| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 4.988354430379747, |
| "eval_steps": 500, |
| "global_step": 1230, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.004050632911392405, |
| "grad_norm": 8.94901334306286, |
| "learning_rate": 3.2520325203252037e-07, |
| "loss": 1.3562, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.00810126582278481, |
| "grad_norm": 9.4428534977171, |
| "learning_rate": 6.504065040650407e-07, |
| "loss": 1.3799, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.012151898734177215, |
| "grad_norm": 9.150069204526321, |
| "learning_rate": 9.75609756097561e-07, |
| "loss": 1.3714, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.01620253164556962, |
| "grad_norm": 8.929228177808707, |
| "learning_rate": 1.3008130081300815e-06, |
| "loss": 1.3572, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.020253164556962026, |
| "grad_norm": 9.014780517237261, |
| "learning_rate": 1.6260162601626018e-06, |
| "loss": 1.3704, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.02430379746835443, |
| "grad_norm": 8.755847829515067, |
| "learning_rate": 1.951219512195122e-06, |
| "loss": 1.3689, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.028354430379746835, |
| "grad_norm": 6.757437877291318, |
| "learning_rate": 2.2764227642276426e-06, |
| "loss": 1.2956, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.03240506329113924, |
| "grad_norm": 6.0297919628512195, |
| "learning_rate": 2.601626016260163e-06, |
| "loss": 1.2669, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03645569620253165, |
| "grad_norm": 3.9482707467546585, |
| "learning_rate": 2.926829268292683e-06, |
| "loss": 1.2101, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.04050632911392405, |
| "grad_norm": 3.616208344560321, |
| "learning_rate": 3.2520325203252037e-06, |
| "loss": 1.2365, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.044556962025316456, |
| "grad_norm": 3.1930792935115297, |
| "learning_rate": 3.577235772357724e-06, |
| "loss": 1.1873, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.04860759493670886, |
| "grad_norm": 4.282005790659612, |
| "learning_rate": 3.902439024390244e-06, |
| "loss": 1.142, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.052658227848101265, |
| "grad_norm": 4.5567656984424865, |
| "learning_rate": 4.227642276422765e-06, |
| "loss": 1.1456, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05670886075949367, |
| "grad_norm": 4.485575458979285, |
| "learning_rate": 4.552845528455285e-06, |
| "loss": 1.1215, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.060759493670886074, |
| "grad_norm": 3.910636206642829, |
| "learning_rate": 4.8780487804878055e-06, |
| "loss": 1.1191, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06481012658227848, |
| "grad_norm": 2.645222454185372, |
| "learning_rate": 5.203252032520326e-06, |
| "loss": 1.0501, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06886075949367089, |
| "grad_norm": 2.7407483953695433, |
| "learning_rate": 5.528455284552846e-06, |
| "loss": 1.0388, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0729113924050633, |
| "grad_norm": 2.3830544567179914, |
| "learning_rate": 5.853658536585366e-06, |
| "loss": 1.0397, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0769620253164557, |
| "grad_norm": 2.003143057328858, |
| "learning_rate": 6.178861788617887e-06, |
| "loss": 1.0101, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.0810126582278481, |
| "grad_norm": 1.6143678573584481, |
| "learning_rate": 6.504065040650407e-06, |
| "loss": 0.9892, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08506329113924051, |
| "grad_norm": 1.6945854072486932, |
| "learning_rate": 6.829268292682928e-06, |
| "loss": 0.9686, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08911392405063291, |
| "grad_norm": 1.6484519990287467, |
| "learning_rate": 7.154471544715448e-06, |
| "loss": 0.953, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.09316455696202532, |
| "grad_norm": 1.395169376265725, |
| "learning_rate": 7.4796747967479676e-06, |
| "loss": 0.9662, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09721518987341772, |
| "grad_norm": 1.3744977959529476, |
| "learning_rate": 7.804878048780489e-06, |
| "loss": 0.956, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.10126582278481013, |
| "grad_norm": 1.2241285451362733, |
| "learning_rate": 8.130081300813009e-06, |
| "loss": 0.951, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10531645569620253, |
| "grad_norm": 1.0345038195772085, |
| "learning_rate": 8.45528455284553e-06, |
| "loss": 0.9637, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10936708860759493, |
| "grad_norm": 0.9707412440123685, |
| "learning_rate": 8.78048780487805e-06, |
| "loss": 0.9548, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.11341772151898734, |
| "grad_norm": 0.9723255565787267, |
| "learning_rate": 9.10569105691057e-06, |
| "loss": 0.9256, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11746835443037974, |
| "grad_norm": 1.0539837937503616, |
| "learning_rate": 9.43089430894309e-06, |
| "loss": 0.9484, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.12151898734177215, |
| "grad_norm": 0.9906216841585511, |
| "learning_rate": 9.756097560975611e-06, |
| "loss": 0.9236, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12556962025316457, |
| "grad_norm": 1.0760273489791128, |
| "learning_rate": 1.008130081300813e-05, |
| "loss": 0.9342, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12962025316455697, |
| "grad_norm": 0.9240925339427987, |
| "learning_rate": 1.0406504065040652e-05, |
| "loss": 0.9134, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.13367088607594937, |
| "grad_norm": 1.0750120804944858, |
| "learning_rate": 1.0731707317073172e-05, |
| "loss": 0.9028, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13772151898734178, |
| "grad_norm": 1.1592919778849058, |
| "learning_rate": 1.1056910569105692e-05, |
| "loss": 0.9291, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.14177215189873418, |
| "grad_norm": 0.9176398278576404, |
| "learning_rate": 1.1382113821138213e-05, |
| "loss": 0.9211, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.1458227848101266, |
| "grad_norm": 1.049596997471436, |
| "learning_rate": 1.1707317073170731e-05, |
| "loss": 0.8833, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.149873417721519, |
| "grad_norm": 1.133848640868933, |
| "learning_rate": 1.2032520325203254e-05, |
| "loss": 0.9084, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.1539240506329114, |
| "grad_norm": 0.8892996920014129, |
| "learning_rate": 1.2357723577235774e-05, |
| "loss": 0.8949, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.1579746835443038, |
| "grad_norm": 1.0202331340342075, |
| "learning_rate": 1.2682926829268294e-05, |
| "loss": 0.9056, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.1620253164556962, |
| "grad_norm": 1.0387532692997856, |
| "learning_rate": 1.3008130081300815e-05, |
| "loss": 0.9087, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1660759493670886, |
| "grad_norm": 0.9993856054706175, |
| "learning_rate": 1.3333333333333333e-05, |
| "loss": 0.8965, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.17012658227848101, |
| "grad_norm": 0.8644144120463609, |
| "learning_rate": 1.3658536585365855e-05, |
| "loss": 0.8942, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.17417721518987342, |
| "grad_norm": 0.9251375762274978, |
| "learning_rate": 1.3983739837398376e-05, |
| "loss": 0.9048, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17822784810126582, |
| "grad_norm": 0.9096530503536108, |
| "learning_rate": 1.4308943089430896e-05, |
| "loss": 0.8824, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.18227848101265823, |
| "grad_norm": 0.9112993418733991, |
| "learning_rate": 1.4634146341463415e-05, |
| "loss": 0.8929, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.18632911392405063, |
| "grad_norm": 0.9339058293758893, |
| "learning_rate": 1.4959349593495935e-05, |
| "loss": 0.8786, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.19037974683544304, |
| "grad_norm": 0.9054634028347933, |
| "learning_rate": 1.528455284552846e-05, |
| "loss": 0.8895, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.19443037974683544, |
| "grad_norm": 0.9074799189441457, |
| "learning_rate": 1.5609756097560978e-05, |
| "loss": 0.8823, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19848101265822785, |
| "grad_norm": 1.024811365521385, |
| "learning_rate": 1.5934959349593496e-05, |
| "loss": 0.8747, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.20253164556962025, |
| "grad_norm": 0.8320167861825961, |
| "learning_rate": 1.6260162601626018e-05, |
| "loss": 0.8567, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.20658227848101265, |
| "grad_norm": 0.9850044474676902, |
| "learning_rate": 1.6585365853658537e-05, |
| "loss": 0.8917, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.21063291139240506, |
| "grad_norm": 0.8857911162657562, |
| "learning_rate": 1.691056910569106e-05, |
| "loss": 0.8857, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.21468354430379746, |
| "grad_norm": 1.0957391721326917, |
| "learning_rate": 1.7235772357723578e-05, |
| "loss": 0.8752, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21873417721518987, |
| "grad_norm": 0.9840985191320425, |
| "learning_rate": 1.75609756097561e-05, |
| "loss": 0.8616, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.22278481012658227, |
| "grad_norm": 1.0618408450370336, |
| "learning_rate": 1.788617886178862e-05, |
| "loss": 0.8677, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.22683544303797468, |
| "grad_norm": 1.0250318444200839, |
| "learning_rate": 1.821138211382114e-05, |
| "loss": 0.8781, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.23088607594936708, |
| "grad_norm": 1.0187455582514224, |
| "learning_rate": 1.8536585365853663e-05, |
| "loss": 0.8756, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.23493670886075949, |
| "grad_norm": 1.027569790863175, |
| "learning_rate": 1.886178861788618e-05, |
| "loss": 0.8659, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.2389873417721519, |
| "grad_norm": 0.909840786525502, |
| "learning_rate": 1.91869918699187e-05, |
| "loss": 0.8731, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.2430379746835443, |
| "grad_norm": 1.0090311970696144, |
| "learning_rate": 1.9512195121951222e-05, |
| "loss": 0.8925, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2470886075949367, |
| "grad_norm": 0.9498024821839715, |
| "learning_rate": 1.983739837398374e-05, |
| "loss": 0.8923, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.25113924050632913, |
| "grad_norm": 1.0476567548911917, |
| "learning_rate": 2.016260162601626e-05, |
| "loss": 0.8763, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.25518987341772154, |
| "grad_norm": 1.0431326092854107, |
| "learning_rate": 2.048780487804878e-05, |
| "loss": 0.8791, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.25924050632911394, |
| "grad_norm": 1.0538139132719164, |
| "learning_rate": 2.0813008130081303e-05, |
| "loss": 0.8548, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.26329113924050634, |
| "grad_norm": 0.9046398759481524, |
| "learning_rate": 2.1138211382113822e-05, |
| "loss": 0.8635, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.26734177215189875, |
| "grad_norm": 0.9204309820110606, |
| "learning_rate": 2.1463414634146344e-05, |
| "loss": 0.8618, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.27139240506329115, |
| "grad_norm": 0.8230698649193114, |
| "learning_rate": 2.1788617886178863e-05, |
| "loss": 0.8689, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.27544303797468356, |
| "grad_norm": 0.8385549898483515, |
| "learning_rate": 2.2113821138211385e-05, |
| "loss": 0.887, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.27949367088607596, |
| "grad_norm": 0.9007047886491376, |
| "learning_rate": 2.2439024390243907e-05, |
| "loss": 0.8712, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.28354430379746837, |
| "grad_norm": 0.878514361601931, |
| "learning_rate": 2.2764227642276426e-05, |
| "loss": 0.8625, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.28759493670886077, |
| "grad_norm": 0.8629804221820763, |
| "learning_rate": 2.3089430894308948e-05, |
| "loss": 0.8541, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.2916455696202532, |
| "grad_norm": 0.9496534299401382, |
| "learning_rate": 2.3414634146341463e-05, |
| "loss": 0.8841, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.2956962025316456, |
| "grad_norm": 1.058922980380956, |
| "learning_rate": 2.3739837398373985e-05, |
| "loss": 0.8383, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.299746835443038, |
| "grad_norm": 0.9773455978400799, |
| "learning_rate": 2.4065040650406507e-05, |
| "loss": 0.8754, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.3037974683544304, |
| "grad_norm": 1.2215435307432119, |
| "learning_rate": 2.4390243902439026e-05, |
| "loss": 0.8717, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3078481012658228, |
| "grad_norm": 0.9521743612629814, |
| "learning_rate": 2.4715447154471548e-05, |
| "loss": 0.8613, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.3118987341772152, |
| "grad_norm": 1.0351393224567293, |
| "learning_rate": 2.5040650406504066e-05, |
| "loss": 0.8574, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.3159493670886076, |
| "grad_norm": 1.0215639389644928, |
| "learning_rate": 2.536585365853659e-05, |
| "loss": 0.8491, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 0.9687731071325082, |
| "learning_rate": 2.569105691056911e-05, |
| "loss": 0.8658, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.3240506329113924, |
| "grad_norm": 0.9827524608076812, |
| "learning_rate": 2.601626016260163e-05, |
| "loss": 0.8449, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3281012658227848, |
| "grad_norm": 1.0308350898422396, |
| "learning_rate": 2.634146341463415e-05, |
| "loss": 0.8684, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3321518987341772, |
| "grad_norm": 1.0005539277754403, |
| "learning_rate": 2.6666666666666667e-05, |
| "loss": 0.8692, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3362025316455696, |
| "grad_norm": 1.1437677035852034, |
| "learning_rate": 2.699186991869919e-05, |
| "loss": 0.8669, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.34025316455696203, |
| "grad_norm": 0.9413963500607863, |
| "learning_rate": 2.731707317073171e-05, |
| "loss": 0.8751, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.34430379746835443, |
| "grad_norm": 1.0751316498218904, |
| "learning_rate": 2.764227642276423e-05, |
| "loss": 0.8591, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.34835443037974684, |
| "grad_norm": 0.9435939537471055, |
| "learning_rate": 2.796747967479675e-05, |
| "loss": 0.8565, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.35240506329113924, |
| "grad_norm": 1.0026430798521169, |
| "learning_rate": 2.829268292682927e-05, |
| "loss": 0.8696, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.35645569620253165, |
| "grad_norm": 0.9740967584093266, |
| "learning_rate": 2.8617886178861792e-05, |
| "loss": 0.8663, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.36050632911392405, |
| "grad_norm": 0.8876086447184031, |
| "learning_rate": 2.8943089430894314e-05, |
| "loss": 0.8762, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.36455696202531646, |
| "grad_norm": 0.918623939385348, |
| "learning_rate": 2.926829268292683e-05, |
| "loss": 0.8578, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.36860759493670886, |
| "grad_norm": 1.127263421806182, |
| "learning_rate": 2.959349593495935e-05, |
| "loss": 0.8565, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.37265822784810126, |
| "grad_norm": 0.8118457881197524, |
| "learning_rate": 2.991869918699187e-05, |
| "loss": 0.8447, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.37670886075949367, |
| "grad_norm": 1.159465307139045, |
| "learning_rate": 3.0243902439024392e-05, |
| "loss": 0.8476, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3807594936708861, |
| "grad_norm": 0.91684113245444, |
| "learning_rate": 3.056910569105692e-05, |
| "loss": 0.8793, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.3848101265822785, |
| "grad_norm": 0.894600691958256, |
| "learning_rate": 3.089430894308943e-05, |
| "loss": 0.8645, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.3888607594936709, |
| "grad_norm": 1.0652678537579372, |
| "learning_rate": 3.1219512195121955e-05, |
| "loss": 0.8572, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.3929113924050633, |
| "grad_norm": 1.041042821720491, |
| "learning_rate": 3.154471544715447e-05, |
| "loss": 0.8495, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.3969620253164557, |
| "grad_norm": 0.9816454472348835, |
| "learning_rate": 3.186991869918699e-05, |
| "loss": 0.8547, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.4010126582278481, |
| "grad_norm": 1.1094503070523636, |
| "learning_rate": 3.2195121951219514e-05, |
| "loss": 0.8647, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.4050632911392405, |
| "grad_norm": 0.9803726827479325, |
| "learning_rate": 3.2520325203252037e-05, |
| "loss": 0.8757, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4091139240506329, |
| "grad_norm": 1.0581977316747706, |
| "learning_rate": 3.284552845528456e-05, |
| "loss": 0.8673, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.4131645569620253, |
| "grad_norm": 1.0046544249157943, |
| "learning_rate": 3.3170731707317074e-05, |
| "loss": 0.8533, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.4172151898734177, |
| "grad_norm": 0.9214423338463754, |
| "learning_rate": 3.3495934959349596e-05, |
| "loss": 0.8704, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.4212658227848101, |
| "grad_norm": 0.9088532241770915, |
| "learning_rate": 3.382113821138212e-05, |
| "loss": 0.8339, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4253164556962025, |
| "grad_norm": 0.8617585251298888, |
| "learning_rate": 3.414634146341463e-05, |
| "loss": 0.8775, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4293670886075949, |
| "grad_norm": 0.8279239760295143, |
| "learning_rate": 3.4471544715447155e-05, |
| "loss": 0.827, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.43341772151898733, |
| "grad_norm": 0.9185948063089058, |
| "learning_rate": 3.479674796747968e-05, |
| "loss": 0.8688, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.43746835443037974, |
| "grad_norm": 0.8436215456361902, |
| "learning_rate": 3.51219512195122e-05, |
| "loss": 0.8694, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.44151898734177214, |
| "grad_norm": 0.9435685355443912, |
| "learning_rate": 3.544715447154472e-05, |
| "loss": 0.8652, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.44556962025316454, |
| "grad_norm": 0.8317983015717868, |
| "learning_rate": 3.577235772357724e-05, |
| "loss": 0.8681, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.44962025316455695, |
| "grad_norm": 1.0123601262421673, |
| "learning_rate": 3.609756097560976e-05, |
| "loss": 0.8468, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.45367088607594935, |
| "grad_norm": 0.9075841576415593, |
| "learning_rate": 3.642276422764228e-05, |
| "loss": 0.859, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.45772151898734176, |
| "grad_norm": 1.0740349238147913, |
| "learning_rate": 3.67479674796748e-05, |
| "loss": 0.8538, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.46177215189873416, |
| "grad_norm": 0.9686707022516767, |
| "learning_rate": 3.7073170731707325e-05, |
| "loss": 0.8691, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.46582278481012657, |
| "grad_norm": 0.954551700582292, |
| "learning_rate": 3.739837398373984e-05, |
| "loss": 0.8564, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.46987341772151897, |
| "grad_norm": 1.083238213557681, |
| "learning_rate": 3.772357723577236e-05, |
| "loss": 0.8552, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.4739240506329114, |
| "grad_norm": 0.891652169685144, |
| "learning_rate": 3.804878048780488e-05, |
| "loss": 0.8533, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.4779746835443038, |
| "grad_norm": 1.0915741734473623, |
| "learning_rate": 3.83739837398374e-05, |
| "loss": 0.8542, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.4820253164556962, |
| "grad_norm": 0.9350810681517449, |
| "learning_rate": 3.869918699186992e-05, |
| "loss": 0.8762, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.4860759493670886, |
| "grad_norm": 1.0014944333905844, |
| "learning_rate": 3.9024390243902444e-05, |
| "loss": 0.8557, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.490126582278481, |
| "grad_norm": 1.0742562540202738, |
| "learning_rate": 3.9349593495934966e-05, |
| "loss": 0.8931, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.4941772151898734, |
| "grad_norm": 0.8865085616044852, |
| "learning_rate": 3.967479674796748e-05, |
| "loss": 0.8639, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4982278481012658, |
| "grad_norm": 1.1335372936859516, |
| "learning_rate": 4e-05, |
| "loss": 0.8556, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.5022784810126583, |
| "grad_norm": 1.0559734779301522, |
| "learning_rate": 3.999991946137476e-05, |
| "loss": 0.8788, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.5063291139240507, |
| "grad_norm": 0.9014627007197817, |
| "learning_rate": 3.999967784614766e-05, |
| "loss": 0.8774, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5103797468354431, |
| "grad_norm": 1.0200916260539203, |
| "learning_rate": 3.9999275156264656e-05, |
| "loss": 0.8682, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.5144303797468355, |
| "grad_norm": 0.9493020039272037, |
| "learning_rate": 3.999871139496895e-05, |
| "loss": 0.8556, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.5184810126582279, |
| "grad_norm": 0.9449939041249844, |
| "learning_rate": 3.9997986566800995e-05, |
| "loss": 0.8569, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.5225316455696203, |
| "grad_norm": 0.8113026282928388, |
| "learning_rate": 3.999710067759846e-05, |
| "loss": 0.8529, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.5265822784810127, |
| "grad_norm": 1.0750917664697488, |
| "learning_rate": 3.999605373449617e-05, |
| "loss": 0.8789, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5306329113924051, |
| "grad_norm": 0.9883307777827266, |
| "learning_rate": 3.9994845745926075e-05, |
| "loss": 0.8473, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.5346835443037975, |
| "grad_norm": 0.9395330580628619, |
| "learning_rate": 3.999347672161713e-05, |
| "loss": 0.8635, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5387341772151899, |
| "grad_norm": 1.1158842091521113, |
| "learning_rate": 3.999194667259528e-05, |
| "loss": 0.8415, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5427848101265823, |
| "grad_norm": 0.9556655332772692, |
| "learning_rate": 3.999025561118334e-05, |
| "loss": 0.8672, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5468354430379747, |
| "grad_norm": 0.9218761830028568, |
| "learning_rate": 3.998840355100086e-05, |
| "loss": 0.848, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5508860759493671, |
| "grad_norm": 0.8966798530296456, |
| "learning_rate": 3.998639050696409e-05, |
| "loss": 0.8418, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5549367088607595, |
| "grad_norm": 0.9523638477811749, |
| "learning_rate": 3.998421649528582e-05, |
| "loss": 0.878, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5589873417721519, |
| "grad_norm": 0.91944543303047, |
| "learning_rate": 3.9981881533475234e-05, |
| "loss": 0.891, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.5630379746835443, |
| "grad_norm": 0.8110752687280077, |
| "learning_rate": 3.997938564033779e-05, |
| "loss": 0.8403, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5670886075949367, |
| "grad_norm": 0.874781810405127, |
| "learning_rate": 3.9976728835975064e-05, |
| "loss": 0.8415, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5711392405063291, |
| "grad_norm": 0.7898025089247983, |
| "learning_rate": 3.9973911141784605e-05, |
| "loss": 0.8426, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.5751898734177215, |
| "grad_norm": 0.8650550394494385, |
| "learning_rate": 3.997093258045973e-05, |
| "loss": 0.8448, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.579240506329114, |
| "grad_norm": 0.8922123829713371, |
| "learning_rate": 3.996779317598936e-05, |
| "loss": 0.8819, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.5832911392405064, |
| "grad_norm": 0.832957269439817, |
| "learning_rate": 3.996449295365782e-05, |
| "loss": 0.8782, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5873417721518988, |
| "grad_norm": 0.8154617118359034, |
| "learning_rate": 3.996103194004467e-05, |
| "loss": 0.8567, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5913924050632912, |
| "grad_norm": 0.9839149463451112, |
| "learning_rate": 3.995741016302441e-05, |
| "loss": 0.8497, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5954430379746836, |
| "grad_norm": 0.8703011771333385, |
| "learning_rate": 3.9953627651766364e-05, |
| "loss": 0.8733, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.599493670886076, |
| "grad_norm": 0.8276020805257416, |
| "learning_rate": 3.9949684436734325e-05, |
| "loss": 0.8426, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.6035443037974684, |
