| "loss": 0.3895, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8546835443037974, | |
| "grad_norm": 0.5293990791351674, | |
| "learning_rate": 3.937954332854371e-05, | |
| "loss": 0.3951, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8587341772151899, | |
| "grad_norm": 0.5076050553962923, | |
| "learning_rate": 3.9365437386044016e-05, | |
| "loss": 0.3601, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8627848101265823, | |
| "grad_norm": 0.5416135495238383, | |
| "learning_rate": 3.935117547697387e-05, | |
| "loss": 0.3825, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.8668354430379747, | |
| "grad_norm": 0.5767869396798332, | |
| "learning_rate": 3.933675771619675e-05, | |
| "loss": 0.3851, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8708860759493671, | |
| "grad_norm": 0.6113555146913673, | |
| "learning_rate": 3.932218421983131e-05, | |
| "loss": 0.3794, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8749367088607595, | |
| "grad_norm": 0.5837146828976643, | |
| "learning_rate": 3.9307455105250484e-05, | |
| "loss": 0.3809, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8789873417721519, | |
| "grad_norm": 0.5461040428474638, | |
| "learning_rate": 3.929257049108054e-05, | |
| "loss": 0.3894, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8830379746835443, | |
| "grad_norm": 0.5739833745235723, | |
| "learning_rate": 3.927753049720011e-05, | |
| "loss": 0.37, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8870886075949367, | |
| "grad_norm": 0.5211102975387855, | |
| "learning_rate": 3.9262335244739234e-05, | |
| "loss": 0.4052, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8911392405063291, | |
| "grad_norm": 0.4991560526378414, | |
| "learning_rate": 3.92469848560784e-05, | |
| "loss": 0.3857, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8951898734177215, | |
| "grad_norm": 0.5224618995860465, | |
| "learning_rate": 3.923147945484751e-05, | |
| "loss": 0.3874, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8992405063291139, | |
| "grad_norm": 0.477421510689121, | |
| "learning_rate": 3.9215819165924956e-05, | |
| "loss": 0.3676, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9032911392405063, | |
| "grad_norm": 0.5536086543558858, | |
| "learning_rate": 3.920000411543654e-05, | |
| "loss": 0.3897, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9073417721518987, | |
| "grad_norm": 0.6051175749098449, | |
| "learning_rate": 3.9184034430754495e-05, | |
| "loss": 0.3846, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 0.4937348232737511, | |
| "learning_rate": 3.916791024049648e-05, | |
| "loss": 0.3716, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9154430379746835, | |
| "grad_norm": 0.6812222133960181, | |
| "learning_rate": 3.91516316745245e-05, | |
| "loss": 0.3831, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9194936708860759, | |
| "grad_norm": 0.5748314934477763, | |
| "learning_rate": 3.913519886394389e-05, | |
| "loss": 0.4002, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.9235443037974683, | |
| "grad_norm": 0.5741647730713366, | |
| "learning_rate": 3.911861194110225e-05, | |
| "loss": 0.3866, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.9275949367088607, | |
| "grad_norm": 0.5784934565821027, | |
| "learning_rate": 3.910187103958837e-05, | |
| "loss": 0.3952, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9316455696202531, | |
| "grad_norm": 0.4814101675273128, | |
| "learning_rate": 3.908497629423117e-05, | |
| "loss": 0.3935, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9356962025316455, | |
| "grad_norm": 0.546979357729137, | |
| "learning_rate": 3.9067927841098614e-05, | |
| "loss": 0.3902, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.9397468354430379, | |
| "grad_norm": 0.5025770892517757, | |
| "learning_rate": 3.9050725817496594e-05, | |
| "loss": 0.4029, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9437974683544303, | |
| "grad_norm": 0.5821917598271854, | |
| "learning_rate": 3.9033370361967844e-05, | |
| "loss": 0.3778, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9478481012658228, | |
| "grad_norm": 0.5024023510252028, | |
| "learning_rate": 3.901586161429081e-05, | |
| "loss": 0.388, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.9518987341772152, | |
| "grad_norm": 0.5267212767720472, | |
| "learning_rate": 3.8998199715478545e-05, | |
| "loss": 0.3841, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.9559493670886076, | |
| "grad_norm": 0.6238268101858241, | |
| "learning_rate": 3.8980384807777564e-05, | |
| "loss": 0.3983, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.5893079145288155, | |
| "learning_rate": 3.896241703466667e-05, | |
| "loss": 0.3854, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.9640506329113924, | |
| "grad_norm": 0.6468641889206062, | |
| "learning_rate": 3.894429654085585e-05, | |
| "loss": 0.374, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9681012658227848, | |
| "grad_norm": 0.7482777713379926, | |
| "learning_rate": 3.892602347228505e-05, | |
| "loss": 0.4045, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.9721518987341772, | |
| "grad_norm": 0.7603982438389965, | |
| "learning_rate": 3.890759797612307e-05, | |
| "loss": 0.3923, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9762025316455696, | |
| "grad_norm": 0.7035831852665279, | |
| "learning_rate": 3.888902020076632e-05, | |
| "loss": 0.3825, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.980253164556962, | |
| "grad_norm": 0.630457224867576, | |
| "learning_rate": 3.887029029583764e-05, | |
| "loss": 0.3666, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9843037974683544, | |
| "grad_norm": 0.6541685968753982, | |
| "learning_rate": 3.8851408412185125e-05, | |
| "loss": 0.3893, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.9883544303797468, | |
| "grad_norm": 0.5169956458557505, | |
| "learning_rate": 3.8832374701880855e-05, | |
| "loss": 0.3869, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9924050632911392, | |
| "grad_norm": 0.6314309521273052, | |
| "learning_rate": 3.881318931821972e-05, | |
| "loss": 0.3948, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.9964556962025316, | |
| "grad_norm": 0.535320674141279, | |
| "learning_rate": 3.879385241571817e-05, | |
| "loss": 0.3715, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.0020253164556963, | |
| "grad_norm": 0.5164533448301665, | |
| "learning_rate": 3.8774364150112955e-05, | |
| "loss": 0.3425, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.0060759493670886, | |
| "grad_norm": 0.5592549340340781, | |
| "learning_rate": 3.8754724678359884e-05, | |
| "loss": 0.3335, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.010126582278481, | |
| "grad_norm": 0.5644720693824358, | |
| "learning_rate": 3.873493415863256e-05, | |
| "loss": 0.3206, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.0141772151898734, | |
| "grad_norm": 0.5392486693276395, | |
| "learning_rate": 3.871499275032111e-05, | |
| "loss": 0.3091, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.0182278481012659, | |
| "grad_norm": 0.6250190756057589, | |
| "learning_rate": 3.869490061403091e-05, | |
| "loss": 0.3231, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.0222784810126582, | |
| "grad_norm": 0.6578556331248738, | |
| "learning_rate": 3.867465791158124e-05, | |
| "loss": 0.3034, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.0263291139240507, | |
| "grad_norm": 0.5861216982210773, | |
| "learning_rate": 3.865426480600407e-05, | |
| "loss": 0.297, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.030379746835443, | |
| "grad_norm": 0.4679572877875589, | |
| "learning_rate": 3.863372146154264e-05, | |
| "loss": 0.299, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.0344303797468355, | |
| "grad_norm": 0.5254754729170253, | |
| "learning_rate": 3.861302804365024e-05, | |
| "loss": 0.3146, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.0384810126582278, | |
| "grad_norm": 0.44179135648516277, | |
| "learning_rate": 3.85921847189888e-05, | |
| "loss": 0.3024, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.0425316455696203, | |
| "grad_norm": 0.5365553543616408, | |
| "learning_rate": 3.85711916554276e-05, | |
| "loss": 0.2992, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.0465822784810126, | |
| "grad_norm": 0.44319249687819556, | |
| "learning_rate": 3.85500490220419e-05, | |
| "loss": 0.2916, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 0.6088096214837235, | |
| "learning_rate": 3.852875698911154e-05, | |
| "loss": 0.315, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.0546835443037974, | |
| "grad_norm": 0.5629201466279181, | |
| "learning_rate": 3.850731572811963e-05, | |
| "loss": 0.2952, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.05873417721519, | |
| "grad_norm": 0.5021276983948468, | |
| "learning_rate": 3.848572541175116e-05, | |
| "loss": 0.3047, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.0627848101265822, | |
| "grad_norm": 0.506068631982397, | |
| "learning_rate": 3.846398621389154e-05, | |
| "loss": 0.2994, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.0668354430379747, | |
| "grad_norm": 0.5664146463951643, | |
| "learning_rate": 3.84420983096253e-05, | |
| "loss": 0.318, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.070886075949367, | |
| "grad_norm": 0.4598292036720209, | |
| "learning_rate": 3.8420061875234606e-05, | |
| "loss": 0.3009, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.0749367088607595, | |
| "grad_norm": 0.510651462323841, | |
| "learning_rate": 3.839787708819787e-05, | |
| "loss": 0.3107, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.0789873417721518, | |
| "grad_norm": 0.5027113365645945, | |
| "learning_rate": 3.8375544127188325e-05, | |
| "loss": 0.3021, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.0830379746835443, | |
| "grad_norm": 0.48465394473909684, | |
| "learning_rate": 3.8353063172072564e-05, | |
| "loss": 0.3051, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.0870886075949366, | |
| "grad_norm": 0.5405613279523053, | |
| "learning_rate": 3.8330434403909105e-05, | |
| "loss": 0.2761, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.0911392405063292, | |
| "grad_norm": 0.4567198601679914, | |
| "learning_rate": 3.8307658004946934e-05, | |
| "loss": 0.3091, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.0951898734177214, | |
| "grad_norm": 0.5503718624580656, | |
| "learning_rate": 3.8284734158624046e-05, | |
| "loss": 0.3041, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.099240506329114, | |
| "grad_norm": 0.4834827350126005, | |
| "learning_rate": 3.826166304956594e-05, | |
| "loss": 0.3088, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.1032911392405063, | |
| "grad_norm": 0.5775195786064837, | |
| "learning_rate": 3.8238444863584164e-05, | |
| "loss": 0.3032, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.1073417721518988, | |
| "grad_norm": 0.4950940625639673, | |
| "learning_rate": 3.821507978767479e-05, | |
| "loss": 0.3009, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.111392405063291, | |
| "grad_norm": 0.5123529643997159, | |
| "learning_rate": 3.819156801001693e-05, | |
| "loss": 0.2937, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.1154430379746836, | |
| "grad_norm": 0.43424600594738544, | |
| "learning_rate": 3.816790971997121e-05, | |
| "loss": 0.3165, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.1194936708860759, | |
| "grad_norm": 0.4970351393374756, | |
| "learning_rate": 3.8144105108078246e-05, | |
| "loss": 0.3138, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.1235443037974684, | |
| "grad_norm": 0.41729999191856626, | |
| "learning_rate": 3.81201543660571e-05, | |
| "loss": 0.2977, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.1275949367088607, | |
| "grad_norm": 0.5258220774894454, | |
| "learning_rate": 3.809605768680377e-05, | |
| "loss": 0.3165, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.1316455696202532, | |
| "grad_norm": 0.4108430373565763, | |
| "learning_rate": 3.807181526438958e-05, | |
| "loss": 0.2806, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.1356962025316455, | |
| "grad_norm": 0.48025281960978455, | |
| "learning_rate": 3.8047427294059697e-05, | |
| "loss": 0.3226, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.139746835443038, | |
| "grad_norm": 0.4357672445006471, | |
| "learning_rate": 3.802289397223145e-05, | |
| "loss": 0.3184, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.1437974683544303, | |
| "grad_norm": 0.4547078756712048, | |
| "learning_rate": 3.7998215496492854e-05, | |
| "loss": 0.3103, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.1478481012658228, | |
| "grad_norm": 0.44172524199836344, | |
| "learning_rate": 3.797339206560096e-05, | |
| "loss": 0.3147, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 0.4844533428555339, | |
| "learning_rate": 3.794842387948027e-05, | |
| "loss": 0.2959, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.1559493670886076, | |
| "grad_norm": 0.43214922188382254, | |
| "learning_rate": 3.7923311139221114e-05, | |
| "loss": 0.3181, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.43981339020259286, | |
| "learning_rate": 3.7898054047078054e-05, | |
| "loss": 0.3028, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.1640506329113924, | |
| "grad_norm": 0.4368289955860144, | |
| "learning_rate": 3.787265280646825e-05, | |
| "loss": 0.3084, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.1681012658227847, | |
| "grad_norm": 0.45297876162045597, | |
| "learning_rate": 3.7847107621969786e-05, | |
| "loss": 0.3111, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.1721518987341772, | |
| "grad_norm": 0.44752395062595585, | |
| "learning_rate": 3.7821418699320064e-05, | |
| "loss": 0.3076, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.1762025316455695, | |
| "grad_norm": 0.47280266641836266, | |
| "learning_rate": 3.7795586245414145e-05, | |
| "loss": 0.3025, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.180253164556962, | |
| "grad_norm": 0.4853612453654955, | |
| "learning_rate": 3.776961046830306e-05, | |
| "loss": 0.298, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.1843037974683543, | |
| "grad_norm": 0.40143778402570945, | |
| "learning_rate": 3.774349157719215e-05, | |
| "loss": 0.2974, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.1883544303797469, | |
| "grad_norm": 0.4672989950521805, | |
| "learning_rate": 3.7717229782439365e-05, | |
| "loss": 0.3184, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.1924050632911392, | |
| "grad_norm": 0.44049883038681187, | |
| "learning_rate": 3.769082529555359e-05, | |
| "loss": 0.3147, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.1964556962025317, | |
| "grad_norm": 0.49367433162506474, | |
| "learning_rate": 3.766427832919294e-05, | |
| "loss": 0.2937, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.200506329113924, | |
| "grad_norm": 0.4491648318223995, | |
| "learning_rate": 3.7637589097163024e-05, | |
| "loss": 0.3146, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.2045569620253165, | |
| "grad_norm": 0.5371846820522248, | |
| "learning_rate": 3.761075781441526e-05, | |
| "loss": 0.3071, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.2086075949367088, | |
| "grad_norm": 0.4808698836643849, | |
| "learning_rate": 3.75837846970451e-05, | |
| "loss": 0.2912, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.2126582278481013, | |
| "grad_norm": 0.47650161247651346, | |
| "learning_rate": 3.755666996229032e-05, | |
| "loss": 0.3058, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.2167088607594936, | |
| "grad_norm": 0.5547297298491616, | |
| "learning_rate": 3.752941382852927e-05, | |
| "loss": 0.3368, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.220759493670886, | |
| "grad_norm": 0.4542426868379709, | |
| "learning_rate": 3.7502016515279115e-05, | |
| "loss": 0.2907, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.2248101265822784, | |
| "grad_norm": 0.5378088230305111, | |
| "learning_rate": 3.7474478243194043e-05, | |
| "loss": 0.3074, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.228860759493671, | |
| "grad_norm": 0.4160504345832747, | |
| "learning_rate": 3.744679923406351e-05, | |
| "loss": 0.2963, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.2329113924050632, | |
| "grad_norm": 0.4855991560286833, | |
| "learning_rate": 3.741897971081043e-05, | |
| "loss": 0.3183, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.2369620253164557, | |
| "grad_norm": 0.4446042305980396, | |
| "learning_rate": 3.739101989748946e-05, | |
| "loss": 0.2984, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.241012658227848, | |
| "grad_norm": 0.546713863566404, | |
| "learning_rate": 3.7362920019285066e-05, | |
| "loss": 0.3041, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.2450632911392405, | |
| "grad_norm": 0.5218340895234224, | |
| "learning_rate": 3.73346803025098e-05, | |
| "loss": 0.3286, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.2491139240506328, | |
| "grad_norm": 0.441881575940641, | |
| "learning_rate": 3.730630097460247e-05, | |
| "loss": 0.3033, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 0.5693961197715379, | |
| "learning_rate": 3.727778226412628e-05, | |
| "loss": 0.3097, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.2572151898734178, | |
| "grad_norm": 0.4641541465095693, | |
| "learning_rate": 3.7249124400767006e-05, | |
| "loss": 0.3023, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.2612658227848101, | |
| "grad_norm": 0.45862318812423725, | |
| "learning_rate": 3.722032761533114e-05, | |
| "loss": 0.2997, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.2653164556962024, | |
| "grad_norm": 0.5177697843435217, | |
| "learning_rate": 3.719139213974403e-05, | |
| "loss": 0.3131, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.269367088607595, | |
| "grad_norm": 0.4393532033605689, | |
| "learning_rate": 3.7162318207048006e-05, | |
| "loss": 0.2879, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.2734177215189875, | |
| "grad_norm": 0.5039055430711514, | |
| "learning_rate": 3.713310605140055e-05, | |
| "loss": 0.3157, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.2774683544303798, | |
| "grad_norm": 0.5056036520598989, | |
| "learning_rate": 3.710375590807233e-05, | |
| "loss": 0.3198, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.281518987341772, | |
| "grad_norm": 0.4176176306684854, | |
| "learning_rate": 3.7074268013445365e-05, | |
| "loss": 0.3074, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.2855696202531646, | |
| "grad_norm": 0.42603815986213034, | |
| "learning_rate": 3.7044642605011114e-05, | |
| "loss": 0.3086, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.289620253164557, | |
| "grad_norm": 0.47750251654347814, | |
| "learning_rate": 3.701487992136854e-05, | |
| "loss": 0.2979, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.2936708860759494, | |
| "grad_norm": 0.47139483648376257, | |
| "learning_rate": 3.69849802022222e-05, | |
| "loss": 0.3095, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.2977215189873417, | |
| "grad_norm": 0.44039235636652474, | |
| "learning_rate": 3.6954943688380334e-05, | |
| "loss": 0.3268, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.3017721518987342, | |
| "grad_norm": 0.49618368297830245, | |
| "learning_rate": 3.692477062175289e-05, | |
| "loss": 0.3083, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.3058227848101267, | |
| "grad_norm": 0.4131906297132629, | |
| "learning_rate": 3.689446124534958e-05, | |
| "loss": 0.3102, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.309873417721519, | |
| "grad_norm": 0.48608272953682047, | |
| "learning_rate": 3.686401580327799e-05, | |
| "loss": 0.3148, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.3139240506329113, | |
| "grad_norm": 0.4969145691063366, | |
| "learning_rate": 3.683343454074149e-05, | |
| "loss": 0.3022, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.3179746835443038, | |
| "grad_norm": 0.40919833391528654, | |
| "learning_rate": 3.6802717704037386e-05, | |
| "loss": 0.3068, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.3220253164556963, | |
| "grad_norm": 0.4890189843831828, | |
| "learning_rate": 3.6771865540554855e-05, | |
| "loss": 0.3022, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.3260759493670886, | |
| "grad_norm": 0.4348517866691108, | |
| "learning_rate": 3.674087829877297e-05, | |
| "loss": 0.3101, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.330126582278481, | |
| "grad_norm": 0.4297557111316552, | |
| "learning_rate": 3.6709756228258735e-05, | |
| "loss": 0.29, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.3341772151898734, | |
| "grad_norm": 0.43939570848991377, | |
| "learning_rate": 3.667849957966501e-05, | |
| "loss": 0.3074, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.338227848101266, | |
| "grad_norm": 0.4540442985101207, | |
| "learning_rate": 3.6647108604728546e-05, | |
| "loss": 0.2987, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.3422784810126582, | |
| "grad_norm": 0.45956663165167716, | |
| "learning_rate": 3.661558355626795e-05, | |
| "loss": 0.3169, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.3463291139240505, | |
| "grad_norm": 0.4080645592136522, | |
| "learning_rate": 3.658392468818163e-05, | |
| "loss": 0.3025, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.350379746835443, | |
| "grad_norm": 0.44132695661411886, | |
| "learning_rate": 3.655213225544574e-05, | |
| "loss": 0.3068, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 0.47464296474277357, | |
| "learning_rate": 3.652020651411218e-05, | |
| "loss": 0.3095, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.3584810126582278, | |
| "grad_norm": 0.4083555164993775, | |
| "learning_rate": 3.6488147721306474e-05, | |
| "loss": 0.3228, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.3625316455696201, | |
| "grad_norm": 0.48079579824608454, | |
| "learning_rate": 3.645595613522574e-05, | |
| "loss": 0.3094, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.3665822784810127, | |
| "grad_norm": 0.44539215584011166, | |
| "learning_rate": 3.642363201513657e-05, | |
| "loss": 0.3262, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.3706329113924052, | |
| "grad_norm": 0.48287435115966737, | |
| "learning_rate": 3.6391175621373006e-05, | |
| "loss": 0.3149, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.3746835443037975, | |
| "grad_norm": 0.4468999388655897, | |