| "grad_norm": 1.0580452696406282, |
| "learning_rate": 3.994558054968643e-05, |
| "loss": 0.8628, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.6075949367088608, |
| "grad_norm": 0.7687779641415857, |
| "learning_rate": 3.994131602367481e-05, |
| "loss": 0.8642, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6116455696202532, |
| "grad_norm": 1.0817623884644487, |
| "learning_rate": 3.9936890893045376e-05, |
| "loss": 0.8697, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.6156962025316456, |
| "grad_norm": 0.8211400560907584, |
| "learning_rate": 3.993230519343752e-05, |
| "loss": 0.8934, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.619746835443038, |
| "grad_norm": 0.8284707333189426, |
| "learning_rate": 3.992755896178383e-05, |
| "loss": 0.8576, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.6237974683544304, |
| "grad_norm": 0.7458793099404919, |
| "learning_rate": 3.992265223630981e-05, |
| "loss": 0.8538, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.6278481012658228, |
| "grad_norm": 0.8670428918095003, |
| "learning_rate": 3.991758505653355e-05, |
| "loss": 0.8652, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.6318987341772152, |
| "grad_norm": 0.7799836064421156, |
| "learning_rate": 3.991235746326543e-05, |
| "loss": 0.8565, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6359493670886076, |
| "grad_norm": 0.8023422684752723, |
| "learning_rate": 3.9906969498607745e-05, |
| "loss": 0.8676, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 0.8806922368695018, |
| "learning_rate": 3.990142120595444e-05, |
| "loss": 0.8533, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.6440506329113924, |
| "grad_norm": 0.7979991017380487, |
| "learning_rate": 3.98957126299907e-05, |
| "loss": 0.8743, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6481012658227848, |
| "grad_norm": 0.8826907906308917, |
| "learning_rate": 3.9889843816692596e-05, |
| "loss": 0.8687, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6521518987341772, |
| "grad_norm": 0.8738575938578411, |
| "learning_rate": 3.9883814813326766e-05, |
| "loss": 0.8774, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6562025316455696, |
| "grad_norm": 0.859375632782586, |
| "learning_rate": 3.9877625668449956e-05, |
| "loss": 0.8596, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.660253164556962, |
| "grad_norm": 0.9458007435740864, |
| "learning_rate": 3.98712764319087e-05, |
| "loss": 0.8563, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.6643037974683544, |
| "grad_norm": 0.8155806204638141, |
| "learning_rate": 3.9864767154838864e-05, |
| "loss": 0.8552, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6683544303797468, |
| "grad_norm": 0.8070145721732113, |
| "learning_rate": 3.9858097889665277e-05, |
| "loss": 0.8577, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6724050632911392, |
| "grad_norm": 0.764416293902997, |
| "learning_rate": 3.985126869010129e-05, |
| "loss": 0.8438, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6764556962025317, |
| "grad_norm": 0.8189459790198732, |
| "learning_rate": 3.984427961114833e-05, |
| "loss": 0.8447, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.6805063291139241, |
| "grad_norm": 0.8061834980052596, |
| "learning_rate": 3.9837130709095475e-05, |
| "loss": 0.8702, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.6845569620253165, |
| "grad_norm": 0.781933419645468, |
| "learning_rate": 3.982982204151901e-05, |
| "loss": 0.8701, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6886075949367089, |
| "grad_norm": 0.8383318878414807, |
| "learning_rate": 3.982235366728193e-05, |
| "loss": 0.8593, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6926582278481013, |
| "grad_norm": 0.7402097729302649, |
| "learning_rate": 3.9814725646533505e-05, |
| "loss": 0.8517, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6967088607594937, |
| "grad_norm": 1.0480946444078172, |
| "learning_rate": 3.9806938040708746e-05, |
| "loss": 0.8595, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.7007594936708861, |
| "grad_norm": 0.7972549059558025, |
| "learning_rate": 3.9798990912527976e-05, |
| "loss": 0.8627, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.7048101265822785, |
| "grad_norm": 1.1794919578010339, |
| "learning_rate": 3.979088432599627e-05, |
| "loss": 0.8345, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.7088607594936709, |
| "grad_norm": 0.8139819848825086, |
| "learning_rate": 3.9782618346402964e-05, |
| "loss": 0.8527, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7129113924050633, |
| "grad_norm": 1.0378035003347132, |
| "learning_rate": 3.977419304032111e-05, |
| "loss": 0.8546, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.7169620253164557, |
| "grad_norm": 0.802891439863913, |
| "learning_rate": 3.976560847560697e-05, |
| "loss": 0.8618, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.7210126582278481, |
| "grad_norm": 0.8214887488158006, |
| "learning_rate": 3.9756864721399456e-05, |
| "loss": 0.8491, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.7250632911392405, |
| "grad_norm": 0.7883093976180914, |
| "learning_rate": 3.974796184811956e-05, |
| "loss": 0.8631, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.7291139240506329, |
| "grad_norm": 0.905418504249143, |
| "learning_rate": 3.973889992746979e-05, |
| "loss": 0.8652, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7331645569620253, |
| "grad_norm": 0.915026640991035, |
| "learning_rate": 3.972967903243361e-05, |
| "loss": 0.8438, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.7372151898734177, |
| "grad_norm": 0.7623404271284945, |
| "learning_rate": 3.972029923727486e-05, |
| "loss": 0.8341, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7412658227848101, |
| "grad_norm": 0.8915209830327925, |
| "learning_rate": 3.971076061753709e-05, |
| "loss": 0.8429, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7453164556962025, |
| "grad_norm": 0.7630383172219343, |
| "learning_rate": 3.9701063250043066e-05, |
| "loss": 0.856, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7493670886075949, |
| "grad_norm": 0.8716251073033626, |
| "learning_rate": 3.969120721289402e-05, |
| "loss": 0.8644, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.7534177215189873, |
| "grad_norm": 0.7415485849517455, |
| "learning_rate": 3.9681192585469146e-05, |
| "loss": 0.8714, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.7574683544303797, |
| "grad_norm": 0.7391151873608139, |
| "learning_rate": 3.9671019448424865e-05, |
| "loss": 0.8558, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7615189873417721, |
| "grad_norm": 0.7737909990654371, |
| "learning_rate": 3.966068788369422e-05, |
| "loss": 0.8421, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7655696202531646, |
| "grad_norm": 0.7500334112878827, |
| "learning_rate": 3.965019797448622e-05, |
| "loss": 0.8743, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.769620253164557, |
| "grad_norm": 0.8903671399821592, |
| "learning_rate": 3.963954980528515e-05, |
| "loss": 0.8445, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.7736708860759494, |
| "grad_norm": 0.707367840091562, |
| "learning_rate": 3.9628743461849905e-05, |
| "loss": 0.8587, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7777215189873418, |
| "grad_norm": 0.7569753082774013, |
| "learning_rate": 3.961777903121329e-05, |
| "loss": 0.8451, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7817721518987342, |
| "grad_norm": 0.7234456883517504, |
| "learning_rate": 3.960665660168131e-05, |
| "loss": 0.8369, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7858227848101266, |
| "grad_norm": 0.7221120495561383, |
| "learning_rate": 3.9595376262832485e-05, |
| "loss": 0.8235, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.789873417721519, |
| "grad_norm": 0.7245031814825513, |
| "learning_rate": 3.9583938105517127e-05, |
| "loss": 0.8704, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7939240506329114, |
| "grad_norm": 0.6916987370608221, |
| "learning_rate": 3.957234222185657e-05, |
| "loss": 0.8564, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7979746835443038, |
| "grad_norm": 0.6931127950266375, |
| "learning_rate": 3.9560588705242474e-05, |
| "loss": 0.8343, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.8020253164556962, |
| "grad_norm": 0.721062258482943, |
| "learning_rate": 3.954867765033605e-05, |
| "loss": 0.8634, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.8060759493670886, |
| "grad_norm": 0.7443140101922328, |
| "learning_rate": 3.953660915306728e-05, |
| "loss": 0.8683, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.810126582278481, |
| "grad_norm": 0.6480060578122203, |
| "learning_rate": 3.952438331063419e-05, |
| "loss": 0.8422, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8141772151898734, |
| "grad_norm": 0.7562148406028854, |
| "learning_rate": 3.951200022150205e-05, |
| "loss": 0.8313, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.8182278481012658, |
| "grad_norm": 0.6992783628832099, |
| "learning_rate": 3.949945998540253e-05, |
| "loss": 0.8557, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.8222784810126582, |
| "grad_norm": 0.6782396402927564, |
| "learning_rate": 3.9486762703332993e-05, |
| "loss": 0.8448, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.8263291139240506, |
| "grad_norm": 0.7363608591529054, |
| "learning_rate": 3.947390847755559e-05, |
| "loss": 0.8658, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.830379746835443, |
| "grad_norm": 0.755142716529058, |
| "learning_rate": 3.946089741159648e-05, |
| "loss": 0.8507, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8344303797468354, |
| "grad_norm": 0.7550588267420778, |
| "learning_rate": 3.944772961024501e-05, |
| "loss": 0.8791, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.8384810126582278, |
| "grad_norm": 0.7399255244822275, |
| "learning_rate": 3.943440517955285e-05, |
| "loss": 0.8347, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.8425316455696202, |
| "grad_norm": 0.7400594362488464, |
| "learning_rate": 3.9420924226833126e-05, |
| "loss": 0.8603, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8465822784810126, |
| "grad_norm": 0.8057086715198888, |
| "learning_rate": 3.9407286860659566e-05, |
| "loss": 0.8234, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.850632911392405, |
| "grad_norm": 0.6786413402312906, |
| "learning_rate": 3.9393493190865657e-05, |
| "loss": 0.8421, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8546835443037974, |
| "grad_norm": 0.721542705803571, |
| "learning_rate": 3.937954332854371e-05, |
| "loss": 0.8452, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8587341772151899, |
| "grad_norm": 0.6970052631073644, |
| "learning_rate": 3.9365437386044016e-05, |
| "loss": 0.8439, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.8627848101265823, |
| "grad_norm": 0.7272784897494002, |
| "learning_rate": 3.935117547697387e-05, |
| "loss": 0.857, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8668354430379747, |
| "grad_norm": 0.701200083501653, |
| "learning_rate": 3.933675771619675e-05, |
| "loss": 0.8563, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8708860759493671, |
| "grad_norm": 0.731137410752718, |
| "learning_rate": 3.932218421983131e-05, |
| "loss": 0.8509, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8749367088607595, |
| "grad_norm": 0.7721628767847316, |
| "learning_rate": 3.9307455105250484e-05, |
| "loss": 0.8088, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8789873417721519, |
| "grad_norm": 0.7981773577284843, |
| "learning_rate": 3.929257049108054e-05, |
| "loss": 0.8325, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8830379746835443, |
| "grad_norm": 0.8366708552886382, |
| "learning_rate": 3.927753049720011e-05, |
| "loss": 0.8315, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8870886075949367, |
| "grad_norm": 0.7281907666865322, |
| "learning_rate": 3.9262335244739234e-05, |
| "loss": 0.8484, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8911392405063291, |
| "grad_norm": 0.7829410509562663, |
| "learning_rate": 3.92469848560784e-05, |
| "loss": 0.8329, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8951898734177215, |
| "grad_norm": 0.6839492738194843, |
| "learning_rate": 3.923147945484751e-05, |
| "loss": 0.8242, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8992405063291139, |
| "grad_norm": 0.7494928828549796, |
| "learning_rate": 3.9215819165924956e-05, |
| "loss": 0.8499, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.9032911392405063, |
| "grad_norm": 0.705550754042164, |
| "learning_rate": 3.920000411543654e-05, |
| "loss": 0.8246, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.9073417721518987, |
| "grad_norm": 0.7482595276020934, |
| "learning_rate": 3.9184034430754495e-05, |
| "loss": 0.8588, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.9113924050632911, |
| "grad_norm": 0.7474788021332583, |
| "learning_rate": 3.916791024049648e-05, |
| "loss": 0.8564, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9154430379746835, |
| "grad_norm": 0.7314338142328388, |
| "learning_rate": 3.91516316745245e-05, |
| "loss": 0.8476, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.9194936708860759, |
| "grad_norm": 0.7517874183634545, |
| "learning_rate": 3.913519886394389e-05, |
| "loss": 0.8464, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.9235443037974683, |
| "grad_norm": 0.7079606625658917, |
| "learning_rate": 3.911861194110225e-05, |
| "loss": 0.8443, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.9275949367088607, |
| "grad_norm": 0.730078436890439, |
| "learning_rate": 3.910187103958837e-05, |
| "loss": 0.8446, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.9316455696202531, |
| "grad_norm": 0.7484034906501219, |
| "learning_rate": 3.908497629423117e-05, |
| "loss": 0.8428, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9356962025316455, |
| "grad_norm": 0.694780717488654, |
| "learning_rate": 3.9067927841098614e-05, |
| "loss": 0.8273, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.9397468354430379, |
| "grad_norm": 0.7721900791283391, |
| "learning_rate": 3.9050725817496594e-05, |
| "loss": 0.8455, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.9437974683544303, |
| "grad_norm": 0.6960672807736287, |
| "learning_rate": 3.9033370361967844e-05, |
| "loss": 0.859, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.9478481012658228, |
| "grad_norm": 0.7776977738474254, |
| "learning_rate": 3.901586161429081e-05, |
| "loss": 0.842, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.9518987341772152, |
| "grad_norm": 0.6649602884452366, |
| "learning_rate": 3.8998199715478545e-05, |
| "loss": 0.8417, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9559493670886076, |
| "grad_norm": 0.7462289159566555, |
| "learning_rate": 3.8980384807777564e-05, |
| "loss": 0.826, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 0.6801984476399956, |
| "learning_rate": 3.896241703466667e-05, |
| "loss": 0.8315, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9640506329113924, |
| "grad_norm": 0.8106669491776499, |
| "learning_rate": 3.894429654085585e-05, |
| "loss": 0.8565, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9681012658227848, |
| "grad_norm": 0.7030140800396554, |
| "learning_rate": 3.892602347228505e-05, |
| "loss": 0.84, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.9721518987341772, |
| "grad_norm": 0.6955791727837238, |
| "learning_rate": 3.890759797612307e-05, |
| "loss": 0.8318, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9762025316455696, |
| "grad_norm": 0.764189133140535, |
| "learning_rate": 3.888902020076632e-05, |
| "loss": 0.865, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.980253164556962, |
| "grad_norm": 0.7323052338611501, |
| "learning_rate": 3.887029029583764e-05, |
| "loss": 0.84, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9843037974683544, |
| "grad_norm": 0.7520001579645765, |
| "learning_rate": 3.8851408412185125e-05, |
| "loss": 0.8406, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9883544303797468, |
| "grad_norm": 0.6355490110132702, |
| "learning_rate": 3.8832374701880855e-05, |
| "loss": 0.8307, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9924050632911392, |
| "grad_norm": 0.7444397140058374, |
| "learning_rate": 3.881318931821972e-05, |
| "loss": 0.8513, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9964556962025316, |
| "grad_norm": 0.6789536460441383, |
| "learning_rate": 3.879385241571817e-05, |
| "loss": 0.8457, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0020253164556963, |
| "grad_norm": 0.7655366257114053, |
| "learning_rate": 3.8774364150112955e-05, |
| "loss": 0.7406, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.0060759493670886, |
| "grad_norm": 0.8297106551469194, |
| "learning_rate": 3.8754724678359884e-05, |
| "loss": 0.6292, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.010126582278481, |
| "grad_norm": 0.7244821856310985, |
| "learning_rate": 3.873493415863256e-05, |
| "loss": 0.6282, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.0141772151898734, |
| "grad_norm": 0.9971328197550786, |
| "learning_rate": 3.871499275032111e-05, |
| "loss": 0.605, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0182278481012659, |
| "grad_norm": 0.7915170446839865, |
| "learning_rate": 3.869490061403091e-05, |
| "loss": 0.6172, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.0222784810126582, |
| "grad_norm": 0.7997842652969129, |
| "learning_rate": 3.867465791158124e-05, |
| "loss": 0.6065, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.0263291139240507, |
| "grad_norm": 0.8009764570840647, |
| "learning_rate": 3.865426480600407e-05, |
| "loss": 0.5926, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.030379746835443, |
| "grad_norm": 0.7754353878088227, |
| "learning_rate": 3.863372146154264e-05, |
| "loss": 0.6259, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.0344303797468355, |
| "grad_norm": 0.7107725203694388, |
| "learning_rate": 3.861302804365024e-05, |
| "loss": 0.6161, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.0384810126582278, |
| "grad_norm": 0.7687311446107828, |
| "learning_rate": 3.85921847189888e-05, |
| "loss": 0.6262, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.0425316455696203, |
| "grad_norm": 0.8323431284096109, |
| "learning_rate": 3.85711916554276e-05, |
| "loss": 0.6412, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.0465822784810126, |
| "grad_norm": 0.8382726024225052, |
| "learning_rate": 3.85500490220419e-05, |
| "loss": 0.6296, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.0506329113924051, |
| "grad_norm": 0.6856641325422058, |
| "learning_rate": 3.852875698911154e-05, |
| "loss": 0.6178, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.0546835443037974, |
| "grad_norm": 0.7804775974219711, |
| "learning_rate": 3.850731572811963e-05, |
| "loss": 0.6202, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.05873417721519, |
| "grad_norm": 0.8202185062939767, |
| "learning_rate": 3.848572541175116e-05, |
| "loss": 0.605, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.0627848101265822, |
| "grad_norm": 0.762647203087796, |
| "learning_rate": 3.846398621389154e-05, |
| "loss": 0.6169, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.0668354430379747, |
| "grad_norm": 0.7972607015079801, |
| "learning_rate": 3.84420983096253e-05, |
| "loss": 0.6057, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.070886075949367, |
| "grad_norm": 0.7290109615785214, |
| "learning_rate": 3.8420061875234606e-05, |
| "loss": 0.5973, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.0749367088607595, |
| "grad_norm": 0.8047365871642508, |
| "learning_rate": 3.839787708819787e-05, |
| "loss": 0.6378, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.0789873417721518, |
| "grad_norm": 0.7458961861426812, |
| "learning_rate": 3.8375544127188325e-05, |
| "loss": 0.597, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.0830379746835443, |
| "grad_norm": 0.7759609918423381, |
| "learning_rate": 3.8353063172072564e-05, |
| "loss": 0.6277, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.0870886075949366, |
| "grad_norm": 0.7584949217350448, |
| "learning_rate": 3.8330434403909105e-05, |
| "loss": 0.592, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.0911392405063292, |
| "grad_norm": 0.7584147715303052, |
| "learning_rate": 3.8307658004946934e-05, |
| "loss": 0.6144, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.0951898734177214, |
| "grad_norm": 0.7606245372683064, |
| "learning_rate": 3.8284734158624046e-05, |
| "loss": 0.6084, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.099240506329114, |
| "grad_norm": 0.8433086528811893, |
| "learning_rate": 3.826166304956594e-05, |
| "loss": 0.5826, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.1032911392405063, |
| "grad_norm": 0.8018344808175598, |
| "learning_rate": 3.8238444863584164e-05, |
| "loss": 0.6293, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.1073417721518988, |
| "grad_norm": 0.7249443661087615, |
| "learning_rate": 3.821507978767479e-05, |
| "loss": 0.5976, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.111392405063291, |
| "grad_norm": 0.7214717243022701, |
| "learning_rate": 3.819156801001693e-05, |
| "loss": 0.5993, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.1154430379746836, |
| "grad_norm": 0.81065687327827, |
| "learning_rate": 3.816790971997121e-05, |
| "loss": 0.574, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.1194936708860759, |
| "grad_norm": 0.7133422917196783, |
| "learning_rate": 3.8144105108078246e-05, |
| "loss": 0.5868, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.1235443037974684, |
| "grad_norm": 0.7118017118692876, |
| "learning_rate": 3.81201543660571e-05, |
| "loss": 0.6114, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.1275949367088607, |
| "grad_norm": 0.8091800539175845, |
| "learning_rate": 3.809605768680377e-05, |
| "loss": 0.6194, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.1316455696202532, |
| "grad_norm": 0.7017095087245776, |
| "learning_rate": 3.807181526438958e-05, |
| "loss": 0.6075, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.1356962025316455, |
| "grad_norm": 0.7097053953606419, |
| "learning_rate": 3.8047427294059697e-05, |
| "loss": 0.6097, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.139746835443038, |
| "grad_norm": 0.7112130525829526, |
| "learning_rate": 3.802289397223145e-05, |
| "loss": 0.5817, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.1437974683544303, |
| "grad_norm": 0.675940555138587, |
| "learning_rate": 3.7998215496492854e-05, |
| "loss": 0.6252, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.1478481012658228, |
| "grad_norm": 0.668969571129357, |
| "learning_rate": 3.797339206560096e-05, |
| "loss": 0.6109, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.1518987341772151, |
| "grad_norm": 0.7300754912111809, |
| "learning_rate": 3.794842387948027e-05, |
| "loss": 0.5775, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.1559493670886076, |
| "grad_norm": 0.6301176580673331, |
| "learning_rate": 3.7923311139221114e-05, |
| "loss": 0.5869, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.16, |
| "grad_norm": 0.7517596931296197, |
| "learning_rate": 3.7898054047078054e-05, |
| "loss": 0.6104, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.1640506329113924, |
| "grad_norm": 0.6911790436540044, |
| "learning_rate": 3.787265280646825e-05, |
| "loss": 0.5972, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.1681012658227847, |
| "grad_norm": 0.7293202743187903, |
| "learning_rate": 3.7847107621969786e-05, |
| "loss": 0.6104, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.1721518987341772, |
| "grad_norm": 0.7176887847128914, |
| "learning_rate": 3.7821418699320064e-05, |
| "loss": 0.6185, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.1762025316455695, |
| "grad_norm": 0.6959781243529435, |
| "learning_rate": 3.7795586245414145e-05, |
| "loss": 0.6027, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.180253164556962, |
| "grad_norm": 0.7420344749064076, |
| "learning_rate": 3.776961046830306e-05, |
| "loss": 0.6, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.1843037974683543, |
| "grad_norm": 0.685733651200489, |
| "learning_rate": 3.774349157719215e-05, |
| "loss": 0.6143, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.1883544303797469, |
| "grad_norm": 0.7738493931231789, |
| "learning_rate": 3.7717229782439365e-05, |
| "loss": 0.6123, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.1924050632911392, |
| "grad_norm": 0.6706476976479616, |
| "learning_rate": 3.769082529555359e-05, |
| "loss": 0.616, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.1964556962025317, |
| "grad_norm": 0.6577114433208605, |
| "learning_rate": 3.766427832919294e-05, |
| "loss": 0.6077, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.200506329113924, |
| "grad_norm": 0.6909534093579517, |
| "learning_rate": 3.7637589097163024e-05, |
| "loss": 0.6133, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.2045569620253165, |
| "grad_norm": 0.6887079857320298, |
| "learning_rate": 3.761075781441526e-05, |
| "loss": 0.6104, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.2086075949367088, |
| "grad_norm": 0.6886802445082463, |
| "learning_rate": 3.75837846970451e-05, |
| "loss": 0.5691, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.2126582278481013, |