| "learning_rate": 3.6358587215334355e-05, | |
| "loss": 0.3055, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.3787341772151898, | |
| "grad_norm": 0.47635817641821454, | |
| "learning_rate": 3.632586705948318e-05, | |
| "loss": 0.3053, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.3827848101265823, | |
| "grad_norm": 0.5317535122226926, | |
| "learning_rate": 3.629301541734311e-05, | |
| "loss": 0.3161, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.3868354430379748, | |
| "grad_norm": 0.4319972419582864, | |
| "learning_rate": 3.626003255349676e-05, | |
| "loss": 0.3193, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.390886075949367, | |
| "grad_norm": 0.48673340374272767, | |
| "learning_rate": 3.622691873358357e-05, | |
| "loss": 0.3161, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.3949367088607594, | |
| "grad_norm": 0.4676158756906462, | |
| "learning_rate": 3.61936742242977e-05, | |
| "loss": 0.2929, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.398987341772152, | |
| "grad_norm": 0.4777079739026747, | |
| "learning_rate": 3.6160299293385864e-05, | |
| "loss": 0.3106, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.4030379746835444, | |
| "grad_norm": 0.48250872376790244, | |
| "learning_rate": 3.612679420964516e-05, | |
| "loss": 0.3089, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.4070886075949367, | |
| "grad_norm": 0.4323883535771042, | |
| "learning_rate": 3.609315924292092e-05, | |
| "loss": 0.3153, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.411139240506329, | |
| "grad_norm": 0.4647899492616768, | |
| "learning_rate": 3.6059394664104554e-05, | |
| "loss": 0.309, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.4151898734177215, | |
| "grad_norm": 0.3870600079623766, | |
| "learning_rate": 3.602550074513133e-05, | |
| "loss": 0.2831, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.419240506329114, | |
| "grad_norm": 0.44535141224353786, | |
| "learning_rate": 3.599147775897822e-05, | |
| "loss": 0.3262, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.4232911392405063, | |
| "grad_norm": 0.46503067248053326, | |
| "learning_rate": 3.595732597966167e-05, | |
| "loss": 0.3038, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.4273417721518986, | |
| "grad_norm": 0.4063641878640347, | |
| "learning_rate": 3.592304568223542e-05, | |
| "loss": 0.2987, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.4313924050632911, | |
| "grad_norm": 0.45624599070420285, | |
| "learning_rate": 3.588863714278826e-05, | |
| "loss": 0.3008, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.4354430379746836, | |
| "grad_norm": 0.4170625785802956, | |
| "learning_rate": 3.585410063844186e-05, | |
| "loss": 0.3049, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.439493670886076, | |
| "grad_norm": 0.5289420090761957, | |
| "learning_rate": 3.581943644734846e-05, | |
| "loss": 0.3099, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.4435443037974682, | |
| "grad_norm": 0.4167659038003528, | |
| "learning_rate": 3.578464484868869e-05, | |
| "loss": 0.319, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.4475949367088607, | |
| "grad_norm": 0.5072872865128293, | |
| "learning_rate": 3.5749726122669316e-05, | |
| "loss": 0.311, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.4516455696202533, | |
| "grad_norm": 0.3870103004460665, | |
| "learning_rate": 3.5714680550520943e-05, | |
| "loss": 0.305, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 0.47511143558106667, | |
| "learning_rate": 3.5679508414495794e-05, | |
| "loss": 0.3172, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.4597468354430378, | |
| "grad_norm": 0.41994223708805895, | |
| "learning_rate": 3.564420999786543e-05, | |
| "loss": 0.338, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.4637974683544304, | |
| "grad_norm": 0.49549549628308237, | |
| "learning_rate": 3.560878558491842e-05, | |
| "loss": 0.2974, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.4678481012658229, | |
| "grad_norm": 0.4624851648886058, | |
| "learning_rate": 3.5573235460958145e-05, | |
| "loss": 0.3124, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.4718987341772152, | |
| "grad_norm": 0.44516327868740585, | |
| "learning_rate": 3.553755991230039e-05, | |
| "loss": 0.3143, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.4759493670886075, | |
| "grad_norm": 0.424309416082281, | |
| "learning_rate": 3.5501759226271144e-05, | |
| "loss": 0.2848, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.40857347937668276, | |
| "learning_rate": 3.546583369120419e-05, | |
| "loss": 0.3154, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.4840506329113925, | |
| "grad_norm": 0.4137311477701997, | |
| "learning_rate": 3.5429783596438864e-05, | |
| "loss": 0.326, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.4881012658227848, | |
| "grad_norm": 0.4101544446918758, | |
| "learning_rate": 3.539360923231766e-05, | |
| "loss": 0.3048, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.492151898734177, | |
| "grad_norm": 0.43556142741072734, | |
| "learning_rate": 3.535731089018394e-05, | |
| "loss": 0.3071, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.4962025316455696, | |
| "grad_norm": 0.45754356130994844, | |
| "learning_rate": 3.532088886237956e-05, | |
| "loss": 0.2914, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.5002531645569621, | |
| "grad_norm": 0.39992867335820803, | |
| "learning_rate": 3.528434344224253e-05, | |
| "loss": 0.2986, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.5043037974683544, | |
| "grad_norm": 0.43700993960844353, | |
| "learning_rate": 3.524767492410464e-05, | |
| "loss": 0.3231, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.5083544303797467, | |
| "grad_norm": 0.4377836968626936, | |
| "learning_rate": 3.521088360328908e-05, | |
| "loss": 0.3115, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.5124050632911392, | |
| "grad_norm": 0.41820237582265757, | |
| "learning_rate": 3.517396977610811e-05, | |
| "loss": 0.3157, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.5164556962025317, | |
| "grad_norm": 0.45676305605486855, | |
| "learning_rate": 3.5136933739860595e-05, | |
| "loss": 0.326, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.520506329113924, | |
| "grad_norm": 0.422512789226454, | |
| "learning_rate": 3.509977579282971e-05, | |
| "loss": 0.3037, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.5245569620253163, | |
| "grad_norm": 0.4306629071468946, | |
| "learning_rate": 3.5062496234280424e-05, | |
| "loss": 0.2924, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.5286075949367088, | |
| "grad_norm": 0.42468172967877577, | |
| "learning_rate": 3.502509536445719e-05, | |
| "loss": 0.3032, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.5326582278481014, | |
| "grad_norm": 0.4456453309041947, | |
| "learning_rate": 3.498757348458147e-05, | |
| "loss": 0.3203, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.5367088607594936, | |
| "grad_norm": 0.3708664195603814, | |
| "learning_rate": 3.4949930896849324e-05, | |
| "loss": 0.3266, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.540759493670886, | |
| "grad_norm": 0.491989241594006, | |
| "learning_rate": 3.491216790442899e-05, | |
| "loss": 0.3154, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.5448101265822785, | |
| "grad_norm": 0.39041878746539715, | |
| "learning_rate": 3.487428481145839e-05, | |
| "loss": 0.3162, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.548860759493671, | |
| "grad_norm": 0.46495725034364993, | |
| "learning_rate": 3.483628192304278e-05, | |
| "loss": 0.3093, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.5529113924050633, | |
| "grad_norm": 0.4190748862493647, | |
| "learning_rate": 3.479815954525219e-05, | |
| "loss": 0.3043, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 0.5068265965681228, | |
| "learning_rate": 3.475991798511899e-05, | |
| "loss": 0.3202, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.561012658227848, | |
| "grad_norm": 0.4113169265519984, | |
| "learning_rate": 3.4721557550635464e-05, | |
| "loss": 0.3285, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.5650632911392406, | |
| "grad_norm": 0.4549770292647151, | |
| "learning_rate": 3.468307855075128e-05, | |
| "loss": 0.2998, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.5691139240506329, | |
| "grad_norm": 0.4437311141354066, | |
| "learning_rate": 3.4644481295371005e-05, | |
| "loss": 0.3171, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.5731645569620252, | |
| "grad_norm": 0.4066714001931387, | |
| "learning_rate": 3.460576609535163e-05, | |
| "loss": 0.3086, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.5772151898734177, | |
| "grad_norm": 0.38756655756171776, | |
| "learning_rate": 3.456693326250006e-05, | |
| "loss": 0.3076, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.5812658227848102, | |
| "grad_norm": 0.41936695903220483, | |
| "learning_rate": 3.452798310957058e-05, | |
| "loss": 0.2998, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.5853164556962025, | |
| "grad_norm": 0.444088574618765, | |
| "learning_rate": 3.4488915950262386e-05, | |
| "loss": 0.3118, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.5893670886075948, | |
| "grad_norm": 0.44622165778950507, | |
| "learning_rate": 3.4449732099216985e-05, | |
| "loss": 0.3132, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.5934177215189873, | |
| "grad_norm": 0.5061889768463261, | |
| "learning_rate": 3.441043187201574e-05, | |
| "loss": 0.3281, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.5974683544303798, | |
| "grad_norm": 0.459808238368662, | |
| "learning_rate": 3.437101558517728e-05, | |
| "loss": 0.3107, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.6015189873417721, | |
| "grad_norm": 0.5013806074500857, | |
| "learning_rate": 3.433148355615496e-05, | |
| "loss": 0.3023, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.6055696202531644, | |
| "grad_norm": 0.4910973621650801, | |
| "learning_rate": 3.4291836103334294e-05, | |
| "loss": 0.3212, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.609620253164557, | |
| "grad_norm": 0.45307920677843333, | |
| "learning_rate": 3.425207354603043e-05, | |
| "loss": 0.3121, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.6136708860759494, | |
| "grad_norm": 0.46651180562427774, | |
| "learning_rate": 3.421219620448553e-05, | |
| "loss": 0.2925, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.6177215189873417, | |
| "grad_norm": 0.3982778728772528, | |
| "learning_rate": 3.417220439986623e-05, | |
| "loss": 0.3092, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.621772151898734, | |
| "grad_norm": 0.5103005680616458, | |
| "learning_rate": 3.4132098454261024e-05, | |
| "loss": 0.2964, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.6258227848101265, | |
| "grad_norm": 0.44029061008831233, | |
| "learning_rate": 3.4091878690677676e-05, | |
| "loss": 0.3208, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.629873417721519, | |
| "grad_norm": 0.42291993848313253, | |
| "learning_rate": 3.405154543304065e-05, | |
| "loss": 0.3062, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.6339240506329114, | |
| "grad_norm": 0.46831956327580226, | |
| "learning_rate": 3.401109900618843e-05, | |
| "loss": 0.2963, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.6379746835443036, | |
| "grad_norm": 0.4010832834034687, | |
| "learning_rate": 3.3970539735870996e-05, | |
| "loss": 0.3024, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.6420253164556962, | |
| "grad_norm": 0.43295309968728485, | |
| "learning_rate": 3.392986794874714e-05, | |
| "loss": 0.3109, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.6460759493670887, | |
| "grad_norm": 0.46038020301483285, | |
| "learning_rate": 3.388908397238184e-05, | |
| "loss": 0.2993, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.650126582278481, | |
| "grad_norm": 0.4155007025807809, | |
| "learning_rate": 3.384818813524362e-05, | |
| "loss": 0.3074, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.6541772151898733, | |
| "grad_norm": 0.3943764590291214, | |
| "learning_rate": 3.380718076670195e-05, | |
| "loss": 0.3186, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 0.4289651825658174, | |
| "learning_rate": 3.376606219702454e-05, | |
| "loss": 0.318, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.6622784810126583, | |
| "grad_norm": 0.40037993147738593, | |
| "learning_rate": 3.372483275737468e-05, | |
| "loss": 0.3148, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.6663291139240506, | |
| "grad_norm": 0.4168834872763403, | |
| "learning_rate": 3.368349277980861e-05, | |
| "loss": 0.3456, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.6703797468354429, | |
| "grad_norm": 0.36785156331462854, | |
| "learning_rate": 3.3642042597272844e-05, | |
| "loss": 0.3099, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.6744303797468354, | |
| "grad_norm": 0.4054160379460761, | |
| "learning_rate": 3.360048254360144e-05, | |
| "loss": 0.2974, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.678481012658228, | |
| "grad_norm": 0.4013334024918394, | |
| "learning_rate": 3.355881295351336e-05, | |
| "loss": 0.3186, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.6825316455696202, | |
| "grad_norm": 0.37905695160086705, | |
| "learning_rate": 3.351703416260975e-05, | |
| "loss": 0.3104, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.6865822784810125, | |
| "grad_norm": 0.3991854093481229, | |
| "learning_rate": 3.347514650737126e-05, | |
| "loss": 0.3114, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.690632911392405, | |
| "grad_norm": 0.4313796177714192, | |
| "learning_rate": 3.3433150325155295e-05, | |
| "loss": 0.2987, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.6946835443037975, | |
| "grad_norm": 0.4526867223631983, | |
| "learning_rate": 3.339104595419334e-05, | |
| "loss": 0.3217, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.6987341772151898, | |
| "grad_norm": 0.42351720678166177, | |
| "learning_rate": 3.3348833733588204e-05, | |
| "loss": 0.3134, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.7027848101265821, | |
| "grad_norm": 0.41923456233919615, | |
| "learning_rate": 3.3306514003311305e-05, | |
| "loss": 0.2959, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.7068354430379746, | |
| "grad_norm": 0.43320274661915054, | |
| "learning_rate": 3.326408710419996e-05, | |
| "loss": 0.3023, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.7108860759493671, | |
| "grad_norm": 0.46773486990236657, | |
| "learning_rate": 3.322155337795454e-05, | |
| "loss": 0.2988, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.7149367088607594, | |
| "grad_norm": 0.4629709855540852, | |
| "learning_rate": 3.317891316713587e-05, | |
| "loss": 0.3191, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.7189873417721517, | |
| "grad_norm": 0.42523033463550974, | |
| "learning_rate": 3.313616681516231e-05, | |
| "loss": 0.323, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.7230379746835442, | |
| "grad_norm": 0.4212753950613693, | |
| "learning_rate": 3.309331466630713e-05, | |
| "loss": 0.3063, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.7270886075949368, | |
| "grad_norm": 0.4698001060044043, | |
| "learning_rate": 3.305035706569563e-05, | |
| "loss": 0.315, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.7311392405063293, | |
| "grad_norm": 0.4128210270776783, | |
| "learning_rate": 3.3007294359302433e-05, | |
| "loss": 0.3095, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.7351898734177216, | |
| "grad_norm": 0.3996178369080338, | |
| "learning_rate": 3.296412689394864e-05, | |
| "loss": 0.3221, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.7392405063291139, | |
| "grad_norm": 0.4142987432255265, | |
| "learning_rate": 3.292085501729909e-05, | |
| "loss": 0.3196, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.7432911392405064, | |
| "grad_norm": 0.42660489256827006, | |
| "learning_rate": 3.2877479077859534e-05, | |
| "loss": 0.2823, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.747341772151899, | |
| "grad_norm": 0.42611648513146305, | |
| "learning_rate": 3.283399942497381e-05, | |
| "loss": 0.2961, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.7513924050632912, | |
| "grad_norm": 0.4267205507799652, | |
| "learning_rate": 3.279041640882108e-05, | |
| "loss": 0.3117, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.7554430379746835, | |
| "grad_norm": 0.43514715811806953, | |
| "learning_rate": 3.2746730380412964e-05, | |
| "loss": 0.3047, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 0.4464189504599663, | |
| "learning_rate": 3.2702941691590726e-05, | |
| "loss": 0.314, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.7635443037974685, | |
| "grad_norm": 0.38458157864348014, | |
| "learning_rate": 3.265905069502244e-05, | |
| "loss": 0.3018, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.7675949367088608, | |
| "grad_norm": 0.42300565652647076, | |
| "learning_rate": 3.261505774420016e-05, | |
| "loss": 0.2969, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.771645569620253, | |
| "grad_norm": 0.42678099716584383, | |
| "learning_rate": 3.257096319343707e-05, | |
| "loss": 0.3235, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.7756962025316456, | |
| "grad_norm": 0.39870033332200133, | |
| "learning_rate": 3.2526767397864614e-05, | |
| "loss": 0.2998, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.7797468354430381, | |
| "grad_norm": 0.4216743281097242, | |
| "learning_rate": 3.248247071342966e-05, | |
| "loss": 0.2943, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.7837974683544304, | |
| "grad_norm": 0.45262807057670307, | |
| "learning_rate": 3.243807349689161e-05, | |
| "loss": 0.3124, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.7878481012658227, | |
| "grad_norm": 0.47824185822118004, | |
| "learning_rate": 3.2393576105819544e-05, | |
| "loss": 0.3287, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.7918987341772152, | |
| "grad_norm": 0.3887220755961511, | |
| "learning_rate": 3.2348978898589333e-05, | |
| "loss": 0.3255, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.7959493670886078, | |
| "grad_norm": 0.4420223749512171, | |
| "learning_rate": 3.230428223438075e-05, | |
| "loss": 0.3095, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.4227870800624179, | |
| "learning_rate": 3.225948647317459e-05, | |
| "loss": 0.2974, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.8040506329113923, | |
| "grad_norm": 0.42610794978638405, | |
| "learning_rate": 3.2214591975749745e-05, | |
| "loss": 0.3, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.8081012658227849, | |
| "grad_norm": 0.4536151875856935, | |
| "learning_rate": 3.216959910368034e-05, | |
| "loss": 0.3201, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.8121518987341774, | |
| "grad_norm": 0.39451097586735384, | |
| "learning_rate": 3.212450821933277e-05, | |
| "loss": 0.2999, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.8162025316455697, | |
| "grad_norm": 0.4571926973459174, | |
| "learning_rate": 3.207931968586281e-05, | |
| "loss": 0.3072, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.820253164556962, | |
| "grad_norm": 0.4322556821687935, | |
| "learning_rate": 3.203403386721272e-05, | |
| "loss": 0.3081, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.8243037974683545, | |
| "grad_norm": 0.3757001352547789, | |
| "learning_rate": 3.1988651128108245e-05, | |
| "loss": 0.304, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.828354430379747, | |
| "grad_norm": 0.48693963911987126, | |
| "learning_rate": 3.194317183405573e-05, | |
| "loss": 0.3111, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.8324050632911393, | |
| "grad_norm": 0.40925394722184516, | |
| "learning_rate": 3.189759635133914e-05, | |
| "loss": 0.3187, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.8364556962025316, | |
| "grad_norm": 0.4058249653803545, | |
| "learning_rate": 3.185192504701718e-05, | |
| "loss": 0.3097, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.840506329113924, | |
| "grad_norm": 0.45056623655717964, | |
| "learning_rate": 3.1806158288920234e-05, | |
| "loss": 0.3253, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.8445569620253166, | |
| "grad_norm": 0.4043214526860223, | |
| "learning_rate": 3.1760296445647477e-05, | |
| "loss": 0.3028, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.848607594936709, | |
| "grad_norm": 0.49463513972798473, | |
| "learning_rate": 3.1714339886563896e-05, | |
| "loss": 0.3159, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.8526582278481012, | |
| "grad_norm": 0.4189128448967934, | |
| "learning_rate": 3.166828898179731e-05, | |
| "loss": 0.3027, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.8567088607594937, | |
| "grad_norm": 0.45904343005973375, | |
| "learning_rate": 3.162214410223536e-05, | |
| "loss": 0.3078, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 0.40071896453366185, | |
| "learning_rate": 3.157590561952257e-05, | |
| "loss": 0.3042, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.8648101265822785, | |
| "grad_norm": 0.5206755946022007, | |
| "learning_rate": 3.152957390605732e-05, | |
| "loss": 0.3018, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.8688607594936708, | |
| "grad_norm": 0.39206617976677877, | |
| "learning_rate": 3.148314933498886e-05, | |
| "loss": 0.3076, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.8729113924050633, | |
| "grad_norm": 0.5470298495314802, | |
| "learning_rate": 3.143663228021431e-05, | |
| "loss": 0.3121, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.8769620253164558, | |
| "grad_norm": 0.44645890530183624, | |
| "learning_rate": 3.1390023116375624e-05, | |
| "loss": 0.3102, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.8810126582278481, | |
| "grad_norm": 0.4838879063615826, | |
| "learning_rate": 3.134332221885661e-05, | |
| "loss": 0.3206, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.8850632911392404, | |
| "grad_norm": 0.4653188099295025, | |
| "learning_rate": 3.129652996377987e-05, | |
| "loss": 0.3175, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.889113924050633, | |
| "grad_norm": 0.41672128735285635, | |
| "learning_rate": 3.12496467280038e-05, | |
| "loss": 0.2993, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.8931645569620255, | |
| "grad_norm": 0.398779873091646, | |
| "learning_rate": 3.120267288911952e-05, | |
| "loss": 0.2841, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.8972151898734178, | |
| "grad_norm": 0.40598510165801444, | |
| "learning_rate": 3.11556088254479e-05, | |
| "loss": 0.3148, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.90126582278481, | |
| "grad_norm": 0.42356795232155386, | |
| "learning_rate": 3.11084549160364e-05, | |
| "loss": 0.3065, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.9053164556962026, | |
| "grad_norm": 0.3995072532455159, | |
| "learning_rate": 3.106121154065615e-05, | |
| "loss": 0.3136, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.909367088607595, | |
| "grad_norm": 0.4623222396442162, | |
| "learning_rate": 3.1013879079798805e-05, | |
| "loss": 0.3207, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.9134177215189874, | |
| "grad_norm": 0.3628551592034743, | |
| "learning_rate": 3.096645791467348e-05, | |
| "loss": 0.2968, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.9174683544303797, | |
| "grad_norm": 0.4210574772461327, | |
| "learning_rate": 3.091894842720373e-05, | |
| "loss": 0.2879, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.9215189873417722, | |
| "grad_norm": 0.3774893905993087, | |
| "learning_rate": 3.0871351000024425e-05, | |
| "loss": 0.3091, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.9255696202531647, | |
| "grad_norm": 0.4181469643687812, | |
| "learning_rate": 3.0823666016478716e-05, | |
| "loss": 0.3098, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.929620253164557, | |
| "grad_norm": 0.3635064761211531, | |
| "learning_rate": 3.0775893860614896e-05, | |
| "loss": 0.3224, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.9336708860759493, | |
| "grad_norm": 0.4024629255439821, | |
| "learning_rate": 3.0728034917183336e-05, | |
| "loss": 0.3099, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.9377215189873418, | |
| "grad_norm": 0.3862454275690865, | |
| "learning_rate": 3.06800895716334e-05, | |
| "loss": 0.297, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.9417721518987343, | |
| "grad_norm": 0.3977547035243837, | |
| "learning_rate": 3.063205821011029e-05, | |
| "loss": 0.3047, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.9458227848101266, | |
| "grad_norm": 0.42729982692769025, | |
| "learning_rate": 3.0583941219452016e-05, | |
| "loss": 0.3067, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.949873417721519, | |
| "grad_norm": 0.4020953058239602, | |
| "learning_rate": 3.053573898718618e-05, | |
| "loss": 0.308, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.9539240506329114, | |
| "grad_norm": 0.4979356405568446, | |
| "learning_rate": 3.0487451901526956e-05, | |
| "loss": 0.3238, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.957974683544304, | |
| "grad_norm": 0.4278568462736831, | |
| "learning_rate": 3.0439080351371875e-05, | |
| "loss": 0.2984, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 0.3938034228755769, | |
| "learning_rate": 3.0390624726298764e-05, | |
| "loss": 0.3081, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.9660759493670885, | |
| "grad_norm": 0.40761100991197, | |
| "learning_rate": 3.034208541656255e-05, | |
| "loss": 0.3123, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.970126582278481, | |
| "grad_norm": 0.38211233590253735, | |
| "learning_rate": 3.029346281309218e-05, | |
| "loss": 0.3, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.9741772151898735, | |
| "grad_norm": 0.41518735486397157, | |
| "learning_rate": 3.0244757307487415e-05, | |
| "loss": 0.2981, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.9782278481012658, | |
| "grad_norm": 0.37580183308075055, | |
| "learning_rate": 3.019596929201569e-05, | |
| "loss": 0.2963, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.9822784810126581, | |
| "grad_norm": 0.44806960935134776, | |
| "learning_rate": 3.0147099159608985e-05, | |
| "loss": 0.3242, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.9863291139240506, | |
| "grad_norm": 0.4476174926144697, | |
| "learning_rate": 3.0098147303860616e-05, | |
| "loss": 0.3024, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.9903797468354432, | |
| "grad_norm": 0.41857382534192633, | |
| "learning_rate": 3.0049114119022117e-05, | |
| "loss": 0.3046, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.9944303797468355, | |
| "grad_norm": 0.41219208137977315, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 0.3037, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.9984810126582278, | |
| "grad_norm": 0.43180079237832886, | |
| "learning_rate": 2.995080534235264e-05, | |
| "loss": 0.31, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.0040506329113925, | |
| "grad_norm": 0.5725395256362034, | |
| "learning_rate": 2.9901530542287044e-05, | |
| "loss": 0.216, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.008101265822785, | |
| "grad_norm": 0.4314521445054805, | |
| "learning_rate": 2.9852175996655676e-05, | |
| "loss": 0.2168, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.012151898734177, | |
| "grad_norm": 0.9367497337137974, | |
| "learning_rate": 2.980274210295326e-05, | |
| "loss": 0.2204, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.0162025316455696, | |
| "grad_norm": 0.5936768460800603, | |
| "learning_rate": 2.9753229259313578e-05, | |
| "loss": 0.2166, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.020253164556962, | |
| "grad_norm": 0.4877968477132733, | |
| "learning_rate": 2.9703637864506274e-05, | |
| "loss": 0.1964, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.0243037974683546, | |
| "grad_norm": 0.5170552258329841, | |
| "learning_rate": 2.965396831793362e-05, | |
| "loss": 0.2056, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.0283544303797467, | |
| "grad_norm": 0.43250441358017866, | |
| "learning_rate": 2.9604221019627316e-05, | |
| "loss": 0.2107, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0324050632911392, | |
| "grad_norm": 0.49675495490748733, | |
| "learning_rate": 2.955439637024526e-05, | |
| "loss": 0.1873, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.0364556962025318, | |
| "grad_norm": 0.474073621176763, | |
| "learning_rate": 2.9504494771068334e-05, | |
| "loss": 0.2117, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.0405063291139243, | |
| "grad_norm": 0.528440224024352, | |
| "learning_rate": 2.9454516623997156e-05, | |
| "loss": 0.2066, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.0445569620253163, | |
| "grad_norm": 0.3661770311562776, | |
| "learning_rate": 2.9404462331548847e-05, | |
| "loss": 0.2072, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.048607594936709, | |
| "grad_norm": 0.46811950818971765, | |
| "learning_rate": 2.93543322968538e-05, | |
| "loss": 0.2258, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.0526582278481014, | |
| "grad_norm": 0.37464447479118745, | |
| "learning_rate": 2.9304126923652428e-05, | |
| "loss": 0.211, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.056708860759494, | |
| "grad_norm": 0.48812549549202583, | |
| "learning_rate": 2.9253846616291896e-05, | |
| "loss": 0.2157, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.060759493670886, | |
| "grad_norm": 0.4049399193541872, | |
| "learning_rate": 2.9203491779722896e-05, | |
| "loss": 0.186, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.0648101265822785, | |
| "grad_norm": 0.43223005212610277, | |
| "learning_rate": 2.9153062819496357e-05, | |
| "loss": 0.2257, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.068860759493671, | |
| "grad_norm": 0.43642772829492177, | |
| "learning_rate": 2.9102560141760178e-05, | |
| "loss": 0.215, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.0729113924050635, | |
| "grad_norm": 0.456846685614077, | |
| "learning_rate": 2.9051984153256004e-05, | |
| "loss": 0.2168, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.0769620253164556, | |
| "grad_norm": 0.40856616948863017, | |
| "learning_rate": 2.900133526131588e-05, | |
| "loss": 0.1945, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.081012658227848, | |
| "grad_norm": 0.49310827503927745, | |
| "learning_rate": 2.8950613873859025e-05, | |
| "loss": 0.1929, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.0850632911392406, | |
| "grad_norm": 0.40338716138782854, | |
| "learning_rate": 2.8899820399388515e-05, | |
| "loss": 0.2134, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.089113924050633, | |
| "grad_norm": 0.4205473173959613, | |
| "learning_rate": 2.8848955246988012e-05, | |
| "loss": 0.2211, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.093164556962025, | |
| "grad_norm": 0.4142596274311818, | |
| "learning_rate": 2.879801882631847e-05, | |
| "loss": 0.2123, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.0972151898734177, | |
| "grad_norm": 0.4425515540457974, | |
| "learning_rate": 2.8747011547614808e-05, | |
| "loss": 0.2168, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 0.3967030121855441, | |
| "learning_rate": 2.8695933821682635e-05, | |
| "loss": 0.2219, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.1053164556962027, | |
| "grad_norm": 0.42177418854812104, | |
| "learning_rate": 2.864478605989494e-05, | |
| "loss": 0.2005, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.109367088607595, | |
| "grad_norm": 0.42847850206065546, | |
| "learning_rate": 2.8593568674188765e-05, | |
| "loss": 0.211, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.1134177215189873, | |
| "grad_norm": 0.38344879912426566, | |
| "learning_rate": 2.8542282077061892e-05, | |
| "loss": 0.2144, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.11746835443038, | |
| "grad_norm": 0.3842831525286898, | |
| "learning_rate": 2.8490926681569523e-05, | |
| "loss": 0.204, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.1215189873417724, | |
| "grad_norm": 0.4264551572005483, | |
| "learning_rate": 2.8439502901320956e-05, | |
| "loss": 0.194, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 2.1255696202531644, | |
| "grad_norm": 0.38721066153104017, | |
| "learning_rate": 2.8388011150476237e-05, | |
| "loss": 0.2276, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 2.129620253164557, | |
| "grad_norm": 0.4553564725074786, | |
| "learning_rate": 2.8336451843742866e-05, | |
| "loss": 0.2014, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.1336708860759495, | |
| "grad_norm": 0.3765931077122536, | |
| "learning_rate": 2.8284825396372387e-05, | |
| "loss": 0.1984, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 2.137721518987342, | |
| "grad_norm": 0.45504274062659383, | |
| "learning_rate": 2.8233132224157132e-05, | |
| "loss": 0.2171, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 2.141772151898734, | |
| "grad_norm": 0.3979764687011173, | |
| "learning_rate": 2.8181372743426805e-05, | |
| "loss": 0.2034, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 2.1458227848101266, | |
| "grad_norm": 0.3795753874557741, | |
| "learning_rate": 2.8129547371045128e-05, | |
| "loss": 0.231, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 2.149873417721519, | |
| "grad_norm": 0.4129037131720884, | |
| "learning_rate": 2.8077656524406534e-05, | |
| "loss": 0.2077, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.1539240506329116, | |
| "grad_norm": 0.38085808325535997, | |
| "learning_rate": 2.802570062143278e-05, | |
| "loss": 0.2072, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 2.1579746835443037, | |
| "grad_norm": 0.3818719506240145, | |
| "learning_rate": 2.7973680080569555e-05, | |
| "loss": 0.2025, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 2.162025316455696, | |
| "grad_norm": 0.37237378684429884, | |
| "learning_rate": 2.792159532078314e-05, | |
| "loss": 0.232, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 2.1660759493670887, | |
| "grad_norm": 0.38029915377397816, | |
| "learning_rate": 2.7869446761557033e-05, | |
| "loss": 0.2082, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 2.170126582278481, | |
| "grad_norm": 0.3756843224380841, | |
| "learning_rate": 2.781723482288857e-05, | |
| "loss": 0.2261, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.1741772151898733, | |
| "grad_norm": 0.40502370064546106, | |
| "learning_rate": 2.7764959925285517e-05, | |
| "loss": 0.2025, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.178227848101266, | |
| "grad_norm": 0.4171771856890464, | |
| "learning_rate": 2.771262248976272e-05, | |
| "loss": 0.1978, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 2.1822784810126583, | |
| "grad_norm": 0.35402673066818835, | |
| "learning_rate": 2.7660222937838677e-05, | |
| "loss": 0.2273, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 2.186329113924051, | |
| "grad_norm": 0.4937885177933189, | |
| "learning_rate": 2.7607761691532186e-05, | |
| "loss": 0.2105, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 2.190379746835443, | |
| "grad_norm": 0.40812936639504555, | |
| "learning_rate": 2.7555239173358916e-05, | |
| "loss": 0.1925, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.1944303797468354, | |
| "grad_norm": 0.4034000120721872, | |
| "learning_rate": 2.7502655806328e-05, | |
| "loss": 0.2101, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 2.198481012658228, | |
| "grad_norm": 0.45237305654322957, | |
| "learning_rate": 2.7450012013938648e-05, | |
| "loss": 0.2081, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 0.424800943771717, | |
| "learning_rate": 2.739730822017673e-05, | |
| "loss": 0.1999, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 2.2065822784810125, | |
| "grad_norm": 0.4162588950090944, | |
| "learning_rate": 2.7344544849511355e-05, | |
| "loss": 0.2093, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 2.210632911392405, | |
| "grad_norm": 0.42508962425075675, | |
| "learning_rate": 2.7291722326891456e-05, | |
| "loss": 0.2216, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.2146835443037975, | |
| "grad_norm": 0.3604294115035183, | |
| "learning_rate": 2.723884107774236e-05, | |
| "loss": 0.196, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 2.21873417721519, | |
| "grad_norm": 0.37476328681127946, | |
| "learning_rate": 2.718590152796239e-05, | |
| "loss": 0.204, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 2.222784810126582, | |
| "grad_norm": 0.41081145762566534, | |
| "learning_rate": 2.71329041039194e-05, | |
| "loss": 0.2013, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 2.2268354430379746, | |
| "grad_norm": 0.39157831535997734, | |
| "learning_rate": 2.7079849232447357e-05, | |
| "loss": 0.2074, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.230886075949367, | |
| "grad_norm": 0.41910470510636555, | |
| "learning_rate": 2.7026737340842895e-05, | |
| "loss": 0.2109, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.2349367088607597, | |
| "grad_norm": 0.4008552659439744, | |
| "learning_rate": 2.697356885686189e-05, | |
| "loss": 0.2159, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.2389873417721518, | |
| "grad_norm": 0.4052524577054268, | |
| "learning_rate": 2.6920344208716014e-05, | |
| "loss": 0.2399, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.2430379746835443, | |
| "grad_norm": 0.3808445277259207, | |
| "learning_rate": 2.6867063825069252e-05, | |
| "loss": 0.2033, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.247088607594937, | |
| "grad_norm": 0.3945635855458829, | |
| "learning_rate": 2.6813728135034494e-05, | |
| "loss": 0.2139, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.2511392405063293, | |
| "grad_norm": 0.4406390621954702, | |
| "learning_rate": 2.6760337568170056e-05, | |
| "loss": 0.1934, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.2551898734177214, | |
| "grad_norm": 0.3942758420251038, | |
| "learning_rate": 2.6706892554476226e-05, | |
| "loss": 0.2099, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.259240506329114, | |
| "grad_norm": 0.39333400981821004, | |
| "learning_rate": 2.6653393524391795e-05, | |
| "loss": 0.2171, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.2632911392405064, | |
| "grad_norm": 0.3853614953485555, | |
| "learning_rate": 2.6599840908790592e-05, | |
| "loss": 0.2114, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.267341772151899, | |
| "grad_norm": 0.3650357220084446, | |
| "learning_rate": 2.6546235138978028e-05, | |
| "loss": 0.2236, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.271392405063291, | |
| "grad_norm": 0.367902113295224, | |
| "learning_rate": 2.6492576646687597e-05, | |
| "loss": 0.2152, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.2754430379746835, | |
| "grad_norm": 0.3575298210575672, | |
| "learning_rate": 2.6438865864077425e-05, | |
| "loss": 0.2156, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.279493670886076, | |
| "grad_norm": 0.3798678898324423, | |
| "learning_rate": 2.6385103223726766e-05, | |
| "loss": 0.2046, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.2835443037974685, | |
| "grad_norm": 0.44296593548548835, | |
| "learning_rate": 2.6331289158632537e-05, | |
| "loss": 0.2074, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.2875949367088606, | |
| "grad_norm": 0.37322404025041805, | |
| "learning_rate": 2.6277424102205817e-05, | |
| "loss": 0.2143, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.291645569620253, | |
| "grad_norm": 0.43959310288778997, | |
| "learning_rate": 2.6223508488268374e-05, | |
| "loss": 0.2078, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.2956962025316456, | |
| "grad_norm": 0.38498618731112133, | |
| "learning_rate": 2.6169542751049148e-05, | |
| "loss": 0.22, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.299746835443038, | |
| "grad_norm": 0.39782977136213116, | |
| "learning_rate": 2.6115527325180754e-05, | |
| "loss": 0.1973, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 0.41257015235079625, | |
| "learning_rate": 2.606146264569603e-05, | |
| "loss": 0.2137, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.3078481012658227, | |
| "grad_norm": 0.36661160080633753, | |
| "learning_rate": 2.6007349148024447e-05, | |
| "loss": 0.2091, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.3118987341772153, | |
| "grad_norm": 0.4250839883619386, | |
| "learning_rate": 2.5953187267988694e-05, | |
| "loss": 0.2033, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.3159493670886078, | |
| "grad_norm": 0.4118348714360504, | |
| "learning_rate": 2.5898977441801097e-05, | |
| "loss": 0.2054, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.3899223249634352, | |
| "learning_rate": 2.584472010606015e-05, | |
| "loss": 0.1976, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.3240506329113924, | |
| "grad_norm": 0.4213621226751583, | |
| "learning_rate": 2.5790415697746976e-05, | |
| "loss": 0.2202, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.328101265822785, | |
| "grad_norm": 0.4251872823875017, | |
| "learning_rate": 2.5736064654221808e-05, | |
| "loss": 0.2269, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.3321518987341774, | |
| "grad_norm": 0.39863661183287286, | |
| "learning_rate": 2.568166741322048e-05, | |
| "loss": 0.205, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.3362025316455695, | |
| "grad_norm": 0.41527015670244927, | |
| "learning_rate": 2.56272244128509e-05, | |
| "loss": 0.209, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.340253164556962, | |
| "grad_norm": 0.3996565592992165, | |
| "learning_rate": 2.55727360915895e-05, | |
| "loss": 0.2042, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.3443037974683545, | |
| "grad_norm": 0.38630085677859616, | |
| "learning_rate": 2.5518202888277734e-05, | |
| "loss": 0.2073, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.348354430379747, | |
| "grad_norm": 0.4278407034750918, | |
| "learning_rate": 2.5463625242118523e-05, | |
| "loss": 0.2054, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.352405063291139, | |
| "grad_norm": 0.37256496256290317, | |
| "learning_rate": 2.5409003592672723e-05, | |
| "loss": 0.2114, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.3564556962025316, | |
| "grad_norm": 0.3742850456084206, | |
| "learning_rate": 2.535433837985559e-05, | |
| "loss": 0.2076, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.360506329113924, | |
| "grad_norm": 0.42921643621358574, | |
| "learning_rate": 2.529963004393324e-05, | |
| "loss": 0.2089, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.3645569620253166, | |
| "grad_norm": 0.37178260593470686, | |
| "learning_rate": 2.524487902551908e-05, | |
| "loss": 0.2036, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.3686075949367087, | |
| "grad_norm": 0.42828039989502176, | |
| "learning_rate": 2.519008576557029e-05, | |
| "loss": 0.1913, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.372658227848101, | |
| "grad_norm": 0.3895148584694815, | |
| "learning_rate": 2.5135250705384254e-05, | |
| "loss": 0.2045, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.3767088607594937, | |
| "grad_norm": 0.40849677456267947, | |
| "learning_rate": 2.5080374286595007e-05, | |
| "loss": 0.2179, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.3807594936708862, | |
| "grad_norm": 0.4337108990616578, | |
| "learning_rate": 2.5025456951169677e-05, | |
| "loss": 0.2112, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.3848101265822783, | |
| "grad_norm": 0.415804827637662, | |
| "learning_rate": 2.4970499141404942e-05, | |
| "loss": 0.203, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.388860759493671, | |
| "grad_norm": 0.38595560859771944, | |
| "learning_rate": 2.491550129992345e-05, | |
| "loss": 0.2024, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.3929113924050633, | |
| "grad_norm": 0.37194514362042275, | |
| "learning_rate": 2.486046386967024e-05, | |
| "loss": 0.2218, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.396962025316456, | |
| "grad_norm": 0.4112184407484538, | |
| "learning_rate": 2.4805387293909214e-05, | |
| "loss": 0.2118, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.401012658227848, | |
| "grad_norm": 0.40242814411456174, | |
| "learning_rate": 2.4750272016219552e-05, | |
| "loss": 0.2348, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 0.4047632613037, | |
| "learning_rate": 2.4695118480492114e-05, | |
| "loss": 0.217, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.409113924050633, | |
| "grad_norm": 0.39902420963767327, | |
| "learning_rate": 2.4639927130925898e-05, | |
| "loss": 0.2186, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.4131645569620255, | |
| "grad_norm": 0.39423909815183866, | |
| "learning_rate": 2.458469841202444e-05, | |