| "grad_norm": 0.7915795094956625, |
| "learning_rate": 3.755666996229032e-05, |
| "loss": 0.6123, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.2167088607594936, |
| "grad_norm": 0.8041995044273236, |
| "learning_rate": 3.752941382852927e-05, |
| "loss": 0.6329, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.220759493670886, |
| "grad_norm": 0.7827040800419226, |
| "learning_rate": 3.7502016515279115e-05, |
| "loss": 0.5968, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.2248101265822784, |
| "grad_norm": 0.8063341143398766, |
| "learning_rate": 3.7474478243194043e-05, |
| "loss": 0.6107, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.228860759493671, |
| "grad_norm": 0.7102945307436225, |
| "learning_rate": 3.744679923406351e-05, |
| "loss": 0.6123, |
| "step": 303 |
| }, |
| { |
| "epoch": 1.2329113924050632, |
| "grad_norm": 0.7394861329698103, |
| "learning_rate": 3.741897971081043e-05, |
| "loss": 0.6128, |
| "step": 304 |
| }, |
| { |
| "epoch": 1.2369620253164557, |
| "grad_norm": 0.7364432495203813, |
| "learning_rate": 3.739101989748946e-05, |
| "loss": 0.6017, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.241012658227848, |
| "grad_norm": 0.7208102084455967, |
| "learning_rate": 3.7362920019285066e-05, |
| "loss": 0.6036, |
| "step": 306 |
| }, |
| { |
| "epoch": 1.2450632911392405, |
| "grad_norm": 0.6977031647432438, |
| "learning_rate": 3.73346803025098e-05, |
| "loss": 0.6075, |
| "step": 307 |
| }, |
| { |
| "epoch": 1.2491139240506328, |
| "grad_norm": 0.7676598752033653, |
| "learning_rate": 3.730630097460247e-05, |
| "loss": 0.6063, |
| "step": 308 |
| }, |
| { |
| "epoch": 1.2531645569620253, |
| "grad_norm": 0.732920125205381, |
| "learning_rate": 3.727778226412628e-05, |
| "loss": 0.6235, |
| "step": 309 |
| }, |
| { |
| "epoch": 1.2572151898734178, |
| "grad_norm": 0.7461133303444637, |
| "learning_rate": 3.7249124400767006e-05, |
| "loss": 0.6106, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.2612658227848101, |
| "grad_norm": 0.7525959211922907, |
| "learning_rate": 3.722032761533114e-05, |
| "loss": 0.5973, |
| "step": 311 |
| }, |
| { |
| "epoch": 1.2653164556962024, |
| "grad_norm": 0.7997916735108418, |
| "learning_rate": 3.719139213974403e-05, |
| "loss": 0.6267, |
| "step": 312 |
| }, |
| { |
| "epoch": 1.269367088607595, |
| "grad_norm": 0.7143911770173964, |
| "learning_rate": 3.7162318207048006e-05, |
| "loss": 0.6176, |
| "step": 313 |
| }, |
| { |
| "epoch": 1.2734177215189875, |
| "grad_norm": 0.7591934745544948, |
| "learning_rate": 3.713310605140055e-05, |
| "loss": 0.6146, |
| "step": 314 |
| }, |
| { |
| "epoch": 1.2774683544303798, |
| "grad_norm": 0.7676140021665315, |
| "learning_rate": 3.710375590807233e-05, |
| "loss": 0.6163, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.281518987341772, |
| "grad_norm": 0.6999858837361977, |
| "learning_rate": 3.7074268013445365e-05, |
| "loss": 0.628, |
| "step": 316 |
| }, |
| { |
| "epoch": 1.2855696202531646, |
| "grad_norm": 0.7845766534634555, |
| "learning_rate": 3.7044642605011114e-05, |
| "loss": 0.6192, |
| "step": 317 |
| }, |
| { |
| "epoch": 1.289620253164557, |
| "grad_norm": 0.6864039659017112, |
| "learning_rate": 3.701487992136854e-05, |
| "loss": 0.5928, |
| "step": 318 |
| }, |
| { |
| "epoch": 1.2936708860759494, |
| "grad_norm": 0.7797510046142586, |
| "learning_rate": 3.69849802022222e-05, |
| "loss": 0.608, |
| "step": 319 |
| }, |
| { |
| "epoch": 1.2977215189873417, |
| "grad_norm": 0.728341217185414, |
| "learning_rate": 3.6954943688380334e-05, |
| "loss": 0.6087, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.3017721518987342, |
| "grad_norm": 0.7278885358216384, |
| "learning_rate": 3.692477062175289e-05, |
| "loss": 0.6178, |
| "step": 321 |
| }, |
| { |
| "epoch": 1.3058227848101267, |
| "grad_norm": 0.7903133647693169, |
| "learning_rate": 3.689446124534958e-05, |
| "loss": 0.6408, |
| "step": 322 |
| }, |
| { |
| "epoch": 1.309873417721519, |
| "grad_norm": 0.7405757316713661, |
| "learning_rate": 3.686401580327799e-05, |
| "loss": 0.6124, |
| "step": 323 |
| }, |
| { |
| "epoch": 1.3139240506329113, |
| "grad_norm": 0.6611859789711387, |
| "learning_rate": 3.683343454074149e-05, |
| "loss": 0.6347, |
| "step": 324 |
| }, |
| { |
| "epoch": 1.3179746835443038, |
| "grad_norm": 0.7588338282150465, |
| "learning_rate": 3.6802717704037386e-05, |
| "loss": 0.6323, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.3220253164556963, |
| "grad_norm": 0.661643633558132, |
| "learning_rate": 3.6771865540554855e-05, |
| "loss": 0.621, |
| "step": 326 |
| }, |
| { |
| "epoch": 1.3260759493670886, |
| "grad_norm": 0.7091169720497806, |
| "learning_rate": 3.674087829877297e-05, |
| "loss": 0.6218, |
| "step": 327 |
| }, |
| { |
| "epoch": 1.330126582278481, |
| "grad_norm": 0.6520761471239637, |
| "learning_rate": 3.6709756228258735e-05, |
| "loss": 0.6027, |
| "step": 328 |
| }, |
| { |
| "epoch": 1.3341772151898734, |
| "grad_norm": 0.6804984613291785, |
| "learning_rate": 3.667849957966501e-05, |
| "loss": 0.5977, |
| "step": 329 |
| }, |
| { |
| "epoch": 1.338227848101266, |
| "grad_norm": 0.7155820211764714, |
| "learning_rate": 3.6647108604728546e-05, |
| "loss": 0.609, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3422784810126582, |
| "grad_norm": 0.6536975176039188, |
| "learning_rate": 3.661558355626795e-05, |
| "loss": 0.6195, |
| "step": 331 |
| }, |
| { |
| "epoch": 1.3463291139240505, |
| "grad_norm": 0.6777647207134689, |
| "learning_rate": 3.658392468818163e-05, |
| "loss": 0.6223, |
| "step": 332 |
| }, |
| { |
| "epoch": 1.350379746835443, |
| "grad_norm": 0.6705985699880267, |
| "learning_rate": 3.655213225544574e-05, |
| "loss": 0.6075, |
| "step": 333 |
| }, |
| { |
| "epoch": 1.3544303797468356, |
| "grad_norm": 0.6614971917249571, |
| "learning_rate": 3.652020651411218e-05, |
| "loss": 0.6189, |
| "step": 334 |
| }, |
| { |
| "epoch": 1.3584810126582278, |
| "grad_norm": 0.7048297347068064, |
| "learning_rate": 3.6488147721306474e-05, |
| "loss": 0.6027, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.3625316455696201, |
| "grad_norm": 0.6520317421724423, |
| "learning_rate": 3.645595613522574e-05, |
| "loss": 0.6177, |
| "step": 336 |
| }, |
| { |
| "epoch": 1.3665822784810127, |
| "grad_norm": 0.64952470732351, |
| "learning_rate": 3.642363201513657e-05, |
| "loss": 0.6182, |
| "step": 337 |
| }, |
| { |
| "epoch": 1.3706329113924052, |
| "grad_norm": 0.6172869738603578, |
| "learning_rate": 3.6391175621373006e-05, |
| "loss": 0.5944, |
| "step": 338 |
| }, |
| { |
| "epoch": 1.3746835443037975, |
| "grad_norm": 0.7163405335360639, |
| "learning_rate": 3.6358587215334355e-05, |
| "loss": 0.6079, |
| "step": 339 |
| }, |
| { |
| "epoch": 1.3787341772151898, |
| "grad_norm": 0.6962129486378216, |
| "learning_rate": 3.632586705948318e-05, |
| "loss": 0.6005, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.3827848101265823, |
| "grad_norm": 0.7280828556429481, |
| "learning_rate": 3.629301541734311e-05, |
| "loss": 0.5974, |
| "step": 341 |
| }, |
| { |
| "epoch": 1.3868354430379748, |
| "grad_norm": 0.7536946865397655, |
| "learning_rate": 3.626003255349676e-05, |
| "loss": 0.6395, |
| "step": 342 |
| }, |
| { |
| "epoch": 1.390886075949367, |
| "grad_norm": 0.6751274512655943, |
| "learning_rate": 3.622691873358357e-05, |
| "loss": 0.64, |
| "step": 343 |
| }, |
| { |
| "epoch": 1.3949367088607594, |
| "grad_norm": 0.7349357694211599, |
| "learning_rate": 3.61936742242977e-05, |
| "loss": 0.6232, |
| "step": 344 |
| }, |
| { |
| "epoch": 1.398987341772152, |
| "grad_norm": 0.6943957241196343, |
| "learning_rate": 3.6160299293385864e-05, |
| "loss": 0.6221, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4030379746835444, |
| "grad_norm": 0.7608867157590847, |
| "learning_rate": 3.612679420964516e-05, |
| "loss": 0.6331, |
| "step": 346 |
| }, |
| { |
| "epoch": 1.4070886075949367, |
| "grad_norm": 0.7190430297315723, |
| "learning_rate": 3.609315924292092e-05, |
| "loss": 0.6443, |
| "step": 347 |
| }, |
| { |
| "epoch": 1.411139240506329, |
| "grad_norm": 0.7436773332606286, |
| "learning_rate": 3.6059394664104554e-05, |
| "loss": 0.5938, |
| "step": 348 |
| }, |
| { |
| "epoch": 1.4151898734177215, |
| "grad_norm": 0.7102694589847485, |
| "learning_rate": 3.602550074513133e-05, |
| "loss": 0.6136, |
| "step": 349 |
| }, |
| { |
| "epoch": 1.419240506329114, |
| "grad_norm": 0.8041814971569166, |
| "learning_rate": 3.599147775897822e-05, |
| "loss": 0.6318, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4232911392405063, |
| "grad_norm": 0.7871626168431862, |
| "learning_rate": 3.595732597966167e-05, |
| "loss": 0.6241, |
| "step": 351 |
| }, |
| { |
| "epoch": 1.4273417721518986, |
| "grad_norm": 0.7081782425135938, |
| "learning_rate": 3.592304568223542e-05, |
| "loss": 0.6193, |
| "step": 352 |
| }, |
| { |
| "epoch": 1.4313924050632911, |
| "grad_norm": 0.6923230538345254, |
| "learning_rate": 3.588863714278826e-05, |
| "loss": 0.6172, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.4354430379746836, |
| "grad_norm": 0.6720889974591896, |
| "learning_rate": 3.585410063844186e-05, |
| "loss": 0.6073, |
| "step": 354 |
| }, |
| { |
| "epoch": 1.439493670886076, |
| "grad_norm": 0.7621637052770909, |
| "learning_rate": 3.581943644734846e-05, |
| "loss": 0.5924, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.4435443037974682, |
| "grad_norm": 0.707850192988169, |
| "learning_rate": 3.578464484868869e-05, |
| "loss": 0.599, |
| "step": 356 |
| }, |
| { |
| "epoch": 1.4475949367088607, |
| "grad_norm": 0.7739826179502968, |
| "learning_rate": 3.5749726122669316e-05, |
| "loss": 0.6094, |
| "step": 357 |
| }, |
| { |
| "epoch": 1.4516455696202533, |
| "grad_norm": 0.6828737038009289, |
| "learning_rate": 3.5714680550520943e-05, |
| "loss": 0.6019, |
| "step": 358 |
| }, |
| { |
| "epoch": 1.4556962025316456, |
| "grad_norm": 0.7038265112054953, |
| "learning_rate": 3.5679508414495794e-05, |
| "loss": 0.6327, |
| "step": 359 |
| }, |
| { |
| "epoch": 1.4597468354430378, |
| "grad_norm": 0.6872554228332286, |
| "learning_rate": 3.564420999786543e-05, |
| "loss": 0.6069, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.4637974683544304, |
| "grad_norm": 0.7433257557893647, |
| "learning_rate": 3.560878558491842e-05, |
| "loss": 0.5991, |
| "step": 361 |
| }, |
| { |
| "epoch": 1.4678481012658229, |
| "grad_norm": 0.730691910500668, |
| "learning_rate": 3.5573235460958145e-05, |
| "loss": 0.6357, |
| "step": 362 |
| }, |
| { |
| "epoch": 1.4718987341772152, |
| "grad_norm": 0.6702248007087148, |
| "learning_rate": 3.553755991230039e-05, |
| "loss": 0.6118, |
| "step": 363 |
| }, |
| { |
| "epoch": 1.4759493670886075, |
| "grad_norm": 0.6655569911276386, |
| "learning_rate": 3.5501759226271144e-05, |
| "loss": 0.6122, |
| "step": 364 |
| }, |
| { |
| "epoch": 1.48, |
| "grad_norm": 0.7080256002339751, |
| "learning_rate": 3.546583369120419e-05, |
| "loss": 0.6077, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.4840506329113925, |
| "grad_norm": 0.7406796912613067, |
| "learning_rate": 3.5429783596438864e-05, |
| "loss": 0.6183, |
| "step": 366 |
| }, |
| { |
| "epoch": 1.4881012658227848, |
| "grad_norm": 0.6657006138109375, |
| "learning_rate": 3.539360923231766e-05, |
| "loss": 0.5979, |
| "step": 367 |
| }, |
| { |
| "epoch": 1.492151898734177, |
| "grad_norm": 0.6553783968025633, |
| "learning_rate": 3.535731089018394e-05, |
| "loss": 0.5972, |
| "step": 368 |
| }, |
| { |
| "epoch": 1.4962025316455696, |
| "grad_norm": 0.6910829643766525, |
| "learning_rate": 3.532088886237956e-05, |
| "loss": 0.6161, |
| "step": 369 |
| }, |
| { |
| "epoch": 1.5002531645569621, |
| "grad_norm": 0.627612651528352, |
| "learning_rate": 3.528434344224253e-05, |
| "loss": 0.6049, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5043037974683544, |
| "grad_norm": 0.6513560628263751, |
| "learning_rate": 3.524767492410464e-05, |
| "loss": 0.6086, |
| "step": 371 |
| }, |
| { |
| "epoch": 1.5083544303797467, |
| "grad_norm": 0.6490783831730885, |
| "learning_rate": 3.521088360328908e-05, |
| "loss": 0.6306, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.5124050632911392, |
| "grad_norm": 0.6687735340492523, |
| "learning_rate": 3.517396977610811e-05, |
| "loss": 0.6001, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.5164556962025317, |
| "grad_norm": 0.6496715163051524, |
| "learning_rate": 3.5136933739860595e-05, |
| "loss": 0.628, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.520506329113924, |
| "grad_norm": 0.7151619644070843, |
| "learning_rate": 3.509977579282971e-05, |
| "loss": 0.631, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.5245569620253163, |
| "grad_norm": 0.6453375736290162, |
| "learning_rate": 3.5062496234280424e-05, |
| "loss": 0.62, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.5286075949367088, |
| "grad_norm": 0.6983452046415175, |
| "learning_rate": 3.502509536445719e-05, |
| "loss": 0.606, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.5326582278481014, |
| "grad_norm": 0.6318725255141671, |
| "learning_rate": 3.498757348458147e-05, |
| "loss": 0.6352, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.5367088607594936, |
| "grad_norm": 0.6735957460490547, |
| "learning_rate": 3.4949930896849324e-05, |
| "loss": 0.6112, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.540759493670886, |
| "grad_norm": 0.6599738121060649, |
| "learning_rate": 3.491216790442899e-05, |
| "loss": 0.6289, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.5448101265822785, |
| "grad_norm": 0.6550500616546687, |
| "learning_rate": 3.487428481145839e-05, |
| "loss": 0.6081, |
| "step": 381 |
| }, |
| { |
| "epoch": 1.548860759493671, |
| "grad_norm": 0.690569691997852, |
| "learning_rate": 3.483628192304278e-05, |
| "loss": 0.6092, |
| "step": 382 |
| }, |
| { |
| "epoch": 1.5529113924050633, |
| "grad_norm": 0.6512693918034059, |
| "learning_rate": 3.479815954525219e-05, |
| "loss": 0.6329, |
| "step": 383 |
| }, |
| { |
| "epoch": 1.5569620253164556, |
| "grad_norm": 0.715928072155129, |
| "learning_rate": 3.475991798511899e-05, |
| "loss": 0.6333, |
| "step": 384 |
| }, |
| { |
| "epoch": 1.561012658227848, |
| "grad_norm": 0.6564140374565031, |
| "learning_rate": 3.4721557550635464e-05, |
| "loss": 0.6129, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.5650632911392406, |
| "grad_norm": 0.663999376528535, |
| "learning_rate": 3.468307855075128e-05, |
| "loss": 0.6035, |
| "step": 386 |
| }, |
| { |
| "epoch": 1.5691139240506329, |
| "grad_norm": 0.6591196343711756, |
| "learning_rate": 3.4644481295371005e-05, |
| "loss": 0.6085, |
| "step": 387 |
| }, |
| { |
| "epoch": 1.5731645569620252, |
| "grad_norm": 0.629910993507765, |
| "learning_rate": 3.460576609535163e-05, |
| "loss": 0.6178, |
| "step": 388 |
| }, |
| { |
| "epoch": 1.5772151898734177, |
| "grad_norm": 0.6738058856021697, |
| "learning_rate": 3.456693326250006e-05, |
| "loss": 0.606, |
| "step": 389 |
| }, |
| { |
| "epoch": 1.5812658227848102, |
| "grad_norm": 0.6232934751096325, |
| "learning_rate": 3.452798310957058e-05, |
| "loss": 0.6107, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.5853164556962025, |
| "grad_norm": 0.6036050515709918, |
| "learning_rate": 3.4488915950262386e-05, |
| "loss": 0.5975, |
| "step": 391 |
| }, |
| { |
| "epoch": 1.5893670886075948, |
| "grad_norm": 0.6539940094903018, |
| "learning_rate": 3.4449732099216985e-05, |
| "loss": 0.6065, |
| "step": 392 |
| }, |
| { |
| "epoch": 1.5934177215189873, |
| "grad_norm": 0.6379568591932163, |
| "learning_rate": 3.441043187201574e-05, |
| "loss": 0.6245, |
| "step": 393 |
| }, |
| { |
| "epoch": 1.5974683544303798, |
| "grad_norm": 0.6985877034513299, |
| "learning_rate": 3.437101558517728e-05, |
| "loss": 0.6189, |
| "step": 394 |
| }, |
| { |
| "epoch": 1.6015189873417721, |
| "grad_norm": 0.6315339702992091, |
| "learning_rate": 3.433148355615496e-05, |
| "loss": 0.6266, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.6055696202531644, |
| "grad_norm": 0.6581719193462288, |
| "learning_rate": 3.4291836103334294e-05, |
| "loss": 0.6429, |
| "step": 396 |
| }, |
| { |
| "epoch": 1.609620253164557, |
| "grad_norm": 0.6568098249169967, |
| "learning_rate": 3.425207354603043e-05, |
| "loss": 0.629, |
| "step": 397 |
| }, |
| { |
| "epoch": 1.6136708860759494, |
| "grad_norm": 0.6738915491166523, |
| "learning_rate": 3.421219620448553e-05, |
| "loss": 0.6063, |
| "step": 398 |
| }, |
| { |
| "epoch": 1.6177215189873417, |
| "grad_norm": 0.7469613500519527, |
| "learning_rate": 3.417220439986623e-05, |
| "loss": 0.5991, |
| "step": 399 |
| }, |
| { |
| "epoch": 1.621772151898734, |
| "grad_norm": 0.641116069391005, |
| "learning_rate": 3.4132098454261024e-05, |
| "loss": 0.6125, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.6258227848101265, |
| "grad_norm": 0.6746184640786532, |
| "learning_rate": 3.4091878690677676e-05, |
| "loss": 0.6145, |
| "step": 401 |
| }, |
| { |
| "epoch": 1.629873417721519, |
| "grad_norm": 0.7584617828160407, |
| "learning_rate": 3.405154543304065e-05, |
| "loss": 0.6082, |
| "step": 402 |
| }, |
| { |
| "epoch": 1.6339240506329114, |
| "grad_norm": 0.6877018779387423, |
| "learning_rate": 3.401109900618843e-05, |
| "loss": 0.6099, |
| "step": 403 |
| }, |
| { |
| "epoch": 1.6379746835443036, |
| "grad_norm": 0.7268279543509871, |
| "learning_rate": 3.3970539735870996e-05, |
| "loss": 0.6179, |
| "step": 404 |
| }, |
| { |
| "epoch": 1.6420253164556962, |
| "grad_norm": 0.6217279614559913, |
| "learning_rate": 3.392986794874714e-05, |
| "loss": 0.6114, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.6460759493670887, |
| "grad_norm": 0.6574203681751561, |
| "learning_rate": 3.388908397238184e-05, |
| "loss": 0.6309, |
| "step": 406 |
| }, |
| { |
| "epoch": 1.650126582278481, |
| "grad_norm": 0.7039093878693672, |
| "learning_rate": 3.384818813524362e-05, |
| "loss": 0.6086, |
| "step": 407 |
| }, |
| { |
| "epoch": 1.6541772151898733, |
| "grad_norm": 0.6153290618493474, |
| "learning_rate": 3.380718076670195e-05, |
| "loss": 0.5997, |
| "step": 408 |
| }, |
| { |
| "epoch": 1.6582278481012658, |
| "grad_norm": 0.6919087388849511, |
| "learning_rate": 3.376606219702454e-05, |
| "loss": 0.6168, |
| "step": 409 |
| }, |
| { |
| "epoch": 1.6622784810126583, |
| "grad_norm": 0.7088686763800891, |
| "learning_rate": 3.372483275737468e-05, |
| "loss": 0.6163, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.6663291139240506, |
| "grad_norm": 0.6301269791105767, |
| "learning_rate": 3.368349277980861e-05, |
| "loss": 0.635, |
| "step": 411 |
| }, |
| { |
| "epoch": 1.6703797468354429, |
| "grad_norm": 0.692547309699777, |
| "learning_rate": 3.3642042597272844e-05, |
| "loss": 0.6454, |
| "step": 412 |
| }, |
| { |
| "epoch": 1.6744303797468354, |
| "grad_norm": 0.7401212940881702, |
| "learning_rate": 3.360048254360144e-05, |
| "loss": 0.6189, |
| "step": 413 |
| }, |
| { |
| "epoch": 1.678481012658228, |
| "grad_norm": 0.6281309024778625, |
| "learning_rate": 3.355881295351336e-05, |
| "loss": 0.6245, |
| "step": 414 |
| }, |
| { |
| "epoch": 1.6825316455696202, |
| "grad_norm": 0.6785060542749777, |
| "learning_rate": 3.351703416260975e-05, |
| "loss": 0.6205, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.6865822784810125, |
| "grad_norm": 0.6458237410636927, |
| "learning_rate": 3.347514650737126e-05, |
| "loss": 0.6332, |
| "step": 416 |
| }, |
| { |
| "epoch": 1.690632911392405, |
| "grad_norm": 0.6715645668790162, |
| "learning_rate": 3.3433150325155295e-05, |
| "loss": 0.6285, |
| "step": 417 |
| }, |
| { |
| "epoch": 1.6946835443037975, |
| "grad_norm": 0.7329355220284975, |
| "learning_rate": 3.339104595419334e-05, |
| "loss": 0.6129, |
| "step": 418 |
| }, |
| { |
| "epoch": 1.6987341772151898, |
| "grad_norm": 0.6240431802224206, |
| "learning_rate": 3.3348833733588204e-05, |
| "loss": 0.6387, |
| "step": 419 |
| }, |
| { |
| "epoch": 1.7027848101265821, |
| "grad_norm": 0.6614335434835169, |
| "learning_rate": 3.3306514003311305e-05, |
| "loss": 0.6084, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7068354430379746, |
| "grad_norm": 0.6692989116822596, |
| "learning_rate": 3.326408710419996e-05, |
| "loss": 0.6331, |
| "step": 421 |
| }, |
| { |
| "epoch": 1.7108860759493671, |
| "grad_norm": 0.6518049634031123, |
| "learning_rate": 3.322155337795454e-05, |
| "loss": 0.6228, |
| "step": 422 |
| }, |
| { |
| "epoch": 1.7149367088607594, |
| "grad_norm": 0.6635845283127371, |
| "learning_rate": 3.317891316713587e-05, |
| "loss": 0.6338, |
| "step": 423 |
| }, |
| { |
| "epoch": 1.7189873417721517, |
| "grad_norm": 0.6403230662413673, |
| "learning_rate": 3.313616681516231e-05, |
| "loss": 0.6125, |
| "step": 424 |
| }, |
| { |
| "epoch": 1.7230379746835442, |
| "grad_norm": 0.6939557895777593, |
| "learning_rate": 3.309331466630713e-05, |
| "loss": 0.6336, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.7270886075949368, |
| "grad_norm": 0.6574918393731015, |
| "learning_rate": 3.305035706569563e-05, |
| "loss": 0.6092, |
| "step": 426 |
| }, |
| { |
| "epoch": 1.7311392405063293, |
| "grad_norm": 0.70951491378644, |
| "learning_rate": 3.3007294359302433e-05, |
| "loss": 0.6243, |
| "step": 427 |
| }, |
| { |
| "epoch": 1.7351898734177216, |
| "grad_norm": 0.6377161843675162, |
| "learning_rate": 3.296412689394864e-05, |
| "loss": 0.627, |
| "step": 428 |
| }, |
| { |
| "epoch": 1.7392405063291139, |
| "grad_norm": 0.7140057130926681, |
| "learning_rate": 3.292085501729909e-05, |
| "loss": 0.6472, |
| "step": 429 |
| }, |
| { |
| "epoch": 1.7432911392405064, |
| "grad_norm": 0.6617634157388314, |
| "learning_rate": 3.2877479077859534e-05, |
| "loss": 0.607, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.747341772151899, |
| "grad_norm": 0.6766796707545487, |
| "learning_rate": 3.283399942497381e-05, |
| "loss": 0.6341, |
| "step": 431 |
| }, |
| { |
| "epoch": 1.7513924050632912, |
| "grad_norm": 0.6686070310209404, |
| "learning_rate": 3.279041640882108e-05, |
| "loss": 0.6247, |
| "step": 432 |
| }, |
| { |
| "epoch": 1.7554430379746835, |
| "grad_norm": 0.6848770608811894, |
| "learning_rate": 3.2746730380412964e-05, |
| "loss": 0.6059, |
| "step": 433 |
| }, |
| { |
| "epoch": 1.759493670886076, |
| "grad_norm": 0.7077095773075113, |
| "learning_rate": 3.2702941691590726e-05, |
| "loss": 0.6033, |
| "step": 434 |
| }, |
| { |
| "epoch": 1.7635443037974685, |
| "grad_norm": 0.6366676892828314, |
| "learning_rate": 3.265905069502244e-05, |
| "loss": 0.6296, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.7675949367088608, |
| "grad_norm": 0.744940252933961, |
| "learning_rate": 3.261505774420016e-05, |
| "loss": 0.6067, |
| "step": 436 |
| }, |
| { |
| "epoch": 1.771645569620253, |
| "grad_norm": 0.6343669468900107, |
| "learning_rate": 3.257096319343707e-05, |
| "loss": 0.6039, |
| "step": 437 |
| }, |
| { |
| "epoch": 1.7756962025316456, |
| "grad_norm": 0.648506642216185, |
| "learning_rate": 3.2526767397864614e-05, |
| "loss": 0.6326, |
| "step": 438 |
| }, |
| { |
| "epoch": 1.7797468354430381, |
| "grad_norm": 0.6657838694702632, |
| "learning_rate": 3.248247071342966e-05, |
| "loss": 0.6343, |
| "step": 439 |
| }, |
| { |
| "epoch": 1.7837974683544304, |
| "grad_norm": 0.690633908866184, |
| "learning_rate": 3.243807349689161e-05, |
| "loss": 0.6271, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.7878481012658227, |
| "grad_norm": 0.6456310381256866, |
| "learning_rate": 3.2393576105819544e-05, |
| "loss": 0.6411, |
| "step": 441 |
| }, |
| { |
| "epoch": 1.7918987341772152, |
| "grad_norm": 0.6192077854908419, |
| "learning_rate": 3.2348978898589333e-05, |
| "loss": 0.6216, |
| "step": 442 |
| }, |
| { |
| "epoch": 1.7959493670886078, |
| "grad_norm": 0.6099896464714306, |
| "learning_rate": 3.230428223438075e-05, |
| "loss": 0.6089, |
| "step": 443 |
| }, |
| { |
| "epoch": 1.8, |
| "grad_norm": 0.7689555886761359, |
| "learning_rate": 3.225948647317459e-05, |
| "loss": 0.6154, |
| "step": 444 |
| }, |
| { |
| "epoch": 1.8040506329113923, |
| "grad_norm": 0.6643822580947245, |
| "learning_rate": 3.2214591975749745e-05, |
| "loss": 0.6188, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.8081012658227849, |
| "grad_norm": 0.7483973285763583, |
| "learning_rate": 3.216959910368034e-05, |
| "loss": 0.6401, |
| "step": 446 |
| }, |
| { |
| "epoch": 1.8121518987341774, |
| "grad_norm": 0.6559378623730281, |
| "learning_rate": 3.212450821933277e-05, |
| "loss": 0.6219, |
| "step": 447 |
| }, |
| { |
| "epoch": 1.8162025316455697, |
| "grad_norm": 0.6998485330761023, |
| "learning_rate": 3.207931968586281e-05, |
| "loss": 0.6385, |
| "step": 448 |
| }, |
| { |
| "epoch": 1.820253164556962, |
| "grad_norm": 0.6401854676392357, |
| "learning_rate": 3.203403386721272e-05, |
| "loss": 0.6194, |
| "step": 449 |
| }, |
| { |
| "epoch": 1.8243037974683545, |
| "grad_norm": 0.6268936068602418, |
| "learning_rate": 3.1988651128108245e-05, |
| "loss": 0.6137, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.828354430379747, |
| "grad_norm": 0.633513270627628, |
| "learning_rate": 3.194317183405573e-05, |
| "loss": 0.6183, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.8324050632911393, |
| "grad_norm": 0.6490867071339926, |
| "learning_rate": 3.189759635133914e-05, |
| "loss": 0.6193, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.8364556962025316, |
| "grad_norm": 0.6337972205715314, |
| "learning_rate": 3.185192504701718e-05, |
| "loss": 0.6289, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.840506329113924, |