| "loss": 0.2025, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.4172151898734175, | |
| "grad_norm": 0.37394713736129825, | |
| "learning_rate": 2.452943276859226e-05, | |
| "loss": 0.2111, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.42126582278481, | |
| "grad_norm": 0.3788158444307574, | |
| "learning_rate": 2.447413064573125e-05, | |
| "loss": 0.218, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.4253164556962026, | |
| "grad_norm": 0.40109825221415984, | |
| "learning_rate": 2.4418792488837095e-05, | |
| "loss": 0.214, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.429367088607595, | |
| "grad_norm": 0.42086312627703204, | |
| "learning_rate": 2.4363418743595713e-05, | |
| "loss": 0.1959, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 2.433417721518987, | |
| "grad_norm": 0.3577715871107324, | |
| "learning_rate": 2.430800985597963e-05, | |
| "loss": 0.2139, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.4374683544303797, | |
| "grad_norm": 0.3865581706432509, | |
| "learning_rate": 2.4252566272244415e-05, | |
| "loss": 0.219, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 2.441518987341772, | |
| "grad_norm": 0.3958696595207152, | |
| "learning_rate": 2.4197088438925063e-05, | |
| "loss": 0.2222, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 2.4455696202531647, | |
| "grad_norm": 0.37684290414518307, | |
| "learning_rate": 2.4141576802832417e-05, | |
| "loss": 0.2321, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 2.449620253164557, | |
| "grad_norm": 0.399753340169747, | |
| "learning_rate": 2.408603181104957e-05, | |
| "loss": 0.2008, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 2.4536708860759493, | |
| "grad_norm": 0.4263924791446383, | |
| "learning_rate": 2.4030453910928245e-05, | |
| "loss": 0.2013, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.457721518987342, | |
| "grad_norm": 0.3840395770540428, | |
| "learning_rate": 2.397484355008521e-05, | |
| "loss": 0.217, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 2.4617721518987343, | |
| "grad_norm": 0.4777622695034794, | |
| "learning_rate": 2.3919201176398662e-05, | |
| "loss": 0.2156, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 2.4658227848101264, | |
| "grad_norm": 0.38976186972389315, | |
| "learning_rate": 2.3863527238004633e-05, | |
| "loss": 0.2045, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 2.469873417721519, | |
| "grad_norm": 0.42375555865673825, | |
| "learning_rate": 2.380782218329337e-05, | |
| "loss": 0.2184, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 2.4739240506329114, | |
| "grad_norm": 0.3934473306760485, | |
| "learning_rate": 2.3752086460905725e-05, | |
| "loss": 0.2213, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.477974683544304, | |
| "grad_norm": 0.378769530869701, | |
| "learning_rate": 2.3696320519729544e-05, | |
| "loss": 0.2149, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 2.482025316455696, | |
| "grad_norm": 0.402420413340743, | |
| "learning_rate": 2.3640524808896045e-05, | |
| "loss": 0.2004, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 2.4860759493670885, | |
| "grad_norm": 0.38353555360896263, | |
| "learning_rate": 2.3584699777776222e-05, | |
| "loss": 0.1965, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 2.490126582278481, | |
| "grad_norm": 0.38386002581610185, | |
| "learning_rate": 2.3528845875977195e-05, | |
| "loss": 0.2267, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 2.4941772151898736, | |
| "grad_norm": 0.39294322198523934, | |
| "learning_rate": 2.3472963553338614e-05, | |
| "loss": 0.2286, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.4982278481012656, | |
| "grad_norm": 0.3573216076937674, | |
| "learning_rate": 2.341705325992901e-05, | |
| "loss": 0.2065, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.502278481012658, | |
| "grad_norm": 0.3813218672445825, | |
| "learning_rate": 2.336111544604222e-05, | |
| "loss": 0.2027, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 0.37355920894407574, | |
| "learning_rate": 2.33051505621937e-05, | |
| "loss": 0.2053, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.510379746835443, | |
| "grad_norm": 0.3579460403579799, | |
| "learning_rate": 2.324915905911693e-05, | |
| "loss": 0.2139, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 2.5144303797468357, | |
| "grad_norm": 0.41901165407541474, | |
| "learning_rate": 2.319314138775977e-05, | |
| "loss": 0.2111, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.5184810126582278, | |
| "grad_norm": 0.38382613502964325, | |
| "learning_rate": 2.3137097999280856e-05, | |
| "loss": 0.214, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 2.5225316455696203, | |
| "grad_norm": 0.35538819867409627, | |
| "learning_rate": 2.308102934504593e-05, | |
| "loss": 0.2179, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.526582278481013, | |
| "grad_norm": 0.37815812784963637, | |
| "learning_rate": 2.3024935876624222e-05, | |
| "loss": 0.2241, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 2.530632911392405, | |
| "grad_norm": 0.4000620478305346, | |
| "learning_rate": 2.2968818045784813e-05, | |
| "loss": 0.2052, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.5346835443037974, | |
| "grad_norm": 0.37458344062297044, | |
| "learning_rate": 2.2912676304493006e-05, | |
| "loss": 0.2035, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.53873417721519, | |
| "grad_norm": 0.3672952857971511, | |
| "learning_rate": 2.2856511104906668e-05, | |
| "loss": 0.2118, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.5427848101265824, | |
| "grad_norm": 0.3812477879265681, | |
| "learning_rate": 2.2800322899372586e-05, | |
| "loss": 0.205, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 2.546835443037975, | |
| "grad_norm": 0.338608330973011, | |
| "learning_rate": 2.2744112140422844e-05, | |
| "loss": 0.2051, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.550886075949367, | |
| "grad_norm": 0.3833328806872416, | |
| "learning_rate": 2.2687879280771177e-05, | |
| "loss": 0.2073, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 2.5549367088607595, | |
| "grad_norm": 0.37468123631155786, | |
| "learning_rate": 2.26316247733093e-05, | |
| "loss": 0.212, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.558987341772152, | |
| "grad_norm": 0.36116594098102994, | |
| "learning_rate": 2.257534907110328e-05, | |
| "loss": 0.2078, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 2.563037974683544, | |
| "grad_norm": 0.3857568642603925, | |
| "learning_rate": 2.2519052627389882e-05, | |
| "loss": 0.2001, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.5670886075949366, | |
| "grad_norm": 0.390111902887494, | |
| "learning_rate": 2.246273589557294e-05, | |
| "loss": 0.1996, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 2.571139240506329, | |
| "grad_norm": 0.386175446761627, | |
| "learning_rate": 2.240639932921966e-05, | |
| "loss": 0.2135, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.5751898734177217, | |
| "grad_norm": 0.3675482683026408, | |
| "learning_rate": 2.2350043382056995e-05, | |
| "loss": 0.1988, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.579240506329114, | |
| "grad_norm": 0.3817380444265837, | |
| "learning_rate": 2.2293668507968015e-05, | |
| "loss": 0.2146, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.5832911392405062, | |
| "grad_norm": 0.41270581615691004, | |
| "learning_rate": 2.2237275160988186e-05, | |
| "loss": 0.2049, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 2.5873417721518988, | |
| "grad_norm": 0.35713907055575717, | |
| "learning_rate": 2.2180863795301787e-05, | |
| "loss": 0.1956, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.5913924050632913, | |
| "grad_norm": 0.38095097203492273, | |
| "learning_rate": 2.212443486523819e-05, | |
| "loss": 0.2085, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 2.5954430379746833, | |
| "grad_norm": 0.39442091450826666, | |
| "learning_rate": 2.2067988825268243e-05, | |
| "loss": 0.2203, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.599493670886076, | |
| "grad_norm": 0.37531900528394624, | |
| "learning_rate": 2.2011526130000596e-05, | |
| "loss": 0.2167, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 2.6035443037974684, | |
| "grad_norm": 0.3620996661598064, | |
| "learning_rate": 2.1955047234178038e-05, | |
| "loss": 0.1922, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 0.35714860357679973, | |
| "learning_rate": 2.1898552592673825e-05, | |
| "loss": 0.2076, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 2.6116455696202534, | |
| "grad_norm": 0.37165487907606976, | |
| "learning_rate": 2.184204266048803e-05, | |
| "loss": 0.2093, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.6156962025316455, | |
| "grad_norm": 0.3818729971203178, | |
| "learning_rate": 2.1785517892743887e-05, | |
| "loss": 0.2226, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.619746835443038, | |
| "grad_norm": 0.3954630187723401, | |
| "learning_rate": 2.17289787446841e-05, | |
| "loss": 0.2012, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.6237974683544305, | |
| "grad_norm": 0.38650855644431725, | |
| "learning_rate": 2.1672425671667198e-05, | |
| "loss": 0.2173, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 2.6278481012658226, | |
| "grad_norm": 0.3816734174800455, | |
| "learning_rate": 2.161585912916385e-05, | |
| "loss": 0.2146, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.631898734177215, | |
| "grad_norm": 0.40108293667728245, | |
| "learning_rate": 2.1559279572753214e-05, | |
| "loss": 0.2014, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.6359493670886076, | |
| "grad_norm": 0.3887028039703008, | |
| "learning_rate": 2.1502687458119268e-05, | |
| "loss": 0.2288, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.43294791203142613, | |
| "learning_rate": 2.1446083241047116e-05, | |
| "loss": 0.1969, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.6440506329113926, | |
| "grad_norm": 0.4352330896283679, | |
| "learning_rate": 2.1389467377419333e-05, | |
| "loss": 0.2192, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.6481012658227847, | |
| "grad_norm": 0.38799052354610347, | |
| "learning_rate": 2.133284032321232e-05, | |
| "loss": 0.2082, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.6521518987341772, | |
| "grad_norm": 0.4227017320894025, | |
| "learning_rate": 2.1276202534492566e-05, | |
| "loss": 0.2256, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.6562025316455697, | |
| "grad_norm": 0.375872096661127, | |
| "learning_rate": 2.121955446741306e-05, | |
| "loss": 0.2148, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.660253164556962, | |
| "grad_norm": 0.40675807340866127, | |
| "learning_rate": 2.1162896578209517e-05, | |
| "loss": 0.2077, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.6643037974683543, | |
| "grad_norm": 0.36114609557159394, | |
| "learning_rate": 2.1106229323196813e-05, | |
| "loss": 0.215, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.668354430379747, | |
| "grad_norm": 0.38899019247745326, | |
| "learning_rate": 2.1049553158765214e-05, | |
| "loss": 0.2264, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.6724050632911394, | |
| "grad_norm": 0.389836030146083, | |
| "learning_rate": 2.0992868541376764e-05, | |
| "loss": 0.2022, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.676455696202532, | |
| "grad_norm": 0.41186682910669126, | |
| "learning_rate": 2.093617592756158e-05, | |
| "loss": 0.1975, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.680506329113924, | |
| "grad_norm": 0.436220864841521, | |
| "learning_rate": 2.0879475773914167e-05, | |
| "loss": 0.213, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.6845569620253165, | |
| "grad_norm": 0.399031601609028, | |
| "learning_rate": 2.082276853708978e-05, | |
| "loss": 0.2145, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.688607594936709, | |
| "grad_norm": 0.38771598247673533, | |
| "learning_rate": 2.076605467380071e-05, | |
| "loss": 0.2211, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.692658227848101, | |
| "grad_norm": 0.38384311392107223, | |
| "learning_rate": 2.0709334640812613e-05, | |
| "loss": 0.1933, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.6967088607594936, | |
| "grad_norm": 0.40711281305316427, | |
| "learning_rate": 2.0652608894940824e-05, | |
| "loss": 0.209, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.700759493670886, | |
| "grad_norm": 0.3460818231215482, | |
| "learning_rate": 2.0595877893046722e-05, | |
| "loss": 0.22, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.7048101265822786, | |
| "grad_norm": 0.3884422685293419, | |
| "learning_rate": 2.0539142092033985e-05, | |
| "loss": 0.2235, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 0.3410658902231116, | |
| "learning_rate": 2.048240194884496e-05, | |
| "loss": 0.2165, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.712911392405063, | |
| "grad_norm": 0.37064541495886205, | |
| "learning_rate": 2.042565792045695e-05, | |
| "loss": 0.197, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.7169620253164557, | |
| "grad_norm": 0.3651805295210989, | |
| "learning_rate": 2.036891046387857e-05, | |
| "loss": 0.1958, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.721012658227848, | |
| "grad_norm": 0.36434694235957416, | |
| "learning_rate": 2.0312160036146036e-05, | |
| "loss": 0.2219, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.7250632911392403, | |
| "grad_norm": 0.3776560661209156, | |
| "learning_rate": 2.025540709431948e-05, | |
| "loss": 0.242, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.729113924050633, | |
| "grad_norm": 0.3462016024506827, | |
| "learning_rate": 2.0198652095479298e-05, | |
| "loss": 0.2143, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.7331645569620253, | |
| "grad_norm": 0.36016280633734893, | |
| "learning_rate": 2.014189549672245e-05, | |
| "loss": 0.201, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.737215189873418, | |
| "grad_norm": 0.40500663917994817, | |
| "learning_rate": 2.0085137755158776e-05, | |
| "loss": 0.2088, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.7412658227848103, | |
| "grad_norm": 0.37353338140762743, | |
| "learning_rate": 2.0028379327907327e-05, | |
| "loss": 0.2146, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.7453164556962024, | |
| "grad_norm": 0.3912754780627909, | |
| "learning_rate": 1.9971620672092676e-05, | |
| "loss": 0.2089, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.749367088607595, | |
| "grad_norm": 0.3638908767158199, | |
| "learning_rate": 1.991486224484123e-05, | |
| "loss": 0.2181, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.7534177215189874, | |
| "grad_norm": 0.3668714212493011, | |
| "learning_rate": 1.985810450327756e-05, | |
| "loss": 0.2341, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.7574683544303795, | |
| "grad_norm": 0.352345293054932, | |
| "learning_rate": 1.9801347904520706e-05, | |
| "loss": 0.2054, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.761518987341772, | |
| "grad_norm": 0.35266971616383924, | |
| "learning_rate": 1.974459290568053e-05, | |
| "loss": 0.2002, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.7655696202531646, | |
| "grad_norm": 0.3779084014866482, | |
| "learning_rate": 1.968783996385397e-05, | |
| "loss": 0.2119, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.769620253164557, | |
| "grad_norm": 0.3417014537203359, | |
| "learning_rate": 1.963108953612143e-05, | |
| "loss": 0.2238, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.7736708860759496, | |
| "grad_norm": 0.37788300378942935, | |
| "learning_rate": 1.9574342079543056e-05, | |
| "loss": 0.2033, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.7777215189873417, | |
| "grad_norm": 0.3693014235043108, | |
| "learning_rate": 1.9517598051155046e-05, | |
| "loss": 0.2261, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.781772151898734, | |
| "grad_norm": 0.34362737744399285, | |
| "learning_rate": 1.9460857907966025e-05, | |
| "loss": 0.2111, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.7858227848101267, | |
| "grad_norm": 0.3520097144724682, | |
| "learning_rate": 1.9404122106953285e-05, | |
| "loss": 0.1921, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.7898734177215188, | |
| "grad_norm": 0.36526485498840333, | |
| "learning_rate": 1.9347391105059176e-05, | |
| "loss": 0.1985, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.7939240506329113, | |
| "grad_norm": 0.3540901138580736, | |
| "learning_rate": 1.92906653591874e-05, | |
| "loss": 0.2135, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.797974683544304, | |
| "grad_norm": 0.3494768149620793, | |
| "learning_rate": 1.9233945326199295e-05, | |
| "loss": 0.2172, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.8020253164556963, | |
| "grad_norm": 0.3653937786854222, | |
| "learning_rate": 1.917723146291022e-05, | |
| "loss": 0.2059, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.806075949367089, | |
| "grad_norm": 0.34867207473282463, | |
| "learning_rate": 1.912052422608584e-05, | |
| "loss": 0.2186, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 0.35903283278769227, | |
| "learning_rate": 1.9063824072438428e-05, | |
| "loss": 0.2185, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.8141772151898734, | |
| "grad_norm": 0.34478024837907045, | |
| "learning_rate": 1.9007131458623246e-05, | |
| "loss": 0.2254, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.818227848101266, | |
| "grad_norm": 0.34901638064140184, | |
| "learning_rate": 1.895044684123479e-05, | |
| "loss": 0.2338, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.822278481012658, | |
| "grad_norm": 0.3460749152726901, | |
| "learning_rate": 1.8893770676803194e-05, | |
| "loss": 0.2017, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.8263291139240505, | |
| "grad_norm": 0.3489906986498477, | |
| "learning_rate": 1.8837103421790486e-05, | |
| "loss": 0.2268, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.830379746835443, | |
| "grad_norm": 0.36479121091332967, | |
| "learning_rate": 1.8780445532586952e-05, | |
| "loss": 0.2016, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.8344303797468355, | |
| "grad_norm": 0.3710391089530568, | |
| "learning_rate": 1.872379746550743e-05, | |
| "loss": 0.1957, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.838481012658228, | |
| "grad_norm": 0.37674532422573437, | |
| "learning_rate": 1.866715967678769e-05, | |
| "loss": 0.1974, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.84253164556962, | |
| "grad_norm": 0.37440634485340135, | |
| "learning_rate": 1.861053262258067e-05, | |
| "loss": 0.2164, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.8465822784810126, | |
| "grad_norm": 0.3895620073009604, | |
| "learning_rate": 1.8553916758952897e-05, | |
| "loss": 0.205, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.850632911392405, | |
| "grad_norm": 0.34047732476886916, | |
| "learning_rate": 1.8497312541880735e-05, | |
| "loss": 0.1999, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.8546835443037972, | |
| "grad_norm": 0.3752996136958883, | |
| "learning_rate": 1.8440720427246786e-05, | |
| "loss": 0.2079, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.8587341772151897, | |
| "grad_norm": 0.38651080004184135, | |
| "learning_rate": 1.8384140870836157e-05, | |
| "loss": 0.1929, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.8627848101265823, | |
| "grad_norm": 0.3598448082960502, | |
| "learning_rate": 1.8327574328332806e-05, | |
| "loss": 0.2039, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.8668354430379748, | |
| "grad_norm": 0.3741178878271117, | |
| "learning_rate": 1.8271021255315906e-05, | |
| "loss": 0.2133, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.8708860759493673, | |
| "grad_norm": 0.38269574581690285, | |
| "learning_rate": 1.8214482107256117e-05, | |
| "loss": 0.2136, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.8749367088607594, | |
| "grad_norm": 0.36272823150714, | |
| "learning_rate": 1.8157957339511968e-05, | |
| "loss": 0.2066, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.878987341772152, | |
| "grad_norm": 0.3402196024525534, | |
| "learning_rate": 1.8101447407326182e-05, | |
| "loss": 0.2111, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.8830379746835444, | |
| "grad_norm": 0.38220673299042984, | |
| "learning_rate": 1.8044952765821966e-05, | |
| "loss": 0.2219, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.8870886075949365, | |
| "grad_norm": 0.4007796337972502, | |
| "learning_rate": 1.7988473869999407e-05, | |
| "loss": 0.1997, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.891139240506329, | |
| "grad_norm": 0.3645051340151498, | |
| "learning_rate": 1.7932011174731764e-05, | |
| "loss": 0.1962, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.8951898734177215, | |
| "grad_norm": 0.38627890470232307, | |
| "learning_rate": 1.7875565134761817e-05, | |
| "loss": 0.2074, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.899240506329114, | |
| "grad_norm": 0.37164468616514335, | |
| "learning_rate": 1.7819136204698226e-05, | |
| "loss": 0.212, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.9032911392405065, | |
| "grad_norm": 0.3862932904349216, | |
| "learning_rate": 1.776272483901182e-05, | |
| "loss": 0.1999, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.9073417721518986, | |
| "grad_norm": 0.3767721713188255, | |
| "learning_rate": 1.7706331492031995e-05, | |
| "loss": 0.2026, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 0.3462490502722588, | |
| "learning_rate": 1.764995661794301e-05, | |
| "loss": 0.2131, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.9154430379746836, | |
| "grad_norm": 0.3685594913874667, | |
| "learning_rate": 1.759360067078035e-05, | |
| "loss": 0.221, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.9194936708860757, | |
| "grad_norm": 0.36916214803172437, | |
| "learning_rate": 1.7537264104427064e-05, | |
| "loss": 0.1989, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.923544303797468, | |
| "grad_norm": 0.3523457238761439, | |
| "learning_rate": 1.748094737261012e-05, | |
| "loss": 0.2263, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.9275949367088607, | |
| "grad_norm": 0.3556601430486097, | |
| "learning_rate": 1.7424650928896726e-05, | |
| "loss": 0.212, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.9316455696202532, | |
| "grad_norm": 0.3516395461547922, | |
| "learning_rate": 1.7368375226690712e-05, | |
| "loss": 0.2092, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.9356962025316458, | |
| "grad_norm": 0.33916270971414186, | |
| "learning_rate": 1.731212071922883e-05, | |
| "loss": 0.2028, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.939746835443038, | |
| "grad_norm": 0.36143112656132226, | |
| "learning_rate": 1.7255887859577156e-05, | |
| "loss": 0.2246, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.9437974683544303, | |
| "grad_norm": 0.35717912252537776, | |
| "learning_rate": 1.7199677100627427e-05, | |
| "loss": 0.2003, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.947848101265823, | |
| "grad_norm": 0.33736384468960345, | |
| "learning_rate": 1.7143488895093343e-05, | |
| "loss": 0.2197, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.951898734177215, | |
| "grad_norm": 0.3624370206019351, | |
| "learning_rate": 1.7087323695506994e-05, | |
| "loss": 0.2001, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.9559493670886074, | |
| "grad_norm": 0.35627585959966884, | |
| "learning_rate": 1.7031181954215194e-05, | |
| "loss": 0.2108, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.3310352205770688, | |
| "learning_rate": 1.6975064123375788e-05, | |
| "loss": 0.2012, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.9640506329113925, | |
| "grad_norm": 0.32777358315288735, | |
| "learning_rate": 1.6918970654954084e-05, | |
| "loss": 0.1939, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.968101265822785, | |
| "grad_norm": 0.34667869348031316, | |
| "learning_rate": 1.686290200071915e-05, | |
| "loss": 0.2177, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.972151898734177, | |
| "grad_norm": 0.3494520416889407, | |
| "learning_rate": 1.6806858612240234e-05, | |
| "loss": 0.2215, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.9762025316455696, | |
| "grad_norm": 0.349627740640338, | |
| "learning_rate": 1.6750840940883078e-05, | |
| "loss": 0.2127, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.980253164556962, | |
| "grad_norm": 0.3719832447005335, | |
| "learning_rate": 1.6694849437806305e-05, | |
| "loss": 0.1998, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.984303797468354, | |
| "grad_norm": 0.36031093050956375, | |
| "learning_rate": 1.663888455395778e-05, | |
| "loss": 0.1993, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.9883544303797467, | |
| "grad_norm": 0.3403490033086363, | |
| "learning_rate": 1.6582946740070995e-05, | |
| "loss": 0.2006, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.992405063291139, | |
| "grad_norm": 0.3666300584613887, | |
| "learning_rate": 1.6527036446661396e-05, | |
| "loss": 0.1937, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.9964556962025317, | |
| "grad_norm": 0.337653439331429, | |
| "learning_rate": 1.6471154124022818e-05, | |
| "loss": 0.2155, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 3.002025316455696, | |
| "grad_norm": 0.3519862987658842, | |
| "learning_rate": 1.6415300222223788e-05, | |
| "loss": 0.1836, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.0060759493670886, | |
| "grad_norm": 0.41600337433513934, | |
| "learning_rate": 1.6359475191103958e-05, | |
| "loss": 0.1439, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 3.010126582278481, | |
| "grad_norm": 0.3133228019991403, | |
| "learning_rate": 1.6303679480270466e-05, | |
| "loss": 0.1291, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 3.0141772151898736, | |
| "grad_norm": 0.42604307267862185, | |
| "learning_rate": 1.624791353909428e-05, | |
| "loss": 0.1419, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 3.0182278481012657, | |
| "grad_norm": 0.5323898693253035, | |
| "learning_rate": 1.619217781670663e-05, | |
| "loss": 0.1371, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 3.022278481012658, | |
| "grad_norm": 0.42525553447026376, | |
| "learning_rate": 1.6136472761995373e-05, | |
| "loss": 0.142, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.0263291139240507, | |
| "grad_norm": 0.37241066759681546, | |
| "learning_rate": 1.608079882360134e-05, | |
| "loss": 0.1305, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 3.030379746835443, | |
| "grad_norm": 0.439359176951651, | |
| "learning_rate": 1.60251564499148e-05, | |
| "loss": 0.1247, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 3.0344303797468353, | |
| "grad_norm": 0.38378434010884177, | |
| "learning_rate": 1.596954608907176e-05, | |
| "loss": 0.1389, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 3.038481012658228, | |
| "grad_norm": 0.3617620237852232, | |
| "learning_rate": 1.591396818895043e-05, | |
| "loss": 0.1282, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 3.0425316455696203, | |
| "grad_norm": 0.3629496883050939, | |
| "learning_rate": 1.585842319716759e-05, | |
| "loss": 0.1274, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.046582278481013, | |
| "grad_norm": 0.3985342253430489, | |
| "learning_rate": 1.5802911561074944e-05, | |
| "loss": 0.1436, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 0.3655605490045354, | |
| "learning_rate": 1.5747433727755595e-05, | |
| "loss": 0.1142, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 3.0546835443037974, | |
| "grad_norm": 0.38824913767417873, | |
| "learning_rate": 1.5691990144020376e-05, | |
| "loss": 0.1322, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 3.05873417721519, | |
| "grad_norm": 0.3508953319418693, | |
| "learning_rate": 1.5636581256404297e-05, | |
| "loss": 0.1494, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 3.0627848101265824, | |
| "grad_norm": 0.3370770452972283, | |
| "learning_rate": 1.558120751116291e-05, | |
| "loss": 0.1194, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.0668354430379745, | |
| "grad_norm": 0.35667472219952023, | |
| "learning_rate": 1.552586935426876e-05, | |
| "loss": 0.1255, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 3.070886075949367, | |
| "grad_norm": 0.3150110121107263, | |
| "learning_rate": 1.547056723140774e-05, | |
| "loss": 0.1242, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 3.0749367088607595, | |
| "grad_norm": 0.35602908254505095, | |
| "learning_rate": 1.5415301587975565e-05, | |
| "loss": 0.1312, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 3.078987341772152, | |
| "grad_norm": 0.3470233646642732, | |
| "learning_rate": 1.536007286907411e-05, | |
| "loss": 0.1316, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 3.083037974683544, | |
| "grad_norm": 0.3314261332484569, | |
| "learning_rate": 1.5304881519507896e-05, | |
| "loss": 0.1338, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.0870886075949366, | |
| "grad_norm": 0.3543928301338737, | |
| "learning_rate": 1.5249727983780453e-05, | |
| "loss": 0.1275, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 3.091139240506329, | |
| "grad_norm": 0.3474148021654982, | |
| "learning_rate": 1.5194612706090786e-05, | |
| "loss": 0.1275, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 3.0951898734177217, | |
| "grad_norm": 0.3364981535952058, | |
| "learning_rate": 1.5139536130329771e-05, | |
| "loss": 0.128, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 3.0992405063291137, | |
| "grad_norm": 0.33585841377946457, | |
| "learning_rate": 1.508449870007656e-05, | |
| "loss": 0.1274, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 3.1032911392405063, | |
| "grad_norm": 0.339779996523037, | |
| "learning_rate": 1.5029500858595056e-05, | |
| "loss": 0.1235, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 3.1073417721518988, | |
| "grad_norm": 0.3230928336152605, | |
| "learning_rate": 1.4974543048830328e-05, | |
| "loss": 0.1309, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 3.1113924050632913, | |
| "grad_norm": 0.31936883767454977, | |
| "learning_rate": 1.4919625713405e-05, | |
| "loss": 0.1272, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 3.1154430379746834, | |
| "grad_norm": 0.31787597880763524, | |
| "learning_rate": 1.4864749294615756e-05, | |
| "loss": 0.1613, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 3.119493670886076, | |
| "grad_norm": 0.30133637181886985, | |
| "learning_rate": 1.4809914234429716e-05, | |
| "loss": 0.1254, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 3.1235443037974684, | |
| "grad_norm": 0.3164684597211431, | |
| "learning_rate": 1.4755120974480923e-05, | |
| "loss": 0.1327, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.127594936708861, | |
| "grad_norm": 0.31005971419477296, | |
| "learning_rate": 1.4700369956066771e-05, | |
| "loss": 0.1284, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 3.131645569620253, | |
| "grad_norm": 0.31783347036794324, | |
| "learning_rate": 1.4645661620144413e-05, | |
| "loss": 0.1439, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 3.1356962025316455, | |
| "grad_norm": 0.32193370316762066, | |
| "learning_rate": 1.4590996407327284e-05, | |
| "loss": 0.1268, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 3.139746835443038, | |
| "grad_norm": 0.315278245735706, | |
| "learning_rate": 1.4536374757881487e-05, | |
| "loss": 0.1211, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 3.1437974683544305, | |
| "grad_norm": 0.31914331331593854, | |
| "learning_rate": 1.4481797111722271e-05, | |
| "loss": 0.1333, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.1478481012658226, | |
| "grad_norm": 0.3335389743800418, | |
| "learning_rate": 1.4427263908410507e-05, | |
| "loss": 0.1214, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 0.3154319051236716, | |
| "learning_rate": 1.4372775587149108e-05, | |
| "loss": 0.1361, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 3.1559493670886076, | |
| "grad_norm": 0.3191977705962591, | |
| "learning_rate": 1.4318332586779522e-05, | |
| "loss": 0.1189, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "grad_norm": 0.3187504647787026, | |
| "learning_rate": 1.4263935345778202e-05, | |
| "loss": 0.1214, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 3.164050632911392, | |
| "grad_norm": 0.3369110649367975, | |
| "learning_rate": 1.420958430225303e-05, | |
| "loss": 0.1398, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.1681012658227847, | |
| "grad_norm": 0.3227612268626544, | |
| "learning_rate": 1.415527989393985e-05, | |
| "loss": 0.1486, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 3.1721518987341772, | |
| "grad_norm": 0.3423430499768305, | |
| "learning_rate": 1.410102255819891e-05, | |
| "loss": 0.1547, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 3.1762025316455698, | |
| "grad_norm": 0.32441677886075143, | |
| "learning_rate": 1.404681273201131e-05, | |
| "loss": 0.1397, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 3.180253164556962, | |
| "grad_norm": 0.3331903455773248, | |
| "learning_rate": 1.399265085197556e-05, | |
| "loss": 0.1156, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 3.1843037974683543, | |
| "grad_norm": 0.31987841761785607, | |
| "learning_rate": 1.393853735430398e-05, | |
| "loss": 0.1399, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 3.188354430379747, | |
| "grad_norm": 0.32851798742173105, | |
| "learning_rate": 1.3884472674819246e-05, | |
| "loss": 0.1226, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 3.1924050632911394, | |
| "grad_norm": 0.3321245051747501, | |
| "learning_rate": 1.3830457248950864e-05, | |
| "loss": 0.136, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 3.1964556962025314, | |
| "grad_norm": 0.3244029341234907, | |
| "learning_rate": 1.377649151173163e-05, | |
| "loss": 0.1398, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 3.200506329113924, | |
| "grad_norm": 0.3393001424210053, | |
| "learning_rate": 1.3722575897794181e-05, | |
| "loss": 0.1154, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 3.2045569620253165, | |
| "grad_norm": 0.3148681905625056, | |
| "learning_rate": 1.3668710841367472e-05, | |
| "loss": 0.1162, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.208607594936709, | |
| "grad_norm": 0.3358829742637307, | |
| "learning_rate": 1.361489677627324e-05, | |
| "loss": 0.129, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 3.212658227848101, | |
| "grad_norm": 0.3184862812316988, | |
| "learning_rate": 1.3561134135922585e-05, | |
| "loss": 0.1435, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 3.2167088607594936, | |
| "grad_norm": 0.3019633836521196, | |
| "learning_rate": 1.350742335331241e-05, | |
| "loss": 0.1248, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 3.220759493670886, | |
| "grad_norm": 0.305293584519831, | |
| "learning_rate": 1.345376486102198e-05, | |
| "loss": 0.1408, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 3.2248101265822786, | |
| "grad_norm": 0.33930088427150507, | |
| "learning_rate": 1.3400159091209414e-05, | |
| "loss": 0.141, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 3.2288607594936707, | |
| "grad_norm": 0.31584525131619484, | |
| "learning_rate": 1.3346606475608216e-05, | |
| "loss": 0.1196, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 3.232911392405063, | |
| "grad_norm": 0.31370918085422994, | |
| "learning_rate": 1.3293107445523781e-05, | |
| "loss": 0.1365, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 3.2369620253164557, | |
| "grad_norm": 0.32057706636431893, | |
| "learning_rate": 1.3239662431829949e-05, | |
| "loss": 0.1245, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 3.2410126582278482, | |
| "grad_norm": 0.31155739266702365, | |
| "learning_rate": 1.3186271864965509e-05, | |
| "loss": 0.1315, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 3.2450632911392403, | |
| "grad_norm": 0.3130630412307185, | |
| "learning_rate": 1.3132936174930756e-05, | |
| "loss": 0.1321, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.249113924050633, | |
| "grad_norm": 0.3240081400494484, | |
| "learning_rate": 1.3079655791283995e-05, | |
| "loss": 0.1234, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 0.3077729794391428, | |
| "learning_rate": 1.3026431143138108e-05, | |
| "loss": 0.1512, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 3.257215189873418, | |
| "grad_norm": 0.33517996418789686, | |
| "learning_rate": 1.2973262659157114e-05, | |
| "loss": 0.1166, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 3.26126582278481, | |
| "grad_norm": 0.32209835684774346, | |
| "learning_rate": 1.2920150767552651e-05, | |
| "loss": 0.1451, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 3.2653164556962024, | |
| "grad_norm": 0.3164496663715927, | |
| "learning_rate": 1.2867095896080607e-05, | |
| "loss": 0.1361, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.269367088607595, | |
| "grad_norm": 0.3235169478860452, | |
| "learning_rate": 1.2814098472037612e-05, | |
| "loss": 0.1423, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 3.2734177215189875, | |
| "grad_norm": 0.33365619974825983, | |
| "learning_rate": 1.276115892225764e-05, | |
| "loss": 0.1181, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 3.27746835443038, | |
| "grad_norm": 0.30965717741204263, | |
| "learning_rate": 1.2708277673108555e-05, | |
| "loss": 0.1259, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 3.281518987341772, | |
| "grad_norm": 0.3111336679511202, | |
| "learning_rate": 1.2655455150488649e-05, | |
| "loss": 0.1158, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 3.2855696202531646, | |
| "grad_norm": 0.32499115380063265, | |
| "learning_rate": 1.2602691779823272e-05, | |
| "loss": 0.1349, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.289620253164557, | |
| "grad_norm": 0.3162664160974659, | |
| "learning_rate": 1.2549987986061355e-05, | |
| "loss": 0.1118, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 3.293670886075949, | |
| "grad_norm": 0.31154973761692845, | |
| "learning_rate": 1.2497344193672005e-05, | |
| "loss": 0.14, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 3.2977215189873417, | |
| "grad_norm": 0.31980323081762735, | |
| "learning_rate": 1.2444760826641092e-05, | |
| "loss": 0.1294, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 3.301772151898734, | |
| "grad_norm": 0.3259396514554832, | |
| "learning_rate": 1.2392238308467817e-05, | |
| "loss": 0.1427, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 3.3058227848101267, | |
| "grad_norm": 0.32665065853239444, | |
| "learning_rate": 1.2339777062161326e-05, | |
| "loss": 0.1375, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.309873417721519, | |
| "grad_norm": 0.3149590453425815, | |
| "learning_rate": 1.2287377510237293e-05, | |
| "loss": 0.1248, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 3.3139240506329113, | |
| "grad_norm": 0.3175697792159252, | |
| "learning_rate": 1.2235040074714488e-05, | |
| "loss": 0.1466, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 3.317974683544304, | |
| "grad_norm": 0.32618705766030776, | |
| "learning_rate": 1.2182765177111434e-05, | |
| "loss": 0.1349, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 3.3220253164556963, | |
| "grad_norm": 0.31951224267848816, | |
| "learning_rate": 1.213055323844297e-05, | |
| "loss": 0.1274, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 3.3260759493670884, | |
| "grad_norm": 0.3155991583461842, | |
| "learning_rate": 1.2078404679216864e-05, | |
| "loss": 0.1327, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.330126582278481, | |
| "grad_norm": 0.3252106878627646, | |
| "learning_rate": 1.2026319919430458e-05, | |
| "loss": 0.1223, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 3.3341772151898734, | |
| "grad_norm": 0.32731956996833034, | |
| "learning_rate": 1.1974299378567227e-05, | |
| "loss": 0.1278, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 3.338227848101266, | |
| "grad_norm": 0.32326595074857734, | |
| "learning_rate": 1.1922343475593462e-05, | |
| "loss": 0.1182, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 3.3422784810126585, | |
| "grad_norm": 0.33112016202888767, | |
| "learning_rate": 1.187045262895488e-05, | |
| "loss": 0.1276, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 3.3463291139240505, | |
| "grad_norm": 0.3197201042817298, | |
| "learning_rate": 1.1818627256573203e-05, | |
| "loss": 0.1158, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.350379746835443, | |
| "grad_norm": 0.333339829066384, | |
| "learning_rate": 1.1766867775842864e-05, | |
| "loss": 0.136, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.32582561196328913, | |
| "learning_rate": 1.1715174603627615e-05, | |
| "loss": 0.1174, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 3.3584810126582276, | |
| "grad_norm": 0.3074538156124539, | |
| "learning_rate": 1.1663548156257147e-05, | |
| "loss": 0.1253, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 3.36253164556962, | |
| "grad_norm": 0.3060590597470408, | |
| "learning_rate": 1.161198884952377e-05, | |
| "loss": 0.117, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 3.3665822784810127, | |
| "grad_norm": 0.332733838145166, | |
| "learning_rate": 1.1560497098679056e-05, | |
| "loss": 0.1267, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.370632911392405, | |
| "grad_norm": 0.34018462471376454, | |
| "learning_rate": 1.1509073318430479e-05, | |
| "loss": 0.1527, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 3.3746835443037977, | |
| "grad_norm": 0.3081064185782735, | |
| "learning_rate": 1.1457717922938116e-05, | |
| "loss": 0.1378, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 3.3787341772151898, | |
| "grad_norm": 0.30910180883891364, | |
| "learning_rate": 1.1406431325811233e-05, | |
| "loss": 0.1274, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 3.3827848101265823, | |
| "grad_norm": 0.3145137945058816, | |
| "learning_rate": 1.135521394010506e-05, | |
| "loss": 0.1261, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 3.386835443037975, | |
| "grad_norm": 0.3064911360970364, | |
| "learning_rate": 1.1304066178317367e-05, | |
| "loss": 0.1185, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.390886075949367, | |
| "grad_norm": 0.30665358073439447, | |
| "learning_rate": 1.1252988452385199e-05, | |
| "loss": 0.1207, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 3.3949367088607594, | |
| "grad_norm": 0.31339135373062327, | |
| "learning_rate": 1.1201981173681536e-05, | |
| "loss": 0.1225, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 3.398987341772152, | |
| "grad_norm": 0.3143717006313313, | |
| "learning_rate": 1.1151044753011991e-05, | |
| "loss": 0.1323, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 3.4030379746835444, | |
| "grad_norm": 0.3266061921129122, | |
| "learning_rate": 1.1100179600611491e-05, | |
| "loss": 0.1126, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 3.407088607594937, | |
| "grad_norm": 0.30994857205656995, | |
| "learning_rate": 1.1049386126140985e-05, | |
| "loss": 0.108, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.411139240506329, | |
| "grad_norm": 0.31857847710838766, | |
| "learning_rate": 1.0998664738684128e-05, | |
| "loss": 0.1373, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 3.4151898734177215, | |
| "grad_norm": 0.30866029606588813, | |
| "learning_rate": 1.0948015846744e-05, | |
| "loss": 0.1374, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 3.419240506329114, | |
| "grad_norm": 0.2927066814447354, | |
| "learning_rate": 1.0897439858239832e-05, | |
| "loss": 0.1388, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 3.423291139240506, | |
| "grad_norm": 0.29944474980304614, | |
| "learning_rate": 1.0846937180503652e-05, | |
| "loss": 0.1283, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 3.4273417721518986, | |
| "grad_norm": 0.3054323326626897, | |
| "learning_rate": 1.0796508220277117e-05, | |
| "loss": 0.1446, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.431392405063291, | |
| "grad_norm": 0.31371606210655395, | |
| "learning_rate": 1.0746153383708107e-05, | |
| "loss": 0.1389, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 3.4354430379746836, | |
| "grad_norm": 0.3284273791577973, | |
| "learning_rate": 1.0695873076347579e-05, | |
| "loss": 0.1428, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 3.439493670886076, | |
| "grad_norm": 0.30744956786803534, | |
| "learning_rate": 1.0645667703146205e-05, | |
| "loss": 0.1218, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 3.4435443037974682, | |
| "grad_norm": 0.3303081317107967, | |
| "learning_rate": 1.0595537668451161e-05, | |
| "loss": 0.1329, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 3.4475949367088607, | |
| "grad_norm": 0.32316425940050214, | |
| "learning_rate": 1.0545483376002854e-05, | |