| "grad_norm": 0.6567050871274642, |
| "learning_rate": 3.1806158288920234e-05, |
| "loss": 0.6405, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.8445569620253166, |
| "grad_norm": 0.6519735174761796, |
| "learning_rate": 3.1760296445647477e-05, |
| "loss": 0.6174, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.848607594936709, |
| "grad_norm": 0.6294170787472738, |
| "learning_rate": 3.1714339886563896e-05, |
| "loss": 0.6085, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.8526582278481012, |
| "grad_norm": 0.6736357583201322, |
| "learning_rate": 3.166828898179731e-05, |
| "loss": 0.6279, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.8567088607594937, |
| "grad_norm": 0.6506765922489426, |
| "learning_rate": 3.162214410223536e-05, |
| "loss": 0.6218, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.8607594936708862, |
| "grad_norm": 0.6930436527540442, |
| "learning_rate": 3.157590561952257e-05, |
| "loss": 0.5886, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.8648101265822785, |
| "grad_norm": 0.694400234927393, |
| "learning_rate": 3.152957390605732e-05, |
| "loss": 0.6117, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.8688607594936708, |
| "grad_norm": 1.5553544478484371, |
| "learning_rate": 3.148314933498886e-05, |
| "loss": 0.621, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.8729113924050633, |
| "grad_norm": 0.7092860012592993, |
| "learning_rate": 3.143663228021431e-05, |
| "loss": 0.5969, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.8769620253164558, |
| "grad_norm": 0.681844666451108, |
| "learning_rate": 3.1390023116375624e-05, |
| "loss": 0.6056, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.8810126582278481, |
| "grad_norm": 0.7087966610295723, |
| "learning_rate": 3.134332221885661e-05, |
| "loss": 0.6142, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.8850632911392404, |
| "grad_norm": 0.6935740296054392, |
| "learning_rate": 3.129652996377987e-05, |
| "loss": 0.6398, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.889113924050633, |
| "grad_norm": 0.6370638090976234, |
| "learning_rate": 3.12496467280038e-05, |
| "loss": 0.6273, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.8931645569620255, |
| "grad_norm": 0.6870763397035786, |
| "learning_rate": 3.120267288911952e-05, |
| "loss": 0.6268, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.8972151898734178, |
| "grad_norm": 0.6381340988975772, |
| "learning_rate": 3.11556088254479e-05, |
| "loss": 0.6224, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.90126582278481, |
| "grad_norm": 0.658479972183502, |
| "learning_rate": 3.11084549160364e-05, |
| "loss": 0.6129, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.9053164556962026, |
| "grad_norm": 0.6745457241143648, |
| "learning_rate": 3.106121154065615e-05, |
| "loss": 0.6264, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.909367088607595, |
| "grad_norm": 0.6674138735683669, |
| "learning_rate": 3.1013879079798805e-05, |
| "loss": 0.6416, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.9134177215189874, |
| "grad_norm": 0.645435654135225, |
| "learning_rate": 3.096645791467348e-05, |
| "loss": 0.6101, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.9174683544303797, |
| "grad_norm": 0.6245397941927614, |
| "learning_rate": 3.091894842720373e-05, |
| "loss": 0.6227, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.9215189873417722, |
| "grad_norm": 0.6384257215896536, |
| "learning_rate": 3.0871351000024425e-05, |
| "loss": 0.6251, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.9255696202531647, |
| "grad_norm": 0.6564857408336725, |
| "learning_rate": 3.0823666016478716e-05, |
| "loss": 0.6424, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.929620253164557, |
| "grad_norm": 0.6377542608115342, |
| "learning_rate": 3.0775893860614896e-05, |
| "loss": 0.6433, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.9336708860759493, |
| "grad_norm": 0.664772447718219, |
| "learning_rate": 3.0728034917183336e-05, |
| "loss": 0.6353, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.9377215189873418, |
| "grad_norm": 0.6264755472858282, |
| "learning_rate": 3.06800895716334e-05, |
| "loss": 0.6251, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.9417721518987343, |
| "grad_norm": 0.6678374512525068, |
| "learning_rate": 3.063205821011029e-05, |
| "loss": 0.6162, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.9458227848101266, |
| "grad_norm": 0.658053254626636, |
| "learning_rate": 3.0583941219452016e-05, |
| "loss": 0.6365, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.949873417721519, |
| "grad_norm": 0.6237603843196734, |
| "learning_rate": 3.053573898718618e-05, |
| "loss": 0.6206, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.9539240506329114, |
| "grad_norm": 0.6520821472624212, |
| "learning_rate": 3.0487451901526956e-05, |
| "loss": 0.6213, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.957974683544304, |
| "grad_norm": 0.6400611882125555, |
| "learning_rate": 3.0439080351371875e-05, |
| "loss": 0.6255, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.9620253164556962, |
| "grad_norm": 0.6518713033852864, |
| "learning_rate": 3.0390624726298764e-05, |
| "loss": 0.6092, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.9660759493670885, |
| "grad_norm": 0.6330187494680329, |
| "learning_rate": 3.034208541656255e-05, |
| "loss": 0.6261, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.970126582278481, |
| "grad_norm": 0.6726453474482517, |
| "learning_rate": 3.029346281309218e-05, |
| "loss": 0.6009, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.9741772151898735, |
| "grad_norm": 0.6431806277874796, |
| "learning_rate": 3.0244757307487415e-05, |
| "loss": 0.625, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.9782278481012658, |
| "grad_norm": 0.6108476926602743, |
| "learning_rate": 3.019596929201569e-05, |
| "loss": 0.6297, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.9822784810126581, |
| "grad_norm": 0.6600124387583248, |
| "learning_rate": 3.0147099159608985e-05, |
| "loss": 0.6131, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.9863291139240506, |
| "grad_norm": 0.7532452759761153, |
| "learning_rate": 3.0098147303860616e-05, |
| "loss": 0.6182, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.9903797468354432, |
| "grad_norm": 0.6559877409184733, |
| "learning_rate": 3.0049114119022117e-05, |
| "loss": 0.6025, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.9944303797468355, |
| "grad_norm": 0.6514699555626042, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.6125, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.9984810126582278, |
| "grad_norm": 0.6561015505564005, |
| "learning_rate": 2.995080534235264e-05, |
| "loss": 0.617, |
| "step": 493 |
| }, |
| { |
| "epoch": 2.0040506329113925, |
| "grad_norm": 1.1292592988423966, |
| "learning_rate": 2.9901530542287044e-05, |
| "loss": 0.3843, |
| "step": 494 |
| }, |
| { |
| "epoch": 2.008101265822785, |
| "grad_norm": 0.6202369074453559, |
| "learning_rate": 2.9852175996655676e-05, |
| "loss": 0.3551, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.012151898734177, |
| "grad_norm": 1.6354267809082748, |
| "learning_rate": 2.980274210295326e-05, |
| "loss": 0.3803, |
| "step": 496 |
| }, |
| { |
| "epoch": 2.0162025316455696, |
| "grad_norm": 1.0732169983695043, |
| "learning_rate": 2.9753229259313578e-05, |
| "loss": 0.3425, |
| "step": 497 |
| }, |
| { |
| "epoch": 2.020253164556962, |
| "grad_norm": 0.6717958509518657, |
| "learning_rate": 2.9703637864506274e-05, |
| "loss": 0.3609, |
| "step": 498 |
| }, |
| { |
| "epoch": 2.0243037974683546, |
| "grad_norm": 0.7740117385398941, |
| "learning_rate": 2.965396831793362e-05, |
| "loss": 0.3743, |
| "step": 499 |
| }, |
| { |
| "epoch": 2.0283544303797467, |
| "grad_norm": 0.702895884995145, |
| "learning_rate": 2.9604221019627316e-05, |
| "loss": 0.3482, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0324050632911392, |
| "grad_norm": 0.635025021379264, |
| "learning_rate": 2.955439637024526e-05, |
| "loss": 0.3543, |
| "step": 501 |
| }, |
| { |
| "epoch": 2.0364556962025318, |
| "grad_norm": 0.6557015786806156, |
| "learning_rate": 2.9504494771068334e-05, |
| "loss": 0.354, |
| "step": 502 |
| }, |
| { |
| "epoch": 2.0405063291139243, |
| "grad_norm": 0.6923260106193392, |
| "learning_rate": 2.9454516623997156e-05, |
| "loss": 0.3217, |
| "step": 503 |
| }, |
| { |
| "epoch": 2.0445569620253163, |
| "grad_norm": 0.6974043330195073, |
| "learning_rate": 2.9404462331548847e-05, |
| "loss": 0.3651, |
| "step": 504 |
| }, |
| { |
| "epoch": 2.048607594936709, |
| "grad_norm": 0.6834086561115056, |
| "learning_rate": 2.93543322968538e-05, |
| "loss": 0.3419, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.0526582278481014, |
| "grad_norm": 0.6257099746046118, |
| "learning_rate": 2.9304126923652428e-05, |
| "loss": 0.3373, |
| "step": 506 |
| }, |
| { |
| "epoch": 2.056708860759494, |
| "grad_norm": 0.698427255899178, |
| "learning_rate": 2.9253846616291896e-05, |
| "loss": 0.343, |
| "step": 507 |
| }, |
| { |
| "epoch": 2.060759493670886, |
| "grad_norm": 0.6186456611694618, |
| "learning_rate": 2.9203491779722896e-05, |
| "loss": 0.3502, |
| "step": 508 |
| }, |
| { |
| "epoch": 2.0648101265822785, |
| "grad_norm": 0.647954542891851, |
| "learning_rate": 2.9153062819496357e-05, |
| "loss": 0.3498, |
| "step": 509 |
| }, |
| { |
| "epoch": 2.068860759493671, |
| "grad_norm": 0.6046681746287332, |
| "learning_rate": 2.9102560141760178e-05, |
| "loss": 0.3473, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.0729113924050635, |
| "grad_norm": 0.6699710104472597, |
| "learning_rate": 2.9051984153256004e-05, |
| "loss": 0.3517, |
| "step": 511 |
| }, |
| { |
| "epoch": 2.0769620253164556, |
| "grad_norm": 0.5838814248038889, |
| "learning_rate": 2.900133526131588e-05, |
| "loss": 0.3501, |
| "step": 512 |
| }, |
| { |
| "epoch": 2.081012658227848, |
| "grad_norm": 0.6761892766012106, |
| "learning_rate": 2.8950613873859025e-05, |
| "loss": 0.3566, |
| "step": 513 |
| }, |
| { |
| "epoch": 2.0850632911392406, |
| "grad_norm": 0.6041613125046124, |
| "learning_rate": 2.8899820399388515e-05, |
| "loss": 0.3524, |
| "step": 514 |
| }, |
| { |
| "epoch": 2.089113924050633, |
| "grad_norm": 0.6155059870632867, |
| "learning_rate": 2.8848955246988012e-05, |
| "loss": 0.3471, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.093164556962025, |
| "grad_norm": 0.610531220934419, |
| "learning_rate": 2.879801882631847e-05, |
| "loss": 0.3354, |
| "step": 516 |
| }, |
| { |
| "epoch": 2.0972151898734177, |
| "grad_norm": 0.5871144355336497, |
| "learning_rate": 2.8747011547614808e-05, |
| "loss": 0.3437, |
| "step": 517 |
| }, |
| { |
| "epoch": 2.1012658227848102, |
| "grad_norm": 0.6378624610494644, |
| "learning_rate": 2.8695933821682635e-05, |
| "loss": 0.3449, |
| "step": 518 |
| }, |
| { |
| "epoch": 2.1053164556962027, |
| "grad_norm": 0.6073425846058068, |
| "learning_rate": 2.864478605989494e-05, |
| "loss": 0.3519, |
| "step": 519 |
| }, |
| { |
| "epoch": 2.109367088607595, |
| "grad_norm": 0.6016157740941189, |
| "learning_rate": 2.8593568674188765e-05, |
| "loss": 0.3319, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.1134177215189873, |
| "grad_norm": 0.619647909427643, |
| "learning_rate": 2.8542282077061892e-05, |
| "loss": 0.3336, |
| "step": 521 |
| }, |
| { |
| "epoch": 2.11746835443038, |
| "grad_norm": 0.6044337634701333, |
| "learning_rate": 2.8490926681569523e-05, |
| "loss": 0.3532, |
| "step": 522 |
| }, |
| { |
| "epoch": 2.1215189873417724, |
| "grad_norm": 0.5631112364832452, |
| "learning_rate": 2.8439502901320956e-05, |
| "loss": 0.3552, |
| "step": 523 |
| }, |
| { |
| "epoch": 2.1255696202531644, |
| "grad_norm": 0.6137347119321382, |
| "learning_rate": 2.8388011150476237e-05, |
| "loss": 0.3513, |
| "step": 524 |
| }, |
| { |
| "epoch": 2.129620253164557, |
| "grad_norm": 0.586939840644733, |
| "learning_rate": 2.8336451843742866e-05, |
| "loss": 0.3479, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.1336708860759495, |
| "grad_norm": 0.6085741472891468, |
| "learning_rate": 2.8284825396372387e-05, |
| "loss": 0.3444, |
| "step": 526 |
| }, |
| { |
| "epoch": 2.137721518987342, |
| "grad_norm": 0.5850435389444315, |
| "learning_rate": 2.8233132224157132e-05, |
| "loss": 0.3384, |
| "step": 527 |
| }, |
| { |
| "epoch": 2.141772151898734, |
| "grad_norm": 0.6497951153847392, |
| "learning_rate": 2.8181372743426805e-05, |
| "loss": 0.3379, |
| "step": 528 |
| }, |
| { |
| "epoch": 2.1458227848101266, |
| "grad_norm": 0.5900528177839782, |
| "learning_rate": 2.8129547371045128e-05, |
| "loss": 0.3624, |
| "step": 529 |
| }, |
| { |
| "epoch": 2.149873417721519, |
| "grad_norm": 0.6196359947649575, |
| "learning_rate": 2.8077656524406534e-05, |
| "loss": 0.3347, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.1539240506329116, |
| "grad_norm": 0.5995914300748625, |
| "learning_rate": 2.802570062143278e-05, |
| "loss": 0.3419, |
| "step": 531 |
| }, |
| { |
| "epoch": 2.1579746835443037, |
| "grad_norm": 0.5705158488518188, |
| "learning_rate": 2.7973680080569555e-05, |
| "loss": 0.3545, |
| "step": 532 |
| }, |
| { |
| "epoch": 2.162025316455696, |
| "grad_norm": 0.6250084024065977, |
| "learning_rate": 2.792159532078314e-05, |
| "loss": 0.3623, |
| "step": 533 |
| }, |
| { |
| "epoch": 2.1660759493670887, |
| "grad_norm": 0.5566098344295728, |
| "learning_rate": 2.7869446761557033e-05, |
| "loss": 0.3611, |
| "step": 534 |
| }, |
| { |
| "epoch": 2.170126582278481, |
| "grad_norm": 0.6259001872388088, |
| "learning_rate": 2.781723482288857e-05, |
| "loss": 0.3705, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.1741772151898733, |
| "grad_norm": 0.553956029793548, |
| "learning_rate": 2.7764959925285517e-05, |
| "loss": 0.3568, |
| "step": 536 |
| }, |
| { |
| "epoch": 2.178227848101266, |
| "grad_norm": 0.5965780139151817, |
| "learning_rate": 2.771262248976272e-05, |
| "loss": 0.3419, |
| "step": 537 |
| }, |
| { |
| "epoch": 2.1822784810126583, |
| "grad_norm": 0.5602829550440678, |
| "learning_rate": 2.7660222937838677e-05, |
| "loss": 0.3606, |
| "step": 538 |
| }, |
| { |
| "epoch": 2.186329113924051, |
| "grad_norm": 0.5568711337769925, |
| "learning_rate": 2.7607761691532186e-05, |
| "loss": 0.3438, |
| "step": 539 |
| }, |
| { |
| "epoch": 2.190379746835443, |
| "grad_norm": 0.581608541993894, |
| "learning_rate": 2.7555239173358916e-05, |
| "loss": 0.3334, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.1944303797468354, |
| "grad_norm": 0.5899403275198342, |
| "learning_rate": 2.7502655806328e-05, |
| "loss": 0.3393, |
| "step": 541 |
| }, |
| { |
| "epoch": 2.198481012658228, |
| "grad_norm": 0.5969318095295478, |
| "learning_rate": 2.7450012013938648e-05, |
| "loss": 0.3568, |
| "step": 542 |
| }, |
| { |
| "epoch": 2.2025316455696204, |
| "grad_norm": 0.5821813313989359, |
| "learning_rate": 2.739730822017673e-05, |
| "loss": 0.3431, |
| "step": 543 |
| }, |
| { |
| "epoch": 2.2065822784810125, |
| "grad_norm": 0.6239840386647869, |
| "learning_rate": 2.7344544849511355e-05, |
| "loss": 0.3535, |
| "step": 544 |
| }, |
| { |
| "epoch": 2.210632911392405, |
| "grad_norm": 0.5901092489529592, |
| "learning_rate": 2.7291722326891456e-05, |
| "loss": 0.354, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.2146835443037975, |
| "grad_norm": 0.5824308686825536, |
| "learning_rate": 2.723884107774236e-05, |
| "loss": 0.337, |
| "step": 546 |
| }, |
| { |
| "epoch": 2.21873417721519, |
| "grad_norm": 0.649410382730811, |
| "learning_rate": 2.718590152796239e-05, |
| "loss": 0.354, |
| "step": 547 |
| }, |
| { |
| "epoch": 2.222784810126582, |
| "grad_norm": 0.6027871374148092, |
| "learning_rate": 2.71329041039194e-05, |
| "loss": 0.3374, |
| "step": 548 |
| }, |
| { |
| "epoch": 2.2268354430379746, |
| "grad_norm": 0.6171606517504709, |
| "learning_rate": 2.7079849232447357e-05, |
| "loss": 0.3406, |
| "step": 549 |
| }, |
| { |
| "epoch": 2.230886075949367, |
| "grad_norm": 0.6197202041358174, |
| "learning_rate": 2.7026737340842895e-05, |
| "loss": 0.3356, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.2349367088607597, |
| "grad_norm": 0.6331205281718484, |
| "learning_rate": 2.697356885686189e-05, |
| "loss": 0.3663, |
| "step": 551 |
| }, |
| { |
| "epoch": 2.2389873417721518, |
| "grad_norm": 0.6182120945152672, |
| "learning_rate": 2.6920344208716014e-05, |
| "loss": 0.3446, |
| "step": 552 |
| }, |
| { |
| "epoch": 2.2430379746835443, |
| "grad_norm": 0.5871104932969851, |
| "learning_rate": 2.6867063825069252e-05, |
| "loss": 0.3535, |
| "step": 553 |
| }, |
| { |
| "epoch": 2.247088607594937, |
| "grad_norm": 0.6353373953233129, |
| "learning_rate": 2.6813728135034494e-05, |
| "loss": 0.3655, |
| "step": 554 |
| }, |
| { |
| "epoch": 2.2511392405063293, |
| "grad_norm": 0.5765376499237801, |
| "learning_rate": 2.6760337568170056e-05, |
| "loss": 0.3337, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.2551898734177214, |
| "grad_norm": 0.5986602909539711, |
| "learning_rate": 2.6706892554476226e-05, |
| "loss": 0.3364, |
| "step": 556 |
| }, |
| { |
| "epoch": 2.259240506329114, |
| "grad_norm": 0.6441552368488795, |
| "learning_rate": 2.6653393524391795e-05, |
| "loss": 0.3437, |
| "step": 557 |
| }, |
| { |
| "epoch": 2.2632911392405064, |
| "grad_norm": 0.5866369619127298, |
| "learning_rate": 2.6599840908790592e-05, |
| "loss": 0.3415, |
| "step": 558 |
| }, |
| { |
| "epoch": 2.267341772151899, |
| "grad_norm": 0.6148037575189962, |
| "learning_rate": 2.6546235138978028e-05, |
| "loss": 0.3567, |
| "step": 559 |
| }, |
| { |
| "epoch": 2.271392405063291, |
| "grad_norm": 0.5738735509513457, |
| "learning_rate": 2.6492576646687597e-05, |
| "loss": 0.3518, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.2754430379746835, |
| "grad_norm": 0.6235818751358062, |
| "learning_rate": 2.6438865864077425e-05, |
| "loss": 0.3582, |
| "step": 561 |
| }, |
| { |
| "epoch": 2.279493670886076, |
| "grad_norm": 0.5988620413243119, |
| "learning_rate": 2.6385103223726766e-05, |
| "loss": 0.3554, |
| "step": 562 |
| }, |
| { |
| "epoch": 2.2835443037974685, |
| "grad_norm": 0.5957983246232195, |
| "learning_rate": 2.6331289158632537e-05, |
| "loss": 0.3521, |
| "step": 563 |
| }, |
| { |
| "epoch": 2.2875949367088606, |
| "grad_norm": 0.6051197678734996, |
| "learning_rate": 2.6277424102205817e-05, |
| "loss": 0.3409, |
| "step": 564 |
| }, |
| { |
| "epoch": 2.291645569620253, |
| "grad_norm": 0.5950658552710962, |
| "learning_rate": 2.6223508488268374e-05, |
| "loss": 0.3474, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.2956962025316456, |
| "grad_norm": 0.5872201893510471, |
| "learning_rate": 2.6169542751049148e-05, |
| "loss": 0.355, |
| "step": 566 |
| }, |
| { |
| "epoch": 2.299746835443038, |
| "grad_norm": 0.6199475336642354, |
| "learning_rate": 2.6115527325180754e-05, |
| "loss": 0.3506, |
| "step": 567 |
| }, |
| { |
| "epoch": 2.3037974683544302, |
| "grad_norm": 0.6161687829284143, |
| "learning_rate": 2.606146264569603e-05, |
| "loss": 0.3406, |
| "step": 568 |
| }, |
| { |
| "epoch": 2.3078481012658227, |
| "grad_norm": 0.6152280739823517, |
| "learning_rate": 2.6007349148024447e-05, |
| "loss": 0.3476, |
| "step": 569 |
| }, |
| { |
| "epoch": 2.3118987341772153, |
| "grad_norm": 0.6168941353372234, |
| "learning_rate": 2.5953187267988694e-05, |
| "loss": 0.3281, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.3159493670886078, |
| "grad_norm": 0.6047632932271637, |
| "learning_rate": 2.5898977441801097e-05, |
| "loss": 0.3698, |
| "step": 571 |
| }, |
| { |
| "epoch": 2.32, |
| "grad_norm": 0.6229079279659953, |
| "learning_rate": 2.584472010606015e-05, |
| "loss": 0.3637, |
| "step": 572 |
| }, |
| { |
| "epoch": 2.3240506329113924, |
| "grad_norm": 0.5973319014930505, |
| "learning_rate": 2.5790415697746976e-05, |
| "loss": 0.36, |
| "step": 573 |
| }, |
| { |
| "epoch": 2.328101265822785, |
| "grad_norm": 0.6250366224441422, |
| "learning_rate": 2.5736064654221808e-05, |
| "loss": 0.3595, |
| "step": 574 |
| }, |
| { |
| "epoch": 2.3321518987341774, |
| "grad_norm": 0.6115867862925441, |
| "learning_rate": 2.568166741322048e-05, |
| "loss": 0.3528, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.3362025316455695, |
| "grad_norm": 0.6335125432659232, |
| "learning_rate": 2.56272244128509e-05, |
| "loss": 0.3501, |
| "step": 576 |
| }, |
| { |
| "epoch": 2.340253164556962, |
| "grad_norm": 0.5981262604362354, |
| "learning_rate": 2.55727360915895e-05, |
| "loss": 0.3529, |
| "step": 577 |
| }, |
| { |
| "epoch": 2.3443037974683545, |
| "grad_norm": 0.6043312043948148, |
| "learning_rate": 2.5518202888277734e-05, |
| "loss": 0.3414, |
| "step": 578 |
| }, |
| { |
| "epoch": 2.348354430379747, |
| "grad_norm": 0.6006032212526693, |
| "learning_rate": 2.5463625242118523e-05, |
| "loss": 0.3719, |
| "step": 579 |
| }, |
| { |
| "epoch": 2.352405063291139, |
| "grad_norm": 0.6199024499567237, |
| "learning_rate": 2.5409003592672723e-05, |
| "loss": 0.3359, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.3564556962025316, |
| "grad_norm": 0.5742379880661546, |
| "learning_rate": 2.535433837985559e-05, |
| "loss": 0.3438, |
| "step": 581 |
| }, |
| { |
| "epoch": 2.360506329113924, |
| "grad_norm": 0.6381646123718941, |
| "learning_rate": 2.529963004393324e-05, |
| "loss": 0.3427, |
| "step": 582 |
| }, |
| { |
| "epoch": 2.3645569620253166, |
| "grad_norm": 0.597811690867079, |
| "learning_rate": 2.524487902551908e-05, |
| "loss": 0.3565, |
| "step": 583 |
| }, |
| { |
| "epoch": 2.3686075949367087, |
| "grad_norm": 0.6095628434480715, |
| "learning_rate": 2.519008576557029e-05, |
| "loss": 0.3552, |
| "step": 584 |
| }, |
| { |
| "epoch": 2.372658227848101, |
| "grad_norm": 0.5976238667283068, |
| "learning_rate": 2.5135250705384254e-05, |
| "loss": 0.3555, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.3767088607594937, |
| "grad_norm": 0.5947480901819823, |
| "learning_rate": 2.5080374286595007e-05, |
| "loss": 0.3587, |
| "step": 586 |
| }, |
| { |
| "epoch": 2.3807594936708862, |
| "grad_norm": 0.6126461573403814, |
| "learning_rate": 2.5025456951169677e-05, |
| "loss": 0.3602, |
| "step": 587 |
| }, |
| { |
| "epoch": 2.3848101265822783, |
| "grad_norm": 0.5849889958997857, |
| "learning_rate": 2.4970499141404942e-05, |
| "loss": 0.3569, |
| "step": 588 |
| }, |
| { |
| "epoch": 2.388860759493671, |
| "grad_norm": 0.6065054603321423, |
| "learning_rate": 2.491550129992345e-05, |
| "loss": 0.357, |
| "step": 589 |
| }, |
| { |
| "epoch": 2.3929113924050633, |
| "grad_norm": 0.6076949529093999, |
| "learning_rate": 2.486046386967024e-05, |
| "loss": 0.3597, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.396962025316456, |
| "grad_norm": 0.6099321060158417, |
| "learning_rate": 2.4805387293909214e-05, |
| "loss": 0.3522, |
| "step": 591 |
| }, |
| { |
| "epoch": 2.401012658227848, |
| "grad_norm": 0.5872636879588808, |
| "learning_rate": 2.4750272016219552e-05, |
| "loss": 0.3451, |
| "step": 592 |
| }, |
| { |
| "epoch": 2.4050632911392404, |
| "grad_norm": 0.6042633672671855, |
| "learning_rate": 2.4695118480492114e-05, |
| "loss": 0.3587, |
| "step": 593 |
| }, |
| { |
| "epoch": 2.409113924050633, |
| "grad_norm": 0.6041213531228166, |
| "learning_rate": 2.4639927130925898e-05, |
| "loss": 0.3657, |
| "step": 594 |
| }, |
| { |
| "epoch": 2.4131645569620255, |
| "grad_norm": 0.6040171213254879, |
| "learning_rate": 2.458469841202444e-05, |
| "loss": 0.3313, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.4172151898734175, |
| "grad_norm": 0.5945306777083931, |
| "learning_rate": 2.452943276859226e-05, |
| "loss": 0.3584, |
| "step": 596 |
| }, |
| { |
| "epoch": 2.42126582278481, |
| "grad_norm": 0.6054654618236216, |
| "learning_rate": 2.447413064573125e-05, |
| "loss": 0.339, |
| "step": 597 |
| }, |
| { |
| "epoch": 2.4253164556962026, |
| "grad_norm": 0.5926953998154738, |
| "learning_rate": 2.4418792488837095e-05, |
| "loss": 0.363, |
| "step": 598 |
| }, |
| { |
| "epoch": 2.429367088607595, |
| "grad_norm": 0.5973584213610953, |
| "learning_rate": 2.4363418743595713e-05, |
| "loss": 0.3309, |
| "step": 599 |
| }, |
| { |
| "epoch": 2.433417721518987, |
| "grad_norm": 0.6234700911644003, |
| "learning_rate": 2.430800985597963e-05, |
| "loss": 0.351, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4374683544303797, |
| "grad_norm": 0.6118747104113249, |
| "learning_rate": 2.4252566272244415e-05, |
| "loss": 0.3617, |
| "step": 601 |
| }, |
| { |
| "epoch": 2.441518987341772, |
| "grad_norm": 0.5864316411707934, |
| "learning_rate": 2.4197088438925063e-05, |
| "loss": 0.348, |
| "step": 602 |
| }, |
| { |
| "epoch": 2.4455696202531647, |
| "grad_norm": 0.6373129232923905, |
| "learning_rate": 2.4141576802832417e-05, |
| "loss": 0.3479, |
| "step": 603 |
| }, |
| { |
| "epoch": 2.449620253164557, |
| "grad_norm": 0.5817262717786971, |
| "learning_rate": 2.408603181104957e-05, |
| "loss": 0.3617, |
| "step": 604 |
| }, |
| { |
| "epoch": 2.4536708860759493, |
| "grad_norm": 0.5989873865461219, |
| "learning_rate": 2.4030453910928245e-05, |
| "loss": 0.3561, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.457721518987342, |
| "grad_norm": 0.6150839662311799, |
| "learning_rate": 2.397484355008521e-05, |
| "loss": 0.3685, |
| "step": 606 |
| }, |
| { |
| "epoch": 2.4617721518987343, |
| "grad_norm": 0.5778410027873472, |
| "learning_rate": 2.3919201176398662e-05, |
| "loss": 0.3656, |
| "step": 607 |
| }, |
| { |
| "epoch": 2.4658227848101264, |
| "grad_norm": 0.598952356323796, |
| "learning_rate": 2.3863527238004633e-05, |
| "loss": 0.3545, |
| "step": 608 |
| }, |
| { |
| "epoch": 2.469873417721519, |
| "grad_norm": 0.603002483959968, |
| "learning_rate": 2.380782218329337e-05, |
| "loss": 0.3526, |
| "step": 609 |
| }, |
| { |
| "epoch": 2.4739240506329114, |
| "grad_norm": 0.5906737918895564, |
| "learning_rate": 2.3752086460905725e-05, |
| "loss": 0.3608, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.477974683544304, |
| "grad_norm": 0.6281474700920565, |
| "learning_rate": 2.3696320519729544e-05, |
| "loss": 0.3485, |
| "step": 611 |
| }, |
| { |
| "epoch": 2.482025316455696, |
| "grad_norm": 0.5914638481297856, |
| "learning_rate": 2.3640524808896045e-05, |
| "loss": 0.3472, |
| "step": 612 |
| }, |
| { |
| "epoch": 2.4860759493670885, |
| "grad_norm": 0.5859002898923953, |
| "learning_rate": 2.3584699777776222e-05, |
| "loss": 0.3571, |
| "step": 613 |
| }, |
| { |
| "epoch": 2.490126582278481, |
| "grad_norm": 0.5713121079453284, |
| "learning_rate": 2.3528845875977195e-05, |
| "loss": 0.3529, |
| "step": 614 |
| }, |
| { |
| "epoch": 2.4941772151898736, |
| "grad_norm": 0.5941609190086, |
| "learning_rate": 2.3472963553338614e-05, |
| "loss": 0.3664, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.4982278481012656, |
| "grad_norm": 0.5641207173193086, |
| "learning_rate": 2.341705325992901e-05, |
| "loss": 0.3436, |
| "step": 616 |
| }, |
| { |
| "epoch": 2.502278481012658, |
| "grad_norm": 0.5800338728863861, |
| "learning_rate": 2.336111544604222e-05, |
| "loss": 0.3561, |
| "step": 617 |
| }, |
| { |
| "epoch": 2.5063291139240507, |
| "grad_norm": 0.6074160321050159, |
| "learning_rate": 2.33051505621937e-05, |
| "loss": 0.3509, |
| "step": 618 |
| }, |
| { |
| "epoch": 2.510379746835443, |
| "grad_norm": 0.582270646634559, |
| "learning_rate": 2.324915905911693e-05, |
| "loss": 0.3413, |
| "step": 619 |
| }, |
| { |
| "epoch": 2.5144303797468357, |
| "grad_norm": 0.5996849191370386, |
| "learning_rate": 2.319314138775977e-05, |
| "loss": 0.35, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.5184810126582278, |
| "grad_norm": 0.5612657320288447, |
| "learning_rate": 2.3137097999280856e-05, |
| "loss": 0.342, |
| "step": 621 |
| }, |
| { |
| "epoch": 2.5225316455696203, |
| "grad_norm": 0.561640516466556, |
| "learning_rate": 2.308102934504593e-05, |
| "loss": 0.3655, |
| "step": 622 |
| }, |
| { |
| "epoch": 2.526582278481013, |
| "grad_norm": 0.5995140063310823, |
| "learning_rate": 2.3024935876624222e-05, |
| "loss": 0.3214, |
| "step": 623 |
| }, |
| { |
| "epoch": 2.530632911392405, |
| "grad_norm": 0.5537281469025281, |
| "learning_rate": 2.2968818045784813e-05, |
| "loss": 0.3447, |
| "step": 624 |
| }, |
| { |
| "epoch": 2.5346835443037974, |
| "grad_norm": 0.5806768448859754, |
| "learning_rate": 2.2912676304493006e-05, |
| "loss": 0.3473, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.53873417721519, |
| "grad_norm": 0.5927146326200241, |
| "learning_rate": 2.2856511104906668e-05, |
| "loss": 0.3481, |
| "step": 626 |
| }, |
| { |
| "epoch": 2.5427848101265824, |
| "grad_norm": 0.5809237430765053, |
| "learning_rate": 2.2800322899372586e-05, |
| "loss": 0.3377, |
| "step": 627 |
| }, |
| { |
| "epoch": 2.546835443037975, |
| "grad_norm": 0.5812455382508629, |
| "learning_rate": 2.2744112140422844e-05, |
| "loss": 0.3546, |
| "step": 628 |
| }, |
| { |
| "epoch": 2.550886075949367, |
| "grad_norm": 0.5895776475510459, |
| "learning_rate": 2.2687879280771177e-05, |
| "loss": 0.3511, |
| "step": 629 |
| }, |
| { |
| "epoch": 2.5549367088607595, |
| "grad_norm": 0.5851153694265145, |
| "learning_rate": 2.26316247733093e-05, |
| "loss": 0.334, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.558987341772152, |
| "grad_norm": 0.5593699232182908, |
| "learning_rate": 2.257534907110328e-05, |
| "loss": 0.3486, |
| "step": 631 |
| }, |
| { |
| "epoch": 2.563037974683544, |
| "grad_norm": 0.577453227320582, |
| "learning_rate": 2.2519052627389882e-05, |
| "loss": 0.3574, |
| "step": 632 |
| }, |
| { |
| "epoch": 2.5670886075949366, |
| "grad_norm": 0.5661809474505712, |
| "learning_rate": 2.246273589557294e-05, |
| "loss": 0.3598, |
| "step": 633 |
| }, |
| { |
| "epoch": 2.571139240506329, |
| "grad_norm": 0.5784966008032585, |
| "learning_rate": 2.240639932921966e-05, |
| "loss": 0.3393, |
| "step": 634 |
| }, |
| { |
| "epoch": 2.5751898734177217, |
| "grad_norm": 0.5857867542701056, |
| "learning_rate": 2.2350043382056995e-05, |
| "loss": 0.3571, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.579240506329114, |
| "grad_norm": 0.5649725226961273, |
| "learning_rate": 2.2293668507968015e-05, |
| "loss": 0.3622, |
| "step": 636 |
| }, |
| { |
| "epoch": 2.5832911392405062, |
| "grad_norm": 0.5903321065413912, |
| "learning_rate": 2.2237275160988186e-05, |
| "loss": 0.3547, |
| "step": 637 |
| }, |
| { |
| "epoch": 2.5873417721518988, |
| "grad_norm": 0.5742249582610803, |
| "learning_rate": 2.2180863795301787e-05, |
| "loss": 0.3584, |
| "step": 638 |
| }, |
| { |
| "epoch": 2.5913924050632913, |
| "grad_norm": 0.6171899408945792, |
| "learning_rate": 2.212443486523819e-05, |
| "loss": 0.3172, |
| "step": 639 |
| }, |
| { |
| "epoch": 2.5954430379746833, |
| "grad_norm": 0.583714281319652, |
| "learning_rate": 2.2067988825268243e-05, |
| "loss": 0.3485, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.599493670886076, |
| "grad_norm": 0.5723864800488129, |
| "learning_rate": 2.2011526130000596e-05, |
| "loss": 0.3474, |
| "step": 641 |
| }, |
| { |
| "epoch": 2.6035443037974684, |
| "grad_norm": 0.5930214245616171, |
| "learning_rate": 2.1955047234178038e-05, |
| "loss": 0.3692, |
| "step": 642 |
| }, |
| { |
| "epoch": 2.607594936708861, |
| "grad_norm": 0.586855696424102, |
| "learning_rate": 2.1898552592673825e-05, |
| "loss": 0.3584, |
| "step": 643 |
| }, |
| { |
| "epoch": 2.6116455696202534, |
| "grad_norm": 0.571486510856587, |
| "learning_rate": 2.184204266048803e-05, |
| "loss": 0.3333, |
| "step": 644 |
| }, |
| { |
| "epoch": 2.6156962025316455, |
| "grad_norm": 0.579070078746352, |
| "learning_rate": 2.1785517892743887e-05, |
| "loss": 0.3622, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.619746835443038, |
| "grad_norm": 0.564543111055874, |
| "learning_rate": 2.17289787446841e-05, |
| "loss": 0.3519, |
| "step": 646 |
| }, |
| { |
| "epoch": 2.6237974683544305, |
| "grad_norm": 0.5597891445429104, |
| "learning_rate": 2.1672425671667198e-05, |
| "loss": 0.3677, |
| "step": 647 |
| }, |
| { |
| "epoch": 2.6278481012658226, |
| "grad_norm": 0.5844618146606766, |
| "learning_rate": 2.161585912916385e-05, |
| "loss": 0.3291, |
| "step": 648 |
| }, |
| { |
| "epoch": 2.631898734177215, |
| "grad_norm": 0.5714238417964623, |
| "learning_rate": 2.1559279572753214e-05, |
| "loss": 0.3657, |
| "step": 649 |
| }, |
| { |
| "epoch": 2.6359493670886076, |
| "grad_norm": 0.5724043950687794, |
| "learning_rate": 2.1502687458119268e-05, |
| "loss": 0.3543, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.64, |
| "grad_norm": 0.5943443234004183, |
| "learning_rate": 2.1446083241047116e-05, |
| "loss": 0.3392, |
| "step": 651 |
| }, |
| { |
| "epoch": 2.6440506329113926, |
| "grad_norm": 0.5846283318819872, |
| "learning_rate": 2.1389467377419333e-05, |
| "loss": 0.3356, |
| "step": 652 |
| }, |
| { |
| "epoch": 2.6481012658227847, |
| "grad_norm": 0.5632839923714534, |
| "learning_rate": 2.133284032321232e-05, |
| "loss": 0.3345, |
| "step": 653 |
| }, |
| { |
| "epoch": 2.6521518987341772, |
| "grad_norm": 0.5629600265128976, |
| "learning_rate": 2.1276202534492566e-05, |
| "loss": 0.358, |
| "step": 654 |
| }, |
| { |
| "epoch": 2.6562025316455697, |
| "grad_norm": 0.5884479305139232, |
| "learning_rate": 2.121955446741306e-05, |
| "loss": 0.3177, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.660253164556962, |
| "grad_norm": 0.5893496234672007, |
| "learning_rate": 2.1162896578209517e-05, |
| "loss": 0.3432, |
| "step": 656 |
| }, |
| { |
| "epoch": 2.6643037974683543, |
| "grad_norm": 0.5857467954538601, |
| "learning_rate": 2.1106229323196813e-05, |
| "loss": 0.3472, |
| "step": 657 |
| }, |
| { |
| "epoch": 2.668354430379747, |
| "grad_norm": 0.5597796124525102, |
| "learning_rate": 2.1049553158765214e-05, |
| "loss": 0.3461, |
| "step": 658 |
| }, |
| { |
| "epoch": 2.6724050632911394, |
| "grad_norm": 0.5686080203482157, |
| "learning_rate": 2.0992868541376764e-05, |
| "loss": 0.3673, |
| "step": 659 |
| }, |
| { |
| "epoch": 2.676455696202532, |
| "grad_norm": 0.5950270461228224, |
| "learning_rate": 2.093617592756158e-05, |
| "loss": 0.3364, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.680506329113924, |
| "grad_norm": 0.5851931293322576, |
| "learning_rate": 2.0879475773914167e-05, |
| "loss": 0.3516, |
| "step": 661 |
| }, |
| { |
| "epoch": 2.6845569620253165, |
| "grad_norm": 0.5811821550039225, |
| "learning_rate": 2.082276853708978e-05, |
| "loss": 0.3451, |
| "step": 662 |
| }, |
| { |
| "epoch": 2.688607594936709, |
| "grad_norm": 0.6035047176086105, |
| "learning_rate": 2.076605467380071e-05, |
| "loss": 0.3521, |
| "step": 663 |
| }, |
| { |
| "epoch": 2.692658227848101, |
| "grad_norm": 0.5722623364658898, |
| "learning_rate": 2.0709334640812613e-05, |
| "loss": 0.3389, |
| "step": 664 |
| }, |
| { |
| "epoch": 2.6967088607594936, |
| "grad_norm": 0.6460949193835274, |
| "learning_rate": 2.0652608894940824e-05, |
| "loss": 0.3559, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.700759493670886, |
| "grad_norm": 0.6386656014857899, |
| "learning_rate": 2.0595877893046722e-05, |
| "loss": 0.3487, |
| "step": 666 |
| }, |
| { |
| "epoch": 2.7048101265822786, |
| "grad_norm": 0.5960013562142185, |
| "learning_rate": 2.0539142092033985e-05, |
| "loss": 0.3439, |
| "step": 667 |
| }, |
| { |
| "epoch": 2.708860759493671, |
| "grad_norm": 0.6062178100353875, |
| "learning_rate": 2.048240194884496e-05, |
| "loss": 0.3513, |
| "step": 668 |
| }, |
| { |
| "epoch": 2.712911392405063, |
| "grad_norm": 0.5841595632537837, |
| "learning_rate": 2.042565792045695e-05, |
| "loss": 0.337, |
| "step": 669 |
| }, |
| { |
| "epoch": 2.7169620253164557, |
| "grad_norm": 0.6002282582182404, |
| "learning_rate": 2.036891046387857e-05, |
| "loss": 0.3489, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.721012658227848, |
| "grad_norm": 0.5756247554524372, |
| "learning_rate": 2.0312160036146036e-05, |
| "loss": 0.3449, |
| "step": 671 |
| }, |
| { |
| "epoch": 2.7250632911392403, |
| "grad_norm": 0.6106408206369549, |
| "learning_rate": 2.025540709431948e-05, |
| "loss": 0.3407, |
| "step": 672 |
| }, |
| { |
| "epoch": 2.729113924050633, |
| "grad_norm": 0.6018277321507436, |
| "learning_rate": 2.0198652095479298e-05, |
| "loss": 0.3438, |
| "step": 673 |
| }, |
| { |
| "epoch": 2.7331645569620253, |
| "grad_norm": 0.5771765717869998, |
| "learning_rate": 2.014189549672245e-05, |
| "loss": 0.3523, |
| "step": 674 |
| }, |
| { |
| "epoch": 2.737215189873418, |
| "grad_norm": 0.5814397213073315, |
| "learning_rate": 2.0085137755158776e-05, |
| "loss": 0.3496, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.7412658227848103, |
| "grad_norm": 0.5847264817580617, |
| "learning_rate": 2.0028379327907327e-05, |
| "loss": 0.3444, |
| "step": 676 |
| }, |
| { |
| "epoch": 2.7453164556962024, |
| "grad_norm": 0.5747260993762996, |
| "learning_rate": 1.9971620672092676e-05, |
| "loss": 0.3562, |
| "step": 677 |
| }, |
| { |
| "epoch": 2.749367088607595, |
| "grad_norm": 0.585432502142789, |
| "learning_rate": 1.991486224484123e-05, |
| "loss": 0.3481, |
| "step": 678 |
| }, |
| { |
| "epoch": 2.7534177215189874, |
| "grad_norm": 0.5599489078327557, |
| "learning_rate": 1.985810450327756e-05, |
| "loss": 0.3634, |
| "step": 679 |
| }, |
| { |
| "epoch": 2.7574683544303795, |
| "grad_norm": 0.5629876690617696, |
| "learning_rate": 1.9801347904520706e-05, |
| "loss": 0.3399, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.761518987341772, |
| "grad_norm": 0.5753922949063179, |
| "learning_rate": 1.974459290568053e-05, |
| "loss": 0.3697, |
| "step": 681 |
| }, |
| { |
| "epoch": 2.7655696202531646, |
| "grad_norm": 0.5666841806426111, |
| "learning_rate": 1.968783996385397e-05, |
| "loss": 0.3493, |
| "step": 682 |
| }, |
| { |
| "epoch": 2.769620253164557, |
| "grad_norm": 0.5772024785351368, |
| "learning_rate": 1.963108953612143e-05, |
| "loss": 0.3417, |
| "step": 683 |
| }, |
| { |
| "epoch": 2.7736708860759496, |
| "grad_norm": 0.5843122977904314, |
| "learning_rate": 1.9574342079543056e-05, |
| "loss": 0.3544, |
| "step": 684 |
| }, |
| { |
| "epoch": 2.7777215189873417, |
| "grad_norm": 0.5617270414609679, |
| "learning_rate": 1.9517598051155046e-05, |
| "loss": 0.3567, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.781772151898734, |
| "grad_norm": 0.5800033482183932, |
| "learning_rate": 1.9460857907966025e-05, |
| "loss": 0.3527, |
| "step": 686 |
| }, |
| { |
| "epoch": 2.7858227848101267, |
| "grad_norm": 0.584722385391435, |
| "learning_rate": 1.9404122106953285e-05, |
| "loss": 0.3548, |
| "step": 687 |
| }, |
| { |
| "epoch": 2.7898734177215188, |
| "grad_norm": 0.5881100600878657, |
| "learning_rate": 1.9347391105059176e-05, |
| "loss": 0.3569, |
| "step": 688 |
| }, |
| { |
| "epoch": 2.7939240506329113, |
| "grad_norm": 0.5623944318371624, |
| "learning_rate": 1.92906653591874e-05, |
| "loss": 0.3393, |
| "step": 689 |
| }, |
| { |
| "epoch": 2.797974683544304, |
| "grad_norm": 0.5863241349345181, |
| "learning_rate": 1.9233945326199295e-05, |
| "loss": 0.3375, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.8020253164556963, |
| "grad_norm": 0.5624904954122713, |
| "learning_rate": 1.917723146291022e-05, |
| "loss": 0.3567, |
| "step": 691 |
| }, |
| { |
| "epoch": 2.806075949367089, |
| "grad_norm": 0.5868797129828841, |
| "learning_rate": 1.912052422608584e-05, |
| "loss": 0.3518, |
| "step": 692 |
| }, |
| { |
| "epoch": 2.810126582278481, |
| "grad_norm": 0.5722072329940924, |
| "learning_rate": 1.9063824072438428e-05, |
| "loss": 0.3396, |
| "step": 693 |
| }, |
| { |
| "epoch": 2.8141772151898734, |
| "grad_norm": 0.59715184074535, |
| "learning_rate": 1.9007131458623246e-05, |
| "loss": 0.3395, |
| "step": 694 |
| }, |
| { |
| "epoch": 2.818227848101266, |
| "grad_norm": 0.5723949753593742, |
| "learning_rate": 1.895044684123479e-05, |
| "loss": 0.3621, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.822278481012658, |
| "grad_norm": 0.5444249475607552, |
| "learning_rate": 1.8893770676803194e-05, |
| "loss": 0.3716, |
| "step": 696 |
| }, |
| { |
| "epoch": 2.8263291139240505, |
| "grad_norm": 0.5663641829260911, |
| "learning_rate": 1.8837103421790486e-05, |
| "loss": 0.3686, |
| "step": 697 |
| }, |
| { |
| "epoch": 2.830379746835443, |
| "grad_norm": 0.5712162181744459, |
| "learning_rate": 1.8780445532586952e-05, |
| "loss": 0.3399, |
| "step": 698 |
| }, |
| { |
| "epoch": 2.8344303797468355, |
| "grad_norm": 0.5667811420815281, |
| "learning_rate": 1.872379746550743e-05, |
| "loss": 0.3316, |
| "step": 699 |
| }, |
| { |
| "epoch": 2.838481012658228, |
| "grad_norm": 0.5613151052247454, |
| "learning_rate": 1.866715967678769e-05, |
| "loss": 0.3497, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.84253164556962, |
| "grad_norm": 0.5497355076437666, |
| "learning_rate": 1.861053262258067e-05, |
| "loss": 0.3641, |
| "step": 701 |
| }, |
| { |
| "epoch": 2.8465822784810126, |
| "grad_norm": 0.564413294608856, |
| "learning_rate": 1.8553916758952897e-05, |
| "loss": 0.3406, |
| "step": 702 |
| }, |
| { |
| "epoch": 2.850632911392405, |
| "grad_norm": 0.5657029873626861, |
| "learning_rate": 1.8497312541880735e-05, |
| "loss": 0.3556, |
| "step": 703 |
| }, |
| { |
| "epoch": 2.8546835443037972, |
| "grad_norm": 0.5604448441964516, |
| "learning_rate": 1.8440720427246786e-05, |
| "loss": 0.3373, |
| "step": 704 |
| }, |
| { |
| "epoch": 2.8587341772151897, |
| "grad_norm": 0.5520113003182913, |
| "learning_rate": 1.8384140870836157e-05, |
| "loss": 0.3335, |
| "step": 705 |
| }, |
| { |
| "epoch": 2.8627848101265823, |
| "grad_norm": 0.5502765254808101, |
| "learning_rate": 1.8327574328332806e-05, |
| "loss": 0.3435, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.8668354430379748, |
| "grad_norm": 0.5545461121580751, |
| "learning_rate": 1.8271021255315906e-05, |
| "loss": 0.3479, |
| "step": 707 |
| }, |
| { |
| "epoch": 2.8708860759493673, |
| "grad_norm": 0.5589009668537787, |
| "learning_rate": 1.8214482107256117e-05, |
| "loss": 0.3363, |
| "step": 708 |
| }, |
| { |
| "epoch": 2.8749367088607594, |
| "grad_norm": 0.5556537683415241, |
| "learning_rate": 1.8157957339511968e-05, |
| "loss": 0.3522, |
| "step": 709 |
| }, |
| { |
| "epoch": 2.878987341772152, |
| "grad_norm": 0.575001535232473, |
| "learning_rate": 1.8101447407326182e-05, |
| "loss": 0.3402, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.8830379746835444, |
| "grad_norm": 0.5831016946731767, |
| "learning_rate": 1.8044952765821966e-05, |
| "loss": 0.3263, |
| "step": 711 |
| }, |
| { |
| "epoch": 2.8870886075949365, |
| "grad_norm": 0.5515645581402768, |
| "learning_rate": 1.7988473869999407e-05, |
| "loss": 0.3552, |
| "step": 712 |
| }, |
| { |
| "epoch": 2.891139240506329, |
| "grad_norm": 0.580414857869068, |
| "learning_rate": 1.7932011174731764e-05, |
| "loss": 0.3404, |
| "step": 713 |
| }, |
| { |
| "epoch": 2.8951898734177215, |
| "grad_norm": 0.545243227764143, |
| "learning_rate": 1.7875565134761817e-05, |
| "loss": 0.3383, |
| "step": 714 |
| }, |
| { |
| "epoch": 2.899240506329114, |
| "grad_norm": 0.5678608023500457, |
| "learning_rate": 1.7819136204698226e-05, |
| "loss": 0.3566, |
| "step": 715 |
| }, |
| { |
| "epoch": 2.9032911392405065, |
| "grad_norm": 0.5645559680295348, |
| "learning_rate": 1.776272483901182e-05, |
| "loss": 0.3426, |
| "step": 716 |
| }, |
| { |
| "epoch": 2.9073417721518986, |
| "grad_norm": 0.5587403590106867, |
| "learning_rate": 1.7706331492031995e-05, |
| "loss": 0.3464, |
| "step": 717 |
| }, |
| { |
| "epoch": 2.911392405063291, |
| "grad_norm": 0.5433570629247826, |
| "learning_rate": 1.764995661794301e-05, |
| "loss": 0.3542, |
| "step": 718 |
| }, |
| { |
| "epoch": 2.9154430379746836, |
| "grad_norm": 0.5699145196234169, |
| "learning_rate": 1.759360067078035e-05, |
| "loss": 0.3329, |
| "step": 719 |
| }, |
| { |
| "epoch": 2.9194936708860757, |
| "grad_norm": 0.5497391767100702, |
| "learning_rate": 1.7537264104427064e-05, |
| "loss": 0.3529, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.923544303797468, |
| "grad_norm": 0.5723544189826021, |
| "learning_rate": 1.748094737261012e-05, |
| "loss": 0.3433, |
| "step": 721 |
| }, |
| { |
| "epoch": 2.9275949367088607, |
| "grad_norm": 0.6004704232070618, |
| "learning_rate": 1.7424650928896726e-05, |
| "loss": 0.3453, |
| "step": 722 |
| }, |
| { |
| "epoch": 2.9316455696202532, |
| "grad_norm": 0.5531582668568056, |
| "learning_rate": 1.7368375226690712e-05, |
| "loss": 0.3545, |
| "step": 723 |
| }, |
| { |
| "epoch": 2.9356962025316458, |
| "grad_norm": 0.5608301178140697, |
| "learning_rate": 1.731212071922883e-05, |
| "loss": 0.3431, |
| "step": 724 |
| }, |
| { |
| "epoch": 2.939746835443038, |
| "grad_norm": 0.5988214093770067, |
| "learning_rate": 1.7255887859577156e-05, |
| "loss": 0.3554, |
| "step": 725 |
| }, |
| { |
| "epoch": 2.9437974683544303, |
| "grad_norm": 0.6118745893574294, |
| "learning_rate": 1.7199677100627427e-05, |
| "loss": 0.346, |
| "step": 726 |
| }, |
| { |
| "epoch": 2.947848101265823, |
| "grad_norm": 0.5437786736505098, |
| "learning_rate": 1.7143488895093343e-05, |
| "loss": 0.3665, |
| "step": 727 |
| }, |
| { |
| "epoch": 2.951898734177215, |
| "grad_norm": 0.5871716757775023, |
| "learning_rate": 1.7087323695506994e-05, |
| "loss": 0.3453, |
| "step": 728 |
| }, |
| { |
| "epoch": 2.9559493670886074, |
| "grad_norm": 0.5525198427031702, |
| "learning_rate": 1.7031181954215194e-05, |
| "loss": 0.3502, |
| "step": 729 |
| }, |
| { |
| "epoch": 2.96, |
| "grad_norm": 0.5493854419536928, |
| "learning_rate": 1.6975064123375788e-05, |
| "loss": 0.36, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.9640506329113925, |
| "grad_norm": 0.5947124674730585, |
| "learning_rate": 1.6918970654954084e-05, |
| "loss": 0.3433, |
| "step": 731 |
| }, |
| { |
| "epoch": 2.968101265822785, |
| "grad_norm": 0.5627073978922057, |
| "learning_rate": 1.686290200071915e-05, |
| "loss": 0.3448, |
| "step": 732 |
| }, |
| { |
| "epoch": 2.972151898734177, |
| "grad_norm": 0.5460184276197433, |
| "learning_rate": 1.6806858612240234e-05, |
| "loss": 0.3329, |
| "step": 733 |
| }, |
| { |
| "epoch": 2.9762025316455696, |
| "grad_norm": 0.5847568469767621, |
| "learning_rate": 1.6750840940883078e-05, |
| "loss": 0.3596, |
| "step": 734 |
| }, |
| { |
| "epoch": 2.980253164556962, |
| "grad_norm": 0.5590419085152337, |
| "learning_rate": 1.6694849437806305e-05, |
| "loss": 0.3534, |
| "step": 735 |
| }, |
| { |
| "epoch": 2.984303797468354, |
| "grad_norm": 0.5754563619421876, |
| "learning_rate": 1.663888455395778e-05, |
| "loss": 0.352, |
| "step": 736 |
| }, |
| { |
| "epoch": 2.9883544303797467, |
| "grad_norm": 0.5566155733607459, |
| "learning_rate": 1.6582946740070995e-05, |
| "loss": 0.3471, |
| "step": 737 |
| }, |
| { |
| "epoch": 2.992405063291139, |
| "grad_norm": 0.5384662268932392, |
| "learning_rate": 1.6527036446661396e-05, |
| "loss": 0.3489, |
| "step": 738 |
| }, |
| { |
| "epoch": 2.9964556962025317, |
| "grad_norm": 0.5618867158636884, |
| "learning_rate": 1.6471154124022818e-05, |
| "loss": 0.3564, |
| "step": 739 |
| }, |
| { |
| "epoch": 3.002025316455696, |
| "grad_norm": 0.5907132100489139, |
| "learning_rate": 1.6415300222223788e-05, |
| "loss": 0.2464, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0060759493670886, |
| "grad_norm": 0.661210122067426, |
| "learning_rate": 1.6359475191103958e-05, |
| "loss": 0.1817, |
| "step": 741 |
| }, |
| { |
| "epoch": 3.010126582278481, |
| "grad_norm": 0.48704107937976737, |
| "learning_rate": 1.6303679480270466e-05, |
| "loss": 0.1695, |
| "step": 742 |
| }, |
| { |
| "epoch": 3.0141772151898736, |
| "grad_norm": 0.5344197928173583, |
| "learning_rate": 1.624791353909428e-05, |
| "loss": 0.1866, |
| "step": 743 |
| }, |
| { |
| "epoch": 3.0182278481012657, |
| "grad_norm": 0.7013574827371979, |
| "learning_rate": 1.619217781670663e-05, |
| "loss": 0.1732, |
| "step": 744 |
| }, |
| { |
| "epoch": 3.022278481012658, |
| "grad_norm": 0.7996733860035391, |
| "learning_rate": 1.6136472761995373e-05, |
| "loss": 0.1756, |
| "step": 745 |
| }, |
| { |
| "epoch": 3.0263291139240507, |
| "grad_norm": 0.645557727017967, |
| "learning_rate": 1.608079882360134e-05, |
| "loss": 0.174, |
| "step": 746 |
| }, |
| { |
| "epoch": 3.030379746835443, |
| "grad_norm": 0.5255390943970383, |
| "learning_rate": 1.60251564499148e-05, |
| "loss": 0.1736, |
| "step": 747 |
| }, |
| { |
| "epoch": 3.0344303797468353, |
| "grad_norm": 0.5532272799123754, |
| "learning_rate": 1.596954608907176e-05, |
| "loss": 0.1639, |
| "step": 748 |
| }, |
| { |
| "epoch": 3.038481012658228, |
| "grad_norm": 0.5888045552652929, |
| "learning_rate": 1.591396818895043e-05, |
| "loss": 0.1836, |
| "step": 749 |
| }, |
| { |
| "epoch": 3.0425316455696203, |
| "grad_norm": 0.5955427658447567, |
| "learning_rate": 1.585842319716759e-05, |
| "loss": 0.1926, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.046582278481013, |
| "grad_norm": 0.4847705653539123, |
| "learning_rate": 1.5802911561074944e-05, |
| "loss": 0.1875, |
| "step": 751 |
| }, |
| { |
| "epoch": 3.050632911392405, |
| "grad_norm": 0.4506878509981733, |
| "learning_rate": 1.5747433727755595e-05, |
| "loss": 0.1734, |
| "step": 752 |
| }, |
| { |
| "epoch": 3.0546835443037974, |
| "grad_norm": 0.4990379809093988, |
| "learning_rate": 1.5691990144020376e-05, |
| "loss": 0.1823, |
| "step": 753 |
| }, |
| { |
| "epoch": 3.05873417721519, |
| "grad_norm": 0.546641592198263, |
| "learning_rate": 1.5636581256404297e-05, |
| "loss": 0.1662, |
| "step": 754 |
| }, |
| { |
| "epoch": 3.0627848101265824, |
| "grad_norm": 0.5456911796353802, |
| "learning_rate": 1.558120751116291e-05, |
| "loss": 0.1665, |
| "step": 755 |
| }, |
| { |
| "epoch": 3.0668354430379745, |
| "grad_norm": 0.5139979453759241, |
| "learning_rate": 1.552586935426876e-05, |
| "loss": 0.1823, |
| "step": 756 |
| }, |
| { |
| "epoch": 3.070886075949367, |
| "grad_norm": 0.49084871441769895, |
| "learning_rate": 1.547056723140774e-05, |
| "loss": 0.1743, |
| "step": 757 |
| }, |
| { |
| "epoch": 3.0749367088607595, |
| "grad_norm": 0.4660286810598985, |
| "learning_rate": 1.5415301587975565e-05, |
| "loss": 0.166, |
| "step": 758 |
| }, |
| { |
| "epoch": 3.078987341772152, |
| "grad_norm": 0.46575132947127407, |
| "learning_rate": 1.536007286907411e-05, |
| "loss": 0.1644, |
| "step": 759 |
| }, |
| { |
| "epoch": 3.083037974683544, |
| "grad_norm": 0.4753270218116812, |
| "learning_rate": 1.5304881519507896e-05, |
| "loss": 0.154, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.0870886075949366, |
| "grad_norm": 0.49695604547779887, |
| "learning_rate": 1.5249727983780453e-05, |
| "loss": 0.1764, |
| "step": 761 |
| }, |
| { |
| "epoch": 3.091139240506329, |
| "grad_norm": 0.4692968436241622, |
| "learning_rate": 1.5194612706090786e-05, |
| "loss": 0.1654, |
| "step": 762 |
| }, |
| { |
| "epoch": 3.0951898734177217, |
| "grad_norm": 0.4481760514866764, |
| "learning_rate": 1.5139536130329771e-05, |
| "loss": 0.1465, |
| "step": 763 |
| }, |
| { |
| "epoch": 3.0992405063291137, |