| "loss": 0.1346, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.4516455696202533, | |
| "grad_norm": 0.32089304603652385, | |
| "learning_rate": 1.0495505228931676e-05, | |
| "loss": 0.1348, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.298867264671637, | |
| "learning_rate": 1.044560362975474e-05, | |
| "loss": 0.1345, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 3.459746835443038, | |
| "grad_norm": 0.3130193066594174, | |
| "learning_rate": 1.0395778980372695e-05, | |
| "loss": 0.1337, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 3.4637974683544304, | |
| "grad_norm": 0.31970615720381806, | |
| "learning_rate": 1.0346031682066381e-05, | |
| "loss": 0.1195, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 3.467848101265823, | |
| "grad_norm": 0.3210271901689455, | |
| "learning_rate": 1.0296362135493724e-05, | |
| "loss": 0.1353, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.4718987341772154, | |
| "grad_norm": 0.3040363307593174, | |
| "learning_rate": 1.0246770740686422e-05, | |
| "loss": 0.1248, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 3.4759493670886075, | |
| "grad_norm": 0.3010202773697072, | |
| "learning_rate": 1.0197257897046743e-05, | |
| "loss": 0.134, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "grad_norm": 0.3173401668750672, | |
| "learning_rate": 1.014782400334433e-05, | |
| "loss": 0.1296, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 3.4840506329113925, | |
| "grad_norm": 0.3171078470258126, | |
| "learning_rate": 1.009846945771296e-05, | |
| "loss": 0.1308, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 3.4881012658227846, | |
| "grad_norm": 0.2988031506456112, | |
| "learning_rate": 1.0049194657647363e-05, | |
| "loss": 0.129, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.492151898734177, | |
| "grad_norm": 0.32374178131669507, | |
| "learning_rate": 1.0000000000000006e-05, | |
| "loss": 0.1155, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 3.4962025316455696, | |
| "grad_norm": 0.3200165709969216, | |
| "learning_rate": 9.950885880977891e-06, | |
| "loss": 0.143, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 3.500253164556962, | |
| "grad_norm": 0.3086795891446176, | |
| "learning_rate": 9.901852696139382e-06, | |
| "loss": 0.1399, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 3.5043037974683546, | |
| "grad_norm": 0.30843094678940697, | |
| "learning_rate": 9.852900840391027e-06, | |
| "loss": 0.1311, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 3.5083544303797467, | |
| "grad_norm": 0.31127899609463033, | |
| "learning_rate": 9.804030707984313e-06, | |
| "loss": 0.1236, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.512405063291139, | |
| "grad_norm": 0.3153783199876622, | |
| "learning_rate": 9.755242692512599e-06, | |
| "loss": 0.1324, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 3.5164556962025317, | |
| "grad_norm": 0.3118322700652409, | |
| "learning_rate": 9.70653718690782e-06, | |
| "loss": 0.1187, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 3.520506329113924, | |
| "grad_norm": 0.3234226502748595, | |
| "learning_rate": 9.657914583437454e-06, | |
| "loss": 0.1295, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 3.5245569620253163, | |
| "grad_norm": 0.3049650663578277, | |
| "learning_rate": 9.609375273701246e-06, | |
| "loss": 0.1248, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 3.528607594936709, | |
| "grad_norm": 0.29862359121242166, | |
| "learning_rate": 9.560919648628133e-06, | |
| "loss": 0.1376, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.5326582278481014, | |
| "grad_norm": 0.3194411374109816, | |
| "learning_rate": 9.512548098473047e-06, | |
| "loss": 0.1201, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 3.536708860759494, | |
| "grad_norm": 0.30251913319529883, | |
| "learning_rate": 9.464261012813825e-06, | |
| "loss": 0.1351, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 3.540759493670886, | |
| "grad_norm": 0.29865373290885083, | |
| "learning_rate": 9.416058780547987e-06, | |
| "loss": 0.1207, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 3.5448101265822785, | |
| "grad_norm": 0.314490646180787, | |
| "learning_rate": 9.367941789889714e-06, | |
| "loss": 0.1405, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 3.548860759493671, | |
| "grad_norm": 0.33089334260265774, | |
| "learning_rate": 9.319910428366607e-06, | |
| "loss": 0.1313, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.552911392405063, | |
| "grad_norm": 0.30228931763627115, | |
| "learning_rate": 9.271965082816667e-06, | |
| "loss": 0.1497, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 0.3056886714010732, | |
| "learning_rate": 9.224106139385111e-06, | |
| "loss": 0.1261, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 3.561012658227848, | |
| "grad_norm": 0.3123366591925014, | |
| "learning_rate": 9.176333983521291e-06, | |
| "loss": 0.138, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 3.5650632911392406, | |
| "grad_norm": 0.29917137505624497, | |
| "learning_rate": 9.12864899997558e-06, | |
| "loss": 0.1301, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 3.569113924050633, | |
| "grad_norm": 0.3029893277877981, | |
| "learning_rate": 9.08105157279628e-06, | |
| "loss": 0.1178, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.573164556962025, | |
| "grad_norm": 0.3147064554316612, | |
| "learning_rate": 9.03354208532653e-06, | |
| "loss": 0.1111, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 3.5772151898734177, | |
| "grad_norm": 0.3103024520680784, | |
| "learning_rate": 8.986120920201205e-06, | |
| "loss": 0.1208, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 3.58126582278481, | |
| "grad_norm": 0.30376967402750576, | |
| "learning_rate": 8.938788459343852e-06, | |
| "loss": 0.16, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 3.5853164556962023, | |
| "grad_norm": 0.30786199599142283, | |
| "learning_rate": 8.8915450839636e-06, | |
| "loss": 0.121, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 3.589367088607595, | |
| "grad_norm": 0.3105354356877044, | |
| "learning_rate": 8.844391174552116e-06, | |
| "loss": 0.1288, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.5934177215189873, | |
| "grad_norm": 0.3213464313400196, | |
| "learning_rate": 8.797327110880479e-06, | |
| "loss": 0.1249, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 3.59746835443038, | |
| "grad_norm": 0.30387457199112056, | |
| "learning_rate": 8.750353271996206e-06, | |
| "loss": 0.1158, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 3.6015189873417723, | |
| "grad_norm": 0.31324951585244865, | |
| "learning_rate": 8.703470036220132e-06, | |
| "loss": 0.1309, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 3.6055696202531644, | |
| "grad_norm": 0.30575570213302206, | |
| "learning_rate": 8.656677781143394e-06, | |
| "loss": 0.1277, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 3.609620253164557, | |
| "grad_norm": 0.285594117338751, | |
| "learning_rate": 8.609976883624377e-06, | |
| "loss": 0.1167, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.6136708860759494, | |
| "grad_norm": 0.2916430363468624, | |
| "learning_rate": 8.563367719785698e-06, | |
| "loss": 0.1294, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 3.6177215189873415, | |
| "grad_norm": 0.31132433518429126, | |
| "learning_rate": 8.516850665011138e-06, | |
| "loss": 0.126, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 3.621772151898734, | |
| "grad_norm": 0.30300431565336317, | |
| "learning_rate": 8.47042609394269e-06, | |
| "loss": 0.1257, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 3.6258227848101265, | |
| "grad_norm": 0.3043804903373158, | |
| "learning_rate": 8.424094380477432e-06, | |
| "loss": 0.1271, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 3.629873417721519, | |
| "grad_norm": 0.29196655745867833, | |
| "learning_rate": 8.37785589776465e-06, | |
| "loss": 0.1001, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.6339240506329116, | |
| "grad_norm": 0.325697608847397, | |
| "learning_rate": 8.331711018202694e-06, | |
| "loss": 0.1305, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 3.6379746835443036, | |
| "grad_norm": 0.30405853431355007, | |
| "learning_rate": 8.285660113436104e-06, | |
| "loss": 0.1353, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 3.642025316455696, | |
| "grad_norm": 0.3299601873697739, | |
| "learning_rate": 8.239703554352527e-06, | |
| "loss": 0.1439, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 3.6460759493670887, | |
| "grad_norm": 0.332497353472452, | |
| "learning_rate": 8.193841711079775e-06, | |
| "loss": 0.1149, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 3.6501265822784807, | |
| "grad_norm": 0.3153178898983026, | |
| "learning_rate": 8.148074952982828e-06, | |
| "loss": 0.1153, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.6541772151898733, | |
| "grad_norm": 0.32084667613327345, | |
| "learning_rate": 8.102403648660859e-06, | |
| "loss": 0.1245, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 0.31306838937914155, | |
| "learning_rate": 8.056828165944282e-06, | |
| "loss": 0.1265, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 3.6622784810126583, | |
| "grad_norm": 0.3104307872509285, | |
| "learning_rate": 8.011348871891762e-06, | |
| "loss": 0.136, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 3.666329113924051, | |
| "grad_norm": 0.30784406516133245, | |
| "learning_rate": 7.965966132787287e-06, | |
| "loss": 0.1195, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 3.670379746835443, | |
| "grad_norm": 0.31642988927923654, | |
| "learning_rate": 7.920680314137189e-06, | |
| "loss": 0.1291, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.6744303797468354, | |
| "grad_norm": 0.30294982039149965, | |
| "learning_rate": 7.875491780667246e-06, | |
| "loss": 0.1258, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 3.678481012658228, | |
| "grad_norm": 0.28752219058067957, | |
| "learning_rate": 7.830400896319667e-06, | |
| "loss": 0.1349, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 3.68253164556962, | |
| "grad_norm": 0.3209291877972472, | |
| "learning_rate": 7.785408024250259e-06, | |
| "loss": 0.121, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 3.6865822784810125, | |
| "grad_norm": 0.31758778870115795, | |
| "learning_rate": 7.74051352682542e-06, | |
| "loss": 0.1334, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 3.690632911392405, | |
| "grad_norm": 0.31371410061713356, | |
| "learning_rate": 7.695717765619257e-06, | |
| "loss": 0.1204, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.6946835443037975, | |
| "grad_norm": 0.3242115124671574, | |
| "learning_rate": 7.651021101410673e-06, | |
| "loss": 0.1258, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 3.69873417721519, | |
| "grad_norm": 0.30799080059334544, | |
| "learning_rate": 7.606423894180464e-06, | |
| "loss": 0.1257, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 3.702784810126582, | |
| "grad_norm": 0.32766812037626963, | |
| "learning_rate": 7.56192650310839e-06, | |
| "loss": 0.1175, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 3.7068354430379746, | |
| "grad_norm": 0.29519214508846414, | |
| "learning_rate": 7.517529286570349e-06, | |
| "loss": 0.1271, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 3.710886075949367, | |
| "grad_norm": 0.3107560572303716, | |
| "learning_rate": 7.473232602135387e-06, | |
| "loss": 0.1265, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.714936708860759, | |
| "grad_norm": 0.3144426043455609, | |
| "learning_rate": 7.429036806562935e-06, | |
| "loss": 0.1249, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 3.7189873417721517, | |
| "grad_norm": 0.31336374208628803, | |
| "learning_rate": 7.3849422557998455e-06, | |
| "loss": 0.1397, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 3.7230379746835442, | |
| "grad_norm": 0.29989811325592486, | |
| "learning_rate": 7.340949304977567e-06, | |
| "loss": 0.1363, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 3.7270886075949368, | |
| "grad_norm": 0.3030772307430105, | |
| "learning_rate": 7.297058308409282e-06, | |
| "loss": 0.1234, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 3.7311392405063293, | |
| "grad_norm": 0.31436689176623717, | |
| "learning_rate": 7.25326961958704e-06, | |
| "loss": 0.1262, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.7351898734177214, | |
| "grad_norm": 0.29822373404254554, | |
| "learning_rate": 7.209583591178921e-06, | |
| "loss": 0.1274, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 3.739240506329114, | |
| "grad_norm": 0.3081251143766001, | |
| "learning_rate": 7.1660005750261925e-06, | |
| "loss": 0.1142, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 3.7432911392405064, | |
| "grad_norm": 0.3135354529407981, | |
| "learning_rate": 7.1225209221404765e-06, | |
| "loss": 0.1123, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 3.747341772151899, | |
| "grad_norm": 0.29822388805375954, | |
| "learning_rate": 7.079144982700909e-06, | |
| "loss": 0.0971, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 3.7513924050632914, | |
| "grad_norm": 0.3167073703032467, | |
| "learning_rate": 7.0358731060513695e-06, | |
| "loss": 0.1215, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.7554430379746835, | |
| "grad_norm": 0.3043939154370372, | |
| "learning_rate": 6.99270564069757e-06, | |
| "loss": 0.1171, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 0.30697893601448345, | |
| "learning_rate": 6.949642934304375e-06, | |
| "loss": 0.12, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 3.7635443037974685, | |
| "grad_norm": 0.3031130723450904, | |
| "learning_rate": 6.906685333692871e-06, | |
| "loss": 0.1208, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 3.7675949367088606, | |
| "grad_norm": 0.3128005849690602, | |
| "learning_rate": 6.86383318483769e-06, | |
| "loss": 0.1197, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 3.771645569620253, | |
| "grad_norm": 0.3134868887895072, | |
| "learning_rate": 6.821086832864139e-06, | |
| "loss": 0.1275, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.7756962025316456, | |
| "grad_norm": 0.3105137902007916, | |
| "learning_rate": 6.77844662204546e-06, | |
| "loss": 0.1303, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 3.779746835443038, | |
| "grad_norm": 0.3059834008139773, | |
| "learning_rate": 6.7359128958000455e-06, | |
| "loss": 0.1188, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 3.7837974683544306, | |
| "grad_norm": 0.3015210121494231, | |
| "learning_rate": 6.693485996688695e-06, | |
| "loss": 0.1308, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 3.7878481012658227, | |
| "grad_norm": 0.3121491506725443, | |
| "learning_rate": 6.651166266411801e-06, | |
| "loss": 0.1282, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 3.7918987341772152, | |
| "grad_norm": 0.3146620203358685, | |
| "learning_rate": 6.6089540458066725e-06, | |
| "loss": 0.1158, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.7959493670886078, | |
| "grad_norm": 0.3167300802923521, | |
| "learning_rate": 6.566849674844711e-06, | |
| "loss": 0.1228, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "grad_norm": 0.30983575865389484, | |
| "learning_rate": 6.524853492628747e-06, | |
| "loss": 0.1063, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 3.8040506329113923, | |
| "grad_norm": 0.2941874213111484, | |
| "learning_rate": 6.4829658373902536e-06, | |
| "loss": 0.1505, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 3.808101265822785, | |
| "grad_norm": 0.29326357360589256, | |
| "learning_rate": 6.441187046486648e-06, | |
| "loss": 0.1184, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 3.8121518987341774, | |
| "grad_norm": 0.3237179084506855, | |
| "learning_rate": 6.399517456398567e-06, | |
| "loss": 0.1473, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.81620253164557, | |
| "grad_norm": 0.3100241762914471, | |
| "learning_rate": 6.357957402727164e-06, | |
| "loss": 0.1357, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 3.820253164556962, | |
| "grad_norm": 0.3167130702062046, | |
| "learning_rate": 6.316507220191395e-06, | |
| "loss": 0.133, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 3.8243037974683545, | |
| "grad_norm": 0.31345394650729574, | |
| "learning_rate": 6.275167242625331e-06, | |
| "loss": 0.1266, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 3.828354430379747, | |
| "grad_norm": 0.2949278523492214, | |
| "learning_rate": 6.233937802975471e-06, | |
| "loss": 0.1174, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 3.832405063291139, | |
| "grad_norm": 0.3088184323451111, | |
| "learning_rate": 6.192819233298046e-06, | |
| "loss": 0.1148, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.8364556962025316, | |
| "grad_norm": 0.3055765803054699, | |
| "learning_rate": 6.151811864756383e-06, | |
| "loss": 0.1224, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 3.840506329113924, | |
| "grad_norm": 0.30986950047819306, | |
| "learning_rate": 6.1109160276181655e-06, | |
| "loss": 0.1287, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 3.8445569620253166, | |
| "grad_norm": 0.29472049055545824, | |
| "learning_rate": 6.070132051252868e-06, | |
| "loss": 0.1144, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 3.848607594936709, | |
| "grad_norm": 0.2897431316754821, | |
| "learning_rate": 6.0294602641290034e-06, | |
| "loss": 0.1128, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 3.852658227848101, | |
| "grad_norm": 0.3021263782452781, | |
| "learning_rate": 5.988900993811575e-06, | |
| "loss": 0.1166, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.8567088607594937, | |
| "grad_norm": 0.3062785465611099, | |
| "learning_rate": 5.948454566959363e-06, | |
| "loss": 0.1216, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.30945790751255287, | |
| "learning_rate": 5.908121309322328e-06, | |
| "loss": 0.1282, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 3.8648101265822783, | |
| "grad_norm": 0.30189073560442575, | |
| "learning_rate": 5.867901545738976e-06, | |
| "loss": 0.1335, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 3.868860759493671, | |
| "grad_norm": 0.303984341650507, | |
| "learning_rate": 5.827795600133774e-06, | |
| "loss": 0.1306, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 3.8729113924050633, | |
| "grad_norm": 0.3006607398409471, | |
| "learning_rate": 5.787803795514466e-06, | |
| "loss": 0.122, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.876962025316456, | |
| "grad_norm": 0.28869907431373293, | |
| "learning_rate": 5.747926453969576e-06, | |
| "loss": 0.1057, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 3.8810126582278484, | |
| "grad_norm": 0.3097442282882525, | |
| "learning_rate": 5.708163896665708e-06, | |
| "loss": 0.1323, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 3.8850632911392404, | |
| "grad_norm": 0.30902240401706826, | |
| "learning_rate": 5.668516443845047e-06, | |
| "loss": 0.1217, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 3.889113924050633, | |
| "grad_norm": 0.29782246453057515, | |
| "learning_rate": 5.6289844148227225e-06, | |
| "loss": 0.1103, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 3.8931645569620255, | |
| "grad_norm": 0.29160932538059076, | |
| "learning_rate": 5.5895681279842615e-06, | |
| "loss": 0.1369, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.8972151898734175, | |
| "grad_norm": 0.29928729152040806, | |
| "learning_rate": 5.550267900783019e-06, | |
| "loss": 0.1327, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 3.90126582278481, | |
| "grad_norm": 0.3005406182540866, | |
| "learning_rate": 5.511084049737623e-06, | |
| "loss": 0.1255, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 3.9053164556962026, | |
| "grad_norm": 0.3014823390578764, | |
| "learning_rate": 5.4720168904294215e-06, | |
| "loss": 0.1276, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 3.909367088607595, | |
| "grad_norm": 0.2999799474443782, | |
| "learning_rate": 5.433066737499948e-06, | |
| "loss": 0.1123, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 3.9134177215189876, | |
| "grad_norm": 0.29449142311964654, | |
| "learning_rate": 5.394233904648376e-06, | |
| "loss": 0.1197, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.9174683544303797, | |
| "grad_norm": 0.30043707602447134, | |
| "learning_rate": 5.355518704628997e-06, | |
| "loss": 0.1275, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 3.921518987341772, | |
| "grad_norm": 0.3075800349066706, | |
| "learning_rate": 5.316921449248731e-06, | |
| "loss": 0.1387, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 3.9255696202531647, | |
| "grad_norm": 0.28763153954937704, | |
| "learning_rate": 5.278442449364538e-06, | |
| "loss": 0.1254, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 3.9296202531645568, | |
| "grad_norm": 0.29433773135456964, | |
| "learning_rate": 5.240082014881016e-06, | |
| "loss": 0.1316, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 3.9336708860759493, | |
| "grad_norm": 0.2996460475577223, | |
| "learning_rate": 5.201840454747822e-06, | |
| "loss": 0.1328, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.937721518987342, | |
| "grad_norm": 0.28968220881862894, | |
| "learning_rate": 5.163718076957223e-06, | |
| "loss": 0.1116, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 3.9417721518987343, | |
| "grad_norm": 0.28131026439160156, | |
| "learning_rate": 5.125715188541609e-06, | |
| "loss": 0.1358, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 3.945822784810127, | |
| "grad_norm": 0.2939472018520688, | |
| "learning_rate": 5.087832095571021e-06, | |
| "loss": 0.133, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 3.949873417721519, | |
| "grad_norm": 0.29518589676933493, | |
| "learning_rate": 5.0500691031506766e-06, | |
| "loss": 0.1267, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.9539240506329114, | |
| "grad_norm": 0.304022263145898, | |
| "learning_rate": 5.01242651541854e-06, | |
| "loss": 0.1065, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.957974683544304, | |
| "grad_norm": 0.29183811608897, | |
| "learning_rate": 4.974904635542815e-06, | |
| "loss": 0.1095, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.3041308295985149, | |
| "learning_rate": 4.937503765719582e-06, | |
| "loss": 0.1274, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 3.9660759493670885, | |
| "grad_norm": 0.30207586646381407, | |
| "learning_rate": 4.900224207170299e-06, | |
| "loss": 0.1191, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.970126582278481, | |
| "grad_norm": 0.31307936478502135, | |
| "learning_rate": 4.8630662601394065e-06, | |
| "loss": 0.1348, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 3.9741772151898735, | |
| "grad_norm": 0.3057253153428158, | |
| "learning_rate": 4.8260302238918995e-06, | |
| "loss": 0.1404, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.978227848101266, | |
| "grad_norm": 0.29708398241273376, | |
| "learning_rate": 4.789116396710924e-06, | |
| "loss": 0.1134, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 3.982278481012658, | |
| "grad_norm": 0.29536660835191775, | |
| "learning_rate": 4.752325075895368e-06, | |
| "loss": 0.1416, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.9863291139240506, | |
| "grad_norm": 0.2928870664914971, | |
| "learning_rate": 4.715656557757473e-06, | |
| "loss": 0.1308, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 3.990379746835443, | |
| "grad_norm": 0.3088950446242622, | |
| "learning_rate": 4.679111137620442e-06, | |
| "loss": 0.1255, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.9944303797468352, | |
| "grad_norm": 0.2791259439336513, | |
| "learning_rate": 4.6426891098160585e-06, | |
| "loss": 0.1234, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.9984810126582278, | |
| "grad_norm": 0.292288693560467, | |
| "learning_rate": 4.6063907676823474e-06, | |
| "loss": 0.1384, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 4.004050632911392, | |
| "grad_norm": 0.34180571206580673, | |
| "learning_rate": 4.570216403561141e-06, | |
| "loss": 0.105, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 4.008101265822785, | |
| "grad_norm": 0.31410495249481873, | |
| "learning_rate": 4.534166308795815e-06, | |
| "loss": 0.0863, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 4.012151898734177, | |
| "grad_norm": 0.2654279546002604, | |
| "learning_rate": 4.498240773728859e-06, | |
| "loss": 0.0952, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 4.01620253164557, | |
| "grad_norm": 0.2228228194600407, | |
| "learning_rate": 4.462440087699609e-06, | |
| "loss": 0.0769, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.020253164556962, | |
| "grad_norm": 0.2308525651111481, | |
| "learning_rate": 4.426764539041861e-06, | |
| "loss": 0.0777, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 4.024303797468354, | |
| "grad_norm": 0.24611213826656947, | |
| "learning_rate": 4.391214415081582e-06, | |
| "loss": 0.0832, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 4.028354430379747, | |
| "grad_norm": 0.2997272742793593, | |
| "learning_rate": 4.355790002134579e-06, | |
| "loss": 0.0756, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 4.032405063291139, | |
| "grad_norm": 0.34574412521009307, | |
| "learning_rate": 4.320491585504207e-06, | |
| "loss": 0.0844, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 4.036455696202531, | |
| "grad_norm": 0.31209378309671904, | |
| "learning_rate": 4.2853194494790615e-06, | |
| "loss": 0.0732, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 4.040506329113924, | |
| "grad_norm": 0.3024486459365821, | |
| "learning_rate": 4.250273877330691e-06, | |
| "loss": 0.09, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 4.044556962025316, | |
| "grad_norm": 0.2664140255616994, | |
| "learning_rate": 4.215355151311313e-06, | |
| "loss": 0.0706, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 4.048607594936709, | |
| "grad_norm": 0.24643027714233431, | |
| "learning_rate": 4.180563552651542e-06, | |
| "loss": 0.0928, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 4.052658227848101, | |
| "grad_norm": 0.24561422891345355, | |
| "learning_rate": 4.145899361558147e-06, | |
| "loss": 0.0691, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 4.056708860759493, | |
| "grad_norm": 0.25429318845712195, | |
| "learning_rate": 4.111362857211738e-06, | |
| "loss": 0.0858, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.060759493670886, | |
| "grad_norm": 0.26807250214471184, | |
| "learning_rate": 4.076954317764592e-06, | |
| "loss": 0.0731, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 4.0648101265822785, | |
| "grad_norm": 0.2528222291343804, | |
| "learning_rate": 4.042674020338335e-06, | |
| "loss": 0.0984, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 4.0688607594936705, | |
| "grad_norm": 0.2612749967549681, | |
| "learning_rate": 4.0085222410217835e-06, | |
| "loss": 0.0902, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 4.0729113924050635, | |
| "grad_norm": 0.2547479800339324, | |
| "learning_rate": 3.974499254868674e-06, | |
| "loss": 0.0785, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 4.076962025316456, | |
| "grad_norm": 0.2382237961643944, | |
| "learning_rate": 3.940605335895451e-06, | |
| "loss": 0.0889, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 4.0810126582278485, | |
| "grad_norm": 0.24585033815451582, | |
| "learning_rate": 3.90684075707908e-06, | |
| "loss": 0.0836, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 4.085063291139241, | |
| "grad_norm": 0.2466004536696518, | |
| "learning_rate": 3.8732057903548505e-06, | |
| "loss": 0.0804, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 4.089113924050633, | |
| "grad_norm": 0.2536491525206007, | |
| "learning_rate": 3.8397007066141375e-06, | |
| "loss": 0.0878, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 4.093164556962026, | |
| "grad_norm": 0.25287777957136287, | |
| "learning_rate": 3.806325775702304e-06, | |
| "loss": 0.0807, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 4.097215189873418, | |
| "grad_norm": 0.24085733185936958, | |
| "learning_rate": 3.773081266416434e-06, | |
| "loss": 0.0768, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.24605898835776938, | |
| "learning_rate": 3.739967446503245e-06, | |
| "loss": 0.0794, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 4.105316455696203, | |
| "grad_norm": 0.26299947376676797, | |
| "learning_rate": 3.706984582656894e-06, | |
| "loss": 0.0842, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 4.109367088607595, | |
| "grad_norm": 0.22999814650707148, | |
| "learning_rate": 3.6741329405168237e-06, | |
| "loss": 0.0843, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 4.113417721518988, | |
| "grad_norm": 0.23765220317488436, | |
| "learning_rate": 3.641412784665648e-06, | |
| "loss": 0.0784, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 4.11746835443038, | |
| "grad_norm": 0.22498214218610316, | |
| "learning_rate": 3.608824378627005e-06, | |
| "loss": 0.0925, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 4.121518987341772, | |
| "grad_norm": 0.2324719740186393, | |
| "learning_rate": 3.5763679848634337e-06, | |
| "loss": 0.0906, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 4.125569620253165, | |
| "grad_norm": 0.23408207007018292, | |
| "learning_rate": 3.544043864774269e-06, | |
| "loss": 0.0875, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 4.129620253164557, | |
| "grad_norm": 0.2416716612386286, | |
| "learning_rate": 3.5118522786935282e-06, | |
| "loss": 0.075, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 4.133670886075949, | |
| "grad_norm": 0.2569300119035158, | |
| "learning_rate": 3.479793485887819e-06, | |
| "loss": 0.0723, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 4.137721518987342, | |
| "grad_norm": 0.2485155371639945, | |
| "learning_rate": 3.4478677445542653e-06, | |
| "loss": 0.0768, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.141772151898734, | |
| "grad_norm": 0.2395969568815722, | |
| "learning_rate": 3.4160753118183767e-06, | |
| "loss": 0.0864, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 4.145822784810127, | |
| "grad_norm": 0.23123800172991366, | |
| "learning_rate": 3.3844164437320527e-06, | |
| "loss": 0.0864, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 4.149873417721519, | |
| "grad_norm": 0.23478239318669067, | |
| "learning_rate": 3.3528913952714558e-06, | |
| "loss": 0.0672, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 4.153924050632911, | |
| "grad_norm": 0.24826104736355342, | |
| "learning_rate": 3.321500420335e-06, | |
| "loss": 0.078, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 4.157974683544304, | |
| "grad_norm": 0.2456518558983129, | |
| "learning_rate": 3.290243771741275e-06, | |
| "loss": 0.0961, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.162025316455696, | |
| "grad_norm": 0.2497671810972341, | |
| "learning_rate": 3.2591217012270325e-06, | |
| "loss": 0.0885, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 4.166075949367088, | |
| "grad_norm": 0.2361117932790613, | |
| "learning_rate": 3.228134459445149e-06, | |
| "loss": 0.1043, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 4.170126582278481, | |
| "grad_norm": 0.247914952953322, | |
| "learning_rate": 3.1972822959626205e-06, | |
| "loss": 0.0704, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 4.174177215189873, | |
| "grad_norm": 0.252377531418107, | |
| "learning_rate": 3.166565459258513e-06, | |
| "loss": 0.0889, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 4.178227848101266, | |
| "grad_norm": 0.23933567427911429, | |
| "learning_rate": 3.1359841967220193e-06, | |
| "loss": 0.0756, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.182278481012658, | |
| "grad_norm": 0.25077469319557244, | |
| "learning_rate": 3.105538754650419e-06, | |
| "loss": 0.086, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 4.18632911392405, | |
| "grad_norm": 0.24137623385720489, | |
| "learning_rate": 3.07522937824712e-06, | |
| "loss": 0.0808, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 4.190379746835443, | |
| "grad_norm": 0.2365710875992418, | |
| "learning_rate": 3.0450563116196697e-06, | |
| "loss": 0.0655, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 4.194430379746835, | |
| "grad_norm": 0.23223373824651078, | |
| "learning_rate": 3.0150197977778008e-06, | |
| "loss": 0.073, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 4.1984810126582275, | |
| "grad_norm": 0.22825452527347284, | |
| "learning_rate": 2.985120078631465e-06, | |
| "loss": 0.0703, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.2470151157969707, | |
| "learning_rate": 2.9553573949888893e-06, | |
| "loss": 0.077, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 4.2065822784810125, | |
| "grad_norm": 0.23770287224358044, | |
| "learning_rate": 2.9257319865546384e-06, | |
| "loss": 0.0736, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 4.2106329113924055, | |
| "grad_norm": 0.2213693278455982, | |
| "learning_rate": 2.896244091927678e-06, | |
| "loss": 0.0771, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 4.2146835443037975, | |
| "grad_norm": 0.2480849257472046, | |
| "learning_rate": 2.8668939485994584e-06, | |
| "loss": 0.0961, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 4.21873417721519, | |
| "grad_norm": 0.2382952328661818, | |
| "learning_rate": 2.837681792951994e-06, | |
| "loss": 0.0849, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.222784810126583, | |
| "grad_norm": 0.24660553186936118, | |
| "learning_rate": 2.808607860255981e-06, | |
| "loss": 0.0885, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 4.226835443037975, | |
| "grad_norm": 0.22556699421498907, | |
| "learning_rate": 2.7796723846688634e-06, | |
| "loss": 0.0758, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 4.230886075949367, | |
| "grad_norm": 0.23512437999226576, | |
| "learning_rate": 2.7508755992329937e-06, | |
| "loss": 0.0947, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 4.23493670886076, | |
| "grad_norm": 0.24052264718709396, | |
| "learning_rate": 2.722217735873718e-06, | |
| "loss": 0.0859, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 4.238987341772152, | |
| "grad_norm": 0.22716309246626507, | |
| "learning_rate": 2.6936990253975315e-06, | |
| "loss": 0.0829, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 4.243037974683545, | |
| "grad_norm": 0.25452675528395324, | |
| "learning_rate": 2.665319697490205e-06, | |
| "loss": 0.0769, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 4.247088607594937, | |
| "grad_norm": 0.23256697946462962, | |
| "learning_rate": 2.637079980714945e-06, | |
| "loss": 0.0944, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 4.251139240506329, | |
| "grad_norm": 0.2300764100146653, | |
| "learning_rate": 2.6089801025105453e-06, | |
| "loss": 0.0986, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 4.255189873417722, | |
| "grad_norm": 0.23587912296041694, | |
| "learning_rate": 2.581020289189571e-06, | |
| "loss": 0.0847, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 4.259240506329114, | |
| "grad_norm": 0.2429668741193643, | |
| "learning_rate": 2.553200765936501e-06, | |
| "loss": 0.0695, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.263291139240506, | |
| "grad_norm": 0.250808083809732, | |
| "learning_rate": 2.525521756805962e-06, | |
| "loss": 0.0811, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 4.267341772151899, | |
| "grad_norm": 0.23335164280297868, | |
| "learning_rate": 2.497983484720885e-06, | |
| "loss": 0.0857, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 4.271392405063291, | |
| "grad_norm": 0.22832419302263285, | |
| "learning_rate": 2.470586171470728e-06, | |
| "loss": 0.0741, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 4.275443037974684, | |
| "grad_norm": 0.23729738081089838, | |
| "learning_rate": 2.4433300377096836e-06, | |
| "loss": 0.068, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 4.279493670886076, | |
| "grad_norm": 0.2421328691019601, | |
| "learning_rate": 2.4162153029549073e-06, | |
| "loss": 0.0741, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 4.283544303797468, | |
| "grad_norm": 0.23532999173860664, | |
| "learning_rate": 2.3892421855847458e-06, | |
| "loss": 0.085, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 4.287594936708861, | |
| "grad_norm": 0.23143812575627878, | |
| "learning_rate": 2.362410902836978e-06, | |
| "loss": 0.0798, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 4.291645569620253, | |
| "grad_norm": 0.22880664477009666, | |
| "learning_rate": 2.3357216708070653e-06, | |
| "loss": 0.0815, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 4.295696202531645, | |
| "grad_norm": 0.24416536533340885, | |
| "learning_rate": 2.309174704446411e-06, | |
| "loss": 0.0704, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 4.299746835443038, | |
| "grad_norm": 0.23864082416403726, | |
| "learning_rate": 2.2827702175606437e-06, | |
| "loss": 0.0748, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.24248130912774987, | |
| "learning_rate": 2.256508422807855e-06, | |
| "loss": 0.0972, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 4.307848101265823, | |
| "grad_norm": 0.2410860075224067, | |
| "learning_rate": 2.230389531696946e-06, | |
| "loss": 0.0807, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 4.311898734177215, | |
| "grad_norm": 0.23920792174142702, | |
| "learning_rate": 2.204413754585857e-06, | |
| "loss": 0.1092, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 4.315949367088607, | |
| "grad_norm": 0.22040344607432644, | |
| "learning_rate": 2.1785813006799406e-06, | |
| "loss": 0.0676, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "grad_norm": 0.2367872370912161, | |
| "learning_rate": 2.1528923780302224e-06, | |
| "loss": 0.0817, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 4.324050632911392, | |
| "grad_norm": 0.24644615519184934, | |
| "learning_rate": 2.127347193531757e-06, | |
| "loss": 0.0739, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 4.328101265822784, | |
| "grad_norm": 0.2315161806191221, | |
| "learning_rate": 2.101945952921942e-06, | |
| "loss": 0.0755, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 4.332151898734177, | |
| "grad_norm": 0.23838185545562585, | |
| "learning_rate": 2.0766888607788906e-06, | |
| "loss": 0.0796, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 4.3362025316455695, | |
| "grad_norm": 0.23222822345423624, | |
| "learning_rate": 2.0515761205197337e-06, | |
| "loss": 0.0831, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 4.340253164556962, | |
| "grad_norm": 0.25229118593461325, | |
| "learning_rate": 2.0266079343990453e-06, | |
| "loss": 0.0812, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.3443037974683545, | |
| "grad_norm": 0.25129835443569865, | |
| "learning_rate": 2.0017845035071494e-06, | |
| "loss": 0.0807, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 4.348354430379747, | |
| "grad_norm": 0.22782185103891361, | |
| "learning_rate": 1.9771060277685537e-06, | |
| "loss": 0.094, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 4.3524050632911395, | |
| "grad_norm": 0.2509293675802797, | |
| "learning_rate": 1.95257270594031e-06, | |
| "loss": 0.094, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 4.356455696202532, | |
| "grad_norm": 0.24860728709178492, | |
| "learning_rate": 1.9281847356104188e-06, | |
| "loss": 0.0656, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 4.360506329113924, | |
| "grad_norm": 0.23876993333685054, | |
| "learning_rate": 1.9039423131962365e-06, | |
| "loss": 0.0807, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.364556962025317, | |
| "grad_norm": 0.24024166601449953, | |
| "learning_rate": 1.8798456339429027e-06, | |
| "loss": 0.0738, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 4.368607594936709, | |
| "grad_norm": 0.23142216771263383, | |
| "learning_rate": 1.8558948919217612e-06, | |
| "loss": 0.0728, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 4.372658227848102, | |
| "grad_norm": 0.22933491662564082, | |
| "learning_rate": 1.8320902800287954e-06, | |
| "loss": 0.0821, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 4.376708860759494, | |
| "grad_norm": 0.2338120362844697, | |
| "learning_rate": 1.8084319899830726e-06, | |
| "loss": 0.0873, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 4.380759493670886, | |
| "grad_norm": 0.2286707240384324, | |
| "learning_rate": 1.7849202123252097e-06, | |
| "loss": 0.086, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.384810126582279, | |
| "grad_norm": 0.24348039765299545, | |
| "learning_rate": 1.7615551364158401e-06, | |
| "loss": 0.0732, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 4.388860759493671, | |
| "grad_norm": 0.24119149456752978, | |
| "learning_rate": 1.738336950434061e-06, | |
| "loss": 0.0751, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 4.392911392405063, | |
| "grad_norm": 0.24220096779620673, | |
| "learning_rate": 1.715265841375957e-06, | |
| "loss": 0.0853, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 4.396962025316456, | |
| "grad_norm": 0.22558048400478425, | |
| "learning_rate": 1.6923419950530684e-06, | |
| "loss": 0.0816, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 4.401012658227848, | |
| "grad_norm": 0.24028166555615774, | |
| "learning_rate": 1.6695655960909008e-06, | |
| "loss": 0.0728, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.2249133402804275, | |
| "learning_rate": 1.646936827927441e-06, | |
| "loss": 0.0801, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 4.409113924050633, | |
| "grad_norm": 0.22959583507283082, | |
| "learning_rate": 1.6244558728116766e-06, | |
| "loss": 0.0845, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 4.413164556962025, | |
| "grad_norm": 0.22227960757536686, | |
| "learning_rate": 1.6021229118021265e-06, | |
| "loss": 0.0736, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 4.417215189873418, | |
| "grad_norm": 0.23235064914700748, | |
| "learning_rate": 1.5799381247653967e-06, | |
| "loss": 0.0808, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 4.42126582278481, | |
| "grad_norm": 0.23037313528324965, | |
| "learning_rate": 1.5579016903747013e-06, | |
| "loss": 0.0831, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.425316455696202, | |
| "grad_norm": 0.24890128795170835, | |
| "learning_rate": 1.5360137861084656e-06, | |
| "loss": 0.0728, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 4.429367088607595, | |
| "grad_norm": 0.23682487155736542, | |
| "learning_rate": 1.5142745882488475e-06, | |
| "loss": 0.0817, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 4.433417721518987, | |
| "grad_norm": 0.23396304105416, | |
| "learning_rate": 1.4926842718803691e-06, | |
| "loss": 0.0837, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 4.43746835443038, | |
| "grad_norm": 0.23488525616123407, | |
| "learning_rate": 1.4712430108884657e-06, | |
| "loss": 0.0692, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 4.441518987341772, | |
| "grad_norm": 0.24379344554698001, | |
| "learning_rate": 1.4499509779581078e-06, | |
| "loss": 0.0799, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.445569620253164, | |
| "grad_norm": 0.22234741353186283, | |
| "learning_rate": 1.4288083445723988e-06, | |
| "loss": 0.073, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 4.449620253164557, | |
| "grad_norm": 0.24244563625097137, | |
| "learning_rate": 1.4078152810112045e-06, | |
| "loss": 0.0912, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 4.453670886075949, | |
| "grad_norm": 0.228940227641687, | |
| "learning_rate": 1.3869719563497697e-06, | |
| "loss": 0.0721, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 4.457721518987341, | |
| "grad_norm": 0.23023213307784898, | |
| "learning_rate": 1.3662785384573663e-06, | |
| "loss": 0.0844, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 4.461772151898734, | |
| "grad_norm": 0.23661611449752168, | |
| "learning_rate": 1.3457351939959383e-06, | |
| "loss": 0.096, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.465822784810126, | |
| "grad_norm": 0.2397230570156505, | |
| "learning_rate": 1.3253420884187551e-06, | |
| "loss": 0.0988, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 4.469873417721519, | |
| "grad_norm": 0.23097416155358336, | |
| "learning_rate": 1.3050993859690953e-06, | |
| "loss": 0.0863, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 4.473924050632911, | |
| "grad_norm": 0.23778793854195163, | |
| "learning_rate": 1.2850072496788869e-06, | |
| "loss": 0.0773, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 4.4779746835443035, | |
| "grad_norm": 0.2500513755821008, | |
| "learning_rate": 1.2650658413674434e-06, | |
| "loss": 0.0816, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 4.4820253164556965, | |
| "grad_norm": 0.2433499222748311, | |
| "learning_rate": 1.2452753216401226e-06, | |