| "grad_norm": 0.4662230460891209, |
| "learning_rate": 1.508449870007656e-05, |
| "loss": 0.1717, |
| "step": 764 |
| }, |
| { |
| "epoch": 3.1032911392405063, |
| "grad_norm": 0.4774558364310024, |
| "learning_rate": 1.5029500858595056e-05, |
| "loss": 0.1659, |
| "step": 765 |
| }, |
| { |
| "epoch": 3.1073417721518988, |
| "grad_norm": 0.48326410431798655, |
| "learning_rate": 1.4974543048830328e-05, |
| "loss": 0.1935, |
| "step": 766 |
| }, |
| { |
| "epoch": 3.1113924050632913, |
| "grad_norm": 0.46633286322638556, |
| "learning_rate": 1.4919625713405e-05, |
| "loss": 0.1802, |
| "step": 767 |
| }, |
| { |
| "epoch": 3.1154430379746834, |
| "grad_norm": 0.5034441131771464, |
| "learning_rate": 1.4864749294615756e-05, |
| "loss": 0.1849, |
| "step": 768 |
| }, |
| { |
| "epoch": 3.119493670886076, |
| "grad_norm": 0.46084470230818586, |
| "learning_rate": 1.4809914234429716e-05, |
| "loss": 0.1621, |
| "step": 769 |
| }, |
| { |
| "epoch": 3.1235443037974684, |
| "grad_norm": 0.4389684613485898, |
| "learning_rate": 1.4755120974480923e-05, |
| "loss": 0.1637, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.127594936708861, |
| "grad_norm": 0.4636245708419567, |
| "learning_rate": 1.4700369956066771e-05, |
| "loss": 0.1707, |
| "step": 771 |
| }, |
| { |
| "epoch": 3.131645569620253, |
| "grad_norm": 0.47438750232360716, |
| "learning_rate": 1.4645661620144413e-05, |
| "loss": 0.1788, |
| "step": 772 |
| }, |
| { |
| "epoch": 3.1356962025316455, |
| "grad_norm": 0.4594604066840488, |
| "learning_rate": 1.4590996407327284e-05, |
| "loss": 0.1724, |
| "step": 773 |
| }, |
| { |
| "epoch": 3.139746835443038, |
| "grad_norm": 0.45134342617325746, |
| "learning_rate": 1.4536374757881487e-05, |
| "loss": 0.1636, |
| "step": 774 |
| }, |
| { |
| "epoch": 3.1437974683544305, |
| "grad_norm": 0.4349113852418833, |
| "learning_rate": 1.4481797111722271e-05, |
| "loss": 0.1623, |
| "step": 775 |
| }, |
| { |
| "epoch": 3.1478481012658226, |
| "grad_norm": 0.44491062658803704, |
| "learning_rate": 1.4427263908410507e-05, |
| "loss": 0.161, |
| "step": 776 |
| }, |
| { |
| "epoch": 3.151898734177215, |
| "grad_norm": 0.4541235385809739, |
| "learning_rate": 1.4372775587149108e-05, |
| "loss": 0.1635, |
| "step": 777 |
| }, |
| { |
| "epoch": 3.1559493670886076, |
| "grad_norm": 0.4827339737581685, |
| "learning_rate": 1.4318332586779522e-05, |
| "loss": 0.1624, |
| "step": 778 |
| }, |
| { |
| "epoch": 3.16, |
| "grad_norm": 0.46017040004743703, |
| "learning_rate": 1.4263935345778202e-05, |
| "loss": 0.1869, |
| "step": 779 |
| }, |
| { |
| "epoch": 3.164050632911392, |
| "grad_norm": 0.4466153906512416, |
| "learning_rate": 1.420958430225303e-05, |
| "loss": 0.1744, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.1681012658227847, |
| "grad_norm": 0.43986118803837754, |
| "learning_rate": 1.415527989393985e-05, |
| "loss": 0.1591, |
| "step": 781 |
| }, |
| { |
| "epoch": 3.1721518987341772, |
| "grad_norm": 0.4579401622855876, |
| "learning_rate": 1.410102255819891e-05, |
| "loss": 0.1683, |
| "step": 782 |
| }, |
| { |
| "epoch": 3.1762025316455698, |
| "grad_norm": 0.4544369845113158, |
| "learning_rate": 1.404681273201131e-05, |
| "loss": 0.1578, |
| "step": 783 |
| }, |
| { |
| "epoch": 3.180253164556962, |
| "grad_norm": 0.4436543512236902, |
| "learning_rate": 1.399265085197556e-05, |
| "loss": 0.1627, |
| "step": 784 |
| }, |
| { |
| "epoch": 3.1843037974683543, |
| "grad_norm": 0.45659729486378303, |
| "learning_rate": 1.393853735430398e-05, |
| "loss": 0.1598, |
| "step": 785 |
| }, |
| { |
| "epoch": 3.188354430379747, |
| "grad_norm": 0.46155185647190927, |
| "learning_rate": 1.3884472674819246e-05, |
| "loss": 0.1649, |
| "step": 786 |
| }, |
| { |
| "epoch": 3.1924050632911394, |
| "grad_norm": 0.4614525274346001, |
| "learning_rate": 1.3830457248950864e-05, |
| "loss": 0.1549, |
| "step": 787 |
| }, |
| { |
| "epoch": 3.1964556962025314, |
| "grad_norm": 0.4678810646973461, |
| "learning_rate": 1.377649151173163e-05, |
| "loss": 0.1766, |
| "step": 788 |
| }, |
| { |
| "epoch": 3.200506329113924, |
| "grad_norm": 0.4694493741547763, |
| "learning_rate": 1.3722575897794181e-05, |
| "loss": 0.1521, |
| "step": 789 |
| }, |
| { |
| "epoch": 3.2045569620253165, |
| "grad_norm": 0.475350527113205, |
| "learning_rate": 1.3668710841367472e-05, |
| "loss": 0.1673, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.208607594936709, |
| "grad_norm": 0.4522595793078942, |
| "learning_rate": 1.361489677627324e-05, |
| "loss": 0.1736, |
| "step": 791 |
| }, |
| { |
| "epoch": 3.212658227848101, |
| "grad_norm": 0.4482299126555434, |
| "learning_rate": 1.3561134135922585e-05, |
| "loss": 0.1591, |
| "step": 792 |
| }, |
| { |
| "epoch": 3.2167088607594936, |
| "grad_norm": 0.4371003809215436, |
| "learning_rate": 1.350742335331241e-05, |
| "loss": 0.1511, |
| "step": 793 |
| }, |
| { |
| "epoch": 3.220759493670886, |
| "grad_norm": 0.4468787097193015, |
| "learning_rate": 1.345376486102198e-05, |
| "loss": 0.1665, |
| "step": 794 |
| }, |
| { |
| "epoch": 3.2248101265822786, |
| "grad_norm": 0.44721966825628856, |
| "learning_rate": 1.3400159091209414e-05, |
| "loss": 0.1772, |
| "step": 795 |
| }, |
| { |
| "epoch": 3.2288607594936707, |
| "grad_norm": 0.4517692034393115, |
| "learning_rate": 1.3346606475608216e-05, |
| "loss": 0.1795, |
| "step": 796 |
| }, |
| { |
| "epoch": 3.232911392405063, |
| "grad_norm": 0.44364851166782027, |
| "learning_rate": 1.3293107445523781e-05, |
| "loss": 0.1558, |
| "step": 797 |
| }, |
| { |
| "epoch": 3.2369620253164557, |
| "grad_norm": 0.453312347836684, |
| "learning_rate": 1.3239662431829949e-05, |
| "loss": 0.17, |
| "step": 798 |
| }, |
| { |
| "epoch": 3.2410126582278482, |
| "grad_norm": 0.4460106283248485, |
| "learning_rate": 1.3186271864965509e-05, |
| "loss": 0.1537, |
| "step": 799 |
| }, |
| { |
| "epoch": 3.2450632911392403, |
| "grad_norm": 0.4583916643459183, |
| "learning_rate": 1.3132936174930756e-05, |
| "loss": 0.1514, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.249113924050633, |
| "grad_norm": 0.47738587780208536, |
| "learning_rate": 1.3079655791283995e-05, |
| "loss": 0.1648, |
| "step": 801 |
| }, |
| { |
| "epoch": 3.2531645569620253, |
| "grad_norm": 0.4765772705099482, |
| "learning_rate": 1.3026431143138108e-05, |
| "loss": 0.1635, |
| "step": 802 |
| }, |
| { |
| "epoch": 3.257215189873418, |
| "grad_norm": 0.4468996488706342, |
| "learning_rate": 1.2973262659157114e-05, |
| "loss": 0.1592, |
| "step": 803 |
| }, |
| { |
| "epoch": 3.26126582278481, |
| "grad_norm": 0.4554840372748846, |
| "learning_rate": 1.2920150767552651e-05, |
| "loss": 0.1708, |
| "step": 804 |
| }, |
| { |
| "epoch": 3.2653164556962024, |
| "grad_norm": 0.4561094141385018, |
| "learning_rate": 1.2867095896080607e-05, |
| "loss": 0.1643, |
| "step": 805 |
| }, |
| { |
| "epoch": 3.269367088607595, |
| "grad_norm": 0.45929461287433154, |
| "learning_rate": 1.2814098472037612e-05, |
| "loss": 0.1701, |
| "step": 806 |
| }, |
| { |
| "epoch": 3.2734177215189875, |
| "grad_norm": 0.4592668456202418, |
| "learning_rate": 1.276115892225764e-05, |
| "loss": 0.1691, |
| "step": 807 |
| }, |
| { |
| "epoch": 3.27746835443038, |
| "grad_norm": 0.46278013909037985, |
| "learning_rate": 1.2708277673108555e-05, |
| "loss": 0.1707, |
| "step": 808 |
| }, |
| { |
| "epoch": 3.281518987341772, |
| "grad_norm": 0.44235907229392224, |
| "learning_rate": 1.2655455150488649e-05, |
| "loss": 0.1637, |
| "step": 809 |
| }, |
| { |
| "epoch": 3.2855696202531646, |
| "grad_norm": 0.43750353994955377, |
| "learning_rate": 1.2602691779823272e-05, |
| "loss": 0.1522, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.289620253164557, |
| "grad_norm": 0.46475553171975786, |
| "learning_rate": 1.2549987986061355e-05, |
| "loss": 0.1767, |
| "step": 811 |
| }, |
| { |
| "epoch": 3.293670886075949, |
| "grad_norm": 0.4629887754098841, |
| "learning_rate": 1.2497344193672005e-05, |
| "loss": 0.155, |
| "step": 812 |
| }, |
| { |
| "epoch": 3.2977215189873417, |
| "grad_norm": 0.45543542552010646, |
| "learning_rate": 1.2444760826641092e-05, |
| "loss": 0.1709, |
| "step": 813 |
| }, |
| { |
| "epoch": 3.301772151898734, |
| "grad_norm": 0.4804289627209647, |
| "learning_rate": 1.2392238308467817e-05, |
| "loss": 0.1586, |
| "step": 814 |
| }, |
| { |
| "epoch": 3.3058227848101267, |
| "grad_norm": 0.4598869848143127, |
| "learning_rate": 1.2339777062161326e-05, |
| "loss": 0.1612, |
| "step": 815 |
| }, |
| { |
| "epoch": 3.309873417721519, |
| "grad_norm": 0.44667679804744925, |
| "learning_rate": 1.2287377510237293e-05, |
| "loss": 0.1675, |
| "step": 816 |
| }, |
| { |
| "epoch": 3.3139240506329113, |
| "grad_norm": 0.4613184177660385, |
| "learning_rate": 1.2235040074714488e-05, |
| "loss": 0.1443, |
| "step": 817 |
| }, |
| { |
| "epoch": 3.317974683544304, |
| "grad_norm": 0.47265364237388546, |
| "learning_rate": 1.2182765177111434e-05, |
| "loss": 0.1654, |
| "step": 818 |
| }, |
| { |
| "epoch": 3.3220253164556963, |
| "grad_norm": 0.4546192721417468, |
| "learning_rate": 1.213055323844297e-05, |
| "loss": 0.1582, |
| "step": 819 |
| }, |
| { |
| "epoch": 3.3260759493670884, |
| "grad_norm": 0.4404220042392021, |
| "learning_rate": 1.2078404679216864e-05, |
| "loss": 0.1718, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.330126582278481, |
| "grad_norm": 0.466575921623242, |
| "learning_rate": 1.2026319919430458e-05, |
| "loss": 0.1697, |
| "step": 821 |
| }, |
| { |
| "epoch": 3.3341772151898734, |
| "grad_norm": 0.44352369248313833, |
| "learning_rate": 1.1974299378567227e-05, |
| "loss": 0.1553, |
| "step": 822 |
| }, |
| { |
| "epoch": 3.338227848101266, |
| "grad_norm": 0.43894258285317833, |
| "learning_rate": 1.1922343475593462e-05, |
| "loss": 0.1425, |
| "step": 823 |
| }, |
| { |
| "epoch": 3.3422784810126585, |
| "grad_norm": 0.4609722470105194, |
| "learning_rate": 1.187045262895488e-05, |
| "loss": 0.161, |
| "step": 824 |
| }, |
| { |
| "epoch": 3.3463291139240505, |
| "grad_norm": 0.44539600175628047, |
| "learning_rate": 1.1818627256573203e-05, |
| "loss": 0.1608, |
| "step": 825 |
| }, |
| { |
| "epoch": 3.350379746835443, |
| "grad_norm": 0.4519380950028345, |
| "learning_rate": 1.1766867775842864e-05, |
| "loss": 0.1788, |
| "step": 826 |
| }, |
| { |
| "epoch": 3.3544303797468356, |
| "grad_norm": 0.4504424912600629, |
| "learning_rate": 1.1715174603627615e-05, |
| "loss": 0.1738, |
| "step": 827 |
| }, |
| { |
| "epoch": 3.3584810126582276, |
| "grad_norm": 0.44893737955134244, |
| "learning_rate": 1.1663548156257147e-05, |
| "loss": 0.1574, |
| "step": 828 |
| }, |
| { |
| "epoch": 3.36253164556962, |
| "grad_norm": 0.44332933791675594, |
| "learning_rate": 1.161198884952377e-05, |
| "loss": 0.1594, |
| "step": 829 |
| }, |
| { |
| "epoch": 3.3665822784810127, |
| "grad_norm": 0.453799743691885, |
| "learning_rate": 1.1560497098679056e-05, |
| "loss": 0.1762, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.370632911392405, |
| "grad_norm": 0.46731368140177604, |
| "learning_rate": 1.1509073318430479e-05, |
| "loss": 0.1689, |
| "step": 831 |
| }, |
| { |
| "epoch": 3.3746835443037977, |
| "grad_norm": 0.44097512050137, |
| "learning_rate": 1.1457717922938116e-05, |
| "loss": 0.1654, |
| "step": 832 |
| }, |
| { |
| "epoch": 3.3787341772151898, |
| "grad_norm": 0.45200206457098024, |
| "learning_rate": 1.1406431325811233e-05, |
| "loss": 0.1722, |
| "step": 833 |
| }, |
| { |
| "epoch": 3.3827848101265823, |
| "grad_norm": 0.4228206587263341, |
| "learning_rate": 1.135521394010506e-05, |
| "loss": 0.1591, |
| "step": 834 |
| }, |
| { |
| "epoch": 3.386835443037975, |
| "grad_norm": 0.4583689545813303, |
| "learning_rate": 1.1304066178317367e-05, |
| "loss": 0.1664, |
| "step": 835 |
| }, |
| { |
| "epoch": 3.390886075949367, |
| "grad_norm": 0.46001557361660333, |
| "learning_rate": 1.1252988452385199e-05, |
| "loss": 0.1566, |
| "step": 836 |
| }, |
| { |
| "epoch": 3.3949367088607594, |
| "grad_norm": 0.44938962748259287, |
| "learning_rate": 1.1201981173681536e-05, |
| "loss": 0.162, |
| "step": 837 |
| }, |
| { |
| "epoch": 3.398987341772152, |
| "grad_norm": 0.4453946793688315, |
| "learning_rate": 1.1151044753011991e-05, |
| "loss": 0.1629, |
| "step": 838 |
| }, |
| { |
| "epoch": 3.4030379746835444, |
| "grad_norm": 0.44880802883808113, |
| "learning_rate": 1.1100179600611491e-05, |
| "loss": 0.1734, |
| "step": 839 |
| }, |
| { |
| "epoch": 3.407088607594937, |
| "grad_norm": 0.44749004647352575, |
| "learning_rate": 1.1049386126140985e-05, |
| "loss": 0.1639, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.411139240506329, |
| "grad_norm": 0.4467954638026324, |
| "learning_rate": 1.0998664738684128e-05, |
| "loss": 0.1902, |
| "step": 841 |
| }, |
| { |
| "epoch": 3.4151898734177215, |
| "grad_norm": 0.435554924238635, |
| "learning_rate": 1.0948015846744e-05, |
| "loss": 0.167, |
| "step": 842 |
| }, |
| { |
| "epoch": 3.419240506329114, |
| "grad_norm": 0.4656031423494464, |
| "learning_rate": 1.0897439858239832e-05, |
| "loss": 0.144, |
| "step": 843 |
| }, |
| { |
| "epoch": 3.423291139240506, |
| "grad_norm": 0.43418642423081766, |
| "learning_rate": 1.0846937180503652e-05, |
| "loss": 0.1547, |
| "step": 844 |
| }, |
| { |
| "epoch": 3.4273417721518986, |
| "grad_norm": 0.4581704739561118, |
| "learning_rate": 1.0796508220277117e-05, |
| "loss": 0.1566, |
| "step": 845 |
| }, |
| { |
| "epoch": 3.431392405063291, |
| "grad_norm": 0.4610676585119131, |
| "learning_rate": 1.0746153383708107e-05, |
| "loss": 0.165, |
| "step": 846 |
| }, |
| { |
| "epoch": 3.4354430379746836, |
| "grad_norm": 0.4452668600136283, |
| "learning_rate": 1.0695873076347579e-05, |
| "loss": 0.1748, |
| "step": 847 |
| }, |
| { |
| "epoch": 3.439493670886076, |
| "grad_norm": 0.44307559880692515, |
| "learning_rate": 1.0645667703146205e-05, |
| "loss": 0.175, |
| "step": 848 |
| }, |
| { |
| "epoch": 3.4435443037974682, |
| "grad_norm": 0.4478002660706502, |
| "learning_rate": 1.0595537668451161e-05, |
| "loss": 0.1645, |
| "step": 849 |
| }, |
| { |
| "epoch": 3.4475949367088607, |
| "grad_norm": 0.45641338798328196, |
| "learning_rate": 1.0545483376002854e-05, |
| "loss": 0.1574, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.4516455696202533, |
| "grad_norm": 0.43632204729685453, |
| "learning_rate": 1.0495505228931676e-05, |
| "loss": 0.1516, |
| "step": 851 |
| }, |
| { |
| "epoch": 3.4556962025316453, |
| "grad_norm": 0.4497875651159521, |
| "learning_rate": 1.044560362975474e-05, |
| "loss": 0.1582, |
| "step": 852 |
| }, |
| { |
| "epoch": 3.459746835443038, |
| "grad_norm": 0.43567170418375495, |
| "learning_rate": 1.0395778980372695e-05, |
| "loss": 0.1964, |
| "step": 853 |
| }, |
| { |
| "epoch": 3.4637974683544304, |
| "grad_norm": 0.4370021873441559, |
| "learning_rate": 1.0346031682066381e-05, |
| "loss": 0.1836, |
| "step": 854 |
| }, |
| { |
| "epoch": 3.467848101265823, |
| "grad_norm": 0.43175346777243867, |
| "learning_rate": 1.0296362135493724e-05, |
| "loss": 0.1537, |
| "step": 855 |
| }, |
| { |
| "epoch": 3.4718987341772154, |
| "grad_norm": 0.4407711586557324, |
| "learning_rate": 1.0246770740686422e-05, |
| "loss": 0.1718, |
| "step": 856 |
| }, |
| { |
| "epoch": 3.4759493670886075, |
| "grad_norm": 0.4519647268090711, |
| "learning_rate": 1.0197257897046743e-05, |
| "loss": 0.1752, |
| "step": 857 |
| }, |
| { |
| "epoch": 3.48, |
| "grad_norm": 0.4538767371056951, |
| "learning_rate": 1.014782400334433e-05, |
| "loss": 0.1636, |
| "step": 858 |
| }, |
| { |
| "epoch": 3.4840506329113925, |
| "grad_norm": 0.4598085989037832, |
| "learning_rate": 1.009846945771296e-05, |
| "loss": 0.165, |
| "step": 859 |
| }, |
| { |
| "epoch": 3.4881012658227846, |
| "grad_norm": 0.4565944668139176, |
| "learning_rate": 1.0049194657647363e-05, |
| "loss": 0.1478, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.492151898734177, |
| "grad_norm": 0.4596655771463825, |
| "learning_rate": 1.0000000000000006e-05, |
| "loss": 0.1614, |
| "step": 861 |
| }, |
| { |
| "epoch": 3.4962025316455696, |
| "grad_norm": 0.4567023823911369, |
| "learning_rate": 9.950885880977891e-06, |
| "loss": 0.1655, |
| "step": 862 |
| }, |
| { |
| "epoch": 3.500253164556962, |
| "grad_norm": 0.45163489072127455, |
| "learning_rate": 9.901852696139382e-06, |
| "loss": 0.1674, |
| "step": 863 |
| }, |
| { |
| "epoch": 3.5043037974683546, |
| "grad_norm": 0.43916569549845214, |
| "learning_rate": 9.852900840391027e-06, |
| "loss": 0.1691, |
| "step": 864 |
| }, |
| { |
| "epoch": 3.5083544303797467, |
| "grad_norm": 0.4473361739612025, |
| "learning_rate": 9.804030707984313e-06, |
| "loss": 0.1809, |
| "step": 865 |
| }, |
| { |
| "epoch": 3.512405063291139, |
| "grad_norm": 0.4426108837606294, |
| "learning_rate": 9.755242692512599e-06, |
| "loss": 0.172, |
| "step": 866 |
| }, |
| { |
| "epoch": 3.5164556962025317, |
| "grad_norm": 0.43287292027784835, |
| "learning_rate": 9.70653718690782e-06, |
| "loss": 0.15, |
| "step": 867 |
| }, |
| { |
| "epoch": 3.520506329113924, |
| "grad_norm": 0.42544975924351214, |
| "learning_rate": 9.657914583437454e-06, |
| "loss": 0.1625, |
| "step": 868 |
| }, |
| { |
| "epoch": 3.5245569620253163, |
| "grad_norm": 0.44030446322646505, |
| "learning_rate": 9.609375273701246e-06, |
| "loss": 0.1475, |
| "step": 869 |
| }, |
| { |
| "epoch": 3.528607594936709, |
| "grad_norm": 0.43742056301158216, |
| "learning_rate": 9.560919648628133e-06, |
| "loss": 0.1519, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.5326582278481014, |
| "grad_norm": 0.4445867228245256, |
| "learning_rate": 9.512548098473047e-06, |
| "loss": 0.1676, |
| "step": 871 |
| }, |
| { |
| "epoch": 3.536708860759494, |
| "grad_norm": 0.4509282994512908, |
| "learning_rate": 9.464261012813825e-06, |
| "loss": 0.1639, |
| "step": 872 |
| }, |
| { |
| "epoch": 3.540759493670886, |
| "grad_norm": 0.46326625703634433, |
| "learning_rate": 9.416058780547987e-06, |
| "loss": 0.1757, |
| "step": 873 |
| }, |
| { |
| "epoch": 3.5448101265822785, |
| "grad_norm": 0.4472372572221488, |
| "learning_rate": 9.367941789889714e-06, |
| "loss": 0.1776, |
| "step": 874 |
| }, |
| { |
| "epoch": 3.548860759493671, |
| "grad_norm": 0.4476256283900259, |
| "learning_rate": 9.319910428366607e-06, |
| "loss": 0.1603, |
| "step": 875 |
| }, |
| { |
| "epoch": 3.552911392405063, |
| "grad_norm": 0.4386259731440994, |
| "learning_rate": 9.271965082816667e-06, |
| "loss": 0.1567, |
| "step": 876 |
| }, |
| { |
| "epoch": 3.5569620253164556, |
| "grad_norm": 0.4477189563607, |
| "learning_rate": 9.224106139385111e-06, |
| "loss": 0.1598, |
| "step": 877 |
| }, |
| { |
| "epoch": 3.561012658227848, |
| "grad_norm": 0.45515655726209564, |
| "learning_rate": 9.176333983521291e-06, |
| "loss": 0.1742, |
| "step": 878 |
| }, |
| { |
| "epoch": 3.5650632911392406, |
| "grad_norm": 0.43009167455030134, |
| "learning_rate": 9.12864899997558e-06, |
| "loss": 0.164, |
| "step": 879 |
| }, |
| { |
| "epoch": 3.569113924050633, |
| "grad_norm": 0.436449048548147, |
| "learning_rate": 9.08105157279628e-06, |
| "loss": 0.1549, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.573164556962025, |
| "grad_norm": 0.4345138744543701, |
| "learning_rate": 9.03354208532653e-06, |
| "loss": 0.1596, |
| "step": 881 |
| }, |
| { |
| "epoch": 3.5772151898734177, |
| "grad_norm": 0.44083467149103056, |
| "learning_rate": 8.986120920201205e-06, |
| "loss": 0.1669, |
| "step": 882 |
| }, |
| { |
| "epoch": 3.58126582278481, |
| "grad_norm": 0.444900998470013, |
| "learning_rate": 8.938788459343852e-06, |
| "loss": 0.1662, |
| "step": 883 |
| }, |
| { |
| "epoch": 3.5853164556962023, |
| "grad_norm": 0.4357644042576251, |
| "learning_rate": 8.8915450839636e-06, |
| "loss": 0.1676, |
| "step": 884 |
| }, |
| { |
| "epoch": 3.589367088607595, |
| "grad_norm": 0.44511191682447165, |
| "learning_rate": 8.844391174552116e-06, |
| "loss": 0.1622, |
| "step": 885 |
| }, |
| { |
| "epoch": 3.5934177215189873, |
| "grad_norm": 0.4394248021325751, |
| "learning_rate": 8.797327110880479e-06, |
| "loss": 0.1612, |
| "step": 886 |
| }, |
| { |
| "epoch": 3.59746835443038, |
| "grad_norm": 0.4334838204790154, |
| "learning_rate": 8.750353271996206e-06, |
| "loss": 0.1613, |
| "step": 887 |
| }, |
| { |
| "epoch": 3.6015189873417723, |
| "grad_norm": 0.4326104104407368, |
| "learning_rate": 8.703470036220132e-06, |
| "loss": 0.1599, |
| "step": 888 |
| }, |
| { |
| "epoch": 3.6055696202531644, |
| "grad_norm": 0.4421592311196267, |
| "learning_rate": 8.656677781143394e-06, |
| "loss": 0.1661, |
| "step": 889 |
| }, |
| { |
| "epoch": 3.609620253164557, |
| "grad_norm": 0.45073787596299203, |
| "learning_rate": 8.609976883624377e-06, |
| "loss": 0.1655, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.6136708860759494, |
| "grad_norm": 0.45193272554808395, |
| "learning_rate": 8.563367719785698e-06, |
| "loss": 0.1704, |
| "step": 891 |
| }, |
| { |
| "epoch": 3.6177215189873415, |
| "grad_norm": 0.4446490727930166, |
| "learning_rate": 8.516850665011138e-06, |
| "loss": 0.1895, |
| "step": 892 |
| }, |
| { |
| "epoch": 3.621772151898734, |
| "grad_norm": 0.4483889128386641, |
| "learning_rate": 8.47042609394269e-06, |
| "loss": 0.1548, |
| "step": 893 |
| }, |
| { |
| "epoch": 3.6258227848101265, |
| "grad_norm": 0.442263751409046, |
| "learning_rate": 8.424094380477432e-06, |
| "loss": 0.188, |
| "step": 894 |
| }, |
| { |
| "epoch": 3.629873417721519, |
| "grad_norm": 0.4494949819861269, |
| "learning_rate": 8.37785589776465e-06, |
| "loss": 0.1806, |
| "step": 895 |
| }, |
| { |
| "epoch": 3.6339240506329116, |
| "grad_norm": 0.4540129813257601, |
| "learning_rate": 8.331711018202694e-06, |
| "loss": 0.169, |
| "step": 896 |
| }, |
| { |
| "epoch": 3.6379746835443036, |
| "grad_norm": 0.4518843482412336, |
| "learning_rate": 8.285660113436104e-06, |
| "loss": 0.1372, |
| "step": 897 |
| }, |
| { |
| "epoch": 3.642025316455696, |
| "grad_norm": 0.4473610575025946, |
| "learning_rate": 8.239703554352527e-06, |
| "loss": 0.1621, |
| "step": 898 |
| }, |
| { |
| "epoch": 3.6460759493670887, |
| "grad_norm": 0.4645974932336988, |
| "learning_rate": 8.193841711079775e-06, |
| "loss": 0.157, |
| "step": 899 |
| }, |
| { |
| "epoch": 3.6501265822784807, |
| "grad_norm": 0.45773046847448984, |
| "learning_rate": 8.148074952982828e-06, |
| "loss": 0.1552, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.6541772151898733, |
| "grad_norm": 0.44205615187155917, |
| "learning_rate": 8.102403648660859e-06, |
| "loss": 0.1435, |
| "step": 901 |
| }, |
| { |
| "epoch": 3.6582278481012658, |
| "grad_norm": 0.45680014006138214, |
| "learning_rate": 8.056828165944282e-06, |
| "loss": 0.1746, |
| "step": 902 |
| }, |
| { |
| "epoch": 3.6622784810126583, |
| "grad_norm": 0.4462292323845422, |
| "learning_rate": 8.011348871891762e-06, |
| "loss": 0.1482, |
| "step": 903 |
| }, |
| { |
| "epoch": 3.666329113924051, |
| "grad_norm": 0.44862249517664193, |
| "learning_rate": 7.965966132787287e-06, |
| "loss": 0.1882, |
| "step": 904 |
| }, |
| { |
| "epoch": 3.670379746835443, |
| "grad_norm": 0.436511287742748, |
| "learning_rate": 7.920680314137189e-06, |
| "loss": 0.1758, |
| "step": 905 |
| }, |
| { |
| "epoch": 3.6744303797468354, |
| "grad_norm": 0.43768600229284704, |
| "learning_rate": 7.875491780667246e-06, |
| "loss": 0.1606, |
| "step": 906 |
| }, |
| { |
| "epoch": 3.678481012658228, |
| "grad_norm": 0.4502079466899219, |
| "learning_rate": 7.830400896319667e-06, |
| "loss": 0.1566, |
| "step": 907 |
| }, |
| { |
| "epoch": 3.68253164556962, |
| "grad_norm": 0.44768784351153845, |
| "learning_rate": 7.785408024250259e-06, |
| "loss": 0.1735, |
| "step": 908 |
| }, |
| { |
| "epoch": 3.6865822784810125, |
| "grad_norm": 0.43274319215863305, |
| "learning_rate": 7.74051352682542e-06, |
| "loss": 0.1556, |
| "step": 909 |
| }, |
| { |
| "epoch": 3.690632911392405, |
| "grad_norm": 0.4537658577536865, |
| "learning_rate": 7.695717765619257e-06, |
| "loss": 0.1851, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.6946835443037975, |
| "grad_norm": 0.43537699097507754, |
| "learning_rate": 7.651021101410673e-06, |
| "loss": 0.1544, |
| "step": 911 |
| }, |
| { |
| "epoch": 3.69873417721519, |
| "grad_norm": 0.4476222792458484, |
| "learning_rate": 7.606423894180464e-06, |
| "loss": 0.1495, |
| "step": 912 |
| }, |
| { |
| "epoch": 3.702784810126582, |
| "grad_norm": 0.44974985262271916, |
| "learning_rate": 7.56192650310839e-06, |
| "loss": 0.1743, |
| "step": 913 |
| }, |
| { |
| "epoch": 3.7068354430379746, |
| "grad_norm": 0.45840780140298426, |
| "learning_rate": 7.517529286570349e-06, |
| "loss": 0.1694, |
| "step": 914 |
| }, |
| { |
| "epoch": 3.710886075949367, |
| "grad_norm": 0.4409619679237105, |
| "learning_rate": 7.473232602135387e-06, |
| "loss": 0.1569, |
| "step": 915 |
| }, |
| { |
| "epoch": 3.714936708860759, |
| "grad_norm": 0.44879295387471263, |
| "learning_rate": 7.429036806562935e-06, |
| "loss": 0.1417, |
| "step": 916 |
| }, |
| { |
| "epoch": 3.7189873417721517, |
| "grad_norm": 0.43194632725585297, |
| "learning_rate": 7.3849422557998455e-06, |
| "loss": 0.1604, |
| "step": 917 |
| }, |
| { |
| "epoch": 3.7230379746835442, |
| "grad_norm": 0.4425184002267637, |
| "learning_rate": 7.340949304977567e-06, |
| "loss": 0.1693, |
| "step": 918 |
| }, |
| { |
| "epoch": 3.7270886075949368, |
| "grad_norm": 0.4461048404044285, |
| "learning_rate": 7.297058308409282e-06, |
| "loss": 0.1486, |
| "step": 919 |
| }, |
| { |
| "epoch": 3.7311392405063293, |
| "grad_norm": 0.43546415557585305, |
| "learning_rate": 7.25326961958704e-06, |
| "loss": 0.1732, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.7351898734177214, |
| "grad_norm": 0.4391794750048748, |
| "learning_rate": 7.209583591178921e-06, |
| "loss": 0.1843, |
| "step": 921 |
| }, |
| { |
| "epoch": 3.739240506329114, |
| "grad_norm": 0.44253740804906877, |
| "learning_rate": 7.1660005750261925e-06, |
| "loss": 0.1653, |
| "step": 922 |
| }, |
| { |
| "epoch": 3.7432911392405064, |
| "grad_norm": 0.4469183205244511, |
| "learning_rate": 7.1225209221404765e-06, |
| "loss": 0.1632, |
| "step": 923 |
| }, |
| { |
| "epoch": 3.747341772151899, |
| "grad_norm": 0.44494901642851514, |
| "learning_rate": 7.079144982700909e-06, |
| "loss": 0.1559, |
| "step": 924 |
| }, |
| { |
| "epoch": 3.7513924050632914, |
| "grad_norm": 0.445336587563813, |
| "learning_rate": 7.0358731060513695e-06, |
| "loss": 0.1581, |
| "step": 925 |
| }, |
| { |
| "epoch": 3.7554430379746835, |
| "grad_norm": 0.4342913133927004, |
| "learning_rate": 6.99270564069757e-06, |
| "loss": 0.1452, |
| "step": 926 |
| }, |
| { |
| "epoch": 3.759493670886076, |
| "grad_norm": 0.5307317623105801, |
| "learning_rate": 6.949642934304375e-06, |
| "loss": 0.1788, |
| "step": 927 |
| }, |
| { |
| "epoch": 3.7635443037974685, |
| "grad_norm": 0.44417214610767686, |
| "learning_rate": 6.906685333692871e-06, |
| "loss": 0.1835, |
| "step": 928 |
| }, |
| { |
| "epoch": 3.7675949367088606, |
| "grad_norm": 0.4421194609597355, |
| "learning_rate": 6.86383318483769e-06, |
| "loss": 0.1456, |
| "step": 929 |
| }, |
| { |
| "epoch": 3.771645569620253, |
| "grad_norm": 0.445531413841371, |
| "learning_rate": 6.821086832864139e-06, |
| "loss": 0.1725, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.7756962025316456, |
| "grad_norm": 0.426002251199808, |
| "learning_rate": 6.77844662204546e-06, |
| "loss": 0.1537, |
| "step": 931 |
| }, |
| { |
| "epoch": 3.779746835443038, |
| "grad_norm": 0.429223159298163, |
| "learning_rate": 6.7359128958000455e-06, |
| "loss": 0.1708, |
| "step": 932 |
| }, |
| { |
| "epoch": 3.7837974683544306, |
| "grad_norm": 0.4335803137467169, |
| "learning_rate": 6.693485996688695e-06, |
| "loss": 0.1739, |
| "step": 933 |
| }, |
| { |
| "epoch": 3.7878481012658227, |
| "grad_norm": 0.4402083005904946, |
| "learning_rate": 6.651166266411801e-06, |
| "loss": 0.1509, |
| "step": 934 |
| }, |
| { |
| "epoch": 3.7918987341772152, |
| "grad_norm": 0.42052205636484424, |
| "learning_rate": 6.6089540458066725e-06, |
| "loss": 0.1609, |
| "step": 935 |
| }, |
| { |
| "epoch": 3.7959493670886078, |
| "grad_norm": 0.4480301616790334, |
| "learning_rate": 6.566849674844711e-06, |
| "loss": 0.1527, |
| "step": 936 |
| }, |
| { |
| "epoch": 3.8, |
| "grad_norm": 0.4355883335480955, |
| "learning_rate": 6.524853492628747e-06, |
| "loss": 0.1606, |
| "step": 937 |
| }, |
| { |
| "epoch": 3.8040506329113923, |
| "grad_norm": 0.45707921409036906, |
| "learning_rate": 6.4829658373902536e-06, |
| "loss": 0.1783, |
| "step": 938 |
| }, |
| { |
| "epoch": 3.808101265822785, |
| "grad_norm": 0.4487151376495278, |
| "learning_rate": 6.441187046486648e-06, |
| "loss": 0.1577, |
| "step": 939 |
| }, |
| { |
| "epoch": 3.8121518987341774, |
| "grad_norm": 0.4286307454134509, |
| "learning_rate": 6.399517456398567e-06, |
| "loss": 0.15, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.81620253164557, |
| "grad_norm": 0.4447594884309322, |
| "learning_rate": 6.357957402727164e-06, |
| "loss": 0.1852, |
| "step": 941 |
| }, |
| { |
| "epoch": 3.820253164556962, |
| "grad_norm": 0.4455353490755155, |
| "learning_rate": 6.316507220191395e-06, |
| "loss": 0.1533, |
| "step": 942 |
| }, |
| { |
| "epoch": 3.8243037974683545, |
| "grad_norm": 0.43375124442284074, |
| "learning_rate": 6.275167242625331e-06, |
| "loss": 0.1398, |
| "step": 943 |
| }, |
| { |
| "epoch": 3.828354430379747, |
| "grad_norm": 0.4370587608512517, |
| "learning_rate": 6.233937802975471e-06, |
| "loss": 0.162, |
| "step": 944 |
| }, |
| { |
| "epoch": 3.832405063291139, |
| "grad_norm": 0.4308435286381364, |
| "learning_rate": 6.192819233298046e-06, |
| "loss": 0.1551, |
| "step": 945 |
| }, |
| { |
| "epoch": 3.8364556962025316, |
| "grad_norm": 0.42179317648310577, |
| "learning_rate": 6.151811864756383e-06, |
| "loss": 0.1592, |
| "step": 946 |
| }, |
| { |
| "epoch": 3.840506329113924, |
| "grad_norm": 0.44344326285480185, |
| "learning_rate": 6.1109160276181655e-06, |
| "loss": 0.1688, |
| "step": 947 |
| }, |
| { |
| "epoch": 3.8445569620253166, |
| "grad_norm": 0.4373050817625722, |
| "learning_rate": 6.070132051252868e-06, |
| "loss": 0.1511, |
| "step": 948 |
| }, |
| { |
| "epoch": 3.848607594936709, |
| "grad_norm": 0.4332515250652569, |
| "learning_rate": 6.0294602641290034e-06, |
| "loss": 0.1771, |
| "step": 949 |
| }, |
| { |
| "epoch": 3.852658227848101, |
| "grad_norm": 0.4325398584992558, |
| "learning_rate": 5.988900993811575e-06, |
| "loss": 0.151, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.8567088607594937, |
| "grad_norm": 0.4236551195893073, |
| "learning_rate": 5.948454566959363e-06, |
| "loss": 0.1583, |
| "step": 951 |
| }, |
| { |
| "epoch": 3.8607594936708862, |
| "grad_norm": 0.44304506503460583, |
| "learning_rate": 5.908121309322328e-06, |
| "loss": 0.1621, |
| "step": 952 |
| }, |
| { |
| "epoch": 3.8648101265822783, |
| "grad_norm": 0.43097732545765294, |
| "learning_rate": 5.867901545738976e-06, |
| "loss": 0.1702, |
| "step": 953 |
| }, |
| { |
| "epoch": 3.868860759493671, |
| "grad_norm": 0.42840303793077, |
| "learning_rate": 5.827795600133774e-06, |
| "loss": 0.1707, |
| "step": 954 |
| }, |
| { |
| "epoch": 3.8729113924050633, |
| "grad_norm": 0.43124097095242286, |
| "learning_rate": 5.787803795514466e-06, |
| "loss": 0.1493, |
| "step": 955 |
| }, |
| { |
| "epoch": 3.876962025316456, |
| "grad_norm": 0.43476471050461224, |
| "learning_rate": 5.747926453969576e-06, |
| "loss": 0.1437, |
| "step": 956 |
| }, |
| { |
| "epoch": 3.8810126582278484, |
| "grad_norm": 0.4296898018522635, |
| "learning_rate": 5.708163896665708e-06, |
| "loss": 0.1691, |
| "step": 957 |
| }, |
| { |
| "epoch": 3.8850632911392404, |
| "grad_norm": 0.4252489850712542, |
| "learning_rate": 5.668516443845047e-06, |
| "loss": 0.1674, |
| "step": 958 |
| }, |
| { |
| "epoch": 3.889113924050633, |
| "grad_norm": 0.431977197311288, |
| "learning_rate": 5.6289844148227225e-06, |
| "loss": 0.1739, |
| "step": 959 |
| }, |
| { |
| "epoch": 3.8931645569620255, |
| "grad_norm": 0.4319335264824662, |
| "learning_rate": 5.5895681279842615e-06, |
| "loss": 0.1681, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.8972151898734175, |
| "grad_norm": 0.4257933202082392, |
| "learning_rate": 5.550267900783019e-06, |
| "loss": 0.163, |
| "step": 961 |
| }, |
| { |
| "epoch": 3.90126582278481, |
| "grad_norm": 0.4388916979751932, |
| "learning_rate": 5.511084049737623e-06, |
| "loss": 0.1509, |
| "step": 962 |
| }, |
| { |
| "epoch": 3.9053164556962026, |
| "grad_norm": 0.4479729125940824, |
| "learning_rate": 5.4720168904294215e-06, |
| "loss": 0.1685, |
| "step": 963 |
| }, |
| { |
| "epoch": 3.909367088607595, |
| "grad_norm": 0.4388614253430693, |
| "learning_rate": 5.433066737499948e-06, |
| "loss": 0.1654, |
| "step": 964 |
| }, |
| { |
| "epoch": 3.9134177215189876, |
| "grad_norm": 0.42703883563150746, |
| "learning_rate": 5.394233904648376e-06, |
| "loss": 0.1523, |
| "step": 965 |
| }, |
| { |
| "epoch": 3.9174683544303797, |
| "grad_norm": 0.4396117621832177, |
| "learning_rate": 5.355518704628997e-06, |
| "loss": 0.1614, |
| "step": 966 |
| }, |
| { |
| "epoch": 3.921518987341772, |
| "grad_norm": 0.43850500712320317, |
| "learning_rate": 5.316921449248731e-06, |
| "loss": 0.1628, |
| "step": 967 |
| }, |
| { |
| "epoch": 3.9255696202531647, |
| "grad_norm": 0.42473173199472664, |
| "learning_rate": 5.278442449364538e-06, |
| "loss": 0.165, |
| "step": 968 |
| }, |
| { |
| "epoch": 3.9296202531645568, |
| "grad_norm": 0.42256447029379, |
| "learning_rate": 5.240082014881016e-06, |
| "loss": 0.157, |
| "step": 969 |
| }, |
| { |
| "epoch": 3.9336708860759493, |
| "grad_norm": 0.42339210635375435, |
| "learning_rate": 5.201840454747822e-06, |
| "loss": 0.1782, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.937721518987342, |
| "grad_norm": 0.4323652289332351, |
| "learning_rate": 5.163718076957223e-06, |
| "loss": 0.1578, |
| "step": 971 |
| }, |
| { |
| "epoch": 3.9417721518987343, |
| "grad_norm": 0.4362206738819554, |
| "learning_rate": 5.125715188541609e-06, |
| "loss": 0.1536, |
| "step": 972 |
| }, |
| { |
| "epoch": 3.945822784810127, |
| "grad_norm": 0.4383492420169514, |
| "learning_rate": 5.087832095571021e-06, |
| "loss": 0.1619, |
| "step": 973 |
| }, |
| { |
| "epoch": 3.949873417721519, |
| "grad_norm": 0.42970563482354945, |
| "learning_rate": 5.0500691031506766e-06, |
| "loss": 0.1578, |
| "step": 974 |
| }, |
| { |
| "epoch": 3.9539240506329114, |
| "grad_norm": 0.41387734979146934, |
| "learning_rate": 5.01242651541854e-06, |
| "loss": 0.1636, |
| "step": 975 |
| }, |
| { |
| "epoch": 3.957974683544304, |
| "grad_norm": 0.4378257711385319, |
| "learning_rate": 4.974904635542815e-06, |
| "loss": 0.1568, |
| "step": 976 |
| }, |
| { |
| "epoch": 3.962025316455696, |
| "grad_norm": 0.44678864430672727, |
| "learning_rate": 4.937503765719582e-06, |
| "loss": 0.1732, |
| "step": 977 |
| }, |
| { |
| "epoch": 3.9660759493670885, |
| "grad_norm": 0.422467505400545, |
| "learning_rate": 4.900224207170299e-06, |
| "loss": 0.1627, |
| "step": 978 |
| }, |
| { |
| "epoch": 3.970126582278481, |
| "grad_norm": 0.4433730861123042, |
| "learning_rate": 4.8630662601394065e-06, |
| "loss": 0.1727, |
| "step": 979 |
| }, |
| { |
| "epoch": 3.9741772151898735, |
| "grad_norm": 0.4290438473376259, |
| "learning_rate": 4.8260302238918995e-06, |
| "loss": 0.1696, |
| "step": 980 |
| }, |
| { |
| "epoch": 3.978227848101266, |
| "grad_norm": 0.4693405755886944, |
| "learning_rate": 4.789116396710924e-06, |
| "loss": 0.1399, |
| "step": 981 |
| }, |
| { |
| "epoch": 3.982278481012658, |
| "grad_norm": 0.4397368672727158, |
| "learning_rate": 4.752325075895368e-06, |
| "loss": 0.1555, |
| "step": 982 |
| }, |
| { |
| "epoch": 3.9863291139240506, |
| "grad_norm": 0.4204473096254341, |
| "learning_rate": 4.715656557757473e-06, |
| "loss": 0.16, |
| "step": 983 |
| }, |
| { |
| "epoch": 3.990379746835443, |
| "grad_norm": 0.42890487842286984, |
| "learning_rate": 4.679111137620442e-06, |
| "loss": 0.1507, |
| "step": 984 |
| }, |
| { |
| "epoch": 3.9944303797468352, |
| "grad_norm": 0.4348503203294379, |
| "learning_rate": 4.6426891098160585e-06, |
| "loss": 0.175, |
| "step": 985 |
| }, |
| { |
| "epoch": 3.9984810126582278, |
| "grad_norm": 0.42944307917462266, |
| "learning_rate": 4.6063907676823474e-06, |
| "loss": 0.1815, |
| "step": 986 |
| }, |
| { |
| "epoch": 4.004050632911392, |
| "grad_norm": 0.48903034558206, |
| "learning_rate": 4.570216403561141e-06, |
| "loss": 0.0908, |
| "step": 987 |
| }, |
| { |
| "epoch": 4.008101265822785, |
| "grad_norm": 0.3876685788517385, |
| "learning_rate": 4.534166308795815e-06, |
| "loss": 0.0772, |
| "step": 988 |
| }, |
| { |
| "epoch": 4.012151898734177, |
| "grad_norm": 0.3939799184180199, |
| "learning_rate": 4.498240773728859e-06, |
| "loss": 0.1093, |
| "step": 989 |
| }, |
| { |
| "epoch": 4.01620253164557, |
| "grad_norm": 0.33036039396107747, |
| "learning_rate": 4.462440087699609e-06, |
| "loss": 0.0921, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.020253164556962, |
| "grad_norm": 0.3080086213474653, |
| "learning_rate": 4.426764539041861e-06, |
| "loss": 0.108, |
| "step": 991 |
| }, |
| { |
| "epoch": 4.024303797468354, |
| "grad_norm": 0.29913698841610586, |
| "learning_rate": 4.391214415081582e-06, |
| "loss": 0.0859, |
| "step": 992 |
| }, |
| { |
| "epoch": 4.028354430379747, |
| "grad_norm": 0.30398389491586025, |
| "learning_rate": 4.355790002134579e-06, |
| "loss": 0.0821, |
| "step": 993 |
| }, |
| { |
| "epoch": 4.032405063291139, |
| "grad_norm": 0.35767312095131215, |
| "learning_rate": 4.320491585504207e-06, |
| "loss": 0.0906, |
| "step": 994 |
| }, |
| { |
| "epoch": 4.036455696202531, |
| "grad_norm": 0.43157914491452565, |
| "learning_rate": 4.2853194494790615e-06, |
| "loss": 0.1, |
| "step": 995 |
| }, |
| { |
| "epoch": 4.040506329113924, |
| "grad_norm": 0.3512919388650817, |
| "learning_rate": 4.250273877330691e-06, |
| "loss": 0.0723, |
| "step": 996 |
| }, |
| { |
| "epoch": 4.044556962025316, |
| "grad_norm": 0.4295761096347376, |
| "learning_rate": 4.215355151311313e-06, |
| "loss": 0.091, |
| "step": 997 |
| }, |
| { |
| "epoch": 4.048607594936709, |
| "grad_norm": 0.4367526429530527, |
| "learning_rate": 4.180563552651542e-06, |
| "loss": 0.1057, |
| "step": 998 |
| }, |
| { |
| "epoch": 4.052658227848101, |
| "grad_norm": 0.3811467500828999, |
| "learning_rate": 4.145899361558147e-06, |
| "loss": 0.0796, |
| "step": 999 |
| }, |
| { |
| "epoch": 4.056708860759493, |
| "grad_norm": 0.34829989111771464, |
| "learning_rate": 4.111362857211738e-06, |
| "loss": 0.0945, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.060759493670886, |
| "grad_norm": 0.3359000579315526, |
| "learning_rate": 4.076954317764592e-06, |
| "loss": 0.0857, |
| "step": 1001 |
| }, |
| { |
| "epoch": 4.0648101265822785, |
| "grad_norm": 0.3094556629647863, |
| "learning_rate": 4.042674020338335e-06, |
| "loss": 0.0923, |
| "step": 1002 |
| }, |
| { |
| "epoch": 4.0688607594936705, |
| "grad_norm": 0.30277342342667224, |
| "learning_rate": 4.0085222410217835e-06, |
| "loss": 0.0875, |
| "step": 1003 |
| }, |
| { |
| "epoch": 4.0729113924050635, |
| "grad_norm": 0.31598960511201934, |
| "learning_rate": 3.974499254868674e-06, |
| "loss": 0.088, |
| "step": 1004 |
| }, |
| { |
| "epoch": 4.076962025316456, |
| "grad_norm": 0.323811915125567, |
| "learning_rate": 3.940605335895451e-06, |
| "loss": 0.0828, |
| "step": 1005 |
| }, |
| { |
| "epoch": 4.0810126582278485, |
| "grad_norm": 0.32711087495592994, |
| "learning_rate": 3.90684075707908e-06, |
| "loss": 0.0876, |
| "step": 1006 |
| }, |
| { |
| "epoch": 4.085063291139241, |
| "grad_norm": 0.3109673159770605, |
| "learning_rate": 3.8732057903548505e-06, |
| "loss": 0.0905, |
| "step": 1007 |
| }, |
| { |
| "epoch": 4.089113924050633, |
| "grad_norm": 0.31781091137848144, |
| "learning_rate": 3.8397007066141375e-06, |
| "loss": 0.0956, |
| "step": 1008 |
| }, |
| { |
| "epoch": 4.093164556962026, |
| "grad_norm": 0.31063755209093574, |
| "learning_rate": 3.806325775702304e-06, |
| "loss": 0.0934, |
| "step": 1009 |
| }, |
| { |
| "epoch": 4.097215189873418, |
| "grad_norm": 0.2995065975843054, |
| "learning_rate": 3.773081266416434e-06, |
| "loss": 0.0982, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.10126582278481, |
| "grad_norm": 0.29190327836893737, |
| "learning_rate": 3.739967446503245e-06, |
| "loss": 0.0795, |
| "step": 1011 |
| }, |
| { |
| "epoch": 4.105316455696203, |
| "grad_norm": 0.2814205341428212, |
| "learning_rate": 3.706984582656894e-06, |
| "loss": 0.0734, |
| "step": 1012 |
| }, |
| { |
| "epoch": 4.109367088607595, |
| "grad_norm": 0.2964348002463654, |
| "learning_rate": 3.6741329405168237e-06, |
| "loss": 0.0834, |
| "step": 1013 |
| }, |
| { |
| "epoch": 4.113417721518988, |
| "grad_norm": 0.3185847884530159, |
| "learning_rate": 3.641412784665648e-06, |
| "loss": 0.0937, |
| "step": 1014 |
| }, |
| { |
| "epoch": 4.11746835443038, |
| "grad_norm": 0.32859838198868146, |
| "learning_rate": 3.608824378627005e-06, |
| "loss": 0.097, |
| "step": 1015 |
| }, |
| { |
| "epoch": 4.121518987341772, |
| "grad_norm": 0.3103120155002663, |
| "learning_rate": 3.5763679848634337e-06, |
| "loss": 0.0828, |
| "step": 1016 |
| }, |
| { |
| "epoch": 4.125569620253165, |
| "grad_norm": 0.31174869672025396, |
| "learning_rate": 3.544043864774269e-06, |
| "loss": 0.0829, |
| "step": 1017 |
| }, |
| { |
| "epoch": 4.129620253164557, |
| "grad_norm": 0.3070035304893173, |
| "learning_rate": 3.5118522786935282e-06, |
| "loss": 0.0879, |
| "step": 1018 |
| }, |
| { |
| "epoch": 4.133670886075949, |
| "grad_norm": 0.33388824655242416, |
| "learning_rate": 3.479793485887819e-06, |
| "loss": 0.0892, |
| "step": 1019 |
| }, |
| { |
| "epoch": 4.137721518987342, |
| "grad_norm": 0.2996352478020263, |
| "learning_rate": 3.4478677445542653e-06, |
| "loss": 0.0797, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.141772151898734, |
| "grad_norm": 0.2895911252200642, |
| "learning_rate": 3.4160753118183767e-06, |
| "loss": 0.081, |
| "step": 1021 |
| }, |
| { |
| "epoch": 4.145822784810127, |
| "grad_norm": 0.3092598102088199, |
| "learning_rate": 3.3844164437320527e-06, |
| "loss": 0.0849, |
| "step": 1022 |
| }, |
| { |
| "epoch": 4.149873417721519, |
| "grad_norm": 0.30858518350838177, |
| "learning_rate": 3.3528913952714558e-06, |
| "loss": 0.098, |
| "step": 1023 |
| }, |
| { |
| "epoch": 4.153924050632911, |
| "grad_norm": 0.29631549119580924, |
| "learning_rate": 3.321500420335e-06, |
| "loss": 0.0786, |
| "step": 1024 |
| }, |
| { |
| "epoch": 4.157974683544304, |
| "grad_norm": 0.3009490647546524, |
| "learning_rate": 3.290243771741275e-06, |
| "loss": 0.0899, |
| "step": 1025 |
| }, |
| { |
| "epoch": 4.162025316455696, |
| "grad_norm": 0.3011775880626061, |
| "learning_rate": 3.2591217012270325e-06, |
| "loss": 0.0859, |
| "step": 1026 |
| }, |
| { |
| "epoch": 4.166075949367088, |
| "grad_norm": 0.28752724022693166, |
| "learning_rate": 3.228134459445149e-06, |
| "loss": 0.082, |
| "step": 1027 |
| }, |
| { |
| "epoch": 4.170126582278481, |
| "grad_norm": 0.29596421199523926, |
| "learning_rate": 3.1972822959626205e-06, |
| "loss": 0.0771, |
| "step": 1028 |
| }, |
| { |
| "epoch": 4.174177215189873, |
| "grad_norm": 0.31043225854209217, |
| "learning_rate": 3.166565459258513e-06, |
| "loss": 0.1054, |
| "step": 1029 |
| }, |
| { |
| "epoch": 4.178227848101266, |
| "grad_norm": 0.3040500558435447, |
| "learning_rate": 3.1359841967220193e-06, |
| "loss": 0.1004, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.182278481012658, |
| "grad_norm": 0.2882601944982037, |
| "learning_rate": 3.105538754650419e-06, |
| "loss": 0.0768, |
| "step": 1031 |
| }, |
| { |
| "epoch": 4.18632911392405, |
| "grad_norm": 0.3022976310082866, |
| "learning_rate": 3.07522937824712e-06, |
| "loss": 0.0775, |
| "step": 1032 |
| }, |
| { |
| "epoch": 4.190379746835443, |
| "grad_norm": 0.3032382575117942, |
| "learning_rate": 3.0450563116196697e-06, |
| "loss": 0.0677, |
| "step": 1033 |
| }, |
| { |
| "epoch": 4.194430379746835, |
| "grad_norm": 0.2997781351045828, |
| "learning_rate": 3.0150197977778008e-06, |
| "loss": 0.0825, |
| "step": 1034 |
| }, |
| { |
| "epoch": 4.1984810126582275, |
| "grad_norm": 0.2875121143013046, |
| "learning_rate": 2.985120078631465e-06, |
| "loss": 0.0788, |
| "step": 1035 |
| }, |
| { |
| "epoch": 4.2025316455696204, |
| "grad_norm": 0.29347041971937243, |
| "learning_rate": 2.9553573949888893e-06, |
| "loss": 0.0859, |
| "step": 1036 |
| }, |
| { |
| "epoch": 4.2065822784810125, |
| "grad_norm": 0.3114121419403578, |
| "learning_rate": 2.9257319865546384e-06, |
| "loss": 0.0917, |
| "step": 1037 |
| }, |
| { |
| "epoch": 4.2106329113924055, |
| "grad_norm": 0.2842313929718759, |
| "learning_rate": 2.896244091927678e-06, |
| "loss": 0.0913, |
| "step": 1038 |
| }, |
| { |
| "epoch": 4.2146835443037975, |
| "grad_norm": 0.2985926422534801, |
| "learning_rate": 2.8668939485994584e-06, |
| "loss": 0.0865, |
| "step": 1039 |
| }, |
| { |
| "epoch": 4.21873417721519, |
| "grad_norm": 0.2931731630859391, |
| "learning_rate": 2.837681792951994e-06, |
| "loss": 0.073, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.222784810126583, |
| "grad_norm": 0.30520199643869905, |
| "learning_rate": 2.808607860255981e-06, |
| "loss": 0.0862, |
| "step": 1041 |
| }, |
| { |
| "epoch": 4.226835443037975, |
| "grad_norm": 0.30485690448687675, |
| "learning_rate": 2.7796723846688634e-06, |
| "loss": 0.0809, |
| "step": 1042 |
| }, |
| { |
| "epoch": 4.230886075949367, |
| "grad_norm": 0.30166475163176365, |
| "learning_rate": 2.7508755992329937e-06, |
| "loss": 0.0811, |
| "step": 1043 |
| }, |
| { |
| "epoch": 4.23493670886076, |
| "grad_norm": 0.28626784648093917, |
| "learning_rate": 2.722217735873718e-06, |
| "loss": 0.0805, |
| "step": 1044 |
| }, |
| { |
| "epoch": 4.238987341772152, |
| "grad_norm": 0.29525347081889325, |
| "learning_rate": 2.6936990253975315e-06, |
| "loss": 0.0832, |
| "step": 1045 |
| }, |
| { |
| "epoch": 4.243037974683545, |
| "grad_norm": 0.32296960626570503, |
| "learning_rate": 2.665319697490205e-06, |
| "loss": 0.1174, |
| "step": 1046 |
| }, |
| { |
| "epoch": 4.247088607594937, |
| "grad_norm": 0.29809442973381983, |
| "learning_rate": 2.637079980714945e-06, |
| "loss": 0.0968, |
| "step": 1047 |
| }, |
| { |
| "epoch": 4.251139240506329, |
| "grad_norm": 0.2975622942174841, |
| "learning_rate": 2.6089801025105453e-06, |
| "loss": 0.0768, |
| "step": 1048 |
| }, |
| { |
| "epoch": 4.255189873417722, |
| "grad_norm": 0.2881177162448149, |
| "learning_rate": 2.581020289189571e-06, |
| "loss": 0.0832, |
| "step": 1049 |
| }, |
| { |
| "epoch": 4.259240506329114, |
| "grad_norm": 0.3085103226362676, |
| "learning_rate": 2.553200765936501e-06, |
| "loss": 0.082, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.263291139240506, |
| "grad_norm": 0.31184472574587474, |
| "learning_rate": 2.525521756805962e-06, |
| "loss": 0.1123, |
| "step": 1051 |
| }, |
| { |
| "epoch": 4.267341772151899, |
| "grad_norm": 0.29786547781764805, |
| "learning_rate": 2.497983484720885e-06, |
| "loss": 0.092, |
| "step": 1052 |
| }, |
| { |
| "epoch": 4.271392405063291, |
| "grad_norm": 0.29115987308082175, |
| "learning_rate": 2.470586171470728e-06, |
| "loss": 0.0799, |
| "step": 1053 |
| }, |
| { |
| "epoch": 4.275443037974684, |
| "grad_norm": 0.31110932459369584, |
| "learning_rate": 2.4433300377096836e-06, |
| "loss": 0.0915, |
| "step": 1054 |
| }, |
| { |
| "epoch": 4.279493670886076, |
| "grad_norm": 0.2800748578428545, |
| "learning_rate": 2.4162153029549073e-06, |
| "loss": 0.0699, |
| "step": 1055 |
| }, |
| { |
| "epoch": 4.283544303797468, |
| "grad_norm": 0.29777781773700895, |
| "learning_rate": 2.3892421855847458e-06, |
| "loss": 0.0941, |
| "step": 1056 |
| }, |
| { |
| "epoch": 4.287594936708861, |
| "grad_norm": 0.30517811281208945, |
| "learning_rate": 2.362410902836978e-06, |
| "loss": 0.0992, |
| "step": 1057 |
| }, |
| { |
| "epoch": 4.291645569620253, |
| "grad_norm": 0.28311748176612805, |
| "learning_rate": 2.3357216708070653e-06, |
| "loss": 0.0885, |
| "step": 1058 |
| }, |
| { |
| "epoch": 4.295696202531645, |
| "grad_norm": 0.30475024753544094, |
| "learning_rate": 2.309174704446411e-06, |
| "loss": 0.0994, |
| "step": 1059 |
| }, |
| { |
| "epoch": 4.299746835443038, |
| "grad_norm": 0.3024547903734781, |
| "learning_rate": 2.2827702175606437e-06, |
| "loss": 0.0834, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.30379746835443, |
| "grad_norm": 0.30469561181089955, |
| "learning_rate": 2.256508422807855e-06, |
| "loss": 0.0898, |
| "step": 1061 |
| }, |
| { |
| "epoch": 4.307848101265823, |
| "grad_norm": 0.29511893747180257, |
| "learning_rate": 2.230389531696946e-06, |
| "loss": 0.0699, |
| "step": 1062 |
| }, |
| { |
| "epoch": 4.311898734177215, |
| "grad_norm": 0.29838264887016674, |
| "learning_rate": 2.204413754585857e-06, |
| "loss": 0.0738, |
| "step": 1063 |
| }, |
| { |
| "epoch": 4.315949367088607, |
| "grad_norm": 0.31746463845270906, |
| "learning_rate": 2.1785813006799406e-06, |
| "loss": 0.1078, |
| "step": 1064 |
| }, |
| { |
| "epoch": 4.32, |
| "grad_norm": 0.3105536915888808, |
| "learning_rate": 2.1528923780302224e-06, |
| "loss": 0.0847, |
| "step": 1065 |
| }, |
| { |
| "epoch": 4.324050632911392, |
| "grad_norm": 0.2845877826113816, |
| "learning_rate": 2.127347193531757e-06, |
| "loss": 0.0794, |
| "step": 1066 |
| }, |
| { |
| "epoch": 4.328101265822784, |
| "grad_norm": 0.30270773588493216, |
| "learning_rate": 2.101945952921942e-06, |
| "loss": 0.0976, |
| "step": 1067 |
| }, |
| { |
| "epoch": 4.332151898734177, |
| "grad_norm": 0.2815810023745531, |
| "learning_rate": 2.0766888607788906e-06, |
| "loss": 0.0834, |
| "step": 1068 |
| }, |
| { |
| "epoch": 4.3362025316455695, |
| "grad_norm": 0.290879794424829, |
| "learning_rate": 2.0515761205197337e-06, |
| "loss": 0.0712, |
| "step": 1069 |
| }, |
| { |
| "epoch": 4.340253164556962, |
| "grad_norm": 0.29006299048957773, |
| "learning_rate": 2.0266079343990453e-06, |
| "loss": 0.0809, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.3443037974683545, |
| "grad_norm": 0.3006641789228888, |
| "learning_rate": 2.0017845035071494e-06, |
| "loss": 0.0916, |
| "step": 1071 |
| }, |
| { |
| "epoch": 4.348354430379747, |
| "grad_norm": 0.275408398510679, |
| "learning_rate": 1.9771060277685537e-06, |
| "loss": 0.0832, |
| "step": 1072 |
| }, |
| { |
| "epoch": 4.3524050632911395, |
| "grad_norm": 0.2873332060397401, |
| "learning_rate": 1.95257270594031e-06, |
| "loss": 0.0831, |
| "step": 1073 |
| }, |
| { |
| "epoch": 4.356455696202532, |
| "grad_norm": 0.29338392670687213, |
| "learning_rate": 1.9281847356104188e-06, |
| "loss": 0.0786, |
| "step": 1074 |
| }, |
| { |
| "epoch": 4.360506329113924, |
| "grad_norm": 0.3017725280258459, |
| "learning_rate": 1.9039423131962365e-06, |
| "loss": 0.0982, |
| "step": 1075 |
| }, |
| { |
| "epoch": 4.364556962025317, |
| "grad_norm": 0.30733137024257884, |
| "learning_rate": 1.8798456339429027e-06, |
| "loss": 0.0909, |
| "step": 1076 |
| }, |
| { |
| "epoch": 4.368607594936709, |
| "grad_norm": 0.30583661357705916, |
| "learning_rate": 1.8558948919217612e-06, |
| "loss": 0.0844, |
| "step": 1077 |
| }, |
| { |
| "epoch": 4.372658227848102, |
| "grad_norm": 0.31699046625170196, |
| "learning_rate": 1.8320902800287954e-06, |
| "loss": 0.0804, |
| "step": 1078 |
| }, |
| { |
| "epoch": 4.376708860759494, |
| "grad_norm": 0.3118089698392865, |
| "learning_rate": 1.8084319899830726e-06, |
| "loss": 0.0953, |
| "step": 1079 |
| }, |
| { |
| "epoch": 4.380759493670886, |
| "grad_norm": 0.30436487939606993, |
| "learning_rate": 1.7849202123252097e-06, |
| "loss": 0.1005, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.384810126582279, |
| "grad_norm": 0.3149297311741284, |
| "learning_rate": 1.7615551364158401e-06, |
| "loss": 0.0954, |
| "step": 1081 |
| }, |
| { |
| "epoch": 4.388860759493671, |
| "grad_norm": 0.2850011981182486, |
| "learning_rate": 1.738336950434061e-06, |
| "loss": 0.088, |
| "step": 1082 |
| }, |
| { |
| "epoch": 4.392911392405063, |
| "grad_norm": 0.30304480484546337, |
| "learning_rate": 1.715265841375957e-06, |
| "loss": 0.0892, |
| "step": 1083 |
| }, |
| { |
| "epoch": 4.396962025316456, |
| "grad_norm": 0.29031826164418967, |
| "learning_rate": 1.6923419950530684e-06, |
| "loss": 0.0634, |
| "step": 1084 |
| }, |
| { |
| "epoch": 4.401012658227848, |
| "grad_norm": 0.3018334648928463, |
| "learning_rate": 1.6695655960909008e-06, |
| "loss": 0.095, |
| "step": 1085 |
| }, |
| { |
| "epoch": 4.405063291139241, |
| "grad_norm": 0.29185664158478597, |
| "learning_rate": 1.646936827927441e-06, |
| "loss": 0.0854, |
| "step": 1086 |
| }, |
| { |
| "epoch": 4.409113924050633, |
| "grad_norm": 0.29308916513695793, |
| "learning_rate": 1.6244558728116766e-06, |
| "loss": 0.0825, |
| "step": 1087 |
| }, |
| { |
| "epoch": 4.413164556962025, |
| "grad_norm": 0.295472061569332, |
| "learning_rate": 1.6021229118021265e-06, |
| "loss": 0.0809, |
| "step": 1088 |
| }, |
| { |
| "epoch": 4.417215189873418, |
| "grad_norm": 0.29822237866428597, |
| "learning_rate": 1.5799381247653967e-06, |
| "loss": 0.0871, |
| "step": 1089 |
| }, |
| { |
| "epoch": 4.42126582278481, |
| "grad_norm": 0.2956173918732139, |
| "learning_rate": 1.5579016903747013e-06, |
| "loss": 0.07, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.425316455696202, |
| "grad_norm": 0.29570443909543764, |
| "learning_rate": 1.5360137861084656e-06, |
| "loss": 0.0911, |
| "step": 1091 |
| }, |
| { |
| "epoch": 4.429367088607595, |
| "grad_norm": 0.3061693239671819, |
| "learning_rate": 1.5142745882488475e-06, |
| "loss": 0.0861, |
| "step": 1092 |
| }, |
| { |
| "epoch": 4.433417721518987, |
| "grad_norm": 0.30290240953405584, |
| "learning_rate": 1.4926842718803691e-06, |
| "loss": 0.1095, |
| "step": 1093 |
| }, |
| { |
| "epoch": 4.43746835443038, |
| "grad_norm": 0.2924550696890672, |
| "learning_rate": 1.4712430108884657e-06, |
| "loss": 0.0873, |
| "step": 1094 |
| }, |
| { |
| "epoch": 4.441518987341772, |
| "grad_norm": 0.2945193363376506, |
| "learning_rate": 1.4499509779581078e-06, |
| "loss": 0.0887, |
| "step": 1095 |
| }, |
| { |
| "epoch": 4.445569620253164, |
| "grad_norm": 0.2837699217715454, |
| "learning_rate": 1.4288083445723988e-06, |
| "loss": 0.0738, |
| "step": 1096 |
| }, |
| { |
| "epoch": 4.449620253164557, |
| "grad_norm": 0.30132133989011106, |
| "learning_rate": 1.4078152810112045e-06, |
| "loss": 0.0848, |
| "step": 1097 |
| }, |
| { |
| "epoch": 4.453670886075949, |
| "grad_norm": 0.30394133210596885, |
| "learning_rate": 1.3869719563497697e-06, |
| "loss": 0.0773, |
| "step": 1098 |
| }, |
| { |
| "epoch": 4.457721518987341, |
| "grad_norm": 0.30732762296879623, |
| "learning_rate": 1.3662785384573663e-06, |
| "loss": 0.0879, |
| "step": 1099 |
| }, |
| { |
| "epoch": 4.461772151898734, |
| "grad_norm": 0.27806594062312306, |
| "learning_rate": 1.3457351939959383e-06, |
| "loss": 0.0671, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.465822784810126, |
| "grad_norm": 0.3011993096182882, |
| "learning_rate": 1.3253420884187551e-06, |
| "loss": 0.0711, |
| "step": 1101 |
| }, |
| { |
| "epoch": 4.469873417721519, |
| "grad_norm": 0.3073752830701084, |
| "learning_rate": 1.3050993859690953e-06, |
| "loss": 0.0829, |
| "step": 1102 |
| }, |
| { |
| "epoch": 4.473924050632911, |
| "grad_norm": 0.3038560400508032, |
| "learning_rate": 1.2850072496788869e-06, |
| "loss": 0.0809, |
| "step": 1103 |
| }, |
| { |
| "epoch": 4.4779746835443035, |
| "grad_norm": 0.2821636290416366, |
| "learning_rate": 1.2650658413674434e-06, |
| "loss": 0.0619, |
| "step": 1104 |
| }, |
| { |
| "epoch": 4.4820253164556965, |
| "grad_norm": 0.30464923795687704, |
| "learning_rate": 1.2452753216401226e-06, |
| "loss": 0.0917, |
| "step": 1105 |
| }, |
| { |
| "epoch": 4.4860759493670885, |
| "grad_norm": 0.30563821487210563, |
| "learning_rate": 1.2256358498870503e-06, |
| "loss": 0.0827, |
| "step": 1106 |
| }, |
| { |
| "epoch": 4.490126582278481, |
| "grad_norm": 0.3077165887046052, |
| "learning_rate": 1.2061475842818337e-06, |
| "loss": 0.0682, |
| "step": 1107 |
| }, |
| { |
| "epoch": 4.494177215189874, |
| "grad_norm": 0.30142349398017265, |
| "learning_rate": 1.1868106817802816e-06, |
| "loss": 0.0886, |
| "step": 1108 |
| }, |
| { |
| "epoch": 4.498227848101266, |
| "grad_norm": 0.3001573063324343, |
| "learning_rate": 1.1676252981191482e-06, |
| "loss": 0.0788, |
| "step": 1109 |
| }, |
| { |
| "epoch": 4.502278481012659, |
| "grad_norm": 0.2884238068789639, |
| "learning_rate": 1.1485915878148823e-06, |
| "loss": 0.0972, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.506329113924051, |
| "grad_norm": 0.29645865155464596, |
| "learning_rate": 1.1297097041623584e-06, |
| "loss": 0.0768, |
| "step": 1111 |
| }, |
| { |
| "epoch": 4.510379746835443, |
| "grad_norm": 0.29275350724817456, |
| "learning_rate": 1.1109797992336847e-06, |
| "loss": 0.0985, |
| "step": 1112 |
| }, |
| { |
| "epoch": 4.514430379746836, |
| "grad_norm": 0.28683545504316843, |
| "learning_rate": 1.092402023876933e-06, |
| "loss": 0.0868, |
| "step": 1113 |
| }, |
| { |
| "epoch": 4.518481012658228, |
| "grad_norm": 0.2881863549182941, |
| "learning_rate": 1.0739765277149527e-06, |
| "loss": 0.0833, |
| "step": 1114 |
| }, |
| { |
| "epoch": 4.52253164556962, |
| "grad_norm": 0.3181706336510317, |
| "learning_rate": 1.0557034591441596e-06, |
| "loss": 0.0914, |
| "step": 1115 |
| }, |
| { |
| "epoch": 4.526582278481013, |
| "grad_norm": 0.2842094993793465, |
| "learning_rate": 1.0375829653333324e-06, |
| "loss": 0.0747, |
| "step": 1116 |
| }, |
| { |
| "epoch": 4.530632911392405, |
| "grad_norm": 0.2934633502495019, |
| "learning_rate": 1.0196151922224385e-06, |
| "loss": 0.0877, |
| "step": 1117 |
| }, |
| { |
| "epoch": 4.534683544303798, |
| "grad_norm": 0.2856016606225609, |
| "learning_rate": 1.0018002845214526e-06, |
| "loss": 0.0841, |
| "step": 1118 |
| }, |
| { |
| "epoch": 4.53873417721519, |
| "grad_norm": 0.2991430056012222, |
| "learning_rate": 9.841383857091947e-07, |
| "loss": 0.0952, |
| "step": 1119 |
| }, |
| { |
| "epoch": 4.542784810126582, |
| "grad_norm": 0.2871136603448637, |
| "learning_rate": 9.666296380321616e-07, |
| "loss": 0.0716, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.546835443037975, |
| "grad_norm": 0.2867125967223642, |
| "learning_rate": 9.492741825034124e-07, |
| "loss": 0.0571, |
| "step": 1121 |
| }, |
| { |
| "epoch": 4.550886075949367, |
| "grad_norm": 0.29885347808049206, |
| "learning_rate": 9.320721589013892e-07, |
| "loss": 0.0942, |
| "step": 1122 |
| }, |
| { |
| "epoch": 4.55493670886076, |
| "grad_norm": 0.310220117463769, |
| "learning_rate": 9.150237057688339e-07, |
| "loss": 0.0911, |
| "step": 1123 |
| }, |
| { |
| "epoch": 4.558987341772152, |
| "grad_norm": 0.28596851708115156, |
| "learning_rate": 8.981289604116328e-07, |
| "loss": 0.0806, |
| "step": 1124 |
| }, |
| { |
| "epoch": 4.563037974683544, |
| "grad_norm": 0.3015866169828308, |
| "learning_rate": 8.813880588977542e-07, |
| "loss": 0.0782, |
| "step": 1125 |
| }, |
| { |
| "epoch": 4.567088607594937, |
| "grad_norm": 0.2899259095519048, |
| "learning_rate": 8.648011360561126e-07, |
| "loss": 0.0719, |
| "step": 1126 |
| }, |
| { |
| "epoch": 4.571139240506329, |
| "grad_norm": 0.2882338722309331, |
| "learning_rate": 8.483683254755037e-07, |
| "loss": 0.0726, |
| "step": 1127 |
| }, |
| { |
| "epoch": 4.575189873417721, |
| "grad_norm": 0.2801382616689189, |
| "learning_rate": 8.320897595035227e-07, |
| "loss": 0.0809, |
| "step": 1128 |
| }, |
| { |
| "epoch": 4.579240506329114, |
| "grad_norm": 0.3060788608830804, |
| "learning_rate": 8.159655692455093e-07, |
| "loss": 0.0757, |
| "step": 1129 |
| }, |
| { |
| "epoch": 4.583291139240506, |
| "grad_norm": 0.3104130288865869, |
| "learning_rate": 7.999958845634648e-07, |
| "loss": 0.0993, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.587341772151898, |
| "grad_norm": 0.30326186643444764, |
| "learning_rate": 7.841808340750478e-07, |
| "loss": 0.1029, |
| "step": 1131 |
| }, |
| { |
| "epoch": 4.591392405063291, |
| "grad_norm": 0.304023756545761, |
| "learning_rate": 7.685205451524869e-07, |
| "loss": 0.0791, |
| "step": 1132 |
| }, |
| { |
| "epoch": 4.595443037974683, |
| "grad_norm": 0.28010956250572894, |
| "learning_rate": 7.530151439216027e-07, |
| "loss": 0.0681, |
| "step": 1133 |
| }, |
| { |
| "epoch": 4.599493670886076, |
| "grad_norm": 0.3031515771371865, |
| "learning_rate": 7.376647552607675e-07, |
| "loss": 0.0908, |
| "step": 1134 |
| }, |
| { |
| "epoch": 4.603544303797468, |
| "grad_norm": 0.3093337917653243, |
| "learning_rate": 7.224695027998963e-07, |
| "loss": 0.103, |
| "step": 1135 |
| }, |
| { |
| "epoch": 4.6075949367088604, |
| "grad_norm": 0.30185749680095536, |
| "learning_rate": 7.07429508919466e-07, |
| "loss": 0.0879, |
| "step": 1136 |
| }, |
| { |
| "epoch": 4.611645569620253, |
| "grad_norm": 0.2949492307615738, |
| "learning_rate": 6.925448947495206e-07, |
| "loss": 0.0922, |
| "step": 1137 |
| }, |
| { |
| "epoch": 4.6156962025316455, |
| "grad_norm": 0.29037405163527297, |
| "learning_rate": 6.778157801686936e-07, |
| "loss": 0.0673, |
| "step": 1138 |
| }, |
| { |
| "epoch": 4.619746835443038, |
| "grad_norm": 0.30330358212035613, |
| "learning_rate": 6.632422838032515e-07, |
| "loss": 0.0814, |
| "step": 1139 |
| }, |
| { |
| "epoch": 4.6237974683544305, |
| "grad_norm": 0.29967661026038295, |
| "learning_rate": 6.488245230261281e-07, |
| "loss": 0.0901, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.627848101265823, |
| "grad_norm": 0.30242033462979845, |
| "learning_rate": 6.345626139559868e-07, |
| "loss": 0.0844, |
| "step": 1141 |
| }, |
| { |
| "epoch": 4.6318987341772155, |
| "grad_norm": 0.3030419425853118, |
| "learning_rate": 6.204566714562866e-07, |
| "loss": 0.0839, |
| "step": 1142 |
| }, |
| { |
| "epoch": 4.635949367088608, |
| "grad_norm": 0.2966345485226554, |
| "learning_rate": 6.06506809134344e-07, |
| "loss": 0.0752, |
| "step": 1143 |
| }, |
| { |
| "epoch": 4.64, |
| "grad_norm": 0.29219743390419817, |
| "learning_rate": 5.927131393404373e-07, |
| "loss": 0.0773, |
| "step": 1144 |
| }, |
| { |
| "epoch": 4.644050632911393, |
| "grad_norm": 0.30276334676694255, |
| "learning_rate": 5.790757731668817e-07, |
| "loss": 0.0956, |
| "step": 1145 |
| }, |
| { |
| "epoch": 4.648101265822785, |
| "grad_norm": 0.2904538868346222, |
| "learning_rate": 5.655948204471507e-07, |
| "loss": 0.0672, |
| "step": 1146 |
| }, |
| { |
| "epoch": 4.652151898734177, |
| "grad_norm": 0.2969479857208302, |
| "learning_rate": 5.522703897549875e-07, |
| "loss": 0.0866, |
| "step": 1147 |
| }, |
| { |
| "epoch": 4.65620253164557, |
| "grad_norm": 0.28425128470113786, |
| "learning_rate": 5.391025884035239e-07, |
| "loss": 0.0744, |
| "step": 1148 |
| }, |
| { |
| "epoch": 4.660253164556962, |
| "grad_norm": 0.30992966972352287, |
| "learning_rate": 5.260915224444207e-07, |
| "loss": 0.0921, |
| "step": 1149 |
| }, |
| { |
| "epoch": 4.664303797468355, |
| "grad_norm": 0.29689194473926056, |
| "learning_rate": 5.132372966670129e-07, |
| "loss": 0.0919, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.668354430379747, |
| "grad_norm": 0.28163013735856113, |
| "learning_rate": 5.005400145974704e-07, |
| "loss": 0.0752, |
| "step": 1151 |
| }, |
| { |
| "epoch": 4.672405063291139, |
| "grad_norm": 0.2865851285135141, |
| "learning_rate": 4.879997784979562e-07, |
| "loss": 0.0856, |
| "step": 1152 |
| }, |
| { |
| "epoch": 4.676455696202532, |
| "grad_norm": 0.30528970878579115, |
| "learning_rate": 4.7561668936580984e-07, |
| "loss": 0.0781, |
| "step": 1153 |
| }, |
| { |
| "epoch": 4.680506329113924, |
| "grad_norm": 0.2957833260043001, |
| "learning_rate": 4.6339084693272306e-07, |
| "loss": 0.0916, |
| "step": 1154 |
| }, |
| { |
| "epoch": 4.684556962025317, |
| "grad_norm": 0.29359004900368174, |
| "learning_rate": 4.5132234966395847e-07, |
| "loss": 0.0824, |
| "step": 1155 |
| }, |
| { |
| "epoch": 4.688607594936709, |
| "grad_norm": 0.30408696531244367, |
| "learning_rate": 4.3941129475752795e-07, |
| "loss": 0.088, |
| "step": 1156 |
| }, |
| { |
| "epoch": 4.692658227848101, |
| "grad_norm": 0.2890208786094657, |
| "learning_rate": 4.27657778143431e-07, |
| "loss": 0.0829, |
| "step": 1157 |
| }, |
| { |
| "epoch": 4.696708860759494, |
| "grad_norm": 0.30628633575033914, |
| "learning_rate": 4.1606189448287757e-07, |
| "loss": 0.0964, |
| "step": 1158 |
| }, |
| { |
| "epoch": 4.700759493670886, |
| "grad_norm": 0.2906214007993461, |
| "learning_rate": 4.046237371675177e-07, |
| "loss": 0.0788, |
| "step": 1159 |
| }, |
| { |
| "epoch": 4.704810126582278, |
| "grad_norm": 0.3030249273303387, |
| "learning_rate": 3.9334339831869963e-07, |
| "loss": 0.0765, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.708860759493671, |
| "grad_norm": 0.29444774565423637, |
| "learning_rate": 3.8222096878671955e-07, |
| "loss": 0.0876, |
| "step": 1161 |
| }, |
| { |
| "epoch": 4.712911392405063, |
| "grad_norm": 0.3055923557849022, |
| "learning_rate": 3.7125653815009545e-07, |
| "loss": 0.0889, |
| "step": 1162 |
| }, |
| { |
| "epoch": 4.716962025316455, |
| "grad_norm": 0.299813341611945, |
| "learning_rate": 3.6045019471484974e-07, |
| "loss": 0.0813, |
| "step": 1163 |
| }, |
| { |
| "epoch": 4.721012658227848, |
| "grad_norm": 0.3053745160465112, |
| "learning_rate": 3.498020255137813e-07, |
| "loss": 0.0797, |
| "step": 1164 |
| }, |
| { |
| "epoch": 4.72506329113924, |
| "grad_norm": 0.2988127439638993, |
| "learning_rate": 3.393121163057811e-07, |
| "loss": 0.1039, |
| "step": 1165 |
| }, |
| { |
| "epoch": 4.729113924050633, |
| "grad_norm": 0.2956192857895059, |
| "learning_rate": 3.289805515751399e-07, |
| "loss": 0.0863, |
| "step": 1166 |
| }, |
| { |
| "epoch": 4.733164556962025, |
| "grad_norm": 0.30238282860969273, |
| "learning_rate": 3.188074145308573e-07, |
| "loss": 0.1056, |
| "step": 1167 |
| }, |
| { |
| "epoch": 4.737215189873417, |
| "grad_norm": 0.29892358731924884, |
| "learning_rate": 3.087927871059804e-07, |
| "loss": 0.0993, |
| "step": 1168 |
| }, |
| { |
| "epoch": 4.74126582278481, |
| "grad_norm": 0.2953403765704961, |
| "learning_rate": 2.989367499569418e-07, |
| "loss": 0.0816, |
| "step": 1169 |
| }, |
| { |
| "epoch": 4.745316455696202, |
| "grad_norm": 0.30238449062573675, |
| "learning_rate": 2.8923938246290917e-07, |
| "loss": 0.0918, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.749367088607595, |
| "grad_norm": 0.3083236598445814, |
| "learning_rate": 2.7970076272514804e-07, |
| "loss": 0.1026, |
| "step": 1171 |
| }, |
| { |
| "epoch": 4.7534177215189874, |
| "grad_norm": 0.29094023889362086, |
| "learning_rate": 2.703209675663887e-07, |
| "loss": 0.092, |
| "step": 1172 |
| }, |
| { |
| "epoch": 4.7574683544303795, |
| "grad_norm": 0.2826353195648466, |
| "learning_rate": 2.6110007253021374e-07, |
| "loss": 0.0833, |
| "step": 1173 |
| }, |
| { |
| "epoch": 4.7615189873417725, |
| "grad_norm": 0.3006964794989746, |
| "learning_rate": 2.520381518804471e-07, |
| "loss": 0.0817, |
| "step": 1174 |
| }, |
| { |
| "epoch": 4.7655696202531646, |
| "grad_norm": 0.30155482686582324, |
| "learning_rate": 2.4313527860054585e-07, |
| "loss": 0.092, |
| "step": 1175 |
| }, |
| { |
| "epoch": 4.769620253164557, |
| "grad_norm": 0.29648054204987195, |
| "learning_rate": 2.343915243930317e-07, |
| "loss": 0.0783, |
| "step": 1176 |
| }, |
| { |
| "epoch": 4.77367088607595, |
| "grad_norm": 0.29282725957403316, |
| "learning_rate": 2.2580695967889367e-07, |
| "loss": 0.0836, |
| "step": 1177 |
| }, |
| { |
| "epoch": 4.777721518987342, |
| "grad_norm": 0.29754689302445225, |
| "learning_rate": 2.1738165359704189e-07, |
| "loss": 0.0889, |
| "step": 1178 |
| }, |
| { |
| "epoch": 4.781772151898734, |
| "grad_norm": 0.2849772143473628, |
| "learning_rate": 2.0911567400373257e-07, |
| "loss": 0.0948, |
| "step": 1179 |
| }, |
| { |
| "epoch": 4.785822784810127, |
| "grad_norm": 0.27659757124134066, |
| "learning_rate": 2.0100908747202607e-07, |
| "loss": 0.0718, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.789873417721519, |
| "grad_norm": 0.3071079306718665, |
| "learning_rate": 1.9306195929125638e-07, |
| "loss": 0.0774, |
| "step": 1181 |
| }, |
| { |
| "epoch": 4.793924050632912, |
| "grad_norm": 0.2842887991247307, |
| "learning_rate": 1.8527435346650247e-07, |
| "loss": 0.0747, |
| "step": 1182 |
| }, |
| { |
| "epoch": 4.797974683544304, |
| "grad_norm": 0.2892577554261885, |
| "learning_rate": 1.7764633271807108e-07, |
| "loss": 0.0918, |
| "step": 1183 |
| }, |
| { |
| "epoch": 4.802025316455696, |
| "grad_norm": 0.2744943565942207, |
| "learning_rate": 1.7017795848099262e-07, |
| "loss": 0.0715, |
| "step": 1184 |
| }, |
| { |
| "epoch": 4.806075949367089, |
| "grad_norm": 0.3039007783204275, |
| "learning_rate": 1.6286929090452596e-07, |
| "loss": 0.0985, |
| "step": 1185 |
| }, |
| { |
| "epoch": 4.810126582278481, |
| "grad_norm": 0.28994230927555664, |
| "learning_rate": 1.557203888516745e-07, |
| "loss": 0.0832, |
| "step": 1186 |
| }, |
| { |
| "epoch": 4.814177215189874, |
| "grad_norm": 0.2910779304756549, |
| "learning_rate": 1.487313098987131e-07, |
| "loss": 0.0735, |
| "step": 1187 |
| }, |
| { |
| "epoch": 4.818227848101266, |
| "grad_norm": 0.3018865317822536, |
| "learning_rate": 1.4190211033472402e-07, |
| "loss": 0.1033, |
| "step": 1188 |
| }, |
| { |
| "epoch": 4.822278481012658, |
| "grad_norm": 0.2891339827813415, |
| "learning_rate": 1.3523284516113955e-07, |
| "loss": 0.0946, |
| "step": 1189 |
| }, |
| { |
| "epoch": 4.826329113924051, |
| "grad_norm": 0.29033702992734794, |
| "learning_rate": 1.2872356809130682e-07, |
| "loss": 0.0835, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.830379746835443, |
| "grad_norm": 0.2992089684244637, |
| "learning_rate": 1.2237433155004807e-07, |
| "loss": 0.0841, |
| "step": 1191 |
| }, |
| { |
| "epoch": 4.834430379746835, |
| "grad_norm": 0.30730955660308173, |
| "learning_rate": 1.1618518667323886e-07, |
| "loss": 0.088, |
| "step": 1192 |
| }, |
| { |
| "epoch": 4.838481012658228, |
| "grad_norm": 0.30708713152646117, |
| "learning_rate": 1.1015618330740385e-07, |
| "loss": 0.1107, |
| "step": 1193 |
| }, |
| { |
| "epoch": 4.84253164556962, |
| "grad_norm": 0.2941959539439714, |
| "learning_rate": 1.042873700093061e-07, |
| "loss": 0.0905, |
| "step": 1194 |
| }, |
| { |
| "epoch": 4.846582278481012, |
| "grad_norm": 0.3257503650973894, |
| "learning_rate": 9.857879404556291e-08, |
| "loss": 0.0992, |
| "step": 1195 |
| }, |
| { |
| "epoch": 4.850632911392405, |
| "grad_norm": 0.29030425153656586, |
| "learning_rate": 9.303050139225722e-08, |
| "loss": 0.0824, |
| "step": 1196 |
| }, |
| { |
| "epoch": 4.854683544303797, |
| "grad_norm": 0.30289610767641323, |
| "learning_rate": 8.76425367345779e-08, |
| "loss": 0.0904, |
| "step": 1197 |
| }, |
| { |
| "epoch": 4.85873417721519, |
| "grad_norm": 0.3015771091374866, |
| "learning_rate": 8.241494346644897e-08, |
| "loss": 0.0817, |
| "step": 1198 |
| }, |
| { |
| "epoch": 4.862784810126582, |
| "grad_norm": 0.29333935885624274, |
| "learning_rate": 7.734776369019204e-08, |
| "loss": 0.0857, |
| "step": 1199 |
| }, |
| { |
| "epoch": 4.866835443037974, |
| "grad_norm": 0.32254594202790776, |
| "learning_rate": 7.244103821617332e-08, |
| "loss": 0.1185, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.870886075949367, |
| "grad_norm": 0.3004022975540334, |
| "learning_rate": 6.769480656248606e-08, |
| "loss": 0.0875, |
| "step": 1201 |
| }, |
| { |
| "epoch": 4.874936708860759, |
| "grad_norm": 0.2775588065719596, |
| "learning_rate": 6.310910695462635e-08, |
| "loss": 0.0812, |
| "step": 1202 |
| }, |
| { |
| "epoch": 4.878987341772152, |
| "grad_norm": 0.29620415958102075, |
| "learning_rate": 5.8683976325191185e-08, |
| "loss": 0.0866, |
| "step": 1203 |
| }, |
| { |
| "epoch": 4.883037974683544, |
| "grad_norm": 0.31175560522296647, |
| "learning_rate": 5.4419450313571984e-08, |
| "loss": 0.0782, |
| "step": 1204 |
| }, |
| { |
| "epoch": 4.8870886075949365, |
| "grad_norm": 0.2853123013659545, |
| "learning_rate": 5.031556326567488e-08, |
| "loss": 0.08, |
| "step": 1205 |
| }, |
| { |
| "epoch": 4.891139240506329, |
| "grad_norm": 0.3138708036433306, |
| "learning_rate": 4.637234823364312e-08, |
| "loss": 0.0822, |
| "step": 1206 |
| }, |
| { |
| "epoch": 4.8951898734177215, |
| "grad_norm": 0.3021081159103301, |
| "learning_rate": 4.258983697558838e-08, |
| "loss": 0.106, |
| "step": 1207 |
| }, |
| { |
| "epoch": 4.899240506329114, |
| "grad_norm": 0.29486527068890717, |
| "learning_rate": 3.896805995533548e-08, |
| "loss": 0.0749, |
| "step": 1208 |
| }, |
| { |
| "epoch": 4.9032911392405065, |
| "grad_norm": 0.2826160198036475, |
| "learning_rate": 3.550704634218028e-08, |
| "loss": 0.0866, |
| "step": 1209 |
| }, |
| { |
| "epoch": 4.907341772151899, |
| "grad_norm": 0.28971984826104263, |
| "learning_rate": 3.2206824010647676e-08, |
| "loss": 0.0926, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.911392405063291, |
| "grad_norm": 0.29406778561385377, |
| "learning_rate": 2.9067419540278476e-08, |
| "loss": 0.0974, |
| "step": 1211 |
| }, |
| { |
| "epoch": 4.915443037974684, |
| "grad_norm": 0.3175299144288119, |
| "learning_rate": 2.6088858215400638e-08, |
| "loss": 0.0969, |
| "step": 1212 |
| }, |
| { |
| "epoch": 4.919493670886076, |
| "grad_norm": 0.2980208328180376, |
| "learning_rate": 2.3271164024940564e-08, |
| "loss": 0.0945, |
| "step": 1213 |
| }, |
| { |
| "epoch": 4.923544303797469, |
| "grad_norm": 0.29525922952033345, |
| "learning_rate": 2.061435966221881e-08, |
| "loss": 0.0901, |
| "step": 1214 |
| }, |
| { |
| "epoch": 4.927594936708861, |
| "grad_norm": 0.2981718723266732, |
| "learning_rate": 1.811846652477245e-08, |
| "loss": 0.081, |
| "step": 1215 |
| }, |
| { |
| "epoch": 4.931645569620253, |
| "grad_norm": 0.28254297084980123, |
| "learning_rate": 1.5783504714184106e-08, |
| "loss": 0.0758, |
| "step": 1216 |
| }, |
| { |
| "epoch": 4.935696202531646, |
| "grad_norm": 0.29593678488870334, |
| "learning_rate": 1.360949303591097e-08, |
| "loss": 0.0861, |
| "step": 1217 |
| }, |
| { |
| "epoch": 4.939746835443038, |
| "grad_norm": 0.29101556877269835, |
| "learning_rate": 1.1596448999144916e-08, |
| "loss": 0.0937, |
| "step": 1218 |
| }, |
| { |
| "epoch": 4.943797468354431, |
| "grad_norm": 0.299772369327956, |
| "learning_rate": 9.744388816668172e-09, |
| "loss": 0.0924, |
| "step": 1219 |
| }, |
| { |
| "epoch": 4.947848101265823, |
| "grad_norm": 0.3012632394891287, |
| "learning_rate": 8.05332740472009e-09, |
| "loss": 0.0997, |
| "step": 1220 |
| }, |
| { |
| "epoch": 4.951898734177215, |
| "grad_norm": 0.29682665747646597, |
| "learning_rate": 6.523278382872811e-09, |
| "loss": 0.0838, |
| "step": 1221 |
| }, |
| { |
| "epoch": 4.955949367088608, |
| "grad_norm": 0.2922227094898111, |
| "learning_rate": 5.15425407393133e-09, |
| "loss": 0.0977, |
| "step": 1222 |
| }, |
| { |
| "epoch": 4.96, |
| "grad_norm": 0.28112977754850815, |
| "learning_rate": 3.94626550383137e-09, |
| "loss": 0.084, |
| "step": 1223 |
| }, |
| { |
| "epoch": 4.964050632911392, |
| "grad_norm": 0.3144957715453971, |
| "learning_rate": 2.899322401546112e-09, |
| "loss": 0.0867, |
| "step": 1224 |
| }, |
| { |
| "epoch": 4.968101265822785, |
| "grad_norm": 0.2900253745683034, |
| "learning_rate": 2.013433199010706e-09, |
| "loss": 0.0914, |
| "step": 1225 |
| }, |
| { |
| "epoch": 4.972151898734177, |
| "grad_norm": 0.2834118355816139, |
| "learning_rate": 1.2886050310556563e-09, |
| "loss": 0.0886, |
| "step": 1226 |
| }, |
| { |
| "epoch": 4.976202531645569, |
| "grad_norm": 0.2944673511278016, |
| "learning_rate": 7.248437353468695e-10, |
| "loss": 0.0975, |
| "step": 1227 |
| }, |
| { |
| "epoch": 4.980253164556962, |
| "grad_norm": 0.305368917685997, |
| "learning_rate": 3.221538523412449e-10, |
| "loss": 0.0901, |
| "step": 1228 |
| }, |
| { |
| "epoch": 4.984303797468354, |
| "grad_norm": 0.30205373771409705, |
| "learning_rate": 8.053862524670663e-11, |
| "loss": 0.0776, |
| "step": 1229 |
| }, |
| { |
| "epoch": 4.988354430379747, |
| "grad_norm": 0.29642034896352676, |
| "learning_rate": 0.0, |
| "loss": 0.0823, |
| "step": 1230 |
| }, |
| { |
| "epoch": 4.988354430379747, |
| "step": 1230, |
| "total_flos": 9.208740243455345e+17, |
| "train_loss": 0.42258189511371824, |
| "train_runtime": 28169.462, |
| "train_samples_per_second": 5.609, |
| "train_steps_per_second": 0.044 |
| } |
| ], |
| "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": 9.208740243455345e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