| "loss": 0.0963, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.4860759493670885, | |
| "grad_norm": 0.22474113673709606, | |
| "learning_rate": 1.2256358498870503e-06, | |
| "loss": 0.0895, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 4.490126582278481, | |
| "grad_norm": 0.2204658751008857, | |
| "learning_rate": 1.2061475842818337e-06, | |
| "loss": 0.0939, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 4.494177215189874, | |
| "grad_norm": 0.2332460950014052, | |
| "learning_rate": 1.1868106817802816e-06, | |
| "loss": 0.079, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 4.498227848101266, | |
| "grad_norm": 0.22980557761433243, | |
| "learning_rate": 1.1676252981191482e-06, | |
| "loss": 0.0912, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 4.502278481012659, | |
| "grad_norm": 0.23598026793990434, | |
| "learning_rate": 1.1485915878148823e-06, | |
| "loss": 0.0859, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.22924124867403953, | |
| "learning_rate": 1.1297097041623584e-06, | |
| "loss": 0.0891, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 4.510379746835443, | |
| "grad_norm": 0.22513199260829228, | |
| "learning_rate": 1.1109797992336847e-06, | |
| "loss": 0.0804, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 4.514430379746836, | |
| "grad_norm": 0.2376333143931441, | |
| "learning_rate": 1.092402023876933e-06, | |
| "loss": 0.0724, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 4.518481012658228, | |
| "grad_norm": 0.23357518013532308, | |
| "learning_rate": 1.0739765277149527e-06, | |
| "loss": 0.0815, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 4.52253164556962, | |
| "grad_norm": 0.2343993556364393, | |
| "learning_rate": 1.0557034591441596e-06, | |
| "loss": 0.0956, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.526582278481013, | |
| "grad_norm": 0.2319950654496008, | |
| "learning_rate": 1.0375829653333324e-06, | |
| "loss": 0.0666, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 4.530632911392405, | |
| "grad_norm": 0.2192282116539, | |
| "learning_rate": 1.0196151922224385e-06, | |
| "loss": 0.0908, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 4.534683544303798, | |
| "grad_norm": 0.22768145247754826, | |
| "learning_rate": 1.0018002845214526e-06, | |
| "loss": 0.083, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 4.53873417721519, | |
| "grad_norm": 0.23917439513415267, | |
| "learning_rate": 9.841383857091947e-07, | |
| "loss": 0.0856, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.542784810126582, | |
| "grad_norm": 0.23523496552302525, | |
| "learning_rate": 9.666296380321616e-07, | |
| "loss": 0.0822, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.546835443037975, | |
| "grad_norm": 0.23896028098590624, | |
| "learning_rate": 9.492741825034124e-07, | |
| "loss": 0.0962, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 4.550886075949367, | |
| "grad_norm": 0.2371001120099644, | |
| "learning_rate": 9.320721589013892e-07, | |
| "loss": 0.0901, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 4.55493670886076, | |
| "grad_norm": 0.2380629564756662, | |
| "learning_rate": 9.150237057688339e-07, | |
| "loss": 0.1042, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 4.558987341772152, | |
| "grad_norm": 0.24528025952360813, | |
| "learning_rate": 8.981289604116328e-07, | |
| "loss": 0.0938, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 4.563037974683544, | |
| "grad_norm": 0.23120095227331558, | |
| "learning_rate": 8.813880588977542e-07, | |
| "loss": 0.0816, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.567088607594937, | |
| "grad_norm": 0.24611114660544411, | |
| "learning_rate": 8.648011360561126e-07, | |
| "loss": 0.0944, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 4.571139240506329, | |
| "grad_norm": 0.23771232915483897, | |
| "learning_rate": 8.483683254755037e-07, | |
| "loss": 0.0922, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 4.575189873417721, | |
| "grad_norm": 0.23612928312383752, | |
| "learning_rate": 8.320897595035227e-07, | |
| "loss": 0.0649, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 4.579240506329114, | |
| "grad_norm": 0.24088412439702084, | |
| "learning_rate": 8.159655692455093e-07, | |
| "loss": 0.0722, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 4.583291139240506, | |
| "grad_norm": 0.22572624096835983, | |
| "learning_rate": 7.999958845634648e-07, | |
| "loss": 0.0784, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.587341772151898, | |
| "grad_norm": 0.23230645969316296, | |
| "learning_rate": 7.841808340750478e-07, | |
| "loss": 0.0773, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 4.591392405063291, | |
| "grad_norm": 0.23258604925461546, | |
| "learning_rate": 7.685205451524869e-07, | |
| "loss": 0.091, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 4.595443037974683, | |
| "grad_norm": 0.23465483685476413, | |
| "learning_rate": 7.530151439216027e-07, | |
| "loss": 0.0865, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 4.599493670886076, | |
| "grad_norm": 0.22654267120890134, | |
| "learning_rate": 7.376647552607675e-07, | |
| "loss": 0.0753, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 4.603544303797468, | |
| "grad_norm": 0.2371115149701726, | |
| "learning_rate": 7.224695027998963e-07, | |
| "loss": 0.0868, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.25373836912615644, | |
| "learning_rate": 7.07429508919466e-07, | |
| "loss": 0.0973, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 4.611645569620253, | |
| "grad_norm": 0.23340423436039498, | |
| "learning_rate": 6.925448947495206e-07, | |
| "loss": 0.0764, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 4.6156962025316455, | |
| "grad_norm": 0.23363671565326075, | |
| "learning_rate": 6.778157801686936e-07, | |
| "loss": 0.066, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 4.619746835443038, | |
| "grad_norm": 0.2335520707746117, | |
| "learning_rate": 6.632422838032515e-07, | |
| "loss": 0.0657, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 4.6237974683544305, | |
| "grad_norm": 0.23278075914048044, | |
| "learning_rate": 6.488245230261281e-07, | |
| "loss": 0.0927, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.627848101265823, | |
| "grad_norm": 0.23381634322967199, | |
| "learning_rate": 6.345626139559868e-07, | |
| "loss": 0.0768, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 4.6318987341772155, | |
| "grad_norm": 0.24050161207071213, | |
| "learning_rate": 6.204566714562866e-07, | |
| "loss": 0.0712, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 4.635949367088608, | |
| "grad_norm": 0.24416586524591632, | |
| "learning_rate": 6.06506809134344e-07, | |
| "loss": 0.0775, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "grad_norm": 0.22762317446272867, | |
| "learning_rate": 5.927131393404373e-07, | |
| "loss": 0.0881, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 4.644050632911393, | |
| "grad_norm": 0.23751719262159832, | |
| "learning_rate": 5.790757731668817e-07, | |
| "loss": 0.0865, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.648101265822785, | |
| "grad_norm": 0.23865313539313046, | |
| "learning_rate": 5.655948204471507e-07, | |
| "loss": 0.0759, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 4.652151898734177, | |
| "grad_norm": 0.2227371364937204, | |
| "learning_rate": 5.522703897549875e-07, | |
| "loss": 0.0587, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 4.65620253164557, | |
| "grad_norm": 0.2251155484760629, | |
| "learning_rate": 5.391025884035239e-07, | |
| "loss": 0.0805, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 4.660253164556962, | |
| "grad_norm": 0.23592185872305735, | |
| "learning_rate": 5.260915224444207e-07, | |
| "loss": 0.096, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 4.664303797468355, | |
| "grad_norm": 0.24746879561417315, | |
| "learning_rate": 5.132372966670129e-07, | |
| "loss": 0.083, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.668354430379747, | |
| "grad_norm": 0.2307580194559123, | |
| "learning_rate": 5.005400145974704e-07, | |
| "loss": 0.0778, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 4.672405063291139, | |
| "grad_norm": 0.22585058267981112, | |
| "learning_rate": 4.879997784979562e-07, | |
| "loss": 0.0821, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 4.676455696202532, | |
| "grad_norm": 0.2352657428604589, | |
| "learning_rate": 4.7561668936580984e-07, | |
| "loss": 0.09, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 4.680506329113924, | |
| "grad_norm": 0.238899407045373, | |
| "learning_rate": 4.6339084693272306e-07, | |
| "loss": 0.0763, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 4.684556962025317, | |
| "grad_norm": 0.23381373319035262, | |
| "learning_rate": 4.5132234966395847e-07, | |
| "loss": 0.0852, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.688607594936709, | |
| "grad_norm": 0.22800410807442115, | |
| "learning_rate": 4.3941129475752795e-07, | |
| "loss": 0.0825, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 4.692658227848101, | |
| "grad_norm": 0.23614113733411066, | |
| "learning_rate": 4.27657778143431e-07, | |
| "loss": 0.0633, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 4.696708860759494, | |
| "grad_norm": 0.22895686542076243, | |
| "learning_rate": 4.1606189448287757e-07, | |
| "loss": 0.085, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 4.700759493670886, | |
| "grad_norm": 0.23096196985537834, | |
| "learning_rate": 4.046237371675177e-07, | |
| "loss": 0.1053, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 4.704810126582278, | |
| "grad_norm": 0.2324181644776875, | |
| "learning_rate": 3.9334339831869963e-07, | |
| "loss": 0.0819, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.22992164234848772, | |
| "learning_rate": 3.8222096878671955e-07, | |
| "loss": 0.0818, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 4.712911392405063, | |
| "grad_norm": 0.23438424967138308, | |
| "learning_rate": 3.7125653815009545e-07, | |
| "loss": 0.0795, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 4.716962025316455, | |
| "grad_norm": 0.23852698163217825, | |
| "learning_rate": 3.6045019471484974e-07, | |
| "loss": 0.0787, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 4.721012658227848, | |
| "grad_norm": 0.2334371267680364, | |
| "learning_rate": 3.498020255137813e-07, | |
| "loss": 0.0701, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 4.72506329113924, | |
| "grad_norm": 0.23478632556767837, | |
| "learning_rate": 3.393121163057811e-07, | |
| "loss": 0.0949, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.729113924050633, | |
| "grad_norm": 0.2276812500151959, | |
| "learning_rate": 3.289805515751399e-07, | |
| "loss": 0.0939, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 4.733164556962025, | |
| "grad_norm": 0.22981905429401506, | |
| "learning_rate": 3.188074145308573e-07, | |
| "loss": 0.0798, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 4.737215189873417, | |
| "grad_norm": 0.21919053671642613, | |
| "learning_rate": 3.087927871059804e-07, | |
| "loss": 0.0811, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 4.74126582278481, | |
| "grad_norm": 0.24500012258217424, | |
| "learning_rate": 2.989367499569418e-07, | |
| "loss": 0.0811, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 4.745316455696202, | |
| "grad_norm": 0.23735500684996658, | |
| "learning_rate": 2.8923938246290917e-07, | |
| "loss": 0.0977, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.749367088607595, | |
| "grad_norm": 0.22249566577112406, | |
| "learning_rate": 2.7970076272514804e-07, | |
| "loss": 0.0812, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 4.7534177215189874, | |
| "grad_norm": 0.22622703071135042, | |
| "learning_rate": 2.703209675663887e-07, | |
| "loss": 0.0916, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 4.7574683544303795, | |
| "grad_norm": 0.21675431348265556, | |
| "learning_rate": 2.6110007253021374e-07, | |
| "loss": 0.0719, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 4.7615189873417725, | |
| "grad_norm": 0.2370259303790961, | |
| "learning_rate": 2.520381518804471e-07, | |
| "loss": 0.0687, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 4.7655696202531646, | |
| "grad_norm": 0.23045518865970002, | |
| "learning_rate": 2.4313527860054585e-07, | |
| "loss": 0.0786, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.769620253164557, | |
| "grad_norm": 0.2301672992008348, | |
| "learning_rate": 2.343915243930317e-07, | |
| "loss": 0.1001, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 4.77367088607595, | |
| "grad_norm": 0.21715634964377228, | |
| "learning_rate": 2.2580695967889367e-07, | |
| "loss": 0.0692, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 4.777721518987342, | |
| "grad_norm": 0.23809329554369366, | |
| "learning_rate": 2.1738165359704189e-07, | |
| "loss": 0.0961, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 4.781772151898734, | |
| "grad_norm": 0.2331406491457119, | |
| "learning_rate": 2.0911567400373257e-07, | |
| "loss": 0.071, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 4.785822784810127, | |
| "grad_norm": 0.22393420125750313, | |
| "learning_rate": 2.0100908747202607e-07, | |
| "loss": 0.0881, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.789873417721519, | |
| "grad_norm": 0.24193676695510472, | |
| "learning_rate": 1.9306195929125638e-07, | |
| "loss": 0.0822, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 4.793924050632912, | |
| "grad_norm": 0.23083208390943188, | |
| "learning_rate": 1.8527435346650247e-07, | |
| "loss": 0.0847, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 4.797974683544304, | |
| "grad_norm": 0.25034077057975807, | |
| "learning_rate": 1.7764633271807108e-07, | |
| "loss": 0.0795, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 4.802025316455696, | |
| "grad_norm": 0.23397458099401014, | |
| "learning_rate": 1.7017795848099262e-07, | |
| "loss": 0.074, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 4.806075949367089, | |
| "grad_norm": 0.23416870118789915, | |
| "learning_rate": 1.6286929090452596e-07, | |
| "loss": 0.0667, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.2364007947992263, | |
| "learning_rate": 1.557203888516745e-07, | |
| "loss": 0.0774, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 4.814177215189874, | |
| "grad_norm": 0.23805592513606416, | |
| "learning_rate": 1.487313098987131e-07, | |
| "loss": 0.0762, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 4.818227848101266, | |
| "grad_norm": 0.23651734736043784, | |
| "learning_rate": 1.4190211033472402e-07, | |
| "loss": 0.0868, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 4.822278481012658, | |
| "grad_norm": 0.23240461599825626, | |
| "learning_rate": 1.3523284516113955e-07, | |
| "loss": 0.0792, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 4.826329113924051, | |
| "grad_norm": 0.22764058198068896, | |
| "learning_rate": 1.2872356809130682e-07, | |
| "loss": 0.0874, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.830379746835443, | |
| "grad_norm": 0.22530129205544844, | |
| "learning_rate": 1.2237433155004807e-07, | |
| "loss": 0.0938, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 4.834430379746835, | |
| "grad_norm": 0.2396954376885287, | |
| "learning_rate": 1.1618518667323886e-07, | |
| "loss": 0.0711, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 4.838481012658228, | |
| "grad_norm": 0.2249970501643855, | |
| "learning_rate": 1.1015618330740385e-07, | |
| "loss": 0.098, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 4.84253164556962, | |
| "grad_norm": 0.21801385708022022, | |
| "learning_rate": 1.042873700093061e-07, | |
| "loss": 0.0739, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 4.846582278481012, | |
| "grad_norm": 0.26792320378677265, | |
| "learning_rate": 9.857879404556291e-08, | |
| "loss": 0.0893, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.850632911392405, | |
| "grad_norm": 0.22677822621856208, | |
| "learning_rate": 9.303050139225722e-08, | |
| "loss": 0.0813, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 4.854683544303797, | |
| "grad_norm": 0.22520640735270447, | |
| "learning_rate": 8.76425367345779e-08, | |
| "loss": 0.0779, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 4.85873417721519, | |
| "grad_norm": 0.2386933859007057, | |
| "learning_rate": 8.241494346644897e-08, | |
| "loss": 0.0776, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 4.862784810126582, | |
| "grad_norm": 0.23467944277845418, | |
| "learning_rate": 7.734776369019204e-08, | |
| "loss": 0.0826, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 4.866835443037974, | |
| "grad_norm": 0.23127899625544712, | |
| "learning_rate": 7.244103821617332e-08, | |
| "loss": 0.0834, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.870886075949367, | |
| "grad_norm": 0.22402359457017906, | |
| "learning_rate": 6.769480656248606e-08, | |
| "loss": 0.0777, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 4.874936708860759, | |
| "grad_norm": 0.22181508930708455, | |
| "learning_rate": 6.310910695462635e-08, | |
| "loss": 0.0873, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 4.878987341772152, | |
| "grad_norm": 0.2225332073501807, | |
| "learning_rate": 5.8683976325191185e-08, | |
| "loss": 0.0848, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 4.883037974683544, | |
| "grad_norm": 0.23795030210450988, | |
| "learning_rate": 5.4419450313571984e-08, | |
| "loss": 0.0934, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 4.8870886075949365, | |
| "grad_norm": 0.2206604692023197, | |
| "learning_rate": 5.031556326567488e-08, | |
| "loss": 0.0855, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.891139240506329, | |
| "grad_norm": 0.23170967178536436, | |
| "learning_rate": 4.637234823364312e-08, | |
| "loss": 0.0912, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 4.8951898734177215, | |
| "grad_norm": 0.23351544053976916, | |
| "learning_rate": 4.258983697558838e-08, | |
| "loss": 0.0853, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 4.899240506329114, | |
| "grad_norm": 0.23141993838882324, | |
| "learning_rate": 3.896805995533548e-08, | |
| "loss": 0.0781, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 4.9032911392405065, | |
| "grad_norm": 0.22997435732365476, | |
| "learning_rate": 3.550704634218028e-08, | |
| "loss": 0.0738, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 4.907341772151899, | |
| "grad_norm": 0.24452200071330524, | |
| "learning_rate": 3.2206824010647676e-08, | |
| "loss": 0.0677, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.22374296816678968, | |
| "learning_rate": 2.9067419540278476e-08, | |
| "loss": 0.082, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 4.915443037974684, | |
| "grad_norm": 0.23068440804447618, | |
| "learning_rate": 2.6088858215400638e-08, | |
| "loss": 0.0808, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 4.919493670886076, | |
| "grad_norm": 0.21981325276174332, | |
| "learning_rate": 2.3271164024940564e-08, | |
| "loss": 0.0675, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 4.923544303797469, | |
| "grad_norm": 0.22410515023056557, | |
| "learning_rate": 2.061435966221881e-08, | |
| "loss": 0.0861, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 4.927594936708861, | |
| "grad_norm": 0.23194172834378676, | |
| "learning_rate": 1.811846652477245e-08, | |
| "loss": 0.0855, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.931645569620253, | |
| "grad_norm": 0.2325020353048279, | |
| "learning_rate": 1.5783504714184106e-08, | |
| "loss": 0.082, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 4.935696202531646, | |
| "grad_norm": 0.22455155325479795, | |
| "learning_rate": 1.360949303591097e-08, | |
| "loss": 0.0693, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 4.939746835443038, | |
| "grad_norm": 0.22441985155077934, | |
| "learning_rate": 1.1596448999144916e-08, | |
| "loss": 0.0867, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 4.943797468354431, | |
| "grad_norm": 0.23022138894826585, | |
| "learning_rate": 9.744388816668172e-09, | |
| "loss": 0.0693, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 4.947848101265823, | |
| "grad_norm": 0.24034259686562143, | |
| "learning_rate": 8.05332740472009e-09, | |
| "loss": 0.0766, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.951898734177215, | |
| "grad_norm": 0.23651497258585444, | |
| "learning_rate": 6.523278382872811e-09, | |
| "loss": 0.0767, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 4.955949367088608, | |
| "grad_norm": 0.22445739455604466, | |
| "learning_rate": 5.15425407393133e-09, | |
| "loss": 0.0834, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "grad_norm": 0.2232800453831962, | |
| "learning_rate": 3.94626550383137e-09, | |
| "loss": 0.0772, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 4.964050632911392, | |
| "grad_norm": 0.2292776301930113, | |
| "learning_rate": 2.899322401546112e-09, | |
| "loss": 0.0751, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 4.968101265822785, | |
| "grad_norm": 0.2288881335942612, | |
| "learning_rate": 2.013433199010706e-09, | |
| "loss": 0.0739, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.972151898734177, | |
| "grad_norm": 0.24416214388734708, | |
| "learning_rate": 1.2886050310556563e-09, | |
| "loss": 0.0862, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 4.976202531645569, | |
| "grad_norm": 0.23875469624629606, | |
| "learning_rate": 7.248437353468695e-10, | |
| "loss": 0.0776, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 4.980253164556962, | |
| "grad_norm": 0.23156184114498335, | |
| "learning_rate": 3.221538523412449e-10, | |
| "loss": 0.0829, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 4.984303797468354, | |
| "grad_norm": 0.22967305399040353, | |
| "learning_rate": 8.053862524670663e-11, | |
| "loss": 0.0685, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 4.988354430379747, | |
| "grad_norm": 0.22644794001984692, | |
| "learning_rate": 0.0, | |
| "loss": 0.0944, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.988354430379747, | |
| "step": 1230, | |
| "total_flos": 1.3274507119060582e+18, | |
| "train_loss": 0.23268839856836854, | |
| "train_runtime": 53095.2409, | |
| "train_samples_per_second": 2.976, | |
| "train_steps_per_second": 0.023 | |
| } | |
| ], | |
| "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": 1.3274507119060582e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |