| { | |
| "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": 8.343612353426472, | |
| "learning_rate": 3.2520325203252037e-07, | |
| "loss": 1.3158, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00810126582278481, | |
| "grad_norm": 8.16425347503262, | |
| "learning_rate": 6.504065040650407e-07, | |
| "loss": 1.3015, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012151898734177215, | |
| "grad_norm": 8.11702783303773, | |
| "learning_rate": 9.75609756097561e-07, | |
| "loss": 1.2989, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01620253164556962, | |
| "grad_norm": 8.077047322654167, | |
| "learning_rate": 1.3008130081300815e-06, | |
| "loss": 1.2925, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.020253164556962026, | |
| "grad_norm": 7.894128254404671, | |
| "learning_rate": 1.6260162601626018e-06, | |
| "loss": 1.2902, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02430379746835443, | |
| "grad_norm": 7.510480917359572, | |
| "learning_rate": 1.951219512195122e-06, | |
| "loss": 1.2832, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028354430379746835, | |
| "grad_norm": 5.753275347013823, | |
| "learning_rate": 2.2764227642276426e-06, | |
| "loss": 1.2293, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03240506329113924, | |
| "grad_norm": 5.289360091324954, | |
| "learning_rate": 2.601626016260163e-06, | |
| "loss": 1.1983, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03645569620253165, | |
| "grad_norm": 3.2227504854252116, | |
| "learning_rate": 2.926829268292683e-06, | |
| "loss": 1.1842, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04050632911392405, | |
| "grad_norm": 2.7932029547233608, | |
| "learning_rate": 3.2520325203252037e-06, | |
| "loss": 1.156, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044556962025316456, | |
| "grad_norm": 2.658220517583248, | |
| "learning_rate": 3.577235772357724e-06, | |
| "loss": 1.1299, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04860759493670886, | |
| "grad_norm": 4.870615792004928, | |
| "learning_rate": 3.902439024390244e-06, | |
| "loss": 1.1147, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.052658227848101265, | |
| "grad_norm": 5.228854890427367, | |
| "learning_rate": 4.227642276422765e-06, | |
| "loss": 1.1041, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05670886075949367, | |
| "grad_norm": 4.8435679745958495, | |
| "learning_rate": 4.552845528455285e-06, | |
| "loss": 1.0908, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.060759493670886074, | |
| "grad_norm": 4.35775799247247, | |
| "learning_rate": 4.8780487804878055e-06, | |
| "loss": 1.0621, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06481012658227848, | |
| "grad_norm": 2.913406153634818, | |
| "learning_rate": 5.203252032520326e-06, | |
| "loss": 1.0423, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06886075949367089, | |
| "grad_norm": 2.891674547281331, | |
| "learning_rate": 5.528455284552846e-06, | |
| "loss": 1.02, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0729113924050633, | |
| "grad_norm": 2.7849022597646305, | |
| "learning_rate": 5.853658536585366e-06, | |
| "loss": 1.0117, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0769620253164557, | |
| "grad_norm": 2.313796832886638, | |
| "learning_rate": 6.178861788617887e-06, | |
| "loss": 0.999, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0810126582278481, | |
| "grad_norm": 1.8838379711624893, | |
| "learning_rate": 6.504065040650407e-06, | |
| "loss": 0.9769, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08506329113924051, | |
| "grad_norm": 1.6682283083828167, | |
| "learning_rate": 6.829268292682928e-06, | |
| "loss": 0.9831, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08911392405063291, | |
| "grad_norm": 1.4638657157883104, | |
| "learning_rate": 7.154471544715448e-06, | |
| "loss": 0.9872, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09316455696202532, | |
| "grad_norm": 1.4545831710418449, | |
| "learning_rate": 7.4796747967479676e-06, | |
| "loss": 0.9652, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09721518987341772, | |
| "grad_norm": 1.413391544692668, | |
| "learning_rate": 7.804878048780489e-06, | |
| "loss": 0.9666, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 1.442819309205835, | |
| "learning_rate": 8.130081300813009e-06, | |
| "loss": 0.9529, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10531645569620253, | |
| "grad_norm": 1.0490031707695895, | |
| "learning_rate": 8.45528455284553e-06, | |
| "loss": 0.9439, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10936708860759493, | |
| "grad_norm": 1.0383645907363823, | |
| "learning_rate": 8.78048780487805e-06, | |
| "loss": 0.9572, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11341772151898734, | |
| "grad_norm": 1.1789078976211094, | |
| "learning_rate": 9.10569105691057e-06, | |
| "loss": 0.9378, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11746835443037974, | |
| "grad_norm": 1.0377692853225864, | |
| "learning_rate": 9.43089430894309e-06, | |
| "loss": 0.9216, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12151898734177215, | |
| "grad_norm": 0.9298853823039268, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 0.921, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12556962025316457, | |
| "grad_norm": 0.9819082012908705, | |
| "learning_rate": 1.008130081300813e-05, | |
| "loss": 0.9073, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12962025316455697, | |
| "grad_norm": 0.9845630521474866, | |
| "learning_rate": 1.0406504065040652e-05, | |
| "loss": 0.9127, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13367088607594937, | |
| "grad_norm": 0.8855420777878603, | |
| "learning_rate": 1.0731707317073172e-05, | |
| "loss": 0.9057, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13772151898734178, | |
| "grad_norm": 0.8695201657020954, | |
| "learning_rate": 1.1056910569105692e-05, | |
| "loss": 0.896, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14177215189873418, | |
| "grad_norm": 0.9247214011304768, | |
| "learning_rate": 1.1382113821138213e-05, | |
| "loss": 0.899, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1458227848101266, | |
| "grad_norm": 1.0768355833283263, | |
| "learning_rate": 1.1707317073170731e-05, | |
| "loss": 0.9044, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.149873417721519, | |
| "grad_norm": 0.9824228099786853, | |
| "learning_rate": 1.2032520325203254e-05, | |
| "loss": 0.896, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1539240506329114, | |
| "grad_norm": 0.7619881117108473, | |
| "learning_rate": 1.2357723577235774e-05, | |
| "loss": 0.8836, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1579746835443038, | |
| "grad_norm": 1.0615594104336317, | |
| "learning_rate": 1.2682926829268294e-05, | |
| "loss": 0.8862, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1620253164556962, | |
| "grad_norm": 0.9798905549962407, | |
| "learning_rate": 1.3008130081300815e-05, | |
| "loss": 0.8841, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1660759493670886, | |
| "grad_norm": 0.787993871325004, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.8672, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17012658227848101, | |
| "grad_norm": 0.8898478304301364, | |
| "learning_rate": 1.3658536585365855e-05, | |
| "loss": 0.8828, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17417721518987342, | |
| "grad_norm": 0.9459104697244063, | |
| "learning_rate": 1.3983739837398376e-05, | |
| "loss": 0.9007, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17822784810126582, | |
| "grad_norm": 0.8661479592289267, | |
| "learning_rate": 1.4308943089430896e-05, | |
| "loss": 0.8812, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18227848101265823, | |
| "grad_norm": 0.89319470895471, | |
| "learning_rate": 1.4634146341463415e-05, | |
| "loss": 0.8804, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18632911392405063, | |
| "grad_norm": 0.8614327918783476, | |
| "learning_rate": 1.4959349593495935e-05, | |
| "loss": 0.8805, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.19037974683544304, | |
| "grad_norm": 0.9955997552021125, | |
| "learning_rate": 1.528455284552846e-05, | |
| "loss": 0.8713, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.19443037974683544, | |
| "grad_norm": 0.8762179225088836, | |
| "learning_rate": 1.5609756097560978e-05, | |
| "loss": 0.8532, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19848101265822785, | |
| "grad_norm": 0.8570962092349507, | |
| "learning_rate": 1.5934959349593496e-05, | |
| "loss": 0.8755, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 0.8573005714140741, | |
| "learning_rate": 1.6260162601626018e-05, | |
| "loss": 0.8603, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.20658227848101265, | |
| "grad_norm": 0.9168199071817832, | |
| "learning_rate": 1.6585365853658537e-05, | |
| "loss": 0.8639, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21063291139240506, | |
| "grad_norm": 0.8520176396098977, | |
| "learning_rate": 1.691056910569106e-05, | |
| "loss": 0.8656, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21468354430379746, | |
| "grad_norm": 0.8812996388430668, | |
| "learning_rate": 1.7235772357723578e-05, | |
| "loss": 0.8725, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21873417721518987, | |
| "grad_norm": 0.8726924298773409, | |
| "learning_rate": 1.75609756097561e-05, | |
| "loss": 0.8596, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22278481012658227, | |
| "grad_norm": 0.8156214162418044, | |
| "learning_rate": 1.788617886178862e-05, | |
| "loss": 0.8453, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22683544303797468, | |
| "grad_norm": 1.1064748027938984, | |
| "learning_rate": 1.821138211382114e-05, | |
| "loss": 0.8665, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23088607594936708, | |
| "grad_norm": 0.9692998884480308, | |
| "learning_rate": 1.8536585365853663e-05, | |
| "loss": 0.8724, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23493670886075949, | |
| "grad_norm": 0.9090728279074842, | |
| "learning_rate": 1.886178861788618e-05, | |
| "loss": 0.8696, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2389873417721519, | |
| "grad_norm": 1.0805209431068907, | |
| "learning_rate": 1.91869918699187e-05, | |
| "loss": 0.8536, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2430379746835443, | |
| "grad_norm": 0.9478570657299719, | |
| "learning_rate": 1.9512195121951222e-05, | |
| "loss": 0.855, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2470886075949367, | |
| "grad_norm": 1.3138154009446834, | |
| "learning_rate": 1.983739837398374e-05, | |
| "loss": 0.8588, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25113924050632913, | |
| "grad_norm": 1.0962920539140328, | |
| "learning_rate": 2.016260162601626e-05, | |
| "loss": 0.8587, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.25518987341772154, | |
| "grad_norm": 1.16875104847529, | |
| "learning_rate": 2.048780487804878e-05, | |
| "loss": 0.8623, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.25924050632911394, | |
| "grad_norm": 1.102790255748164, | |
| "learning_rate": 2.0813008130081303e-05, | |
| "loss": 0.8418, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.26329113924050634, | |
| "grad_norm": 1.0422070607234777, | |
| "learning_rate": 2.1138211382113822e-05, | |
| "loss": 0.846, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26734177215189875, | |
| "grad_norm": 1.0135303046817694, | |
| "learning_rate": 2.1463414634146344e-05, | |
| "loss": 0.8496, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27139240506329115, | |
| "grad_norm": 0.8152417744655771, | |
| "learning_rate": 2.1788617886178863e-05, | |
| "loss": 0.8398, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.27544303797468356, | |
| "grad_norm": 0.9434430388929115, | |
| "learning_rate": 2.2113821138211385e-05, | |
| "loss": 0.8432, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.27949367088607596, | |
| "grad_norm": 0.904429810735032, | |
| "learning_rate": 2.2439024390243907e-05, | |
| "loss": 0.8625, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.28354430379746837, | |
| "grad_norm": 0.8064869055138434, | |
| "learning_rate": 2.2764227642276426e-05, | |
| "loss": 0.8406, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.28759493670886077, | |
| "grad_norm": 0.9680582549271337, | |
| "learning_rate": 2.3089430894308948e-05, | |
| "loss": 0.8598, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2916455696202532, | |
| "grad_norm": 1.0977540971507387, | |
| "learning_rate": 2.3414634146341463e-05, | |
| "loss": 0.8421, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2956962025316456, | |
| "grad_norm": 1.0366663866091461, | |
| "learning_rate": 2.3739837398373985e-05, | |
| "loss": 0.8517, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.299746835443038, | |
| "grad_norm": 0.9895108253484108, | |
| "learning_rate": 2.4065040650406507e-05, | |
| "loss": 0.8608, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 1.129814584136786, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "loss": 0.8574, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3078481012658228, | |
| "grad_norm": 0.9939683021450608, | |
| "learning_rate": 2.4715447154471548e-05, | |
| "loss": 0.8617, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3118987341772152, | |
| "grad_norm": 0.8907969413975944, | |
| "learning_rate": 2.5040650406504066e-05, | |
| "loss": 0.8317, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3159493670886076, | |
| "grad_norm": 0.8821504802861744, | |
| "learning_rate": 2.536585365853659e-05, | |
| "loss": 0.8439, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1.0155417453337001, | |
| "learning_rate": 2.569105691056911e-05, | |
| "loss": 0.8468, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3240506329113924, | |
| "grad_norm": 0.8474312700438981, | |
| "learning_rate": 2.601626016260163e-05, | |
| "loss": 0.8506, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3281012658227848, | |
| "grad_norm": 0.932596284336908, | |
| "learning_rate": 2.634146341463415e-05, | |
| "loss": 0.8382, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3321518987341772, | |
| "grad_norm": 1.1170196456432966, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.8512, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3362025316455696, | |
| "grad_norm": 0.8181724636444612, | |
| "learning_rate": 2.699186991869919e-05, | |
| "loss": 0.8392, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.34025316455696203, | |
| "grad_norm": 0.8122921742009335, | |
| "learning_rate": 2.731707317073171e-05, | |
| "loss": 0.8487, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.34430379746835443, | |
| "grad_norm": 0.9641757997138598, | |
| "learning_rate": 2.764227642276423e-05, | |
| "loss": 0.8469, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.34835443037974684, | |
| "grad_norm": 1.0123433307472793, | |
| "learning_rate": 2.796747967479675e-05, | |
| "loss": 0.8248, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35240506329113924, | |
| "grad_norm": 0.988057154288923, | |
| "learning_rate": 2.829268292682927e-05, | |
| "loss": 0.8455, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.35645569620253165, | |
| "grad_norm": 1.057611278757914, | |
| "learning_rate": 2.8617886178861792e-05, | |
| "loss": 0.8435, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.36050632911392405, | |
| "grad_norm": 1.0542065696201266, | |
| "learning_rate": 2.8943089430894314e-05, | |
| "loss": 0.8331, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36455696202531646, | |
| "grad_norm": 1.2138545423200757, | |
| "learning_rate": 2.926829268292683e-05, | |
| "loss": 0.8455, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.36860759493670886, | |
| "grad_norm": 1.0074287262407686, | |
| "learning_rate": 2.959349593495935e-05, | |
| "loss": 0.8469, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37265822784810126, | |
| "grad_norm": 1.4808077844976184, | |
| "learning_rate": 2.991869918699187e-05, | |
| "loss": 0.834, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.37670886075949367, | |
| "grad_norm": 0.9395044852427558, | |
| "learning_rate": 3.0243902439024392e-05, | |
| "loss": 0.8488, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3807594936708861, | |
| "grad_norm": 1.3803564520329803, | |
| "learning_rate": 3.056910569105692e-05, | |
| "loss": 0.8135, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3848101265822785, | |
| "grad_norm": 1.0359552610821101, | |
| "learning_rate": 3.089430894308943e-05, | |
| "loss": 0.8615, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3888607594936709, | |
| "grad_norm": 1.230027918379461, | |
| "learning_rate": 3.1219512195121955e-05, | |
| "loss": 0.849, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3929113924050633, | |
| "grad_norm": 1.029707583789635, | |
| "learning_rate": 3.154471544715447e-05, | |
| "loss": 0.8221, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3969620253164557, | |
| "grad_norm": 1.0456082378472529, | |
| "learning_rate": 3.186991869918699e-05, | |
| "loss": 0.8475, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4010126582278481, | |
| "grad_norm": 1.0050371852160385, | |
| "learning_rate": 3.2195121951219514e-05, | |
| "loss": 0.8405, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 1.3191384319527482, | |
| "learning_rate": 3.2520325203252037e-05, | |
| "loss": 0.8443, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4091139240506329, | |
| "grad_norm": 1.0962116206928485, | |
| "learning_rate": 3.284552845528456e-05, | |
| "loss": 0.8273, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4131645569620253, | |
| "grad_norm": 1.001890518968171, | |
| "learning_rate": 3.3170731707317074e-05, | |
| "loss": 0.8336, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4172151898734177, | |
| "grad_norm": 1.1411558991121378, | |
| "learning_rate": 3.3495934959349596e-05, | |
| "loss": 0.8199, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4212658227848101, | |
| "grad_norm": 0.9423064860249271, | |
| "learning_rate": 3.382113821138212e-05, | |
| "loss": 0.8324, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4253164556962025, | |
| "grad_norm": 1.0499178157125395, | |
| "learning_rate": 3.414634146341463e-05, | |
| "loss": 0.8259, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4293670886075949, | |
| "grad_norm": 1.085285157685256, | |
| "learning_rate": 3.4471544715447155e-05, | |
| "loss": 0.8419, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.43341772151898733, | |
| "grad_norm": 0.9286581665804918, | |
| "learning_rate": 3.479674796747968e-05, | |
| "loss": 0.8373, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43746835443037974, | |
| "grad_norm": 1.081073162350945, | |
| "learning_rate": 3.51219512195122e-05, | |
| "loss": 0.8384, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44151898734177214, | |
| "grad_norm": 1.1603119304452982, | |
| "learning_rate": 3.544715447154472e-05, | |
| "loss": 0.8359, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44556962025316454, | |
| "grad_norm": 0.9134394208097611, | |
| "learning_rate": 3.577235772357724e-05, | |
| "loss": 0.8401, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44962025316455695, | |
| "grad_norm": 0.8447177804557533, | |
| "learning_rate": 3.609756097560976e-05, | |
| "loss": 0.8372, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.45367088607594935, | |
| "grad_norm": 1.3485809636974844, | |
| "learning_rate": 3.642276422764228e-05, | |
| "loss": 0.852, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.45772151898734176, | |
| "grad_norm": 0.9410827680770264, | |
| "learning_rate": 3.67479674796748e-05, | |
| "loss": 0.8369, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.46177215189873416, | |
| "grad_norm": 0.9979515530568333, | |
| "learning_rate": 3.7073170731707325e-05, | |
| "loss": 0.8326, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46582278481012657, | |
| "grad_norm": 1.1341086838904852, | |
| "learning_rate": 3.739837398373984e-05, | |
| "loss": 0.8191, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.46987341772151897, | |
| "grad_norm": 0.9044683223525496, | |
| "learning_rate": 3.772357723577236e-05, | |
| "loss": 0.8197, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4739240506329114, | |
| "grad_norm": 1.1595514155579196, | |
| "learning_rate": 3.804878048780488e-05, | |
| "loss": 0.8423, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4779746835443038, | |
| "grad_norm": 1.0977167805821404, | |
| "learning_rate": 3.83739837398374e-05, | |
| "loss": 0.8181, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4820253164556962, | |
| "grad_norm": 0.9363022825248438, | |
| "learning_rate": 3.869918699186992e-05, | |
| "loss": 0.8288, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4860759493670886, | |
| "grad_norm": 0.9665639880353534, | |
| "learning_rate": 3.9024390243902444e-05, | |
| "loss": 0.8349, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.490126582278481, | |
| "grad_norm": 1.0278601785376917, | |
| "learning_rate": 3.9349593495934966e-05, | |
| "loss": 0.8312, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4941772151898734, | |
| "grad_norm": 1.1061143067204418, | |
| "learning_rate": 3.967479674796748e-05, | |
| "loss": 0.8539, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.4982278481012658, | |
| "grad_norm": 1.040650077541857, | |
| "learning_rate": 4e-05, | |
| "loss": 0.8354, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5022784810126583, | |
| "grad_norm": 1.3808966975829016, | |
| "learning_rate": 3.999991946137476e-05, | |
| "loss": 0.8333, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 0.9504039532586749, | |
| "learning_rate": 3.999967784614766e-05, | |
| "loss": 0.8365, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5103797468354431, | |
| "grad_norm": 1.2566291580709308, | |
| "learning_rate": 3.9999275156264656e-05, | |
| "loss": 0.8454, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5144303797468355, | |
| "grad_norm": 1.1512432719703223, | |
| "learning_rate": 3.999871139496895e-05, | |
| "loss": 0.8245, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5184810126582279, | |
| "grad_norm": 1.2535670933501428, | |
| "learning_rate": 3.9997986566800995e-05, | |
| "loss": 0.8331, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5225316455696203, | |
| "grad_norm": 1.2196374095898719, | |
| "learning_rate": 3.999710067759846e-05, | |
| "loss": 0.8284, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5265822784810127, | |
| "grad_norm": 1.4555233647318062, | |
| "learning_rate": 3.999605373449617e-05, | |
| "loss": 0.846, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5306329113924051, | |
| "grad_norm": 1.2031302236156332, | |
| "learning_rate": 3.9994845745926075e-05, | |
| "loss": 0.8376, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5346835443037975, | |
| "grad_norm": 1.1646204870885386, | |
| "learning_rate": 3.999347672161713e-05, | |
| "loss": 0.8313, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5387341772151899, | |
| "grad_norm": 1.061388142460045, | |
| "learning_rate": 3.999194667259528e-05, | |
| "loss": 0.8257, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5427848101265823, | |
| "grad_norm": 1.4028112590319723, | |
| "learning_rate": 3.999025561118334e-05, | |
| "loss": 0.8265, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5468354430379747, | |
| "grad_norm": 1.1517171606483585, | |
| "learning_rate": 3.998840355100086e-05, | |
| "loss": 0.8318, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5508860759493671, | |
| "grad_norm": 1.5676829597046367, | |
| "learning_rate": 3.998639050696409e-05, | |
| "loss": 0.831, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5549367088607595, | |
| "grad_norm": 1.2921179806192415, | |
| "learning_rate": 3.998421649528582e-05, | |
| "loss": 0.8309, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5589873417721519, | |
| "grad_norm": 1.5473736324000276, | |
| "learning_rate": 3.9981881533475234e-05, | |
| "loss": 0.8408, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5630379746835443, | |
| "grad_norm": 1.3086352035377558, | |
| "learning_rate": 3.997938564033779e-05, | |
| "loss": 0.8379, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5670886075949367, | |
| "grad_norm": 1.3572169940610046, | |
| "learning_rate": 3.9976728835975064e-05, | |
| "loss": 0.8297, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5711392405063291, | |
| "grad_norm": 1.219291509132751, | |
| "learning_rate": 3.9973911141784605e-05, | |
| "loss": 0.827, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5751898734177215, | |
| "grad_norm": 1.312395466245897, | |
| "learning_rate": 3.997093258045973e-05, | |
| "loss": 0.8275, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.579240506329114, | |
| "grad_norm": 1.1910030328444707, | |
| "learning_rate": 3.996779317598936e-05, | |
| "loss": 0.8271, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5832911392405064, | |
| "grad_norm": 1.0815687000720382, | |
| "learning_rate": 3.996449295365782e-05, | |
| "loss": 0.8112, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5873417721518988, | |
| "grad_norm": 1.1108330268115463, | |
| "learning_rate": 3.996103194004467e-05, | |
| "loss": 0.8268, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5913924050632912, | |
| "grad_norm": 1.0924020280734106, | |
| "learning_rate": 3.995741016302441e-05, | |
| "loss": 0.8232, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5954430379746836, | |
| "grad_norm": 1.0201035550941457, | |
| "learning_rate": 3.9953627651766364e-05, | |
| "loss": 0.8199, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.599493670886076, | |
| "grad_norm": 0.9215770442783521, | |
| "learning_rate": 3.9949684436734325e-05, | |
| "loss": 0.8132, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6035443037974684, | |
| "grad_norm": 0.9196609191155093, | |
| "learning_rate": 3.994558054968643e-05, | |
| "loss": 0.8092, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 0.8661057969551478, | |
| "learning_rate": 3.994131602367481e-05, | |
| "loss": 0.8159, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6116455696202532, | |
| "grad_norm": 1.0912635397292874, | |
| "learning_rate": 3.9936890893045376e-05, | |
| "loss": 0.8318, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6156962025316456, | |
| "grad_norm": 0.8768197821214577, | |
| "learning_rate": 3.993230519343752e-05, | |
| "loss": 0.8217, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.619746835443038, | |
| "grad_norm": 0.9308212445056885, | |
| "learning_rate": 3.992755896178383e-05, | |
| "loss": 0.8167, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6237974683544304, | |
| "grad_norm": 0.8985793963235742, | |
| "learning_rate": 3.992265223630981e-05, | |
| "loss": 0.8209, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6278481012658228, | |
| "grad_norm": 0.9068516044101315, | |
| "learning_rate": 3.991758505653355e-05, | |
| "loss": 0.8127, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6318987341772152, | |
| "grad_norm": 0.7950793729436136, | |
| "learning_rate": 3.991235746326543e-05, | |
| "loss": 0.8067, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6359493670886076, | |
| "grad_norm": 0.8123391601120212, | |
| "learning_rate": 3.9906969498607745e-05, | |
| "loss": 0.8087, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.7450883495772671, | |
| "learning_rate": 3.990142120595444e-05, | |
| "loss": 0.822, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6440506329113924, | |
| "grad_norm": 0.8874648162651937, | |
| "learning_rate": 3.98957126299907e-05, | |
| "loss": 0.825, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6481012658227848, | |
| "grad_norm": 0.8264013897094276, | |
| "learning_rate": 3.9889843816692596e-05, | |
| "loss": 0.8277, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6521518987341772, | |
| "grad_norm": 0.7582075631208012, | |
| "learning_rate": 3.9883814813326766e-05, | |
| "loss": 0.8219, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6562025316455696, | |
| "grad_norm": 0.7910309723811477, | |
| "learning_rate": 3.9877625668449956e-05, | |
| "loss": 0.8071, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.660253164556962, | |
| "grad_norm": 0.7859481191212581, | |
| "learning_rate": 3.98712764319087e-05, | |
| "loss": 0.8313, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6643037974683544, | |
| "grad_norm": 0.8902308036781436, | |
| "learning_rate": 3.9864767154838864e-05, | |
| "loss": 0.8122, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6683544303797468, | |
| "grad_norm": 0.7190377605213393, | |
| "learning_rate": 3.9858097889665277e-05, | |
| "loss": 0.8349, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6724050632911392, | |
| "grad_norm": 0.8929389078433576, | |
| "learning_rate": 3.985126869010129e-05, | |
| "loss": 0.8099, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6764556962025317, | |
| "grad_norm": 0.8229363510017953, | |
| "learning_rate": 3.984427961114833e-05, | |
| "loss": 0.8262, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6805063291139241, | |
| "grad_norm": 0.7964473058251489, | |
| "learning_rate": 3.9837130709095475e-05, | |
| "loss": 0.8326, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6845569620253165, | |
| "grad_norm": 0.7467031715292105, | |
| "learning_rate": 3.982982204151901e-05, | |
| "loss": 0.8373, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6886075949367089, | |
| "grad_norm": 0.8309686318369134, | |
| "learning_rate": 3.982235366728193e-05, | |
| "loss": 0.8297, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6926582278481013, | |
| "grad_norm": 0.8960959933377175, | |
| "learning_rate": 3.9814725646533505e-05, | |
| "loss": 0.8208, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6967088607594937, | |
| "grad_norm": 0.7872051404909569, | |
| "learning_rate": 3.9806938040708746e-05, | |
| "loss": 0.816, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7007594936708861, | |
| "grad_norm": 0.9169211690083529, | |
| "learning_rate": 3.9798990912527976e-05, | |
| "loss": 0.8055, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7048101265822785, | |
| "grad_norm": 0.8149597189053123, | |
| "learning_rate": 3.979088432599627e-05, | |
| "loss": 0.7971, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 0.828433184558364, | |
| "learning_rate": 3.9782618346402964e-05, | |
| "loss": 0.8201, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7129113924050633, | |
| "grad_norm": 0.9646961681492818, | |
| "learning_rate": 3.977419304032111e-05, | |
| "loss": 0.811, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7169620253164557, | |
| "grad_norm": 0.9807292925992036, | |
| "learning_rate": 3.976560847560697e-05, | |
| "loss": 0.8209, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7210126582278481, | |
| "grad_norm": 0.8356035802344153, | |
| "learning_rate": 3.9756864721399456e-05, | |
| "loss": 0.8057, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7250632911392405, | |
| "grad_norm": 0.9763824184547701, | |
| "learning_rate": 3.974796184811956e-05, | |
| "loss": 0.8197, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7291139240506329, | |
| "grad_norm": 0.8263606830973313, | |
| "learning_rate": 3.973889992746979e-05, | |
| "loss": 0.8113, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7331645569620253, | |
| "grad_norm": 0.8628123631658113, | |
| "learning_rate": 3.972967903243361e-05, | |
| "loss": 0.8133, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7372151898734177, | |
| "grad_norm": 0.8096913867018039, | |
| "learning_rate": 3.972029923727486e-05, | |
| "loss": 0.8116, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7412658227848101, | |
| "grad_norm": 0.6991181769192026, | |
| "learning_rate": 3.971076061753709e-05, | |
| "loss": 0.8274, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7453164556962025, | |
| "grad_norm": 0.9587985507133673, | |
| "learning_rate": 3.9701063250043066e-05, | |
| "loss": 0.8147, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7493670886075949, | |
| "grad_norm": 0.7464123786151026, | |
| "learning_rate": 3.969120721289402e-05, | |
| "loss": 0.8138, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7534177215189873, | |
| "grad_norm": 0.9064734523747643, | |
| "learning_rate": 3.9681192585469146e-05, | |
| "loss": 0.7905, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7574683544303797, | |
| "grad_norm": 0.8965441154505833, | |
| "learning_rate": 3.9671019448424865e-05, | |
| "loss": 0.8252, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7615189873417721, | |
| "grad_norm": 0.7366087400980902, | |
| "learning_rate": 3.966068788369422e-05, | |
| "loss": 0.7915, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7655696202531646, | |
| "grad_norm": 0.7352423411380145, | |
| "learning_rate": 3.965019797448622e-05, | |
| "loss": 0.813, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.769620253164557, | |
| "grad_norm": 0.8442479697953837, | |
| "learning_rate": 3.963954980528515e-05, | |
| "loss": 0.8145, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7736708860759494, | |
| "grad_norm": 0.7662871697555996, | |
| "learning_rate": 3.9628743461849905e-05, | |
| "loss": 0.8122, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7777215189873418, | |
| "grad_norm": 0.8508305325498614, | |
| "learning_rate": 3.961777903121329e-05, | |
| "loss": 0.815, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7817721518987342, | |
| "grad_norm": 0.8059404836263478, | |
| "learning_rate": 3.960665660168131e-05, | |
| "loss": 0.8198, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7858227848101266, | |
| "grad_norm": 0.8969648002442475, | |
| "learning_rate": 3.9595376262832485e-05, | |
| "loss": 0.8013, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.789873417721519, | |
| "grad_norm": 0.9426611146178773, | |
| "learning_rate": 3.9583938105517127e-05, | |
| "loss": 0.8097, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7939240506329114, | |
| "grad_norm": 0.8987556320287169, | |
| "learning_rate": 3.957234222185657e-05, | |
| "loss": 0.8074, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7979746835443038, | |
| "grad_norm": 0.8622205026733792, | |
| "learning_rate": 3.9560588705242474e-05, | |
| "loss": 0.7891, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8020253164556962, | |
| "grad_norm": 0.7871035581030087, | |
| "learning_rate": 3.954867765033605e-05, | |
| "loss": 0.8164, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8060759493670886, | |
| "grad_norm": 0.7163867238867948, | |
| "learning_rate": 3.953660915306728e-05, | |
| "loss": 0.7915, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 0.7658181401450425, | |
| "learning_rate": 3.952438331063419e-05, | |
| "loss": 0.8113, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8141772151898734, | |
| "grad_norm": 0.8186237464975145, | |
| "learning_rate": 3.951200022150205e-05, | |
| "loss": 0.8144, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8182278481012658, | |
| "grad_norm": 0.8550732932428305, | |
| "learning_rate": 3.949945998540253e-05, | |
| "loss": 0.8281, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8222784810126582, | |
| "grad_norm": 0.7619938600749018, | |
| "learning_rate": 3.9486762703332993e-05, | |
| "loss": 0.8114, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8263291139240506, | |
| "grad_norm": 0.7662988303514785, | |
| "learning_rate": 3.947390847755559e-05, | |
| "loss": 0.821, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.830379746835443, | |
| "grad_norm": 0.7199420631748139, | |
| "learning_rate": 3.946089741159648e-05, | |
| "loss": 0.8035, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8344303797468354, | |
| "grad_norm": 0.7670079759188053, | |
| "learning_rate": 3.944772961024501e-05, | |
| "loss": 0.7982, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8384810126582278, | |
| "grad_norm": 0.7349503936756709, | |
| "learning_rate": 3.943440517955285e-05, | |
| "loss": 0.8084, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8425316455696202, | |
| "grad_norm": 0.7771424368874337, | |
| "learning_rate": 3.9420924226833126e-05, | |
| "loss": 0.8128, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8465822784810126, | |
| "grad_norm": 0.6777923743878203, | |
| "learning_rate": 3.9407286860659566e-05, | |
| "loss": 0.8035, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.850632911392405, | |
| "grad_norm": 0.8005366363188512, | |
| "learning_rate": 3.9393493190865657e-05, | |
| "loss": 0.8048, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8546835443037974, | |
| "grad_norm": 0.7045670187742422, | |
| "learning_rate": 3.937954332854371e-05, | |
| "loss": 0.8119, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8587341772151899, | |
| "grad_norm": 0.7603859610208397, | |
| "learning_rate": 3.9365437386044016e-05, | |
| "loss": 0.8236, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8627848101265823, | |
| "grad_norm": 0.7720351467207939, | |
| "learning_rate": 3.935117547697387e-05, | |
| "loss": 0.8161, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.8668354430379747, | |
| "grad_norm": 0.7320334461154931, | |
| "learning_rate": 3.933675771619675e-05, | |
| "loss": 0.8048, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8708860759493671, | |
| "grad_norm": 0.9285219425491757, | |
| "learning_rate": 3.932218421983131e-05, | |
| "loss": 0.7876, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8749367088607595, | |
| "grad_norm": 0.7021770042621981, | |
| "learning_rate": 3.9307455105250484e-05, | |
| "loss": 0.7995, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8789873417721519, | |
| "grad_norm": 0.90664929187497, | |
| "learning_rate": 3.929257049108054e-05, | |
| "loss": 0.8254, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8830379746835443, | |
| "grad_norm": 0.7632647802910745, | |
| "learning_rate": 3.927753049720011e-05, | |
| "loss": 0.815, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8870886075949367, | |
| "grad_norm": 0.7984606947412733, | |
| "learning_rate": 3.9262335244739234e-05, | |
| "loss": 0.8036, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8911392405063291, | |
| "grad_norm": 0.8344266407402826, | |
| "learning_rate": 3.92469848560784e-05, | |
| "loss": 0.8045, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8951898734177215, | |
| "grad_norm": 0.698561165157485, | |
| "learning_rate": 3.923147945484751e-05, | |
| "loss": 0.7955, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8992405063291139, | |
| "grad_norm": 0.7327550948117628, | |
| "learning_rate": 3.9215819165924956e-05, | |
| "loss": 0.7827, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9032911392405063, | |
| "grad_norm": 0.6943731749853619, | |
| "learning_rate": 3.920000411543654e-05, | |
| "loss": 0.774, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9073417721518987, | |
| "grad_norm": 0.7052566398335677, | |
| "learning_rate": 3.9184034430754495e-05, | |
| "loss": 0.8143, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 0.8325739583367207, | |
| "learning_rate": 3.916791024049648e-05, | |
| "loss": 0.8034, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9154430379746835, | |
| "grad_norm": 0.7244665093323736, | |
| "learning_rate": 3.91516316745245e-05, | |
| "loss": 0.8069, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9194936708860759, | |
| "grad_norm": 0.8424622507610052, | |
| "learning_rate": 3.913519886394389e-05, | |
| "loss": 0.8087, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.9235443037974683, | |
| "grad_norm": 0.7182346240587765, | |
| "learning_rate": 3.911861194110225e-05, | |
| "loss": 0.7858, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.9275949367088607, | |
| "grad_norm": 0.7256109165898125, | |
| "learning_rate": 3.910187103958837e-05, | |
| "loss": 0.8067, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9316455696202531, | |
| "grad_norm": 0.7609738991234395, | |
| "learning_rate": 3.908497629423117e-05, | |
| "loss": 0.7893, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9356962025316455, | |
| "grad_norm": 0.7424008262178546, | |
| "learning_rate": 3.9067927841098614e-05, | |
| "loss": 0.7965, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.9397468354430379, | |
| "grad_norm": 0.6959036462123632, | |
| "learning_rate": 3.9050725817496594e-05, | |
| "loss": 0.7956, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9437974683544303, | |
| "grad_norm": 0.7586112939966917, | |
| "learning_rate": 3.9033370361967844e-05, | |
| "loss": 0.7851, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9478481012658228, | |
| "grad_norm": 0.7058115503672875, | |
| "learning_rate": 3.901586161429081e-05, | |
| "loss": 0.794, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.9518987341772152, | |
| "grad_norm": 0.7472924286462742, | |
| "learning_rate": 3.8998199715478545e-05, | |
| "loss": 0.7826, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.9559493670886076, | |
| "grad_norm": 0.7062161586079215, | |
| "learning_rate": 3.8980384807777564e-05, | |
| "loss": 0.7935, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.6854082838523912, | |
| "learning_rate": 3.896241703466667e-05, | |
| "loss": 0.7845, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.9640506329113924, | |
| "grad_norm": 0.6966678163913387, | |
| "learning_rate": 3.894429654085585e-05, | |
| "loss": 0.7979, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9681012658227848, | |
| "grad_norm": 0.7291045828333715, | |
| "learning_rate": 3.892602347228505e-05, | |
| "loss": 0.8006, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.9721518987341772, | |
| "grad_norm": 0.7201662056753156, | |
| "learning_rate": 3.890759797612307e-05, | |
| "loss": 0.7948, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9762025316455696, | |
| "grad_norm": 0.7937229385680088, | |
| "learning_rate": 3.888902020076632e-05, | |
| "loss": 0.8032, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.980253164556962, | |
| "grad_norm": 0.6246509623459382, | |
| "learning_rate": 3.887029029583764e-05, | |
| "loss": 0.781, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9843037974683544, | |
| "grad_norm": 0.6755530930612179, | |
| "learning_rate": 3.8851408412185125e-05, | |
| "loss": 0.813, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.9883544303797468, | |
| "grad_norm": 0.6930603027278213, | |
| "learning_rate": 3.8832374701880855e-05, | |
| "loss": 0.7926, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9924050632911392, | |
| "grad_norm": 0.6633999380801953, | |
| "learning_rate": 3.881318931821972e-05, | |
| "loss": 0.8038, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.9964556962025316, | |
| "grad_norm": 0.7563437401188942, | |
| "learning_rate": 3.879385241571817e-05, | |
| "loss": 0.7888, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.0020253164556963, | |
| "grad_norm": 0.864170329029982, | |
| "learning_rate": 3.8774364150112955e-05, | |
| "loss": 0.7116, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.0060759493670886, | |
| "grad_norm": 0.7538438198718099, | |
| "learning_rate": 3.8754724678359884e-05, | |
| "loss": 0.6127, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.010126582278481, | |
| "grad_norm": 0.8443153461464613, | |
| "learning_rate": 3.873493415863256e-05, | |
| "loss": 0.6028, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.0141772151898734, | |
| "grad_norm": 0.9269870113535025, | |
| "learning_rate": 3.871499275032111e-05, | |
| "loss": 0.5802, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.0182278481012659, | |
| "grad_norm": 0.6924378955479644, | |
| "learning_rate": 3.869490061403091e-05, | |
| "loss": 0.5866, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.0222784810126582, | |
| "grad_norm": 0.8421266579223858, | |
| "learning_rate": 3.867465791158124e-05, | |
| "loss": 0.5982, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.0263291139240507, | |
| "grad_norm": 0.7526238836650399, | |
| "learning_rate": 3.865426480600407e-05, | |
| "loss": 0.5829, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.030379746835443, | |
| "grad_norm": 0.7354558920530211, | |
| "learning_rate": 3.863372146154264e-05, | |
| "loss": 0.614, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.0344303797468355, | |
| "grad_norm": 0.922533355216991, | |
| "learning_rate": 3.861302804365024e-05, | |
| "loss": 0.6073, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.0384810126582278, | |
| "grad_norm": 0.7069767883422869, | |
| "learning_rate": 3.85921847189888e-05, | |
| "loss": 0.6044, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.0425316455696203, | |
| "grad_norm": 0.8030200508875822, | |
| "learning_rate": 3.85711916554276e-05, | |
| "loss": 0.6015, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.0465822784810126, | |
| "grad_norm": 0.7892239790495116, | |
| "learning_rate": 3.85500490220419e-05, | |
| "loss": 0.5928, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.0506329113924051, | |
| "grad_norm": 0.8389106593796315, | |
| "learning_rate": 3.852875698911154e-05, | |
| "loss": 0.5974, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.0546835443037974, | |
| "grad_norm": 0.7441170399409961, | |
| "learning_rate": 3.850731572811963e-05, | |
| "loss": 0.6108, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.05873417721519, | |
| "grad_norm": 0.749621241846919, | |
| "learning_rate": 3.848572541175116e-05, | |
| "loss": 0.5958, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.0627848101265822, | |
| "grad_norm": 0.7630910448967563, | |
| "learning_rate": 3.846398621389154e-05, | |
| "loss": 0.5896, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.0668354430379747, | |
| "grad_norm": 0.7167283361101315, | |
| "learning_rate": 3.84420983096253e-05, | |
| "loss": 0.5934, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.070886075949367, | |
| "grad_norm": 0.8505315956477022, | |
| "learning_rate": 3.8420061875234606e-05, | |
| "loss": 0.5916, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.0749367088607595, | |
| "grad_norm": 0.7782539763674439, | |
| "learning_rate": 3.839787708819787e-05, | |
| "loss": 0.6157, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.0789873417721518, | |
| "grad_norm": 0.7749722776162589, | |
| "learning_rate": 3.8375544127188325e-05, | |
| "loss": 0.5991, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.0830379746835443, | |
| "grad_norm": 0.8041300490780838, | |
| "learning_rate": 3.8353063172072564e-05, | |
| "loss": 0.5969, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.0870886075949366, | |
| "grad_norm": 0.7473661024140563, | |
| "learning_rate": 3.8330434403909105e-05, | |
| "loss": 0.5801, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.0911392405063292, | |
| "grad_norm": 0.9111695389650207, | |
| "learning_rate": 3.8307658004946934e-05, | |
| "loss": 0.5859, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.0951898734177214, | |
| "grad_norm": 0.7539251491414497, | |
| "learning_rate": 3.8284734158624046e-05, | |
| "loss": 0.5634, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.099240506329114, | |
| "grad_norm": 0.7355361865476139, | |
| "learning_rate": 3.826166304956594e-05, | |
| "loss": 0.5886, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.1032911392405063, | |
| "grad_norm": 0.8781501097953359, | |
| "learning_rate": 3.8238444863584164e-05, | |
| "loss": 0.6094, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.1073417721518988, | |
| "grad_norm": 0.7304344420476403, | |
| "learning_rate": 3.821507978767479e-05, | |
| "loss": 0.5913, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.111392405063291, | |
| "grad_norm": 0.7044130026206465, | |
| "learning_rate": 3.819156801001693e-05, | |
| "loss": 0.5745, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.1154430379746836, | |
| "grad_norm": 0.6590771114805719, | |
| "learning_rate": 3.816790971997121e-05, | |
| "loss": 0.5867, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.1194936708860759, | |
| "grad_norm": 0.7465995671317784, | |
| "learning_rate": 3.8144105108078246e-05, | |
| "loss": 0.5909, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.1235443037974684, | |
| "grad_norm": 0.7358107380857318, | |
| "learning_rate": 3.81201543660571e-05, | |
| "loss": 0.5912, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.1275949367088607, | |
| "grad_norm": 0.7438374647527917, | |
| "learning_rate": 3.809605768680377e-05, | |
| "loss": 0.5827, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.1316455696202532, | |
| "grad_norm": 0.7081998106501225, | |
| "learning_rate": 3.807181526438958e-05, | |
| "loss": 0.5881, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.1356962025316455, | |
| "grad_norm": 0.7173550555589351, | |
| "learning_rate": 3.8047427294059697e-05, | |
| "loss": 0.6025, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.139746835443038, | |
| "grad_norm": 0.6549692203757064, | |
| "learning_rate": 3.802289397223145e-05, | |
| "loss": 0.5739, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.1437974683544303, | |
| "grad_norm": 0.6705977218822847, | |
| "learning_rate": 3.7998215496492854e-05, | |
| "loss": 0.606, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.1478481012658228, | |
| "grad_norm": 0.6807515005832901, | |
| "learning_rate": 3.797339206560096e-05, | |
| "loss": 0.608, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.1518987341772151, | |
| "grad_norm": 0.6740434825529994, | |
| "learning_rate": 3.794842387948027e-05, | |
| "loss": 0.5934, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.1559493670886076, | |
| "grad_norm": 0.6559138381802385, | |
| "learning_rate": 3.7923311139221114e-05, | |
| "loss": 0.5978, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.7170860111523227, | |
| "learning_rate": 3.7898054047078054e-05, | |
| "loss": 0.612, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.1640506329113924, | |
| "grad_norm": 0.753903283008675, | |
| "learning_rate": 3.787265280646825e-05, | |
| "loss": 0.5913, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.1681012658227847, | |
| "grad_norm": 0.7235558116896846, | |
| "learning_rate": 3.7847107621969786e-05, | |
| "loss": 0.6101, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.1721518987341772, | |
| "grad_norm": 0.6685323134329691, | |
| "learning_rate": 3.7821418699320064e-05, | |
| "loss": 0.5963, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.1762025316455695, | |
| "grad_norm": 0.6740171231224578, | |
| "learning_rate": 3.7795586245414145e-05, | |
| "loss": 0.59, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.180253164556962, | |
| "grad_norm": 0.7402038062138625, | |
| "learning_rate": 3.776961046830306e-05, | |
| "loss": 0.6092, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.1843037974683543, | |
| "grad_norm": 0.7168274401204382, | |
| "learning_rate": 3.774349157719215e-05, | |
| "loss": 0.6169, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.1883544303797469, | |
| "grad_norm": 0.7538374077572204, | |
| "learning_rate": 3.7717229782439365e-05, | |
| "loss": 0.6134, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.1924050632911392, | |
| "grad_norm": 0.7110326673196169, | |
| "learning_rate": 3.769082529555359e-05, | |
| "loss": 0.5789, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.1964556962025317, | |
| "grad_norm": 0.6629042894357543, | |
| "learning_rate": 3.766427832919294e-05, | |
| "loss": 0.6095, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.200506329113924, | |
| "grad_norm": 0.7721104569494721, | |
| "learning_rate": 3.7637589097163024e-05, | |
| "loss": 0.6038, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.2045569620253165, | |
| "grad_norm": 0.6388659618200008, | |
| "learning_rate": 3.761075781441526e-05, | |
| "loss": 0.6025, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.2086075949367088, | |
| "grad_norm": 0.7985966535762967, | |
| "learning_rate": 3.75837846970451e-05, | |
| "loss": 0.5806, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.2126582278481013, | |
| "grad_norm": 0.6848982500284568, | |
| "learning_rate": 3.755666996229032e-05, | |
| "loss": 0.5994, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.2167088607594936, | |
| "grad_norm": 0.731166720977407, | |
| "learning_rate": 3.752941382852927e-05, | |
| "loss": 0.6192, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.220759493670886, | |
| "grad_norm": 0.730303673014515, | |
| "learning_rate": 3.7502016515279115e-05, | |
| "loss": 0.6107, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.2248101265822784, | |
| "grad_norm": 0.7400827624455701, | |
| "learning_rate": 3.7474478243194043e-05, | |
| "loss": 0.6238, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.228860759493671, | |
| "grad_norm": 0.7046503972221974, | |
| "learning_rate": 3.744679923406351e-05, | |
| "loss": 0.5929, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.2329113924050632, | |
| "grad_norm": 0.6849219878732847, | |
| "learning_rate": 3.741897971081043e-05, | |
| "loss": 0.6166, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.2369620253164557, | |
| "grad_norm": 0.7732742772828523, | |
| "learning_rate": 3.739101989748946e-05, | |
| "loss": 0.5968, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.241012658227848, | |
| "grad_norm": 0.6587641288539604, | |
| "learning_rate": 3.7362920019285066e-05, | |
| "loss": 0.6001, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.2450632911392405, | |
| "grad_norm": 0.7184780641599043, | |
| "learning_rate": 3.73346803025098e-05, | |
| "loss": 0.5957, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.2491139240506328, | |
| "grad_norm": 0.6884606724670599, | |
| "learning_rate": 3.730630097460247e-05, | |
| "loss": 0.591, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.2531645569620253, | |
| "grad_norm": 0.6422322247234333, | |
| "learning_rate": 3.727778226412628e-05, | |
| "loss": 0.618, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.2572151898734178, | |
| "grad_norm": 0.6734904463247949, | |
| "learning_rate": 3.7249124400767006e-05, | |
| "loss": 0.613, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.2612658227848101, | |
| "grad_norm": 0.6783958753643822, | |
| "learning_rate": 3.722032761533114e-05, | |
| "loss": 0.5817, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.2653164556962024, | |
| "grad_norm": 0.7106919025079214, | |
| "learning_rate": 3.719139213974403e-05, | |
| "loss": 0.5945, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.269367088607595, | |
| "grad_norm": 0.7304702409815148, | |
| "learning_rate": 3.7162318207048006e-05, | |
| "loss": 0.5938, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.2734177215189875, | |
| "grad_norm": 0.6302087365402924, | |
| "learning_rate": 3.713310605140055e-05, | |
| "loss": 0.6076, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.2774683544303798, | |
| "grad_norm": 0.6858573316949614, | |
| "learning_rate": 3.710375590807233e-05, | |
| "loss": 0.608, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.281518987341772, | |
| "grad_norm": 0.6600076476264306, | |
| "learning_rate": 3.7074268013445365e-05, | |
| "loss": 0.6196, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.2855696202531646, | |
| "grad_norm": 0.6450683723397447, | |
| "learning_rate": 3.7044642605011114e-05, | |
| "loss": 0.5784, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.289620253164557, | |
| "grad_norm": 0.6638440759802613, | |
| "learning_rate": 3.701487992136854e-05, | |
| "loss": 0.5857, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.2936708860759494, | |
| "grad_norm": 0.6296089414897097, | |
| "learning_rate": 3.69849802022222e-05, | |
| "loss": 0.5858, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.2977215189873417, | |
| "grad_norm": 0.5987334388405574, | |
| "learning_rate": 3.6954943688380334e-05, | |
| "loss": 0.5988, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.3017721518987342, | |
| "grad_norm": 0.6151831112251058, | |
| "learning_rate": 3.692477062175289e-05, | |
| "loss": 0.5947, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.3058227848101267, | |
| "grad_norm": 0.6146460095243743, | |
| "learning_rate": 3.689446124534958e-05, | |
| "loss": 0.6247, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.309873417721519, | |
| "grad_norm": 0.6418638762575473, | |
| "learning_rate": 3.686401580327799e-05, | |
| "loss": 0.6172, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.3139240506329113, | |
| "grad_norm": 0.6288546829009947, | |
| "learning_rate": 3.683343454074149e-05, | |
| "loss": 0.6063, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.3179746835443038, | |
| "grad_norm": 0.6694176978602991, | |
| "learning_rate": 3.6802717704037386e-05, | |
| "loss": 0.5941, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.3220253164556963, | |
| "grad_norm": 0.6424767253012746, | |
| "learning_rate": 3.6771865540554855e-05, | |
| "loss": 0.5969, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.3260759493670886, | |
| "grad_norm": 0.6300273694372438, | |
| "learning_rate": 3.674087829877297e-05, | |
| "loss": 0.6009, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.330126582278481, | |
| "grad_norm": 0.6221506010600839, | |
| "learning_rate": 3.6709756228258735e-05, | |
| "loss": 0.5801, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.3341772151898734, | |
| "grad_norm": 0.6592968876253862, | |
| "learning_rate": 3.667849957966501e-05, | |
| "loss": 0.5971, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.338227848101266, | |
| "grad_norm": 0.6574995116484905, | |
| "learning_rate": 3.6647108604728546e-05, | |
| "loss": 0.6084, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.3422784810126582, | |
| "grad_norm": 0.6687701632599151, | |
| "learning_rate": 3.661558355626795e-05, | |
| "loss": 0.6059, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.3463291139240505, | |
| "grad_norm": 0.6930265554756305, | |
| "learning_rate": 3.658392468818163e-05, | |
| "loss": 0.5993, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.350379746835443, | |
| "grad_norm": 0.6435023444206545, | |
| "learning_rate": 3.655213225544574e-05, | |
| "loss": 0.6115, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.3544303797468356, | |
| "grad_norm": 0.6800591829420978, | |
| "learning_rate": 3.652020651411218e-05, | |
| "loss": 0.5862, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.3584810126582278, | |
| "grad_norm": 0.665492852622408, | |
| "learning_rate": 3.6488147721306474e-05, | |
| "loss": 0.5887, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.3625316455696201, | |
| "grad_norm": 0.704501752090632, | |
| "learning_rate": 3.645595613522574e-05, | |
| "loss": 0.6121, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.3665822784810127, | |
| "grad_norm": 0.6411034829144266, | |
| "learning_rate": 3.642363201513657e-05, | |
| "loss": 0.6134, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.3706329113924052, | |
| "grad_norm": 0.6684207595467546, | |
| "learning_rate": 3.6391175621373006e-05, | |
| "loss": 0.6095, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.3746835443037975, | |
| "grad_norm": 0.6401883339229629, | |
| "learning_rate": 3.6358587215334355e-05, | |
| "loss": 0.5903, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.3787341772151898, | |
| "grad_norm": 0.7643636540790698, | |
| "learning_rate": 3.632586705948318e-05, | |
| "loss": 0.5905, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.3827848101265823, | |
| "grad_norm": 0.6865649306624818, | |
| "learning_rate": 3.629301541734311e-05, | |
| "loss": 0.6054, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.3868354430379748, | |
| "grad_norm": 0.8064278075410573, | |
| "learning_rate": 3.626003255349676e-05, | |
| "loss": 0.6217, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.390886075949367, | |
| "grad_norm": 0.7292213416640131, | |
| "learning_rate": 3.622691873358357e-05, | |
| "loss": 0.6035, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.3949367088607594, | |
| "grad_norm": 0.9098446602996306, | |
| "learning_rate": 3.61936742242977e-05, | |
| "loss": 0.6266, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.398987341772152, | |
| "grad_norm": 0.7560544788055907, | |
| "learning_rate": 3.6160299293385864e-05, | |
| "loss": 0.6166, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.4030379746835444, | |
| "grad_norm": 0.8362172679624794, | |
| "learning_rate": 3.612679420964516e-05, | |
| "loss": 0.6184, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.4070886075949367, | |
| "grad_norm": 0.7156056044318765, | |
| "learning_rate": 3.609315924292092e-05, | |
| "loss": 0.6195, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.411139240506329, | |
| "grad_norm": 0.7085769141119255, | |
| "learning_rate": 3.6059394664104554e-05, | |
| "loss": 0.5943, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.4151898734177215, | |
| "grad_norm": 0.8950808448434636, | |
| "learning_rate": 3.602550074513133e-05, | |
| "loss": 0.6097, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.419240506329114, | |
| "grad_norm": 0.682465798054495, | |
| "learning_rate": 3.599147775897822e-05, | |
| "loss": 0.614, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.4232911392405063, | |
| "grad_norm": 0.829888130573614, | |
| "learning_rate": 3.595732597966167e-05, | |
| "loss": 0.6241, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.4273417721518986, | |
| "grad_norm": 0.6761841066472968, | |
| "learning_rate": 3.592304568223542e-05, | |
| "loss": 0.5914, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.4313924050632911, | |
| "grad_norm": 0.6762842010350288, | |
| "learning_rate": 3.588863714278826e-05, | |
| "loss": 0.5932, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.4354430379746836, | |
| "grad_norm": 0.6861283065906502, | |
| "learning_rate": 3.585410063844186e-05, | |
| "loss": 0.5858, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.439493670886076, | |
| "grad_norm": 0.6854923515720026, | |
| "learning_rate": 3.581943644734846e-05, | |
| "loss": 0.596, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.4435443037974682, | |
| "grad_norm": 0.6087158446657662, | |
| "learning_rate": 3.578464484868869e-05, | |
| "loss": 0.5875, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.4475949367088607, | |
| "grad_norm": 0.6050782930522556, | |
| "learning_rate": 3.5749726122669316e-05, | |
| "loss": 0.6117, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.4516455696202533, | |
| "grad_norm": 0.6291300587819175, | |
| "learning_rate": 3.5714680550520943e-05, | |
| "loss": 0.5948, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.4556962025316456, | |
| "grad_norm": 0.6213796833682232, | |
| "learning_rate": 3.5679508414495794e-05, | |
| "loss": 0.5953, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.4597468354430378, | |
| "grad_norm": 0.6107795275496467, | |
| "learning_rate": 3.564420999786543e-05, | |
| "loss": 0.6157, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.4637974683544304, | |
| "grad_norm": 0.6436866227385413, | |
| "learning_rate": 3.560878558491842e-05, | |
| "loss": 0.5871, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.4678481012658229, | |
| "grad_norm": 0.685085646572779, | |
| "learning_rate": 3.5573235460958145e-05, | |
| "loss": 0.6024, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.4718987341772152, | |
| "grad_norm": 0.67603023095013, | |
| "learning_rate": 3.553755991230039e-05, | |
| "loss": 0.6097, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.4759493670886075, | |
| "grad_norm": 0.7086320270941887, | |
| "learning_rate": 3.5501759226271144e-05, | |
| "loss": 0.6065, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.6934548436415007, | |
| "learning_rate": 3.546583369120419e-05, | |
| "loss": 0.5861, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.4840506329113925, | |
| "grad_norm": 0.7710007011228985, | |
| "learning_rate": 3.5429783596438864e-05, | |
| "loss": 0.6029, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.4881012658227848, | |
| "grad_norm": 0.6570399204657505, | |
| "learning_rate": 3.539360923231766e-05, | |
| "loss": 0.6037, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.492151898734177, | |
| "grad_norm": 0.6606825525702773, | |
| "learning_rate": 3.535731089018394e-05, | |
| "loss": 0.5957, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.4962025316455696, | |
| "grad_norm": 0.7227545055326086, | |
| "learning_rate": 3.532088886237956e-05, | |
| "loss": 0.5988, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.5002531645569621, | |
| "grad_norm": 0.668402933598481, | |
| "learning_rate": 3.528434344224253e-05, | |
| "loss": 0.6191, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.5043037974683544, | |
| "grad_norm": 0.6504030049953298, | |
| "learning_rate": 3.524767492410464e-05, | |
| "loss": 0.6043, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.5083544303797467, | |
| "grad_norm": 0.7023043328293829, | |
| "learning_rate": 3.521088360328908e-05, | |
| "loss": 0.6087, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.5124050632911392, | |
| "grad_norm": 0.7313240995138819, | |
| "learning_rate": 3.517396977610811e-05, | |
| "loss": 0.5887, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.5164556962025317, | |
| "grad_norm": 0.648887834909513, | |
| "learning_rate": 3.5136933739860595e-05, | |
| "loss": 0.6017, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.520506329113924, | |
| "grad_norm": 0.7919016887314445, | |
| "learning_rate": 3.509977579282971e-05, | |
| "loss": 0.5959, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.5245569620253163, | |
| "grad_norm": 0.7491150288585879, | |
| "learning_rate": 3.5062496234280424e-05, | |
| "loss": 0.609, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.5286075949367088, | |
| "grad_norm": 0.7129226161966522, | |
| "learning_rate": 3.502509536445719e-05, | |
| "loss": 0.6065, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.5326582278481014, | |
| "grad_norm": 0.7579379016746397, | |
| "learning_rate": 3.498757348458147e-05, | |
| "loss": 0.6143, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.5367088607594936, | |
| "grad_norm": 0.6839842646260479, | |
| "learning_rate": 3.4949930896849324e-05, | |
| "loss": 0.6112, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.540759493670886, | |
| "grad_norm": 0.7053552397324746, | |
| "learning_rate": 3.491216790442899e-05, | |
| "loss": 0.5993, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.5448101265822785, | |
| "grad_norm": 0.7221156125128689, | |
| "learning_rate": 3.487428481145839e-05, | |
| "loss": 0.6072, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.548860759493671, | |
| "grad_norm": 0.6681163984561158, | |
| "learning_rate": 3.483628192304278e-05, | |
| "loss": 0.5834, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.5529113924050633, | |
| "grad_norm": 0.6632262058030448, | |
| "learning_rate": 3.479815954525219e-05, | |
| "loss": 0.6097, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.5569620253164556, | |
| "grad_norm": 0.7780138917380465, | |
| "learning_rate": 3.475991798511899e-05, | |
| "loss": 0.6161, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.561012658227848, | |
| "grad_norm": 0.6816359237498527, | |
| "learning_rate": 3.4721557550635464e-05, | |
| "loss": 0.6174, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.5650632911392406, | |
| "grad_norm": 0.8339919932184552, | |
| "learning_rate": 3.468307855075128e-05, | |
| "loss": 0.5971, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.5691139240506329, | |
| "grad_norm": 0.6813015967036155, | |
| "learning_rate": 3.4644481295371005e-05, | |
| "loss": 0.5821, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.5731645569620252, | |
| "grad_norm": 0.7965120635893493, | |
| "learning_rate": 3.460576609535163e-05, | |
| "loss": 0.6062, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.5772151898734177, | |
| "grad_norm": 0.6435501194239036, | |
| "learning_rate": 3.456693326250006e-05, | |
| "loss": 0.6057, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.5812658227848102, | |
| "grad_norm": 0.707262541233103, | |
| "learning_rate": 3.452798310957058e-05, | |
| "loss": 0.5886, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.5853164556962025, | |
| "grad_norm": 0.6608686423379359, | |
| "learning_rate": 3.4488915950262386e-05, | |
| "loss": 0.5919, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.5893670886075948, | |
| "grad_norm": 0.6699192173674411, | |
| "learning_rate": 3.4449732099216985e-05, | |
| "loss": 0.5925, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.5934177215189873, | |
| "grad_norm": 0.6941242470360076, | |
| "learning_rate": 3.441043187201574e-05, | |
| "loss": 0.6123, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.5974683544303798, | |
| "grad_norm": 0.6923873978608573, | |
| "learning_rate": 3.437101558517728e-05, | |
| "loss": 0.6083, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.6015189873417721, | |
| "grad_norm": 0.6477389513411483, | |
| "learning_rate": 3.433148355615496e-05, | |
| "loss": 0.6089, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.6055696202531644, | |
| "grad_norm": 0.6400973173731054, | |
| "learning_rate": 3.4291836103334294e-05, | |
| "loss": 0.6303, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.609620253164557, | |
| "grad_norm": 0.6159411174627157, | |
| "learning_rate": 3.425207354603043e-05, | |
| "loss": 0.6366, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.6136708860759494, | |
| "grad_norm": 0.6364720288523773, | |
| "learning_rate": 3.421219620448553e-05, | |
| "loss": 0.5969, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.6177215189873417, | |
| "grad_norm": 0.7048961178927856, | |
| "learning_rate": 3.417220439986623e-05, | |
| "loss": 0.6082, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.621772151898734, | |
| "grad_norm": 0.6621424996175125, | |
| "learning_rate": 3.4132098454261024e-05, | |
| "loss": 0.5999, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.6258227848101265, | |
| "grad_norm": 0.6337939011490217, | |
| "learning_rate": 3.4091878690677676e-05, | |
| "loss": 0.6031, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.629873417721519, | |
| "grad_norm": 0.7582353597356026, | |
| "learning_rate": 3.405154543304065e-05, | |
| "loss": 0.5938, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.6339240506329114, | |
| "grad_norm": 0.6686561811275062, | |
| "learning_rate": 3.401109900618843e-05, | |
| "loss": 0.5993, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.6379746835443036, | |
| "grad_norm": 0.6524513526930209, | |
| "learning_rate": 3.3970539735870996e-05, | |
| "loss": 0.6123, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.6420253164556962, | |
| "grad_norm": 0.6356826321199627, | |
| "learning_rate": 3.392986794874714e-05, | |
| "loss": 0.6067, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.6460759493670887, | |
| "grad_norm": 0.6428202826293429, | |
| "learning_rate": 3.388908397238184e-05, | |
| "loss": 0.5892, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.650126582278481, | |
| "grad_norm": 0.6578164731030356, | |
| "learning_rate": 3.384818813524362e-05, | |
| "loss": 0.6109, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.6541772151898733, | |
| "grad_norm": 0.6277860849291302, | |
| "learning_rate": 3.380718076670195e-05, | |
| "loss": 0.6175, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.6582278481012658, | |
| "grad_norm": 0.6894623218037625, | |
| "learning_rate": 3.376606219702454e-05, | |
| "loss": 0.6174, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.6622784810126583, | |
| "grad_norm": 0.6400012878013461, | |
| "learning_rate": 3.372483275737468e-05, | |
| "loss": 0.5957, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.6663291139240506, | |
| "grad_norm": 0.6576054687023889, | |
| "learning_rate": 3.368349277980861e-05, | |
| "loss": 0.595, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.6703797468354429, | |
| "grad_norm": 0.7015087764486826, | |
| "learning_rate": 3.3642042597272844e-05, | |
| "loss": 0.613, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.6744303797468354, | |
| "grad_norm": 0.6471619355879737, | |
| "learning_rate": 3.360048254360144e-05, | |
| "loss": 0.5839, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.678481012658228, | |
| "grad_norm": 0.6550675664685606, | |
| "learning_rate": 3.355881295351336e-05, | |
| "loss": 0.6086, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.6825316455696202, | |
| "grad_norm": 0.6918711791308227, | |
| "learning_rate": 3.351703416260975e-05, | |
| "loss": 0.5989, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.6865822784810125, | |
| "grad_norm": 0.6546648914292846, | |
| "learning_rate": 3.347514650737126e-05, | |
| "loss": 0.6086, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.690632911392405, | |
| "grad_norm": 0.655315540016654, | |
| "learning_rate": 3.3433150325155295e-05, | |
| "loss": 0.6034, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.6946835443037975, | |
| "grad_norm": 0.6510479889937882, | |
| "learning_rate": 3.339104595419334e-05, | |
| "loss": 0.5979, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.6987341772151898, | |
| "grad_norm": 0.6130162695585983, | |
| "learning_rate": 3.3348833733588204e-05, | |
| "loss": 0.6139, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.7027848101265821, | |
| "grad_norm": 0.6518740163133705, | |
| "learning_rate": 3.3306514003311305e-05, | |
| "loss": 0.5864, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.7068354430379746, | |
| "grad_norm": 0.6091006455257595, | |
| "learning_rate": 3.326408710419996e-05, | |
| "loss": 0.5926, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.7108860759493671, | |
| "grad_norm": 0.6889153650357486, | |
| "learning_rate": 3.322155337795454e-05, | |
| "loss": 0.6032, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.7149367088607594, | |
| "grad_norm": 0.6639456223504862, | |
| "learning_rate": 3.317891316713587e-05, | |
| "loss": 0.5895, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.7189873417721517, | |
| "grad_norm": 0.6009700488778719, | |
| "learning_rate": 3.313616681516231e-05, | |
| "loss": 0.5931, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.7230379746835442, | |
| "grad_norm": 0.7177859326735311, | |
| "learning_rate": 3.309331466630713e-05, | |
| "loss": 0.6065, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.7270886075949368, | |
| "grad_norm": 0.6190155995250232, | |
| "learning_rate": 3.305035706569563e-05, | |
| "loss": 0.5921, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.7311392405063293, | |
| "grad_norm": 0.6599529737183828, | |
| "learning_rate": 3.3007294359302433e-05, | |
| "loss": 0.6121, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.7351898734177216, | |
| "grad_norm": 0.6132886720782609, | |
| "learning_rate": 3.296412689394864e-05, | |
| "loss": 0.6127, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.7392405063291139, | |
| "grad_norm": 0.6480714689109799, | |
| "learning_rate": 3.292085501729909e-05, | |
| "loss": 0.5904, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.7432911392405064, | |
| "grad_norm": 0.5945845215755566, | |
| "learning_rate": 3.2877479077859534e-05, | |
| "loss": 0.6135, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.747341772151899, | |
| "grad_norm": 0.6376605172609395, | |
| "learning_rate": 3.283399942497381e-05, | |
| "loss": 0.608, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.7513924050632912, | |
| "grad_norm": 0.6472289648211133, | |
| "learning_rate": 3.279041640882108e-05, | |
| "loss": 0.6166, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.7554430379746835, | |
| "grad_norm": 0.6050357454230106, | |
| "learning_rate": 3.2746730380412964e-05, | |
| "loss": 0.6042, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.759493670886076, | |
| "grad_norm": 0.5787981862858275, | |
| "learning_rate": 3.2702941691590726e-05, | |
| "loss": 0.5977, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.7635443037974685, | |
| "grad_norm": 0.5989057594208751, | |
| "learning_rate": 3.265905069502244e-05, | |
| "loss": 0.6078, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.7675949367088608, | |
| "grad_norm": 0.6020200314162393, | |
| "learning_rate": 3.261505774420016e-05, | |
| "loss": 0.6018, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.771645569620253, | |
| "grad_norm": 0.6292557386853733, | |
| "learning_rate": 3.257096319343707e-05, | |
| "loss": 0.5976, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.7756962025316456, | |
| "grad_norm": 0.6285488855138599, | |
| "learning_rate": 3.2526767397864614e-05, | |
| "loss": 0.6119, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.7797468354430381, | |
| "grad_norm": 0.6186864254838476, | |
| "learning_rate": 3.248247071342966e-05, | |
| "loss": 0.5881, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.7837974683544304, | |
| "grad_norm": 0.67901545052628, | |
| "learning_rate": 3.243807349689161e-05, | |
| "loss": 0.6041, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.7878481012658227, | |
| "grad_norm": 0.6304523059715319, | |
| "learning_rate": 3.2393576105819544e-05, | |
| "loss": 0.5945, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.7918987341772152, | |
| "grad_norm": 0.6527961672625269, | |
| "learning_rate": 3.2348978898589333e-05, | |
| "loss": 0.6053, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.7959493670886078, | |
| "grad_norm": 0.6577741194722845, | |
| "learning_rate": 3.230428223438075e-05, | |
| "loss": 0.5828, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.5944739129986512, | |
| "learning_rate": 3.225948647317459e-05, | |
| "loss": 0.5874, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.8040506329113923, | |
| "grad_norm": 0.6106773164599709, | |
| "learning_rate": 3.2214591975749745e-05, | |
| "loss": 0.5961, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.8081012658227849, | |
| "grad_norm": 0.6177079870768442, | |
| "learning_rate": 3.216959910368034e-05, | |
| "loss": 0.6206, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.8121518987341774, | |
| "grad_norm": 0.6417167647490404, | |
| "learning_rate": 3.212450821933277e-05, | |
| "loss": 0.5986, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.8162025316455697, | |
| "grad_norm": 0.6063568907373491, | |
| "learning_rate": 3.207931968586281e-05, | |
| "loss": 0.5891, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.820253164556962, | |
| "grad_norm": 0.5898483430694693, | |
| "learning_rate": 3.203403386721272e-05, | |
| "loss": 0.5969, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.8243037974683545, | |
| "grad_norm": 0.6166367764774578, | |
| "learning_rate": 3.1988651128108245e-05, | |
| "loss": 0.5969, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.828354430379747, | |
| "grad_norm": 0.63826177996483, | |
| "learning_rate": 3.194317183405573e-05, | |
| "loss": 0.6034, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.8324050632911393, | |
| "grad_norm": 0.6209766946271629, | |
| "learning_rate": 3.189759635133914e-05, | |
| "loss": 0.5964, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.8364556962025316, | |
| "grad_norm": 0.5993630849239893, | |
| "learning_rate": 3.185192504701718e-05, | |
| "loss": 0.6113, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.840506329113924, | |
| "grad_norm": 0.6044437218869078, | |
| "learning_rate": 3.1806158288920234e-05, | |
| "loss": 0.6037, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.8445569620253166, | |
| "grad_norm": 0.6851027411610361, | |
| "learning_rate": 3.1760296445647477e-05, | |
| "loss": 0.5989, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.848607594936709, | |
| "grad_norm": 0.5892475155121027, | |
| "learning_rate": 3.1714339886563896e-05, | |
| "loss": 0.6077, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.8526582278481012, | |
| "grad_norm": 0.6916743078755053, | |
| "learning_rate": 3.166828898179731e-05, | |
| "loss": 0.6198, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.8567088607594937, | |
| "grad_norm": 0.5887522269564013, | |
| "learning_rate": 3.162214410223536e-05, | |
| "loss": 0.59, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.8607594936708862, | |
| "grad_norm": 0.7302108703279462, | |
| "learning_rate": 3.157590561952257e-05, | |
| "loss": 0.5951, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.8648101265822785, | |
| "grad_norm": 0.666384380549875, | |
| "learning_rate": 3.152957390605732e-05, | |
| "loss": 0.5843, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.8688607594936708, | |
| "grad_norm": 0.7118152981953383, | |
| "learning_rate": 3.148314933498886e-05, | |
| "loss": 0.5873, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.8729113924050633, | |
| "grad_norm": 0.6349724291783378, | |
| "learning_rate": 3.143663228021431e-05, | |
| "loss": 0.5895, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.8769620253164558, | |
| "grad_norm": 0.7429541452215366, | |
| "learning_rate": 3.1390023116375624e-05, | |
| "loss": 0.605, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.8810126582278481, | |
| "grad_norm": 0.6428794165673481, | |
| "learning_rate": 3.134332221885661e-05, | |
| "loss": 0.6035, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.8850632911392404, | |
| "grad_norm": 0.6745658818970699, | |
| "learning_rate": 3.129652996377987e-05, | |
| "loss": 0.601, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.889113924050633, | |
| "grad_norm": 0.6278358851536604, | |
| "learning_rate": 3.12496467280038e-05, | |
| "loss": 0.5861, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.8931645569620255, | |
| "grad_norm": 0.6084195127108488, | |
| "learning_rate": 3.120267288911952e-05, | |
| "loss": 0.593, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.8972151898734178, | |
| "grad_norm": 0.7024092965232351, | |
| "learning_rate": 3.11556088254479e-05, | |
| "loss": 0.6117, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.90126582278481, | |
| "grad_norm": 0.6207925970421655, | |
| "learning_rate": 3.11084549160364e-05, | |
| "loss": 0.5739, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.9053164556962026, | |
| "grad_norm": 0.6541245886419225, | |
| "learning_rate": 3.106121154065615e-05, | |
| "loss": 0.5852, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.909367088607595, | |
| "grad_norm": 0.6614131759201899, | |
| "learning_rate": 3.1013879079798805e-05, | |
| "loss": 0.6002, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.9134177215189874, | |
| "grad_norm": 0.6710827510229159, | |
| "learning_rate": 3.096645791467348e-05, | |
| "loss": 0.5925, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.9174683544303797, | |
| "grad_norm": 0.6281404993934403, | |
| "learning_rate": 3.091894842720373e-05, | |
| "loss": 0.6094, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.9215189873417722, | |
| "grad_norm": 0.7372924355622008, | |
| "learning_rate": 3.0871351000024425e-05, | |
| "loss": 0.6019, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.9255696202531647, | |
| "grad_norm": 0.6210138877833387, | |
| "learning_rate": 3.0823666016478716e-05, | |
| "loss": 0.6147, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.929620253164557, | |
| "grad_norm": 0.6224416174728209, | |
| "learning_rate": 3.0775893860614896e-05, | |
| "loss": 0.6171, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.9336708860759493, | |
| "grad_norm": 0.6622525682212437, | |
| "learning_rate": 3.0728034917183336e-05, | |
| "loss": 0.5965, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.9377215189873418, | |
| "grad_norm": 0.6374940079521862, | |
| "learning_rate": 3.06800895716334e-05, | |
| "loss": 0.5963, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.9417721518987343, | |
| "grad_norm": 0.6303040069813604, | |
| "learning_rate": 3.063205821011029e-05, | |
| "loss": 0.5921, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.9458227848101266, | |
| "grad_norm": 0.6379290208714038, | |
| "learning_rate": 3.0583941219452016e-05, | |
| "loss": 0.6073, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.949873417721519, | |
| "grad_norm": 0.6678336659793668, | |
| "learning_rate": 3.053573898718618e-05, | |
| "loss": 0.6014, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.9539240506329114, | |
| "grad_norm": 0.6291780395834737, | |
| "learning_rate": 3.0487451901526956e-05, | |
| "loss": 0.6089, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.957974683544304, | |
| "grad_norm": 0.6383675083068594, | |
| "learning_rate": 3.0439080351371875e-05, | |
| "loss": 0.603, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.9620253164556962, | |
| "grad_norm": 0.6383833922983616, | |
| "learning_rate": 3.0390624726298764e-05, | |
| "loss": 0.5884, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.9660759493670885, | |
| "grad_norm": 0.5984526011191078, | |
| "learning_rate": 3.034208541656255e-05, | |
| "loss": 0.5992, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.970126582278481, | |
| "grad_norm": 0.6124131381590558, | |
| "learning_rate": 3.029346281309218e-05, | |
| "loss": 0.5943, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.9741772151898735, | |
| "grad_norm": 0.6266964300402461, | |
| "learning_rate": 3.0244757307487415e-05, | |
| "loss": 0.5897, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.9782278481012658, | |
| "grad_norm": 0.6095913009029043, | |
| "learning_rate": 3.019596929201569e-05, | |
| "loss": 0.6012, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.9822784810126581, | |
| "grad_norm": 0.6821835917579058, | |
| "learning_rate": 3.0147099159608985e-05, | |
| "loss": 0.6222, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.9863291139240506, | |
| "grad_norm": 0.5969286766897403, | |
| "learning_rate": 3.0098147303860616e-05, | |
| "loss": 0.5822, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.9903797468354432, | |
| "grad_norm": 0.6164522532192707, | |
| "learning_rate": 3.0049114119022117e-05, | |
| "loss": 0.6125, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.9944303797468355, | |
| "grad_norm": 0.5714348869502746, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 0.5955, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.9984810126582278, | |
| "grad_norm": 0.630106288036123, | |
| "learning_rate": 2.995080534235264e-05, | |
| "loss": 0.5909, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.0040506329113925, | |
| "grad_norm": 1.1518979761075145, | |
| "learning_rate": 2.9901530542287044e-05, | |
| "loss": 0.3627, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.008101265822785, | |
| "grad_norm": 0.653179757903636, | |
| "learning_rate": 2.9852175996655676e-05, | |
| "loss": 0.3502, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.012151898734177, | |
| "grad_norm": 1.5176917129711502, | |
| "learning_rate": 2.980274210295326e-05, | |
| "loss": 0.3516, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.0162025316455696, | |
| "grad_norm": 1.009433723409278, | |
| "learning_rate": 2.9753229259313578e-05, | |
| "loss": 0.3534, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.020253164556962, | |
| "grad_norm": 0.658308835742418, | |
| "learning_rate": 2.9703637864506274e-05, | |
| "loss": 0.3578, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.0243037974683546, | |
| "grad_norm": 0.6822254994411749, | |
| "learning_rate": 2.965396831793362e-05, | |
| "loss": 0.3318, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.0283544303797467, | |
| "grad_norm": 0.6421706163197766, | |
| "learning_rate": 2.9604221019627316e-05, | |
| "loss": 0.3517, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0324050632911392, | |
| "grad_norm": 0.6217105822761836, | |
| "learning_rate": 2.955439637024526e-05, | |
| "loss": 0.3432, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.0364556962025318, | |
| "grad_norm": 0.59728217163167, | |
| "learning_rate": 2.9504494771068334e-05, | |
| "loss": 0.343, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.0405063291139243, | |
| "grad_norm": 0.6695204918325587, | |
| "learning_rate": 2.9454516623997156e-05, | |
| "loss": 0.3351, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.0445569620253163, | |
| "grad_norm": 0.6530678889385765, | |
| "learning_rate": 2.9404462331548847e-05, | |
| "loss": 0.3382, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.048607594936709, | |
| "grad_norm": 0.6391474109833464, | |
| "learning_rate": 2.93543322968538e-05, | |
| "loss": 0.343, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.0526582278481014, | |
| "grad_norm": 0.6122975894152145, | |
| "learning_rate": 2.9304126923652428e-05, | |
| "loss": 0.3534, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.056708860759494, | |
| "grad_norm": 0.5931271111354764, | |
| "learning_rate": 2.9253846616291896e-05, | |
| "loss": 0.3453, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.060759493670886, | |
| "grad_norm": 0.5864820241918635, | |
| "learning_rate": 2.9203491779722896e-05, | |
| "loss": 0.3454, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.0648101265822785, | |
| "grad_norm": 0.5993188142300935, | |
| "learning_rate": 2.9153062819496357e-05, | |
| "loss": 0.3321, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.068860759493671, | |
| "grad_norm": 0.6078637859914079, | |
| "learning_rate": 2.9102560141760178e-05, | |
| "loss": 0.3345, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.0729113924050635, | |
| "grad_norm": 0.6363385111439686, | |
| "learning_rate": 2.9051984153256004e-05, | |
| "loss": 0.3574, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.0769620253164556, | |
| "grad_norm": 0.5986947607146489, | |
| "learning_rate": 2.900133526131588e-05, | |
| "loss": 0.336, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.081012658227848, | |
| "grad_norm": 0.604755542933451, | |
| "learning_rate": 2.8950613873859025e-05, | |
| "loss": 0.3317, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.0850632911392406, | |
| "grad_norm": 0.5873502215912892, | |
| "learning_rate": 2.8899820399388515e-05, | |
| "loss": 0.3467, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.089113924050633, | |
| "grad_norm": 0.5891263376539458, | |
| "learning_rate": 2.8848955246988012e-05, | |
| "loss": 0.3541, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.093164556962025, | |
| "grad_norm": 0.6036353056308736, | |
| "learning_rate": 2.879801882631847e-05, | |
| "loss": 0.3419, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.0972151898734177, | |
| "grad_norm": 0.5930073500517279, | |
| "learning_rate": 2.8747011547614808e-05, | |
| "loss": 0.3478, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.1012658227848102, | |
| "grad_norm": 0.5832974160890019, | |
| "learning_rate": 2.8695933821682635e-05, | |
| "loss": 0.3595, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.1053164556962027, | |
| "grad_norm": 0.6638055351899846, | |
| "learning_rate": 2.864478605989494e-05, | |
| "loss": 0.3473, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.109367088607595, | |
| "grad_norm": 0.6028349292002557, | |
| "learning_rate": 2.8593568674188765e-05, | |
| "loss": 0.3443, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.1134177215189873, | |
| "grad_norm": 0.6356993989429873, | |
| "learning_rate": 2.8542282077061892e-05, | |
| "loss": 0.343, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.11746835443038, | |
| "grad_norm": 0.6556173204883355, | |
| "learning_rate": 2.8490926681569523e-05, | |
| "loss": 0.3641, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.1215189873417724, | |
| "grad_norm": 0.5753244710876829, | |
| "learning_rate": 2.8439502901320956e-05, | |
| "loss": 0.3354, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 2.1255696202531644, | |
| "grad_norm": 0.6927049388708466, | |
| "learning_rate": 2.8388011150476237e-05, | |
| "loss": 0.3414, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 2.129620253164557, | |
| "grad_norm": 0.5948077370578037, | |
| "learning_rate": 2.8336451843742866e-05, | |
| "loss": 0.3324, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.1336708860759495, | |
| "grad_norm": 0.6098878368812227, | |
| "learning_rate": 2.8284825396372387e-05, | |
| "loss": 0.3498, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 2.137721518987342, | |
| "grad_norm": 0.5982003815495254, | |
| "learning_rate": 2.8233132224157132e-05, | |
| "loss": 0.3228, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 2.141772151898734, | |
| "grad_norm": 0.6226433262951259, | |
| "learning_rate": 2.8181372743426805e-05, | |
| "loss": 0.3477, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 2.1458227848101266, | |
| "grad_norm": 0.5781419850319266, | |
| "learning_rate": 2.8129547371045128e-05, | |
| "loss": 0.3293, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 2.149873417721519, | |
| "grad_norm": 0.5981056614872015, | |
| "learning_rate": 2.8077656524406534e-05, | |
| "loss": 0.3338, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.1539240506329116, | |
| "grad_norm": 0.5948809085301509, | |
| "learning_rate": 2.802570062143278e-05, | |
| "loss": 0.3462, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 2.1579746835443037, | |
| "grad_norm": 0.617197027366774, | |
| "learning_rate": 2.7973680080569555e-05, | |
| "loss": 0.3567, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 2.162025316455696, | |
| "grad_norm": 0.5662977950545237, | |
| "learning_rate": 2.792159532078314e-05, | |
| "loss": 0.3271, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 2.1660759493670887, | |
| "grad_norm": 0.595761431265803, | |
| "learning_rate": 2.7869446761557033e-05, | |
| "loss": 0.3525, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 2.170126582278481, | |
| "grad_norm": 0.560071374881526, | |
| "learning_rate": 2.781723482288857e-05, | |
| "loss": 0.3415, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.1741772151898733, | |
| "grad_norm": 0.5857838254668501, | |
| "learning_rate": 2.7764959925285517e-05, | |
| "loss": 0.3384, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.178227848101266, | |
| "grad_norm": 0.6026878463417695, | |
| "learning_rate": 2.771262248976272e-05, | |
| "loss": 0.3435, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 2.1822784810126583, | |
| "grad_norm": 0.5584996391655568, | |
| "learning_rate": 2.7660222937838677e-05, | |
| "loss": 0.36, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 2.186329113924051, | |
| "grad_norm": 0.563401922518865, | |
| "learning_rate": 2.7607761691532186e-05, | |
| "loss": 0.337, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 2.190379746835443, | |
| "grad_norm": 0.5684246648420656, | |
| "learning_rate": 2.7555239173358916e-05, | |
| "loss": 0.34, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.1944303797468354, | |
| "grad_norm": 0.5603022933943614, | |
| "learning_rate": 2.7502655806328e-05, | |
| "loss": 0.3506, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 2.198481012658228, | |
| "grad_norm": 0.5436652976661219, | |
| "learning_rate": 2.7450012013938648e-05, | |
| "loss": 0.3347, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 2.2025316455696204, | |
| "grad_norm": 0.5643776475339053, | |
| "learning_rate": 2.739730822017673e-05, | |
| "loss": 0.3433, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 2.2065822784810125, | |
| "grad_norm": 0.5898490493353038, | |
| "learning_rate": 2.7344544849511355e-05, | |
| "loss": 0.3407, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 2.210632911392405, | |
| "grad_norm": 0.5770410449196542, | |
| "learning_rate": 2.7291722326891456e-05, | |
| "loss": 0.3386, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.2146835443037975, | |
| "grad_norm": 0.590406024627372, | |
| "learning_rate": 2.723884107774236e-05, | |
| "loss": 0.3419, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 2.21873417721519, | |
| "grad_norm": 0.5926454091147866, | |
| "learning_rate": 2.718590152796239e-05, | |
| "loss": 0.3456, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 2.222784810126582, | |
| "grad_norm": 0.5456714341846756, | |
| "learning_rate": 2.71329041039194e-05, | |
| "loss": 0.3328, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 2.2268354430379746, | |
| "grad_norm": 0.5854609110123287, | |
| "learning_rate": 2.7079849232447357e-05, | |
| "loss": 0.3437, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.230886075949367, | |
| "grad_norm": 0.5657988988905419, | |
| "learning_rate": 2.7026737340842895e-05, | |
| "loss": 0.3442, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.2349367088607597, | |
| "grad_norm": 0.6056038770812514, | |
| "learning_rate": 2.697356885686189e-05, | |
| "loss": 0.3363, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.2389873417721518, | |
| "grad_norm": 0.6106750211917192, | |
| "learning_rate": 2.6920344208716014e-05, | |
| "loss": 0.3409, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.2430379746835443, | |
| "grad_norm": 0.5818239420615721, | |
| "learning_rate": 2.6867063825069252e-05, | |
| "loss": 0.3523, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.247088607594937, | |
| "grad_norm": 0.6217776566491441, | |
| "learning_rate": 2.6813728135034494e-05, | |
| "loss": 0.361, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.2511392405063293, | |
| "grad_norm": 0.5766409650755651, | |
| "learning_rate": 2.6760337568170056e-05, | |
| "loss": 0.3471, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.2551898734177214, | |
| "grad_norm": 0.6389123268537097, | |
| "learning_rate": 2.6706892554476226e-05, | |
| "loss": 0.3422, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.259240506329114, | |
| "grad_norm": 0.5626642149204923, | |
| "learning_rate": 2.6653393524391795e-05, | |
| "loss": 0.3321, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.2632911392405064, | |
| "grad_norm": 0.6218696495982626, | |
| "learning_rate": 2.6599840908790592e-05, | |
| "loss": 0.3432, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.267341772151899, | |
| "grad_norm": 0.5761521641753026, | |
| "learning_rate": 2.6546235138978028e-05, | |
| "loss": 0.332, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.271392405063291, | |
| "grad_norm": 0.652170311523693, | |
| "learning_rate": 2.6492576646687597e-05, | |
| "loss": 0.3415, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.2754430379746835, | |
| "grad_norm": 0.5911319054809049, | |
| "learning_rate": 2.6438865864077425e-05, | |
| "loss": 0.3579, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.279493670886076, | |
| "grad_norm": 0.7345703302684237, | |
| "learning_rate": 2.6385103223726766e-05, | |
| "loss": 0.3622, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.2835443037974685, | |
| "grad_norm": 0.5692508508706504, | |
| "learning_rate": 2.6331289158632537e-05, | |
| "loss": 0.3319, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.2875949367088606, | |
| "grad_norm": 0.6110264652044216, | |
| "learning_rate": 2.6277424102205817e-05, | |
| "loss": 0.36, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.291645569620253, | |
| "grad_norm": 0.609334670664584, | |
| "learning_rate": 2.6223508488268374e-05, | |
| "loss": 0.3296, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.2956962025316456, | |
| "grad_norm": 0.556712818181301, | |
| "learning_rate": 2.6169542751049148e-05, | |
| "loss": 0.3372, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.299746835443038, | |
| "grad_norm": 0.588581441856511, | |
| "learning_rate": 2.6115527325180754e-05, | |
| "loss": 0.3478, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.3037974683544302, | |
| "grad_norm": 0.5834716518921821, | |
| "learning_rate": 2.606146264569603e-05, | |
| "loss": 0.3415, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.3078481012658227, | |
| "grad_norm": 0.5719098437943086, | |
| "learning_rate": 2.6007349148024447e-05, | |
| "loss": 0.3594, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.3118987341772153, | |
| "grad_norm": 0.5797607208700623, | |
| "learning_rate": 2.5953187267988694e-05, | |
| "loss": 0.3427, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.3159493670886078, | |
| "grad_norm": 0.5623062073115244, | |
| "learning_rate": 2.5898977441801097e-05, | |
| "loss": 0.3378, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.5673674823995074, | |
| "learning_rate": 2.584472010606015e-05, | |
| "loss": 0.3351, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.3240506329113924, | |
| "grad_norm": 0.5652213344761505, | |
| "learning_rate": 2.5790415697746976e-05, | |
| "loss": 0.3473, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.328101265822785, | |
| "grad_norm": 0.577707513709347, | |
| "learning_rate": 2.5736064654221808e-05, | |
| "loss": 0.3345, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.3321518987341774, | |
| "grad_norm": 0.6012070592909171, | |
| "learning_rate": 2.568166741322048e-05, | |
| "loss": 0.3621, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.3362025316455695, | |
| "grad_norm": 0.6030769560056746, | |
| "learning_rate": 2.56272244128509e-05, | |
| "loss": 0.3404, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.340253164556962, | |
| "grad_norm": 0.6023077550691454, | |
| "learning_rate": 2.55727360915895e-05, | |
| "loss": 0.3458, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.3443037974683545, | |
| "grad_norm": 0.5846582979545826, | |
| "learning_rate": 2.5518202888277734e-05, | |
| "loss": 0.3486, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.348354430379747, | |
| "grad_norm": 0.5878963377185566, | |
| "learning_rate": 2.5463625242118523e-05, | |
| "loss": 0.3405, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.352405063291139, | |
| "grad_norm": 0.5887299575383621, | |
| "learning_rate": 2.5409003592672723e-05, | |
| "loss": 0.3381, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.3564556962025316, | |
| "grad_norm": 0.6013596744868673, | |
| "learning_rate": 2.535433837985559e-05, | |
| "loss": 0.3424, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.360506329113924, | |
| "grad_norm": 0.6178388553193682, | |
| "learning_rate": 2.529963004393324e-05, | |
| "loss": 0.3422, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.3645569620253166, | |
| "grad_norm": 0.5758496661180981, | |
| "learning_rate": 2.524487902551908e-05, | |
| "loss": 0.3241, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.3686075949367087, | |
| "grad_norm": 0.6087304176318568, | |
| "learning_rate": 2.519008576557029e-05, | |
| "loss": 0.3473, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.372658227848101, | |
| "grad_norm": 0.5615148730231416, | |
| "learning_rate": 2.5135250705384254e-05, | |
| "loss": 0.3649, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.3767088607594937, | |
| "grad_norm": 0.6250657713983367, | |
| "learning_rate": 2.5080374286595007e-05, | |
| "loss": 0.3592, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.3807594936708862, | |
| "grad_norm": 0.5841685943396147, | |
| "learning_rate": 2.5025456951169677e-05, | |
| "loss": 0.3441, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.3848101265822783, | |
| "grad_norm": 0.5964560897584598, | |
| "learning_rate": 2.4970499141404942e-05, | |
| "loss": 0.3371, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.388860759493671, | |
| "grad_norm": 0.5491380146815262, | |
| "learning_rate": 2.491550129992345e-05, | |
| "loss": 0.3534, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.3929113924050633, | |
| "grad_norm": 0.6031154129884805, | |
| "learning_rate": 2.486046386967024e-05, | |
| "loss": 0.3439, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.396962025316456, | |
| "grad_norm": 0.5963174077856629, | |
| "learning_rate": 2.4805387293909214e-05, | |
| "loss": 0.364, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.401012658227848, | |
| "grad_norm": 0.5674013300812099, | |
| "learning_rate": 2.4750272016219552e-05, | |
| "loss": 0.3511, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.4050632911392404, | |
| "grad_norm": 0.5783183841491331, | |
| "learning_rate": 2.4695118480492114e-05, | |
| "loss": 0.3379, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.409113924050633, | |
| "grad_norm": 0.5528010623073173, | |
| "learning_rate": 2.4639927130925898e-05, | |
| "loss": 0.3407, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.4131645569620255, | |
| "grad_norm": 0.562495107266098, | |
| "learning_rate": 2.458469841202444e-05, | |
| "loss": 0.3493, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.4172151898734175, | |
| "grad_norm": 0.5768503091350035, | |
| "learning_rate": 2.452943276859226e-05, | |
| "loss": 0.3384, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.42126582278481, | |
| "grad_norm": 0.6051804140976867, | |
| "learning_rate": 2.447413064573125e-05, | |
| "loss": 0.359, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.4253164556962026, | |
| "grad_norm": 0.5736602571247281, | |
| "learning_rate": 2.4418792488837095e-05, | |
| "loss": 0.3601, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.429367088607595, | |
| "grad_norm": 0.5671503680262563, | |
| "learning_rate": 2.4363418743595713e-05, | |
| "loss": 0.3468, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 2.433417721518987, | |
| "grad_norm": 0.5744887064186983, | |
| "learning_rate": 2.430800985597963e-05, | |
| "loss": 0.3538, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.4374683544303797, | |
| "grad_norm": 0.558585320762786, | |
| "learning_rate": 2.4252566272244415e-05, | |
| "loss": 0.3414, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 2.441518987341772, | |
| "grad_norm": 0.5763493259726282, | |
| "learning_rate": 2.4197088438925063e-05, | |
| "loss": 0.3355, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 2.4455696202531647, | |
| "grad_norm": 0.5609114839204852, | |
| "learning_rate": 2.4141576802832417e-05, | |
| "loss": 0.345, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 2.449620253164557, | |
| "grad_norm": 0.5460048926706284, | |
| "learning_rate": 2.408603181104957e-05, | |
| "loss": 0.3552, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 2.4536708860759493, | |
| "grad_norm": 0.5713405650838015, | |
| "learning_rate": 2.4030453910928245e-05, | |
| "loss": 0.3362, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.457721518987342, | |
| "grad_norm": 0.5668012235503698, | |
| "learning_rate": 2.397484355008521e-05, | |
| "loss": 0.3571, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 2.4617721518987343, | |
| "grad_norm": 0.5713305112035771, | |
| "learning_rate": 2.3919201176398662e-05, | |
| "loss": 0.3455, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 2.4658227848101264, | |
| "grad_norm": 0.5703946584777596, | |
| "learning_rate": 2.3863527238004633e-05, | |
| "loss": 0.3413, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 2.469873417721519, | |
| "grad_norm": 0.5689553581994804, | |
| "learning_rate": 2.380782218329337e-05, | |
| "loss": 0.3496, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 2.4739240506329114, | |
| "grad_norm": 0.5861506997208186, | |
| "learning_rate": 2.3752086460905725e-05, | |
| "loss": 0.3382, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.477974683544304, | |
| "grad_norm": 0.584308049278221, | |
| "learning_rate": 2.3696320519729544e-05, | |
| "loss": 0.3574, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 2.482025316455696, | |
| "grad_norm": 0.5713187186372324, | |
| "learning_rate": 2.3640524808896045e-05, | |
| "loss": 0.3511, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 2.4860759493670885, | |
| "grad_norm": 0.5338925537995782, | |
| "learning_rate": 2.3584699777776222e-05, | |
| "loss": 0.3382, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 2.490126582278481, | |
| "grad_norm": 0.5805381994829119, | |
| "learning_rate": 2.3528845875977195e-05, | |
| "loss": 0.3587, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 2.4941772151898736, | |
| "grad_norm": 0.5704779712106572, | |
| "learning_rate": 2.3472963553338614e-05, | |
| "loss": 0.3474, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.4982278481012656, | |
| "grad_norm": 0.5603133653130433, | |
| "learning_rate": 2.341705325992901e-05, | |
| "loss": 0.3363, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.502278481012658, | |
| "grad_norm": 0.5369916522275905, | |
| "learning_rate": 2.336111544604222e-05, | |
| "loss": 0.3323, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 2.5063291139240507, | |
| "grad_norm": 0.5551207166969423, | |
| "learning_rate": 2.33051505621937e-05, | |
| "loss": 0.3552, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.510379746835443, | |
| "grad_norm": 0.548270815280531, | |
| "learning_rate": 2.324915905911693e-05, | |
| "loss": 0.3489, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 2.5144303797468357, | |
| "grad_norm": 0.5678602138117044, | |
| "learning_rate": 2.319314138775977e-05, | |
| "loss": 0.3463, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.5184810126582278, | |
| "grad_norm": 0.5688810660569101, | |
| "learning_rate": 2.3137097999280856e-05, | |
| "loss": 0.3521, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 2.5225316455696203, | |
| "grad_norm": 0.5681094736083652, | |
| "learning_rate": 2.308102934504593e-05, | |
| "loss": 0.3377, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.526582278481013, | |
| "grad_norm": 0.5460221631913441, | |
| "learning_rate": 2.3024935876624222e-05, | |
| "loss": 0.3453, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 2.530632911392405, | |
| "grad_norm": 0.5631489381915669, | |
| "learning_rate": 2.2968818045784813e-05, | |
| "loss": 0.3353, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.5346835443037974, | |
| "grad_norm": 0.5621585645622271, | |
| "learning_rate": 2.2912676304493006e-05, | |
| "loss": 0.3337, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.53873417721519, | |
| "grad_norm": 0.5697108456097045, | |
| "learning_rate": 2.2856511104906668e-05, | |
| "loss": 0.3399, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.5427848101265824, | |
| "grad_norm": 0.5897375043130977, | |
| "learning_rate": 2.2800322899372586e-05, | |
| "loss": 0.3494, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 2.546835443037975, | |
| "grad_norm": 0.5445479722494039, | |
| "learning_rate": 2.2744112140422844e-05, | |
| "loss": 0.3455, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.550886075949367, | |
| "grad_norm": 0.5788463758724484, | |
| "learning_rate": 2.2687879280771177e-05, | |
| "loss": 0.3689, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 2.5549367088607595, | |
| "grad_norm": 0.5717572942518326, | |
| "learning_rate": 2.26316247733093e-05, | |
| "loss": 0.3488, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.558987341772152, | |
| "grad_norm": 0.5553248727028708, | |
| "learning_rate": 2.257534907110328e-05, | |
| "loss": 0.3326, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 2.563037974683544, | |
| "grad_norm": 0.5673005533964679, | |
| "learning_rate": 2.2519052627389882e-05, | |
| "loss": 0.3509, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.5670886075949366, | |
| "grad_norm": 0.5443521657605581, | |
| "learning_rate": 2.246273589557294e-05, | |
| "loss": 0.349, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 2.571139240506329, | |
| "grad_norm": 0.562311228070369, | |
| "learning_rate": 2.240639932921966e-05, | |
| "loss": 0.3687, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.5751898734177217, | |
| "grad_norm": 0.5549706753303523, | |
| "learning_rate": 2.2350043382056995e-05, | |
| "loss": 0.3381, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.579240506329114, | |
| "grad_norm": 0.5359033525849632, | |
| "learning_rate": 2.2293668507968015e-05, | |
| "loss": 0.3432, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.5832911392405062, | |
| "grad_norm": 0.5800500561397844, | |
| "learning_rate": 2.2237275160988186e-05, | |
| "loss": 0.351, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 2.5873417721518988, | |
| "grad_norm": 0.5478687832832912, | |
| "learning_rate": 2.2180863795301787e-05, | |
| "loss": 0.3472, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.5913924050632913, | |
| "grad_norm": 0.5496006645932023, | |
| "learning_rate": 2.212443486523819e-05, | |
| "loss": 0.3304, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 2.5954430379746833, | |
| "grad_norm": 0.5646262864337964, | |
| "learning_rate": 2.2067988825268243e-05, | |
| "loss": 0.3502, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.599493670886076, | |
| "grad_norm": 0.5450408365498517, | |
| "learning_rate": 2.2011526130000596e-05, | |
| "loss": 0.3366, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 2.6035443037974684, | |
| "grad_norm": 0.5685895124899917, | |
| "learning_rate": 2.1955047234178038e-05, | |
| "loss": 0.3634, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.607594936708861, | |
| "grad_norm": 0.5583367715702984, | |
| "learning_rate": 2.1898552592673825e-05, | |
| "loss": 0.3267, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 2.6116455696202534, | |
| "grad_norm": 0.5522731876452933, | |
| "learning_rate": 2.184204266048803e-05, | |
| "loss": 0.3549, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.6156962025316455, | |
| "grad_norm": 0.5855100120952624, | |
| "learning_rate": 2.1785517892743887e-05, | |
| "loss": 0.3495, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.619746835443038, | |
| "grad_norm": 0.5546587760987843, | |
| "learning_rate": 2.17289787446841e-05, | |
| "loss": 0.3416, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.6237974683544305, | |
| "grad_norm": 0.5583993794563842, | |
| "learning_rate": 2.1672425671667198e-05, | |
| "loss": 0.3538, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 2.6278481012658226, | |
| "grad_norm": 0.5699326202378778, | |
| "learning_rate": 2.161585912916385e-05, | |
| "loss": 0.3556, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.631898734177215, | |
| "grad_norm": 0.5682081477096774, | |
| "learning_rate": 2.1559279572753214e-05, | |
| "loss": 0.327, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.6359493670886076, | |
| "grad_norm": 0.5706513277451155, | |
| "learning_rate": 2.1502687458119268e-05, | |
| "loss": 0.3328, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.5628142580237316, | |
| "learning_rate": 2.1446083241047116e-05, | |
| "loss": 0.3589, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.6440506329113926, | |
| "grad_norm": 0.5716962112137539, | |
| "learning_rate": 2.1389467377419333e-05, | |
| "loss": 0.3398, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.6481012658227847, | |
| "grad_norm": 0.5390682266860616, | |
| "learning_rate": 2.133284032321232e-05, | |
| "loss": 0.3357, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.6521518987341772, | |
| "grad_norm": 0.5700480462547852, | |
| "learning_rate": 2.1276202534492566e-05, | |
| "loss": 0.3599, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.6562025316455697, | |
| "grad_norm": 0.5634997759870327, | |
| "learning_rate": 2.121955446741306e-05, | |
| "loss": 0.3427, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.660253164556962, | |
| "grad_norm": 0.548592958610737, | |
| "learning_rate": 2.1162896578209517e-05, | |
| "loss": 0.3334, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.6643037974683543, | |
| "grad_norm": 0.54689775046403, | |
| "learning_rate": 2.1106229323196813e-05, | |
| "loss": 0.3434, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.668354430379747, | |
| "grad_norm": 0.5457482607880363, | |
| "learning_rate": 2.1049553158765214e-05, | |
| "loss": 0.3562, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.6724050632911394, | |
| "grad_norm": 0.5644714409701188, | |
| "learning_rate": 2.0992868541376764e-05, | |
| "loss": 0.3527, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.676455696202532, | |
| "grad_norm": 0.5588052053745256, | |
| "learning_rate": 2.093617592756158e-05, | |
| "loss": 0.3562, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.680506329113924, | |
| "grad_norm": 0.5651050727680258, | |
| "learning_rate": 2.0879475773914167e-05, | |
| "loss": 0.3458, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.6845569620253165, | |
| "grad_norm": 0.5484185417333719, | |
| "learning_rate": 2.082276853708978e-05, | |
| "loss": 0.3594, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.688607594936709, | |
| "grad_norm": 0.5711501116241168, | |
| "learning_rate": 2.076605467380071e-05, | |
| "loss": 0.345, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.692658227848101, | |
| "grad_norm": 0.5547864352230584, | |
| "learning_rate": 2.0709334640812613e-05, | |
| "loss": 0.3312, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.6967088607594936, | |
| "grad_norm": 0.5310702219792736, | |
| "learning_rate": 2.0652608894940824e-05, | |
| "loss": 0.3554, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.700759493670886, | |
| "grad_norm": 0.6096481520754055, | |
| "learning_rate": 2.0595877893046722e-05, | |
| "loss": 0.3368, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.7048101265822786, | |
| "grad_norm": 0.5327875994139907, | |
| "learning_rate": 2.0539142092033985e-05, | |
| "loss": 0.367, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.708860759493671, | |
| "grad_norm": 0.544850767259008, | |
| "learning_rate": 2.048240194884496e-05, | |
| "loss": 0.3442, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.712911392405063, | |
| "grad_norm": 0.5574176483877635, | |
| "learning_rate": 2.042565792045695e-05, | |
| "loss": 0.3408, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.7169620253164557, | |
| "grad_norm": 0.5257236600574646, | |
| "learning_rate": 2.036891046387857e-05, | |
| "loss": 0.3434, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.721012658227848, | |
| "grad_norm": 0.5844692681102625, | |
| "learning_rate": 2.0312160036146036e-05, | |
| "loss": 0.3432, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.7250632911392403, | |
| "grad_norm": 0.5580913708299551, | |
| "learning_rate": 2.025540709431948e-05, | |
| "loss": 0.3486, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.729113924050633, | |
| "grad_norm": 0.5296133686083374, | |
| "learning_rate": 2.0198652095479298e-05, | |
| "loss": 0.3416, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.7331645569620253, | |
| "grad_norm": 0.5930780761936966, | |
| "learning_rate": 2.014189549672245e-05, | |
| "loss": 0.353, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.737215189873418, | |
| "grad_norm": 0.5597659319740564, | |
| "learning_rate": 2.0085137755158776e-05, | |
| "loss": 0.3538, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.7412658227848103, | |
| "grad_norm": 0.5677981500657492, | |
| "learning_rate": 2.0028379327907327e-05, | |
| "loss": 0.3386, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.7453164556962024, | |
| "grad_norm": 0.5474937981197034, | |
| "learning_rate": 1.9971620672092676e-05, | |
| "loss": 0.348, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.749367088607595, | |
| "grad_norm": 0.5399841771865885, | |
| "learning_rate": 1.991486224484123e-05, | |
| "loss": 0.3496, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.7534177215189874, | |
| "grad_norm": 0.5540281574017085, | |
| "learning_rate": 1.985810450327756e-05, | |
| "loss": 0.3414, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.7574683544303795, | |
| "grad_norm": 0.5351045063026937, | |
| "learning_rate": 1.9801347904520706e-05, | |
| "loss": 0.3371, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.761518987341772, | |
| "grad_norm": 0.5314665156545249, | |
| "learning_rate": 1.974459290568053e-05, | |
| "loss": 0.3608, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.7655696202531646, | |
| "grad_norm": 0.5538635654635424, | |
| "learning_rate": 1.968783996385397e-05, | |
| "loss": 0.3393, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.769620253164557, | |
| "grad_norm": 0.5470142958206495, | |
| "learning_rate": 1.963108953612143e-05, | |
| "loss": 0.3512, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.7736708860759496, | |
| "grad_norm": 0.5573294296406678, | |
| "learning_rate": 1.9574342079543056e-05, | |
| "loss": 0.3339, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.7777215189873417, | |
| "grad_norm": 0.5451778421505348, | |
| "learning_rate": 1.9517598051155046e-05, | |
| "loss": 0.3558, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.781772151898734, | |
| "grad_norm": 0.5274847935279131, | |
| "learning_rate": 1.9460857907966025e-05, | |
| "loss": 0.3511, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.7858227848101267, | |
| "grad_norm": 0.5622567585931931, | |
| "learning_rate": 1.9404122106953285e-05, | |
| "loss": 0.3406, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.7898734177215188, | |
| "grad_norm": 0.5522627479399809, | |
| "learning_rate": 1.9347391105059176e-05, | |
| "loss": 0.3465, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.7939240506329113, | |
| "grad_norm": 0.5432686531642467, | |
| "learning_rate": 1.92906653591874e-05, | |
| "loss": 0.3396, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.797974683544304, | |
| "grad_norm": 0.567215623436755, | |
| "learning_rate": 1.9233945326199295e-05, | |
| "loss": 0.3351, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.8020253164556963, | |
| "grad_norm": 0.5363347518344822, | |
| "learning_rate": 1.917723146291022e-05, | |
| "loss": 0.3315, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.806075949367089, | |
| "grad_norm": 0.5506404541299195, | |
| "learning_rate": 1.912052422608584e-05, | |
| "loss": 0.3383, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.810126582278481, | |
| "grad_norm": 0.5462390091853541, | |
| "learning_rate": 1.9063824072438428e-05, | |
| "loss": 0.3332, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.8141772151898734, | |
| "grad_norm": 0.5722217444694717, | |
| "learning_rate": 1.9007131458623246e-05, | |
| "loss": 0.3491, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.818227848101266, | |
| "grad_norm": 0.5527651544910991, | |
| "learning_rate": 1.895044684123479e-05, | |
| "loss": 0.3457, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.822278481012658, | |
| "grad_norm": 0.5648419891172669, | |
| "learning_rate": 1.8893770676803194e-05, | |
| "loss": 0.3454, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.8263291139240505, | |
| "grad_norm": 0.5338294041295641, | |
| "learning_rate": 1.8837103421790486e-05, | |
| "loss": 0.345, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.830379746835443, | |
| "grad_norm": 0.5178989387337382, | |
| "learning_rate": 1.8780445532586952e-05, | |
| "loss": 0.3485, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.8344303797468355, | |
| "grad_norm": 0.5663368416629843, | |
| "learning_rate": 1.872379746550743e-05, | |
| "loss": 0.3365, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.838481012658228, | |
| "grad_norm": 0.5331874906472702, | |
| "learning_rate": 1.866715967678769e-05, | |
| "loss": 0.335, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.84253164556962, | |
| "grad_norm": 0.5348817606802119, | |
| "learning_rate": 1.861053262258067e-05, | |
| "loss": 0.3557, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.8465822784810126, | |
| "grad_norm": 0.560949076574053, | |
| "learning_rate": 1.8553916758952897e-05, | |
| "loss": 0.3472, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.850632911392405, | |
| "grad_norm": 0.566111685667874, | |
| "learning_rate": 1.8497312541880735e-05, | |
| "loss": 0.3501, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.8546835443037972, | |
| "grad_norm": 0.5425004841030103, | |
| "learning_rate": 1.8440720427246786e-05, | |
| "loss": 0.329, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.8587341772151897, | |
| "grad_norm": 0.549603733417412, | |
| "learning_rate": 1.8384140870836157e-05, | |
| "loss": 0.3455, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.8627848101265823, | |
| "grad_norm": 0.521482532866843, | |
| "learning_rate": 1.8327574328332806e-05, | |
| "loss": 0.3502, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.8668354430379748, | |
| "grad_norm": 0.5398728849912336, | |
| "learning_rate": 1.8271021255315906e-05, | |
| "loss": 0.3378, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.8708860759493673, | |
| "grad_norm": 0.5402519866925589, | |
| "learning_rate": 1.8214482107256117e-05, | |
| "loss": 0.3256, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.8749367088607594, | |
| "grad_norm": 0.5522158363341279, | |
| "learning_rate": 1.8157957339511968e-05, | |
| "loss": 0.3411, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.878987341772152, | |
| "grad_norm": 0.548417562770566, | |
| "learning_rate": 1.8101447407326182e-05, | |
| "loss": 0.3459, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.8830379746835444, | |
| "grad_norm": 0.5468061404734021, | |
| "learning_rate": 1.8044952765821966e-05, | |
| "loss": 0.3422, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.8870886075949365, | |
| "grad_norm": 0.5638020933236612, | |
| "learning_rate": 1.7988473869999407e-05, | |
| "loss": 0.3437, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.891139240506329, | |
| "grad_norm": 0.5471592860287949, | |
| "learning_rate": 1.7932011174731764e-05, | |
| "loss": 0.3539, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.8951898734177215, | |
| "grad_norm": 0.552867773200172, | |
| "learning_rate": 1.7875565134761817e-05, | |
| "loss": 0.3528, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.899240506329114, | |
| "grad_norm": 0.5614480397322925, | |
| "learning_rate": 1.7819136204698226e-05, | |
| "loss": 0.3386, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.9032911392405065, | |
| "grad_norm": 0.5425471966642801, | |
| "learning_rate": 1.776272483901182e-05, | |
| "loss": 0.3357, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.9073417721518986, | |
| "grad_norm": 0.5621081492594779, | |
| "learning_rate": 1.7706331492031995e-05, | |
| "loss": 0.3433, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.911392405063291, | |
| "grad_norm": 0.548241891812071, | |
| "learning_rate": 1.764995661794301e-05, | |
| "loss": 0.3444, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.9154430379746836, | |
| "grad_norm": 0.5629488187927907, | |
| "learning_rate": 1.759360067078035e-05, | |
| "loss": 0.3376, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.9194936708860757, | |
| "grad_norm": 0.5568155930773715, | |
| "learning_rate": 1.7537264104427064e-05, | |
| "loss": 0.3455, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.923544303797468, | |
| "grad_norm": 0.5626364084070516, | |
| "learning_rate": 1.748094737261012e-05, | |
| "loss": 0.3413, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.9275949367088607, | |
| "grad_norm": 0.5513819356168544, | |
| "learning_rate": 1.7424650928896726e-05, | |
| "loss": 0.3367, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.9316455696202532, | |
| "grad_norm": 0.5320407372179361, | |
| "learning_rate": 1.7368375226690712e-05, | |
| "loss": 0.3302, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.9356962025316458, | |
| "grad_norm": 0.5520784707510994, | |
| "learning_rate": 1.731212071922883e-05, | |
| "loss": 0.3489, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.939746835443038, | |
| "grad_norm": 0.5721154284229713, | |
| "learning_rate": 1.7255887859577156e-05, | |
| "loss": 0.3244, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.9437974683544303, | |
| "grad_norm": 0.5531424952387907, | |
| "learning_rate": 1.7199677100627427e-05, | |
| "loss": 0.3472, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.947848101265823, | |
| "grad_norm": 0.5490862444941942, | |
| "learning_rate": 1.7143488895093343e-05, | |
| "loss": 0.3433, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.951898734177215, | |
| "grad_norm": 0.5863567386403793, | |
| "learning_rate": 1.7087323695506994e-05, | |
| "loss": 0.3461, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.9559493670886074, | |
| "grad_norm": 0.5345202315302394, | |
| "learning_rate": 1.7031181954215194e-05, | |
| "loss": 0.3199, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.547544585838079, | |
| "learning_rate": 1.6975064123375788e-05, | |
| "loss": 0.3366, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.9640506329113925, | |
| "grad_norm": 0.5742226444949811, | |
| "learning_rate": 1.6918970654954084e-05, | |
| "loss": 0.3577, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.968101265822785, | |
| "grad_norm": 0.5144821209803621, | |
| "learning_rate": 1.686290200071915e-05, | |
| "loss": 0.3438, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.972151898734177, | |
| "grad_norm": 0.5226624688011834, | |
| "learning_rate": 1.6806858612240234e-05, | |
| "loss": 0.3448, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.9762025316455696, | |
| "grad_norm": 0.5326043984767954, | |
| "learning_rate": 1.6750840940883078e-05, | |
| "loss": 0.3456, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.980253164556962, | |
| "grad_norm": 0.5649206649747843, | |
| "learning_rate": 1.6694849437806305e-05, | |
| "loss": 0.3577, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.984303797468354, | |
| "grad_norm": 0.5234585367296575, | |
| "learning_rate": 1.663888455395778e-05, | |
| "loss": 0.3422, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.9883544303797467, | |
| "grad_norm": 0.530954644880927, | |
| "learning_rate": 1.6582946740070995e-05, | |
| "loss": 0.3418, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.992405063291139, | |
| "grad_norm": 0.5628050067905472, | |
| "learning_rate": 1.6527036446661396e-05, | |
| "loss": 0.3551, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.9964556962025317, | |
| "grad_norm": 0.53976504437554, | |
| "learning_rate": 1.6471154124022818e-05, | |
| "loss": 0.3337, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 3.002025316455696, | |
| "grad_norm": 0.5847804193946854, | |
| "learning_rate": 1.6415300222223788e-05, | |
| "loss": 0.2576, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.0060759493670886, | |
| "grad_norm": 0.6769023457107671, | |
| "learning_rate": 1.6359475191103958e-05, | |
| "loss": 0.1718, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 3.010126582278481, | |
| "grad_norm": 0.4815557366241773, | |
| "learning_rate": 1.6303679480270466e-05, | |
| "loss": 0.1685, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 3.0141772151898736, | |
| "grad_norm": 0.5142662333746068, | |
| "learning_rate": 1.624791353909428e-05, | |
| "loss": 0.1703, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 3.0182278481012657, | |
| "grad_norm": 0.6337258424418724, | |
| "learning_rate": 1.619217781670663e-05, | |
| "loss": 0.1592, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 3.022278481012658, | |
| "grad_norm": 0.7659616685923202, | |
| "learning_rate": 1.6136472761995373e-05, | |
| "loss": 0.1761, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.0263291139240507, | |
| "grad_norm": 0.5931030341868282, | |
| "learning_rate": 1.608079882360134e-05, | |
| "loss": 0.1867, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 3.030379746835443, | |
| "grad_norm": 0.5098320128588314, | |
| "learning_rate": 1.60251564499148e-05, | |
| "loss": 0.1732, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 3.0344303797468353, | |
| "grad_norm": 0.49831615090684345, | |
| "learning_rate": 1.596954608907176e-05, | |
| "loss": 0.1723, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 3.038481012658228, | |
| "grad_norm": 0.5319285282282624, | |
| "learning_rate": 1.591396818895043e-05, | |
| "loss": 0.1672, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 3.0425316455696203, | |
| "grad_norm": 0.5352087248011316, | |
| "learning_rate": 1.585842319716759e-05, | |
| "loss": 0.1585, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.046582278481013, | |
| "grad_norm": 0.4993731043051351, | |
| "learning_rate": 1.5802911561074944e-05, | |
| "loss": 0.159, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 3.050632911392405, | |
| "grad_norm": 0.478891871749591, | |
| "learning_rate": 1.5747433727755595e-05, | |
| "loss": 0.1578, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 3.0546835443037974, | |
| "grad_norm": 0.5006261240725142, | |
| "learning_rate": 1.5691990144020376e-05, | |
| "loss": 0.1722, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 3.05873417721519, | |
| "grad_norm": 0.5122931292499892, | |
| "learning_rate": 1.5636581256404297e-05, | |
| "loss": 0.1552, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 3.0627848101265824, | |
| "grad_norm": 0.5482685459181504, | |
| "learning_rate": 1.558120751116291e-05, | |
| "loss": 0.1682, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.0668354430379745, | |
| "grad_norm": 0.5257332372163963, | |
| "learning_rate": 1.552586935426876e-05, | |
| "loss": 0.1592, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 3.070886075949367, | |
| "grad_norm": 0.4794175915187234, | |
| "learning_rate": 1.547056723140774e-05, | |
| "loss": 0.1591, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 3.0749367088607595, | |
| "grad_norm": 0.46352692598192047, | |
| "learning_rate": 1.5415301587975565e-05, | |
| "loss": 0.1649, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 3.078987341772152, | |
| "grad_norm": 0.4450937349587019, | |
| "learning_rate": 1.536007286907411e-05, | |
| "loss": 0.1506, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 3.083037974683544, | |
| "grad_norm": 0.463376407204254, | |
| "learning_rate": 1.5304881519507896e-05, | |
| "loss": 0.1647, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.0870886075949366, | |
| "grad_norm": 0.4731052597439788, | |
| "learning_rate": 1.5249727983780453e-05, | |
| "loss": 0.1724, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 3.091139240506329, | |
| "grad_norm": 0.4438911919776663, | |
| "learning_rate": 1.5194612706090786e-05, | |
| "loss": 0.1597, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 3.0951898734177217, | |
| "grad_norm": 0.4510925879455199, | |
| "learning_rate": 1.5139536130329771e-05, | |
| "loss": 0.1691, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 3.0992405063291137, | |
| "grad_norm": 0.4622496922058588, | |
| "learning_rate": 1.508449870007656e-05, | |
| "loss": 0.1504, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 3.1032911392405063, | |
| "grad_norm": 0.4583637453229224, | |
| "learning_rate": 1.5029500858595056e-05, | |
| "loss": 0.1629, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 3.1073417721518988, | |
| "grad_norm": 0.4819116651080537, | |
| "learning_rate": 1.4974543048830328e-05, | |
| "loss": 0.164, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 3.1113924050632913, | |
| "grad_norm": 0.47787490060837584, | |
| "learning_rate": 1.4919625713405e-05, | |
| "loss": 0.1674, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 3.1154430379746834, | |
| "grad_norm": 0.4937259534748153, | |
| "learning_rate": 1.4864749294615756e-05, | |
| "loss": 0.1619, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 3.119493670886076, | |
| "grad_norm": 0.4685813081974994, | |
| "learning_rate": 1.4809914234429716e-05, | |
| "loss": 0.1705, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 3.1235443037974684, | |
| "grad_norm": 0.45158063702049184, | |
| "learning_rate": 1.4755120974480923e-05, | |
| "loss": 0.1528, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.127594936708861, | |
| "grad_norm": 0.44157223296247516, | |
| "learning_rate": 1.4700369956066771e-05, | |
| "loss": 0.1615, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 3.131645569620253, | |
| "grad_norm": 0.463625090492827, | |
| "learning_rate": 1.4645661620144413e-05, | |
| "loss": 0.1595, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 3.1356962025316455, | |
| "grad_norm": 0.44560821656242444, | |
| "learning_rate": 1.4590996407327284e-05, | |
| "loss": 0.1628, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 3.139746835443038, | |
| "grad_norm": 0.4459572608948688, | |
| "learning_rate": 1.4536374757881487e-05, | |
| "loss": 0.1657, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 3.1437974683544305, | |
| "grad_norm": 0.4525583360107206, | |
| "learning_rate": 1.4481797111722271e-05, | |
| "loss": 0.1748, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.1478481012658226, | |
| "grad_norm": 0.45081995907379724, | |
| "learning_rate": 1.4427263908410507e-05, | |
| "loss": 0.1508, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 3.151898734177215, | |
| "grad_norm": 0.4560223255332424, | |
| "learning_rate": 1.4372775587149108e-05, | |
| "loss": 0.153, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 3.1559493670886076, | |
| "grad_norm": 0.4537953625565736, | |
| "learning_rate": 1.4318332586779522e-05, | |
| "loss": 0.1576, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "grad_norm": 0.45152609607229194, | |
| "learning_rate": 1.4263935345778202e-05, | |
| "loss": 0.1716, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 3.164050632911392, | |
| "grad_norm": 0.4566701805458503, | |
| "learning_rate": 1.420958430225303e-05, | |
| "loss": 0.1806, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.1681012658227847, | |
| "grad_norm": 0.42280161497803354, | |
| "learning_rate": 1.415527989393985e-05, | |
| "loss": 0.1702, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 3.1721518987341772, | |
| "grad_norm": 0.4465064794575992, | |
| "learning_rate": 1.410102255819891e-05, | |
| "loss": 0.1771, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 3.1762025316455698, | |
| "grad_norm": 0.4533923868544294, | |
| "learning_rate": 1.404681273201131e-05, | |
| "loss": 0.1592, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 3.180253164556962, | |
| "grad_norm": 0.44834475694947634, | |
| "learning_rate": 1.399265085197556e-05, | |
| "loss": 0.1604, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 3.1843037974683543, | |
| "grad_norm": 0.44169254100226385, | |
| "learning_rate": 1.393853735430398e-05, | |
| "loss": 0.151, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 3.188354430379747, | |
| "grad_norm": 0.4429136981211431, | |
| "learning_rate": 1.3884472674819246e-05, | |
| "loss": 0.1598, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 3.1924050632911394, | |
| "grad_norm": 0.4537622958557298, | |
| "learning_rate": 1.3830457248950864e-05, | |
| "loss": 0.1674, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 3.1964556962025314, | |
| "grad_norm": 0.4392602320248754, | |
| "learning_rate": 1.377649151173163e-05, | |
| "loss": 0.1492, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 3.200506329113924, | |
| "grad_norm": 0.4519861638624284, | |
| "learning_rate": 1.3722575897794181e-05, | |
| "loss": 0.156, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 3.2045569620253165, | |
| "grad_norm": 0.4404570627647797, | |
| "learning_rate": 1.3668710841367472e-05, | |
| "loss": 0.153, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.208607594936709, | |
| "grad_norm": 0.4449085939097247, | |
| "learning_rate": 1.361489677627324e-05, | |
| "loss": 0.1588, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 3.212658227848101, | |
| "grad_norm": 0.46802793556727756, | |
| "learning_rate": 1.3561134135922585e-05, | |
| "loss": 0.1592, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 3.2167088607594936, | |
| "grad_norm": 0.4660090192202216, | |
| "learning_rate": 1.350742335331241e-05, | |
| "loss": 0.15, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 3.220759493670886, | |
| "grad_norm": 0.43813934965267787, | |
| "learning_rate": 1.345376486102198e-05, | |
| "loss": 0.1549, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 3.2248101265822786, | |
| "grad_norm": 0.45852902489434433, | |
| "learning_rate": 1.3400159091209414e-05, | |
| "loss": 0.1601, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 3.2288607594936707, | |
| "grad_norm": 0.4535360253493788, | |
| "learning_rate": 1.3346606475608216e-05, | |
| "loss": 0.1623, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 3.232911392405063, | |
| "grad_norm": 0.44476753867854385, | |
| "learning_rate": 1.3293107445523781e-05, | |
| "loss": 0.1636, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 3.2369620253164557, | |
| "grad_norm": 0.44128745340883685, | |
| "learning_rate": 1.3239662431829949e-05, | |
| "loss": 0.1445, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 3.2410126582278482, | |
| "grad_norm": 0.44250705184559647, | |
| "learning_rate": 1.3186271864965509e-05, | |
| "loss": 0.1568, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 3.2450632911392403, | |
| "grad_norm": 0.44681998372020953, | |
| "learning_rate": 1.3132936174930756e-05, | |
| "loss": 0.1668, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.249113924050633, | |
| "grad_norm": 0.44127899353075867, | |
| "learning_rate": 1.3079655791283995e-05, | |
| "loss": 0.1613, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 3.2531645569620253, | |
| "grad_norm": 0.45913332276250113, | |
| "learning_rate": 1.3026431143138108e-05, | |
| "loss": 0.1514, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 3.257215189873418, | |
| "grad_norm": 0.4563773795602485, | |
| "learning_rate": 1.2973262659157114e-05, | |
| "loss": 0.1588, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 3.26126582278481, | |
| "grad_norm": 0.4573168015112348, | |
| "learning_rate": 1.2920150767552651e-05, | |
| "loss": 0.1665, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 3.2653164556962024, | |
| "grad_norm": 0.4536973878346468, | |
| "learning_rate": 1.2867095896080607e-05, | |
| "loss": 0.1625, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.269367088607595, | |
| "grad_norm": 0.4515877322330883, | |
| "learning_rate": 1.2814098472037612e-05, | |
| "loss": 0.1537, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 3.2734177215189875, | |
| "grad_norm": 0.436392377244569, | |
| "learning_rate": 1.276115892225764e-05, | |
| "loss": 0.163, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 3.27746835443038, | |
| "grad_norm": 0.44019499193643585, | |
| "learning_rate": 1.2708277673108555e-05, | |
| "loss": 0.1502, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 3.281518987341772, | |
| "grad_norm": 0.4429201587164212, | |
| "learning_rate": 1.2655455150488649e-05, | |
| "loss": 0.1616, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 3.2855696202531646, | |
| "grad_norm": 0.44030666167224236, | |
| "learning_rate": 1.2602691779823272e-05, | |
| "loss": 0.1491, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.289620253164557, | |
| "grad_norm": 0.45171755651515266, | |
| "learning_rate": 1.2549987986061355e-05, | |
| "loss": 0.155, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 3.293670886075949, | |
| "grad_norm": 0.4461368224330212, | |
| "learning_rate": 1.2497344193672005e-05, | |
| "loss": 0.1653, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 3.2977215189873417, | |
| "grad_norm": 0.4353445936830784, | |
| "learning_rate": 1.2444760826641092e-05, | |
| "loss": 0.1598, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 3.301772151898734, | |
| "grad_norm": 0.4510636595143857, | |
| "learning_rate": 1.2392238308467817e-05, | |
| "loss": 0.1716, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 3.3058227848101267, | |
| "grad_norm": 0.4454124449375169, | |
| "learning_rate": 1.2339777062161326e-05, | |
| "loss": 0.1518, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.309873417721519, | |
| "grad_norm": 0.43534339109427517, | |
| "learning_rate": 1.2287377510237293e-05, | |
| "loss": 0.1451, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 3.3139240506329113, | |
| "grad_norm": 0.4574461866889834, | |
| "learning_rate": 1.2235040074714488e-05, | |
| "loss": 0.1501, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 3.317974683544304, | |
| "grad_norm": 0.4480741558971031, | |
| "learning_rate": 1.2182765177111434e-05, | |
| "loss": 0.148, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 3.3220253164556963, | |
| "grad_norm": 0.4431796447223421, | |
| "learning_rate": 1.213055323844297e-05, | |
| "loss": 0.1533, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 3.3260759493670884, | |
| "grad_norm": 0.42555162108709327, | |
| "learning_rate": 1.2078404679216864e-05, | |
| "loss": 0.1524, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.330126582278481, | |
| "grad_norm": 0.4473944841358878, | |
| "learning_rate": 1.2026319919430458e-05, | |
| "loss": 0.1483, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 3.3341772151898734, | |
| "grad_norm": 0.42718405421198363, | |
| "learning_rate": 1.1974299378567227e-05, | |
| "loss": 0.1607, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 3.338227848101266, | |
| "grad_norm": 0.4405439296586436, | |
| "learning_rate": 1.1922343475593462e-05, | |
| "loss": 0.1492, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 3.3422784810126585, | |
| "grad_norm": 0.4493405016141931, | |
| "learning_rate": 1.187045262895488e-05, | |
| "loss": 0.1591, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 3.3463291139240505, | |
| "grad_norm": 0.44057659703393176, | |
| "learning_rate": 1.1818627256573203e-05, | |
| "loss": 0.1416, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.350379746835443, | |
| "grad_norm": 0.45180310937777235, | |
| "learning_rate": 1.1766867775842864e-05, | |
| "loss": 0.1701, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 3.3544303797468356, | |
| "grad_norm": 0.4537018502252344, | |
| "learning_rate": 1.1715174603627615e-05, | |
| "loss": 0.16, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 3.3584810126582276, | |
| "grad_norm": 0.42371416808270407, | |
| "learning_rate": 1.1663548156257147e-05, | |
| "loss": 0.1658, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 3.36253164556962, | |
| "grad_norm": 0.44813538259331104, | |
| "learning_rate": 1.161198884952377e-05, | |
| "loss": 0.1686, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 3.3665822784810127, | |
| "grad_norm": 0.4362051021964037, | |
| "learning_rate": 1.1560497098679056e-05, | |
| "loss": 0.161, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.370632911392405, | |
| "grad_norm": 0.4375622240260483, | |
| "learning_rate": 1.1509073318430479e-05, | |
| "loss": 0.1454, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 3.3746835443037977, | |
| "grad_norm": 0.457975026675876, | |
| "learning_rate": 1.1457717922938116e-05, | |
| "loss": 0.1433, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 3.3787341772151898, | |
| "grad_norm": 0.456929558343823, | |
| "learning_rate": 1.1406431325811233e-05, | |
| "loss": 0.1527, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 3.3827848101265823, | |
| "grad_norm": 0.4338318456241126, | |
| "learning_rate": 1.135521394010506e-05, | |
| "loss": 0.1566, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 3.386835443037975, | |
| "grad_norm": 0.47344047952739304, | |
| "learning_rate": 1.1304066178317367e-05, | |
| "loss": 0.1658, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.390886075949367, | |
| "grad_norm": 0.456017062762802, | |
| "learning_rate": 1.1252988452385199e-05, | |
| "loss": 0.167, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 3.3949367088607594, | |
| "grad_norm": 0.42254802971299354, | |
| "learning_rate": 1.1201981173681536e-05, | |
| "loss": 0.1629, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 3.398987341772152, | |
| "grad_norm": 0.4474440965589927, | |
| "learning_rate": 1.1151044753011991e-05, | |
| "loss": 0.1632, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 3.4030379746835444, | |
| "grad_norm": 0.44409381723880714, | |
| "learning_rate": 1.1100179600611491e-05, | |
| "loss": 0.1454, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 3.407088607594937, | |
| "grad_norm": 0.43613728066141716, | |
| "learning_rate": 1.1049386126140985e-05, | |
| "loss": 0.1538, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.411139240506329, | |
| "grad_norm": 0.44498191821534167, | |
| "learning_rate": 1.0998664738684128e-05, | |
| "loss": 0.1644, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 3.4151898734177215, | |
| "grad_norm": 0.453533182053804, | |
| "learning_rate": 1.0948015846744e-05, | |
| "loss": 0.1544, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 3.419240506329114, | |
| "grad_norm": 0.4399501534323009, | |
| "learning_rate": 1.0897439858239832e-05, | |
| "loss": 0.1552, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 3.423291139240506, | |
| "grad_norm": 0.44251729632241865, | |
| "learning_rate": 1.0846937180503652e-05, | |
| "loss": 0.1427, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 3.4273417721518986, | |
| "grad_norm": 0.4402203145757387, | |
| "learning_rate": 1.0796508220277117e-05, | |
| "loss": 0.1582, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.431392405063291, | |
| "grad_norm": 0.4379982254144022, | |
| "learning_rate": 1.0746153383708107e-05, | |
| "loss": 0.149, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 3.4354430379746836, | |
| "grad_norm": 0.43062536632037696, | |
| "learning_rate": 1.0695873076347579e-05, | |
| "loss": 0.1568, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 3.439493670886076, | |
| "grad_norm": 0.43231441982718516, | |
| "learning_rate": 1.0645667703146205e-05, | |
| "loss": 0.1497, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 3.4435443037974682, | |
| "grad_norm": 0.45506533124776843, | |
| "learning_rate": 1.0595537668451161e-05, | |
| "loss": 0.1593, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 3.4475949367088607, | |
| "grad_norm": 0.4412892774259326, | |
| "learning_rate": 1.0545483376002854e-05, | |
| "loss": 0.15, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.4516455696202533, | |
| "grad_norm": 0.44542374185684536, | |
| "learning_rate": 1.0495505228931676e-05, | |
| "loss": 0.172, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 3.4556962025316453, | |
| "grad_norm": 0.4504164840633861, | |
| "learning_rate": 1.044560362975474e-05, | |
| "loss": 0.1532, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 3.459746835443038, | |
| "grad_norm": 0.4290301894383696, | |
| "learning_rate": 1.0395778980372695e-05, | |
| "loss": 0.1546, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 3.4637974683544304, | |
| "grad_norm": 0.42870453803544534, | |
| "learning_rate": 1.0346031682066381e-05, | |
| "loss": 0.1533, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 3.467848101265823, | |
| "grad_norm": 0.42316876639674944, | |
| "learning_rate": 1.0296362135493724e-05, | |
| "loss": 0.1596, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.4718987341772154, | |
| "grad_norm": 0.45002891733545264, | |
| "learning_rate": 1.0246770740686422e-05, | |
| "loss": 0.1555, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 3.4759493670886075, | |
| "grad_norm": 0.4302788584449941, | |
| "learning_rate": 1.0197257897046743e-05, | |
| "loss": 0.1453, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "grad_norm": 0.4482058223136038, | |
| "learning_rate": 1.014782400334433e-05, | |
| "loss": 0.1605, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 3.4840506329113925, | |
| "grad_norm": 0.44170214879019404, | |
| "learning_rate": 1.009846945771296e-05, | |
| "loss": 0.1677, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 3.4881012658227846, | |
| "grad_norm": 0.439402436054417, | |
| "learning_rate": 1.0049194657647363e-05, | |
| "loss": 0.1555, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.492151898734177, | |
| "grad_norm": 0.4466354820211173, | |
| "learning_rate": 1.0000000000000006e-05, | |
| "loss": 0.1665, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 3.4962025316455696, | |
| "grad_norm": 0.4423161080035427, | |
| "learning_rate": 9.950885880977891e-06, | |
| "loss": 0.1541, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 3.500253164556962, | |
| "grad_norm": 0.44411615635056173, | |
| "learning_rate": 9.901852696139382e-06, | |
| "loss": 0.1474, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 3.5043037974683546, | |
| "grad_norm": 0.4354894345853899, | |
| "learning_rate": 9.852900840391027e-06, | |
| "loss": 0.1419, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 3.5083544303797467, | |
| "grad_norm": 0.4336455710246167, | |
| "learning_rate": 9.804030707984313e-06, | |
| "loss": 0.1505, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.512405063291139, | |
| "grad_norm": 0.46038009357380405, | |
| "learning_rate": 9.755242692512599e-06, | |
| "loss": 0.1574, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 3.5164556962025317, | |
| "grad_norm": 0.45707512193642275, | |
| "learning_rate": 9.70653718690782e-06, | |
| "loss": 0.1523, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 3.520506329113924, | |
| "grad_norm": 0.43190944999175385, | |
| "learning_rate": 9.657914583437454e-06, | |
| "loss": 0.1603, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 3.5245569620253163, | |
| "grad_norm": 0.4392656551577909, | |
| "learning_rate": 9.609375273701246e-06, | |
| "loss": 0.1527, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 3.528607594936709, | |
| "grad_norm": 0.45615339118797144, | |
| "learning_rate": 9.560919648628133e-06, | |
| "loss": 0.1449, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.5326582278481014, | |
| "grad_norm": 0.4451095635327418, | |
| "learning_rate": 9.512548098473047e-06, | |
| "loss": 0.1568, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 3.536708860759494, | |
| "grad_norm": 0.44729230318958957, | |
| "learning_rate": 9.464261012813825e-06, | |
| "loss": 0.1616, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 3.540759493670886, | |
| "grad_norm": 0.4402770129438123, | |
| "learning_rate": 9.416058780547987e-06, | |
| "loss": 0.1541, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 3.5448101265822785, | |
| "grad_norm": 0.4378866533122809, | |
| "learning_rate": 9.367941789889714e-06, | |
| "loss": 0.1705, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 3.548860759493671, | |
| "grad_norm": 0.4197980778931241, | |
| "learning_rate": 9.319910428366607e-06, | |
| "loss": 0.1666, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.552911392405063, | |
| "grad_norm": 0.42692251086173044, | |
| "learning_rate": 9.271965082816667e-06, | |
| "loss": 0.1493, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 3.5569620253164556, | |
| "grad_norm": 0.4479814663326769, | |
| "learning_rate": 9.224106139385111e-06, | |
| "loss": 0.1542, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 3.561012658227848, | |
| "grad_norm": 0.4449911231238358, | |
| "learning_rate": 9.176333983521291e-06, | |
| "loss": 0.157, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 3.5650632911392406, | |
| "grad_norm": 0.43414304008591276, | |
| "learning_rate": 9.12864899997558e-06, | |
| "loss": 0.1504, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 3.569113924050633, | |
| "grad_norm": 0.4432404873915249, | |
| "learning_rate": 9.08105157279628e-06, | |
| "loss": 0.1677, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.573164556962025, | |
| "grad_norm": 0.43507677084218077, | |
| "learning_rate": 9.03354208532653e-06, | |
| "loss": 0.1473, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 3.5772151898734177, | |
| "grad_norm": 0.44865148931498466, | |
| "learning_rate": 8.986120920201205e-06, | |
| "loss": 0.1617, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 3.58126582278481, | |
| "grad_norm": 0.4329467851573989, | |
| "learning_rate": 8.938788459343852e-06, | |
| "loss": 0.1456, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 3.5853164556962023, | |
| "grad_norm": 0.441763619658015, | |
| "learning_rate": 8.8915450839636e-06, | |
| "loss": 0.1564, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 3.589367088607595, | |
| "grad_norm": 0.44563967209463656, | |
| "learning_rate": 8.844391174552116e-06, | |
| "loss": 0.1505, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.5934177215189873, | |
| "grad_norm": 0.43628151150807276, | |
| "learning_rate": 8.797327110880479e-06, | |
| "loss": 0.1386, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 3.59746835443038, | |
| "grad_norm": 0.4412878621111705, | |
| "learning_rate": 8.750353271996206e-06, | |
| "loss": 0.1536, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 3.6015189873417723, | |
| "grad_norm": 0.42334622383455506, | |
| "learning_rate": 8.703470036220132e-06, | |
| "loss": 0.1537, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 3.6055696202531644, | |
| "grad_norm": 0.44599099239710205, | |
| "learning_rate": 8.656677781143394e-06, | |
| "loss": 0.1395, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 3.609620253164557, | |
| "grad_norm": 0.4374371311322269, | |
| "learning_rate": 8.609976883624377e-06, | |
| "loss": 0.1414, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.6136708860759494, | |
| "grad_norm": 0.42812694670712226, | |
| "learning_rate": 8.563367719785698e-06, | |
| "loss": 0.1513, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 3.6177215189873415, | |
| "grad_norm": 0.44002139070765367, | |
| "learning_rate": 8.516850665011138e-06, | |
| "loss": 0.1443, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 3.621772151898734, | |
| "grad_norm": 0.4355268148485625, | |
| "learning_rate": 8.47042609394269e-06, | |
| "loss": 0.1438, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 3.6258227848101265, | |
| "grad_norm": 0.4369638750762777, | |
| "learning_rate": 8.424094380477432e-06, | |
| "loss": 0.155, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 3.629873417721519, | |
| "grad_norm": 0.4347005512338693, | |
| "learning_rate": 8.37785589776465e-06, | |
| "loss": 0.155, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.6339240506329116, | |
| "grad_norm": 0.4310939949322397, | |
| "learning_rate": 8.331711018202694e-06, | |
| "loss": 0.1541, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 3.6379746835443036, | |
| "grad_norm": 0.44913269222031643, | |
| "learning_rate": 8.285660113436104e-06, | |
| "loss": 0.1572, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 3.642025316455696, | |
| "grad_norm": 0.4376141268068831, | |
| "learning_rate": 8.239703554352527e-06, | |
| "loss": 0.1538, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 3.6460759493670887, | |
| "grad_norm": 0.45371119773109514, | |
| "learning_rate": 8.193841711079775e-06, | |
| "loss": 0.1728, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 3.6501265822784807, | |
| "grad_norm": 0.43003064670094393, | |
| "learning_rate": 8.148074952982828e-06, | |
| "loss": 0.1601, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.6541772151898733, | |
| "grad_norm": 0.43581058685796953, | |
| "learning_rate": 8.102403648660859e-06, | |
| "loss": 0.154, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 3.6582278481012658, | |
| "grad_norm": 0.4360207368088478, | |
| "learning_rate": 8.056828165944282e-06, | |
| "loss": 0.15, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 3.6622784810126583, | |
| "grad_norm": 0.4373879693451259, | |
| "learning_rate": 8.011348871891762e-06, | |
| "loss": 0.1593, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 3.666329113924051, | |
| "grad_norm": 0.44544020740980966, | |
| "learning_rate": 7.965966132787287e-06, | |
| "loss": 0.1703, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 3.670379746835443, | |
| "grad_norm": 0.4422857733204582, | |
| "learning_rate": 7.920680314137189e-06, | |
| "loss": 0.1545, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.6744303797468354, | |
| "grad_norm": 0.4430018978016549, | |
| "learning_rate": 7.875491780667246e-06, | |
| "loss": 0.1547, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 3.678481012658228, | |
| "grad_norm": 0.42737786779194387, | |
| "learning_rate": 7.830400896319667e-06, | |
| "loss": 0.1594, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 3.68253164556962, | |
| "grad_norm": 0.44507288735115946, | |
| "learning_rate": 7.785408024250259e-06, | |
| "loss": 0.1421, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 3.6865822784810125, | |
| "grad_norm": 0.4246235471772369, | |
| "learning_rate": 7.74051352682542e-06, | |
| "loss": 0.1493, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 3.690632911392405, | |
| "grad_norm": 0.4366272391576363, | |
| "learning_rate": 7.695717765619257e-06, | |
| "loss": 0.137, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.6946835443037975, | |
| "grad_norm": 0.4201375038897027, | |
| "learning_rate": 7.651021101410673e-06, | |
| "loss": 0.1512, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 3.69873417721519, | |
| "grad_norm": 0.4319281563237493, | |
| "learning_rate": 7.606423894180464e-06, | |
| "loss": 0.158, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 3.702784810126582, | |
| "grad_norm": 0.44796209679220006, | |
| "learning_rate": 7.56192650310839e-06, | |
| "loss": 0.1456, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 3.7068354430379746, | |
| "grad_norm": 0.4300876007431019, | |
| "learning_rate": 7.517529286570349e-06, | |
| "loss": 0.13, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 3.710886075949367, | |
| "grad_norm": 0.4217721183381077, | |
| "learning_rate": 7.473232602135387e-06, | |
| "loss": 0.1572, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.714936708860759, | |
| "grad_norm": 0.4233410818626887, | |
| "learning_rate": 7.429036806562935e-06, | |
| "loss": 0.1547, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 3.7189873417721517, | |
| "grad_norm": 0.42989342510630996, | |
| "learning_rate": 7.3849422557998455e-06, | |
| "loss": 0.1573, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 3.7230379746835442, | |
| "grad_norm": 0.4434952925014625, | |
| "learning_rate": 7.340949304977567e-06, | |
| "loss": 0.1538, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 3.7270886075949368, | |
| "grad_norm": 0.4293484755794158, | |
| "learning_rate": 7.297058308409282e-06, | |
| "loss": 0.1559, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 3.7311392405063293, | |
| "grad_norm": 0.4251047148309482, | |
| "learning_rate": 7.25326961958704e-06, | |
| "loss": 0.1578, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.7351898734177214, | |
| "grad_norm": 0.40982864474378106, | |
| "learning_rate": 7.209583591178921e-06, | |
| "loss": 0.1609, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 3.739240506329114, | |
| "grad_norm": 0.4317556498537363, | |
| "learning_rate": 7.1660005750261925e-06, | |
| "loss": 0.1629, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 3.7432911392405064, | |
| "grad_norm": 0.42699362271799274, | |
| "learning_rate": 7.1225209221404765e-06, | |
| "loss": 0.1558, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 3.747341772151899, | |
| "grad_norm": 0.42425475084913084, | |
| "learning_rate": 7.079144982700909e-06, | |
| "loss": 0.1488, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 3.7513924050632914, | |
| "grad_norm": 0.42860474380746616, | |
| "learning_rate": 7.0358731060513695e-06, | |
| "loss": 0.1665, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.7554430379746835, | |
| "grad_norm": 0.41952411773133474, | |
| "learning_rate": 6.99270564069757e-06, | |
| "loss": 0.1605, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 3.759493670886076, | |
| "grad_norm": 0.4272122326212073, | |
| "learning_rate": 6.949642934304375e-06, | |
| "loss": 0.1514, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 3.7635443037974685, | |
| "grad_norm": 0.4314611462253272, | |
| "learning_rate": 6.906685333692871e-06, | |
| "loss": 0.1418, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 3.7675949367088606, | |
| "grad_norm": 0.42087599346794075, | |
| "learning_rate": 6.86383318483769e-06, | |
| "loss": 0.1277, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 3.771645569620253, | |
| "grad_norm": 0.4377001157071647, | |
| "learning_rate": 6.821086832864139e-06, | |
| "loss": 0.1512, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.7756962025316456, | |
| "grad_norm": 0.427400433838755, | |
| "learning_rate": 6.77844662204546e-06, | |
| "loss": 0.1624, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 3.779746835443038, | |
| "grad_norm": 0.430719517775185, | |
| "learning_rate": 6.7359128958000455e-06, | |
| "loss": 0.1404, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 3.7837974683544306, | |
| "grad_norm": 0.43540652152753395, | |
| "learning_rate": 6.693485996688695e-06, | |
| "loss": 0.1565, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 3.7878481012658227, | |
| "grad_norm": 0.42616404925238055, | |
| "learning_rate": 6.651166266411801e-06, | |
| "loss": 0.1591, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 3.7918987341772152, | |
| "grad_norm": 0.4095366344506748, | |
| "learning_rate": 6.6089540458066725e-06, | |
| "loss": 0.1358, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.7959493670886078, | |
| "grad_norm": 0.4339510714980313, | |
| "learning_rate": 6.566849674844711e-06, | |
| "loss": 0.1419, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "grad_norm": 0.4199037876336847, | |
| "learning_rate": 6.524853492628747e-06, | |
| "loss": 0.1474, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 3.8040506329113923, | |
| "grad_norm": 0.4338643260075778, | |
| "learning_rate": 6.4829658373902536e-06, | |
| "loss": 0.1615, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 3.808101265822785, | |
| "grad_norm": 0.4269282901860343, | |
| "learning_rate": 6.441187046486648e-06, | |
| "loss": 0.151, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 3.8121518987341774, | |
| "grad_norm": 0.4331223653043475, | |
| "learning_rate": 6.399517456398567e-06, | |
| "loss": 0.1469, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.81620253164557, | |
| "grad_norm": 0.4226811282483295, | |
| "learning_rate": 6.357957402727164e-06, | |
| "loss": 0.1497, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 3.820253164556962, | |
| "grad_norm": 0.4205105135911889, | |
| "learning_rate": 6.316507220191395e-06, | |
| "loss": 0.1453, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 3.8243037974683545, | |
| "grad_norm": 0.4293972974140739, | |
| "learning_rate": 6.275167242625331e-06, | |
| "loss": 0.1543, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 3.828354430379747, | |
| "grad_norm": 0.4191966421079241, | |
| "learning_rate": 6.233937802975471e-06, | |
| "loss": 0.1525, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 3.832405063291139, | |
| "grad_norm": 0.4404644484085003, | |
| "learning_rate": 6.192819233298046e-06, | |
| "loss": 0.1481, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.8364556962025316, | |
| "grad_norm": 0.4267883288591818, | |
| "learning_rate": 6.151811864756383e-06, | |
| "loss": 0.1401, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 3.840506329113924, | |
| "grad_norm": 0.4373488406840621, | |
| "learning_rate": 6.1109160276181655e-06, | |
| "loss": 0.1472, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 3.8445569620253166, | |
| "grad_norm": 0.42993679136470114, | |
| "learning_rate": 6.070132051252868e-06, | |
| "loss": 0.1614, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 3.848607594936709, | |
| "grad_norm": 0.42373070452972383, | |
| "learning_rate": 6.0294602641290034e-06, | |
| "loss": 0.1658, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 3.852658227848101, | |
| "grad_norm": 0.44097910959977865, | |
| "learning_rate": 5.988900993811575e-06, | |
| "loss": 0.1487, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.8567088607594937, | |
| "grad_norm": 0.4249111274913779, | |
| "learning_rate": 5.948454566959363e-06, | |
| "loss": 0.1555, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 3.8607594936708862, | |
| "grad_norm": 0.4305432661554834, | |
| "learning_rate": 5.908121309322328e-06, | |
| "loss": 0.1465, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 3.8648101265822783, | |
| "grad_norm": 0.4421125393453393, | |
| "learning_rate": 5.867901545738976e-06, | |
| "loss": 0.1526, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 3.868860759493671, | |
| "grad_norm": 0.4426786543828732, | |
| "learning_rate": 5.827795600133774e-06, | |
| "loss": 0.1545, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 3.8729113924050633, | |
| "grad_norm": 0.44664975006533564, | |
| "learning_rate": 5.787803795514466e-06, | |
| "loss": 0.1534, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.876962025316456, | |
| "grad_norm": 0.421371828603597, | |
| "learning_rate": 5.747926453969576e-06, | |
| "loss": 0.1526, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 3.8810126582278484, | |
| "grad_norm": 0.43428283279387936, | |
| "learning_rate": 5.708163896665708e-06, | |
| "loss": 0.1416, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 3.8850632911392404, | |
| "grad_norm": 0.4245311027722247, | |
| "learning_rate": 5.668516443845047e-06, | |
| "loss": 0.1506, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 3.889113924050633, | |
| "grad_norm": 0.426741913627117, | |
| "learning_rate": 5.6289844148227225e-06, | |
| "loss": 0.1455, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 3.8931645569620255, | |
| "grad_norm": 0.4215239190107141, | |
| "learning_rate": 5.5895681279842615e-06, | |
| "loss": 0.1483, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.8972151898734175, | |
| "grad_norm": 0.41182252190395574, | |
| "learning_rate": 5.550267900783019e-06, | |
| "loss": 0.1519, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 3.90126582278481, | |
| "grad_norm": 0.42404033491318216, | |
| "learning_rate": 5.511084049737623e-06, | |
| "loss": 0.1332, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 3.9053164556962026, | |
| "grad_norm": 0.43270946975940455, | |
| "learning_rate": 5.4720168904294215e-06, | |
| "loss": 0.1558, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 3.909367088607595, | |
| "grad_norm": 0.42536161016158247, | |
| "learning_rate": 5.433066737499948e-06, | |
| "loss": 0.153, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 3.9134177215189876, | |
| "grad_norm": 0.43478578246558924, | |
| "learning_rate": 5.394233904648376e-06, | |
| "loss": 0.1503, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.9174683544303797, | |
| "grad_norm": 0.4407149716577981, | |
| "learning_rate": 5.355518704628997e-06, | |
| "loss": 0.1501, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 3.921518987341772, | |
| "grad_norm": 0.4340274670907567, | |
| "learning_rate": 5.316921449248731e-06, | |
| "loss": 0.1657, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 3.9255696202531647, | |
| "grad_norm": 0.42729091444096146, | |
| "learning_rate": 5.278442449364538e-06, | |
| "loss": 0.1414, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 3.9296202531645568, | |
| "grad_norm": 0.41721635639833377, | |
| "learning_rate": 5.240082014881016e-06, | |
| "loss": 0.1393, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 3.9336708860759493, | |
| "grad_norm": 0.42734889060278536, | |
| "learning_rate": 5.201840454747822e-06, | |
| "loss": 0.149, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.937721518987342, | |
| "grad_norm": 0.4185327430760438, | |
| "learning_rate": 5.163718076957223e-06, | |
| "loss": 0.1486, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 3.9417721518987343, | |
| "grad_norm": 0.421886699470847, | |
| "learning_rate": 5.125715188541609e-06, | |
| "loss": 0.1492, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 3.945822784810127, | |
| "grad_norm": 0.43399332931473195, | |
| "learning_rate": 5.087832095571021e-06, | |
| "loss": 0.1467, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 3.949873417721519, | |
| "grad_norm": 0.4224023549637997, | |
| "learning_rate": 5.0500691031506766e-06, | |
| "loss": 0.1435, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.9539240506329114, | |
| "grad_norm": 0.42791930928782695, | |
| "learning_rate": 5.01242651541854e-06, | |
| "loss": 0.1517, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.957974683544304, | |
| "grad_norm": 0.40988577186402325, | |
| "learning_rate": 4.974904635542815e-06, | |
| "loss": 0.1394, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.962025316455696, | |
| "grad_norm": 0.4173357712680607, | |
| "learning_rate": 4.937503765719582e-06, | |
| "loss": 0.1548, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 3.9660759493670885, | |
| "grad_norm": 0.4141499326259297, | |
| "learning_rate": 4.900224207170299e-06, | |
| "loss": 0.1537, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.970126582278481, | |
| "grad_norm": 0.4405861843729818, | |
| "learning_rate": 4.8630662601394065e-06, | |
| "loss": 0.1594, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 3.9741772151898735, | |
| "grad_norm": 0.42543900649093713, | |
| "learning_rate": 4.8260302238918995e-06, | |
| "loss": 0.1477, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.978227848101266, | |
| "grad_norm": 0.40546479126951673, | |
| "learning_rate": 4.789116396710924e-06, | |
| "loss": 0.1281, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 3.982278481012658, | |
| "grad_norm": 0.4246317900358979, | |
| "learning_rate": 4.752325075895368e-06, | |
| "loss": 0.1465, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.9863291139240506, | |
| "grad_norm": 0.4211993854860544, | |
| "learning_rate": 4.715656557757473e-06, | |
| "loss": 0.1479, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 3.990379746835443, | |
| "grad_norm": 0.4330421027940532, | |
| "learning_rate": 4.679111137620442e-06, | |
| "loss": 0.1623, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.9944303797468352, | |
| "grad_norm": 0.43328597286016673, | |
| "learning_rate": 4.6426891098160585e-06, | |
| "loss": 0.1552, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.9984810126582278, | |
| "grad_norm": 0.42420108207079477, | |
| "learning_rate": 4.6063907676823474e-06, | |
| "loss": 0.1567, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 4.004050632911392, | |
| "grad_norm": 0.4642433700883086, | |
| "learning_rate": 4.570216403561141e-06, | |
| "loss": 0.0908, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 4.008101265822785, | |
| "grad_norm": 0.41262944340156893, | |
| "learning_rate": 4.534166308795815e-06, | |
| "loss": 0.0909, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 4.012151898734177, | |
| "grad_norm": 0.389634278286882, | |
| "learning_rate": 4.498240773728859e-06, | |
| "loss": 0.0796, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 4.01620253164557, | |
| "grad_norm": 0.33925685553755414, | |
| "learning_rate": 4.462440087699609e-06, | |
| "loss": 0.0801, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.020253164556962, | |
| "grad_norm": 0.31546609581025, | |
| "learning_rate": 4.426764539041861e-06, | |
| "loss": 0.0738, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 4.024303797468354, | |
| "grad_norm": 0.3038721813119663, | |
| "learning_rate": 4.391214415081582e-06, | |
| "loss": 0.0781, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 4.028354430379747, | |
| "grad_norm": 0.3012574468933216, | |
| "learning_rate": 4.355790002134579e-06, | |
| "loss": 0.0854, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 4.032405063291139, | |
| "grad_norm": 0.3212704613861633, | |
| "learning_rate": 4.320491585504207e-06, | |
| "loss": 0.0747, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 4.036455696202531, | |
| "grad_norm": 0.372231479493523, | |
| "learning_rate": 4.2853194494790615e-06, | |
| "loss": 0.084, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 4.040506329113924, | |
| "grad_norm": 0.3666072671294429, | |
| "learning_rate": 4.250273877330691e-06, | |
| "loss": 0.0656, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 4.044556962025316, | |
| "grad_norm": 0.42517683419513375, | |
| "learning_rate": 4.215355151311313e-06, | |
| "loss": 0.0823, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 4.048607594936709, | |
| "grad_norm": 0.41871538504933953, | |
| "learning_rate": 4.180563552651542e-06, | |
| "loss": 0.0832, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 4.052658227848101, | |
| "grad_norm": 0.38692380535674425, | |
| "learning_rate": 4.145899361558147e-06, | |
| "loss": 0.0806, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 4.056708860759493, | |
| "grad_norm": 0.32236931612536796, | |
| "learning_rate": 4.111362857211738e-06, | |
| "loss": 0.0683, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.060759493670886, | |
| "grad_norm": 0.3124303136516379, | |
| "learning_rate": 4.076954317764592e-06, | |
| "loss": 0.0742, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 4.0648101265822785, | |
| "grad_norm": 0.3132852623378327, | |
| "learning_rate": 4.042674020338335e-06, | |
| "loss": 0.0831, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 4.0688607594936705, | |
| "grad_norm": 0.3261361647187441, | |
| "learning_rate": 4.0085222410217835e-06, | |
| "loss": 0.0749, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 4.0729113924050635, | |
| "grad_norm": 0.2919862422390822, | |
| "learning_rate": 3.974499254868674e-06, | |
| "loss": 0.0736, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 4.076962025316456, | |
| "grad_norm": 0.3053964801817635, | |
| "learning_rate": 3.940605335895451e-06, | |
| "loss": 0.0728, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 4.0810126582278485, | |
| "grad_norm": 0.3003232197089812, | |
| "learning_rate": 3.90684075707908e-06, | |
| "loss": 0.0727, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 4.085063291139241, | |
| "grad_norm": 0.3078801888363416, | |
| "learning_rate": 3.8732057903548505e-06, | |
| "loss": 0.0826, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 4.089113924050633, | |
| "grad_norm": 0.303812493378106, | |
| "learning_rate": 3.8397007066141375e-06, | |
| "loss": 0.0796, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 4.093164556962026, | |
| "grad_norm": 0.3048765256126637, | |
| "learning_rate": 3.806325775702304e-06, | |
| "loss": 0.0758, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 4.097215189873418, | |
| "grad_norm": 0.29955177472215894, | |
| "learning_rate": 3.773081266416434e-06, | |
| "loss": 0.0709, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.10126582278481, | |
| "grad_norm": 0.283857936568104, | |
| "learning_rate": 3.739967446503245e-06, | |
| "loss": 0.0741, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 4.105316455696203, | |
| "grad_norm": 0.28866952709192645, | |
| "learning_rate": 3.706984582656894e-06, | |
| "loss": 0.0666, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 4.109367088607595, | |
| "grad_norm": 0.2872551805111798, | |
| "learning_rate": 3.6741329405168237e-06, | |
| "loss": 0.077, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 4.113417721518988, | |
| "grad_norm": 0.29470699662365984, | |
| "learning_rate": 3.641412784665648e-06, | |
| "loss": 0.0689, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 4.11746835443038, | |
| "grad_norm": 0.29246075317094705, | |
| "learning_rate": 3.608824378627005e-06, | |
| "loss": 0.0671, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 4.121518987341772, | |
| "grad_norm": 0.3091406881236883, | |
| "learning_rate": 3.5763679848634337e-06, | |
| "loss": 0.0738, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 4.125569620253165, | |
| "grad_norm": 0.2988245487952412, | |
| "learning_rate": 3.544043864774269e-06, | |
| "loss": 0.0642, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 4.129620253164557, | |
| "grad_norm": 0.31975316576432367, | |
| "learning_rate": 3.5118522786935282e-06, | |
| "loss": 0.0685, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 4.133670886075949, | |
| "grad_norm": 0.3212369552190634, | |
| "learning_rate": 3.479793485887819e-06, | |
| "loss": 0.0758, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 4.137721518987342, | |
| "grad_norm": 0.3057645173238477, | |
| "learning_rate": 3.4478677445542653e-06, | |
| "loss": 0.0769, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.141772151898734, | |
| "grad_norm": 0.31197840424107165, | |
| "learning_rate": 3.4160753118183767e-06, | |
| "loss": 0.0812, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 4.145822784810127, | |
| "grad_norm": 0.3232232978183221, | |
| "learning_rate": 3.3844164437320527e-06, | |
| "loss": 0.087, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 4.149873417721519, | |
| "grad_norm": 0.3025109889669587, | |
| "learning_rate": 3.3528913952714558e-06, | |
| "loss": 0.0716, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 4.153924050632911, | |
| "grad_norm": 0.2945653778024827, | |
| "learning_rate": 3.321500420335e-06, | |
| "loss": 0.0741, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 4.157974683544304, | |
| "grad_norm": 0.30220689275609564, | |
| "learning_rate": 3.290243771741275e-06, | |
| "loss": 0.0806, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.162025316455696, | |
| "grad_norm": 0.292593000527882, | |
| "learning_rate": 3.2591217012270325e-06, | |
| "loss": 0.078, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 4.166075949367088, | |
| "grad_norm": 0.27805915297765754, | |
| "learning_rate": 3.228134459445149e-06, | |
| "loss": 0.0808, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 4.170126582278481, | |
| "grad_norm": 0.29577892727312977, | |
| "learning_rate": 3.1972822959626205e-06, | |
| "loss": 0.0898, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 4.174177215189873, | |
| "grad_norm": 0.2963290077595325, | |
| "learning_rate": 3.166565459258513e-06, | |
| "loss": 0.0797, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 4.178227848101266, | |
| "grad_norm": 0.2848572613159732, | |
| "learning_rate": 3.1359841967220193e-06, | |
| "loss": 0.0788, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.182278481012658, | |
| "grad_norm": 0.2999958061994636, | |
| "learning_rate": 3.105538754650419e-06, | |
| "loss": 0.0727, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 4.18632911392405, | |
| "grad_norm": 0.2946904272863437, | |
| "learning_rate": 3.07522937824712e-06, | |
| "loss": 0.0837, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 4.190379746835443, | |
| "grad_norm": 0.5297053530921748, | |
| "learning_rate": 3.0450563116196697e-06, | |
| "loss": 0.0897, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 4.194430379746835, | |
| "grad_norm": 0.29043756820794775, | |
| "learning_rate": 3.0150197977778008e-06, | |
| "loss": 0.0696, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 4.1984810126582275, | |
| "grad_norm": 0.27905242410319486, | |
| "learning_rate": 2.985120078631465e-06, | |
| "loss": 0.0633, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 4.2025316455696204, | |
| "grad_norm": 0.2947499911336863, | |
| "learning_rate": 2.9553573949888893e-06, | |
| "loss": 0.078, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 4.2065822784810125, | |
| "grad_norm": 0.29038656974461907, | |
| "learning_rate": 2.9257319865546384e-06, | |
| "loss": 0.073, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 4.2106329113924055, | |
| "grad_norm": 0.2986174007514029, | |
| "learning_rate": 2.896244091927678e-06, | |
| "loss": 0.0713, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 4.2146835443037975, | |
| "grad_norm": 0.28899435444680216, | |
| "learning_rate": 2.8668939485994584e-06, | |
| "loss": 0.0734, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 4.21873417721519, | |
| "grad_norm": 0.2986515340984011, | |
| "learning_rate": 2.837681792951994e-06, | |
| "loss": 0.0816, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.222784810126583, | |
| "grad_norm": 0.2994570034200759, | |
| "learning_rate": 2.808607860255981e-06, | |
| "loss": 0.0782, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 4.226835443037975, | |
| "grad_norm": 0.2972677942281225, | |
| "learning_rate": 2.7796723846688634e-06, | |
| "loss": 0.0811, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 4.230886075949367, | |
| "grad_norm": 0.29365466621682196, | |
| "learning_rate": 2.7508755992329937e-06, | |
| "loss": 0.0714, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 4.23493670886076, | |
| "grad_norm": 0.30094482378410514, | |
| "learning_rate": 2.722217735873718e-06, | |
| "loss": 0.0845, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 4.238987341772152, | |
| "grad_norm": 0.3066784075758663, | |
| "learning_rate": 2.6936990253975315e-06, | |
| "loss": 0.0718, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 4.243037974683545, | |
| "grad_norm": 0.30618183442325825, | |
| "learning_rate": 2.665319697490205e-06, | |
| "loss": 0.0753, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 4.247088607594937, | |
| "grad_norm": 0.30243677074947223, | |
| "learning_rate": 2.637079980714945e-06, | |
| "loss": 0.074, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 4.251139240506329, | |
| "grad_norm": 0.28303064239904635, | |
| "learning_rate": 2.6089801025105453e-06, | |
| "loss": 0.0699, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 4.255189873417722, | |
| "grad_norm": 0.2872966557900058, | |
| "learning_rate": 2.581020289189571e-06, | |
| "loss": 0.0647, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 4.259240506329114, | |
| "grad_norm": 0.2989269892512023, | |
| "learning_rate": 2.553200765936501e-06, | |
| "loss": 0.091, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.263291139240506, | |
| "grad_norm": 0.29028679829309406, | |
| "learning_rate": 2.525521756805962e-06, | |
| "loss": 0.076, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 4.267341772151899, | |
| "grad_norm": 0.2854836243477457, | |
| "learning_rate": 2.497983484720885e-06, | |
| "loss": 0.0687, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 4.271392405063291, | |
| "grad_norm": 0.30686369337496, | |
| "learning_rate": 2.470586171470728e-06, | |
| "loss": 0.0724, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 4.275443037974684, | |
| "grad_norm": 0.2936816093799084, | |
| "learning_rate": 2.4433300377096836e-06, | |
| "loss": 0.0816, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 4.279493670886076, | |
| "grad_norm": 0.2890320903965297, | |
| "learning_rate": 2.4162153029549073e-06, | |
| "loss": 0.0746, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 4.283544303797468, | |
| "grad_norm": 0.28369616316702145, | |
| "learning_rate": 2.3892421855847458e-06, | |
| "loss": 0.0694, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 4.287594936708861, | |
| "grad_norm": 0.29332211999846614, | |
| "learning_rate": 2.362410902836978e-06, | |
| "loss": 0.0816, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 4.291645569620253, | |
| "grad_norm": 0.284480041633391, | |
| "learning_rate": 2.3357216708070653e-06, | |
| "loss": 0.0592, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 4.295696202531645, | |
| "grad_norm": 0.29916295499723305, | |
| "learning_rate": 2.309174704446411e-06, | |
| "loss": 0.0882, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 4.299746835443038, | |
| "grad_norm": 0.28592740504578534, | |
| "learning_rate": 2.2827702175606437e-06, | |
| "loss": 0.0727, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.30379746835443, | |
| "grad_norm": 0.29137375474814653, | |
| "learning_rate": 2.256508422807855e-06, | |
| "loss": 0.0752, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 4.307848101265823, | |
| "grad_norm": 0.29607175807407604, | |
| "learning_rate": 2.230389531696946e-06, | |
| "loss": 0.0653, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 4.311898734177215, | |
| "grad_norm": 0.2941896281185634, | |
| "learning_rate": 2.204413754585857e-06, | |
| "loss": 0.0769, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 4.315949367088607, | |
| "grad_norm": 0.30070152997937827, | |
| "learning_rate": 2.1785813006799406e-06, | |
| "loss": 0.0761, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "grad_norm": 0.289978744608868, | |
| "learning_rate": 2.1528923780302224e-06, | |
| "loss": 0.0814, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 4.324050632911392, | |
| "grad_norm": 0.28765948288667154, | |
| "learning_rate": 2.127347193531757e-06, | |
| "loss": 0.0843, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 4.328101265822784, | |
| "grad_norm": 0.29935773004447935, | |
| "learning_rate": 2.101945952921942e-06, | |
| "loss": 0.0798, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 4.332151898734177, | |
| "grad_norm": 0.29750916221670437, | |
| "learning_rate": 2.0766888607788906e-06, | |
| "loss": 0.0765, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 4.3362025316455695, | |
| "grad_norm": 0.2820966758127444, | |
| "learning_rate": 2.0515761205197337e-06, | |
| "loss": 0.0703, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 4.340253164556962, | |
| "grad_norm": 0.2903533202283874, | |
| "learning_rate": 2.0266079343990453e-06, | |
| "loss": 0.0713, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.3443037974683545, | |
| "grad_norm": 0.2908781157607626, | |
| "learning_rate": 2.0017845035071494e-06, | |
| "loss": 0.0711, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 4.348354430379747, | |
| "grad_norm": 0.2988615408876085, | |
| "learning_rate": 1.9771060277685537e-06, | |
| "loss": 0.0764, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 4.3524050632911395, | |
| "grad_norm": 0.28433325231184087, | |
| "learning_rate": 1.95257270594031e-06, | |
| "loss": 0.0723, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 4.356455696202532, | |
| "grad_norm": 0.3099766012133724, | |
| "learning_rate": 1.9281847356104188e-06, | |
| "loss": 0.0788, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 4.360506329113924, | |
| "grad_norm": 0.28801926599608574, | |
| "learning_rate": 1.9039423131962365e-06, | |
| "loss": 0.0837, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.364556962025317, | |
| "grad_norm": 0.3066568963175479, | |
| "learning_rate": 1.8798456339429027e-06, | |
| "loss": 0.0686, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 4.368607594936709, | |
| "grad_norm": 0.2973157788044432, | |
| "learning_rate": 1.8558948919217612e-06, | |
| "loss": 0.0641, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 4.372658227848102, | |
| "grad_norm": 0.28400877834821103, | |
| "learning_rate": 1.8320902800287954e-06, | |
| "loss": 0.0729, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 4.376708860759494, | |
| "grad_norm": 0.30006121891684406, | |
| "learning_rate": 1.8084319899830726e-06, | |
| "loss": 0.0652, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 4.380759493670886, | |
| "grad_norm": 0.2961602873212686, | |
| "learning_rate": 1.7849202123252097e-06, | |
| "loss": 0.0654, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.384810126582279, | |
| "grad_norm": 0.3015818104294254, | |
| "learning_rate": 1.7615551364158401e-06, | |
| "loss": 0.0745, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 4.388860759493671, | |
| "grad_norm": 0.2909759799476154, | |
| "learning_rate": 1.738336950434061e-06, | |
| "loss": 0.0733, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 4.392911392405063, | |
| "grad_norm": 0.30241359530103956, | |
| "learning_rate": 1.715265841375957e-06, | |
| "loss": 0.068, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 4.396962025316456, | |
| "grad_norm": 0.2985498553020439, | |
| "learning_rate": 1.6923419950530684e-06, | |
| "loss": 0.0792, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 4.401012658227848, | |
| "grad_norm": 0.28281187299557414, | |
| "learning_rate": 1.6695655960909008e-06, | |
| "loss": 0.063, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.405063291139241, | |
| "grad_norm": 0.2888707662551115, | |
| "learning_rate": 1.646936827927441e-06, | |
| "loss": 0.0727, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 4.409113924050633, | |
| "grad_norm": 0.30036138519631583, | |
| "learning_rate": 1.6244558728116766e-06, | |
| "loss": 0.0686, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 4.413164556962025, | |
| "grad_norm": 0.27998986037318085, | |
| "learning_rate": 1.6021229118021265e-06, | |
| "loss": 0.0638, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 4.417215189873418, | |
| "grad_norm": 0.2823129745354462, | |
| "learning_rate": 1.5799381247653967e-06, | |
| "loss": 0.0651, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 4.42126582278481, | |
| "grad_norm": 0.29566520418570125, | |
| "learning_rate": 1.5579016903747013e-06, | |
| "loss": 0.0758, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.425316455696202, | |
| "grad_norm": 0.3015076772790398, | |
| "learning_rate": 1.5360137861084656e-06, | |
| "loss": 0.0842, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 4.429367088607595, | |
| "grad_norm": 0.29413893484669557, | |
| "learning_rate": 1.5142745882488475e-06, | |
| "loss": 0.0697, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 4.433417721518987, | |
| "grad_norm": 0.3064960563092083, | |
| "learning_rate": 1.4926842718803691e-06, | |
| "loss": 0.0783, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 4.43746835443038, | |
| "grad_norm": 0.29391706990020333, | |
| "learning_rate": 1.4712430108884657e-06, | |
| "loss": 0.083, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 4.441518987341772, | |
| "grad_norm": 0.32737179138603545, | |
| "learning_rate": 1.4499509779581078e-06, | |
| "loss": 0.0843, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.445569620253164, | |
| "grad_norm": 0.27736231484950374, | |
| "learning_rate": 1.4288083445723988e-06, | |
| "loss": 0.0718, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 4.449620253164557, | |
| "grad_norm": 0.2994146764788232, | |
| "learning_rate": 1.4078152810112045e-06, | |
| "loss": 0.0771, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 4.453670886075949, | |
| "grad_norm": 0.2995728070535943, | |
| "learning_rate": 1.3869719563497697e-06, | |
| "loss": 0.0841, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 4.457721518987341, | |
| "grad_norm": 0.30065743026811864, | |
| "learning_rate": 1.3662785384573663e-06, | |
| "loss": 0.0746, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 4.461772151898734, | |
| "grad_norm": 0.2878559602236857, | |
| "learning_rate": 1.3457351939959383e-06, | |
| "loss": 0.0748, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.465822784810126, | |
| "grad_norm": 0.29384211642861613, | |
| "learning_rate": 1.3253420884187551e-06, | |
| "loss": 0.0765, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 4.469873417721519, | |
| "grad_norm": 0.2905432364521729, | |
| "learning_rate": 1.3050993859690953e-06, | |
| "loss": 0.0763, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 4.473924050632911, | |
| "grad_norm": 0.2879101239346877, | |
| "learning_rate": 1.2850072496788869e-06, | |
| "loss": 0.0774, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 4.4779746835443035, | |
| "grad_norm": 0.2981532706877797, | |
| "learning_rate": 1.2650658413674434e-06, | |
| "loss": 0.0785, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 4.4820253164556965, | |
| "grad_norm": 0.2925901904900995, | |
| "learning_rate": 1.2452753216401226e-06, | |
| "loss": 0.0662, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.4860759493670885, | |
| "grad_norm": 0.2973316102930347, | |
| "learning_rate": 1.2256358498870503e-06, | |
| "loss": 0.0688, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 4.490126582278481, | |
| "grad_norm": 0.28759693946080717, | |
| "learning_rate": 1.2061475842818337e-06, | |
| "loss": 0.0792, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 4.494177215189874, | |
| "grad_norm": 0.2956499911719464, | |
| "learning_rate": 1.1868106817802816e-06, | |
| "loss": 0.065, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 4.498227848101266, | |
| "grad_norm": 0.2907105390679205, | |
| "learning_rate": 1.1676252981191482e-06, | |
| "loss": 0.0745, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 4.502278481012659, | |
| "grad_norm": 0.28123464058627123, | |
| "learning_rate": 1.1485915878148823e-06, | |
| "loss": 0.0609, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.506329113924051, | |
| "grad_norm": 0.2771931366613535, | |
| "learning_rate": 1.1297097041623584e-06, | |
| "loss": 0.0598, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 4.510379746835443, | |
| "grad_norm": 0.30012058476006365, | |
| "learning_rate": 1.1109797992336847e-06, | |
| "loss": 0.0787, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 4.514430379746836, | |
| "grad_norm": 0.3088629172404503, | |
| "learning_rate": 1.092402023876933e-06, | |
| "loss": 0.0877, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 4.518481012658228, | |
| "grad_norm": 0.2925003110215352, | |
| "learning_rate": 1.0739765277149527e-06, | |
| "loss": 0.0682, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 4.52253164556962, | |
| "grad_norm": 0.29125543157270734, | |
| "learning_rate": 1.0557034591441596e-06, | |
| "loss": 0.0778, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.526582278481013, | |
| "grad_norm": 0.3009549429431239, | |
| "learning_rate": 1.0375829653333324e-06, | |
| "loss": 0.0757, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 4.530632911392405, | |
| "grad_norm": 0.3026704983680009, | |
| "learning_rate": 1.0196151922224385e-06, | |
| "loss": 0.0778, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 4.534683544303798, | |
| "grad_norm": 0.294064623675452, | |
| "learning_rate": 1.0018002845214526e-06, | |
| "loss": 0.0782, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 4.53873417721519, | |
| "grad_norm": 0.2954743734426885, | |
| "learning_rate": 9.841383857091947e-07, | |
| "loss": 0.0765, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.542784810126582, | |
| "grad_norm": 0.28927454048408313, | |
| "learning_rate": 9.666296380321616e-07, | |
| "loss": 0.066, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.546835443037975, | |
| "grad_norm": 0.2916343154979188, | |
| "learning_rate": 9.492741825034124e-07, | |
| "loss": 0.0815, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 4.550886075949367, | |
| "grad_norm": 0.2832889091361109, | |
| "learning_rate": 9.320721589013892e-07, | |
| "loss": 0.0754, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 4.55493670886076, | |
| "grad_norm": 0.2931831473822679, | |
| "learning_rate": 9.150237057688339e-07, | |
| "loss": 0.0909, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 4.558987341772152, | |
| "grad_norm": 0.27698744956036503, | |
| "learning_rate": 8.981289604116328e-07, | |
| "loss": 0.0788, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 4.563037974683544, | |
| "grad_norm": 0.2860488823144332, | |
| "learning_rate": 8.813880588977542e-07, | |
| "loss": 0.0803, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.567088607594937, | |
| "grad_norm": 0.2947253492367951, | |
| "learning_rate": 8.648011360561126e-07, | |
| "loss": 0.0716, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 4.571139240506329, | |
| "grad_norm": 0.2972279870191502, | |
| "learning_rate": 8.483683254755037e-07, | |
| "loss": 0.0815, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 4.575189873417721, | |
| "grad_norm": 0.30777363511923334, | |
| "learning_rate": 8.320897595035227e-07, | |
| "loss": 0.0765, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 4.579240506329114, | |
| "grad_norm": 0.3046034244356588, | |
| "learning_rate": 8.159655692455093e-07, | |
| "loss": 0.0809, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 4.583291139240506, | |
| "grad_norm": 0.2915952680901405, | |
| "learning_rate": 7.999958845634648e-07, | |
| "loss": 0.0734, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.587341772151898, | |
| "grad_norm": 0.3023584789558434, | |
| "learning_rate": 7.841808340750478e-07, | |
| "loss": 0.0705, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 4.591392405063291, | |
| "grad_norm": 0.2839779205327264, | |
| "learning_rate": 7.685205451524869e-07, | |
| "loss": 0.0693, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 4.595443037974683, | |
| "grad_norm": 0.2809560519048446, | |
| "learning_rate": 7.530151439216027e-07, | |
| "loss": 0.074, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 4.599493670886076, | |
| "grad_norm": 0.29210496052013596, | |
| "learning_rate": 7.376647552607675e-07, | |
| "loss": 0.0778, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 4.603544303797468, | |
| "grad_norm": 0.2940221010524311, | |
| "learning_rate": 7.224695027998963e-07, | |
| "loss": 0.0706, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.6075949367088604, | |
| "grad_norm": 0.3062022877292046, | |
| "learning_rate": 7.07429508919466e-07, | |
| "loss": 0.0868, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 4.611645569620253, | |
| "grad_norm": 0.29637374031163616, | |
| "learning_rate": 6.925448947495206e-07, | |
| "loss": 0.0788, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 4.6156962025316455, | |
| "grad_norm": 0.3023474847246503, | |
| "learning_rate": 6.778157801686936e-07, | |
| "loss": 0.0761, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 4.619746835443038, | |
| "grad_norm": 0.29400706196511006, | |
| "learning_rate": 6.632422838032515e-07, | |
| "loss": 0.0704, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 4.6237974683544305, | |
| "grad_norm": 0.2869267630694395, | |
| "learning_rate": 6.488245230261281e-07, | |
| "loss": 0.0713, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.627848101265823, | |
| "grad_norm": 0.2850626124352276, | |
| "learning_rate": 6.345626139559868e-07, | |
| "loss": 0.0653, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 4.6318987341772155, | |
| "grad_norm": 0.30114241227124466, | |
| "learning_rate": 6.204566714562866e-07, | |
| "loss": 0.0698, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 4.635949367088608, | |
| "grad_norm": 0.2937319391470622, | |
| "learning_rate": 6.06506809134344e-07, | |
| "loss": 0.0812, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "grad_norm": 0.29795079294186705, | |
| "learning_rate": 5.927131393404373e-07, | |
| "loss": 0.0873, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 4.644050632911393, | |
| "grad_norm": 0.2831635612321249, | |
| "learning_rate": 5.790757731668817e-07, | |
| "loss": 0.0687, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.648101265822785, | |
| "grad_norm": 0.28864641924282536, | |
| "learning_rate": 5.655948204471507e-07, | |
| "loss": 0.075, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 4.652151898734177, | |
| "grad_norm": 0.2823213463205551, | |
| "learning_rate": 5.522703897549875e-07, | |
| "loss": 0.0681, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 4.65620253164557, | |
| "grad_norm": 0.28651084396331766, | |
| "learning_rate": 5.391025884035239e-07, | |
| "loss": 0.081, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 4.660253164556962, | |
| "grad_norm": 0.29447658947812116, | |
| "learning_rate": 5.260915224444207e-07, | |
| "loss": 0.0785, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 4.664303797468355, | |
| "grad_norm": 0.30388671192072164, | |
| "learning_rate": 5.132372966670129e-07, | |
| "loss": 0.0737, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.668354430379747, | |
| "grad_norm": 0.29119226444984914, | |
| "learning_rate": 5.005400145974704e-07, | |
| "loss": 0.0774, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 4.672405063291139, | |
| "grad_norm": 0.2897454059332075, | |
| "learning_rate": 4.879997784979562e-07, | |
| "loss": 0.0759, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 4.676455696202532, | |
| "grad_norm": 0.2899147725520681, | |
| "learning_rate": 4.7561668936580984e-07, | |
| "loss": 0.0765, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 4.680506329113924, | |
| "grad_norm": 0.2795086254583641, | |
| "learning_rate": 4.6339084693272306e-07, | |
| "loss": 0.0728, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 4.684556962025317, | |
| "grad_norm": 0.2879654583973189, | |
| "learning_rate": 4.5132234966395847e-07, | |
| "loss": 0.0826, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.688607594936709, | |
| "grad_norm": 0.2955303962823585, | |
| "learning_rate": 4.3941129475752795e-07, | |
| "loss": 0.0763, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 4.692658227848101, | |
| "grad_norm": 0.30244663468068617, | |
| "learning_rate": 4.27657778143431e-07, | |
| "loss": 0.0808, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 4.696708860759494, | |
| "grad_norm": 0.2997307244394497, | |
| "learning_rate": 4.1606189448287757e-07, | |
| "loss": 0.0781, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 4.700759493670886, | |
| "grad_norm": 0.2853268676094933, | |
| "learning_rate": 4.046237371675177e-07, | |
| "loss": 0.0714, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 4.704810126582278, | |
| "grad_norm": 0.2952831057026446, | |
| "learning_rate": 3.9334339831869963e-07, | |
| "loss": 0.0634, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.708860759493671, | |
| "grad_norm": 0.29551170230381874, | |
| "learning_rate": 3.8222096878671955e-07, | |
| "loss": 0.0727, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 4.712911392405063, | |
| "grad_norm": 0.2944626725790237, | |
| "learning_rate": 3.7125653815009545e-07, | |
| "loss": 0.0672, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 4.716962025316455, | |
| "grad_norm": 0.3113257851597562, | |
| "learning_rate": 3.6045019471484974e-07, | |
| "loss": 0.0858, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 4.721012658227848, | |
| "grad_norm": 0.2872378598729664, | |
| "learning_rate": 3.498020255137813e-07, | |
| "loss": 0.0693, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 4.72506329113924, | |
| "grad_norm": 0.29330565951262316, | |
| "learning_rate": 3.393121163057811e-07, | |
| "loss": 0.0757, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.729113924050633, | |
| "grad_norm": 0.2848855792758256, | |
| "learning_rate": 3.289805515751399e-07, | |
| "loss": 0.0707, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 4.733164556962025, | |
| "grad_norm": 0.2887135426937249, | |
| "learning_rate": 3.188074145308573e-07, | |
| "loss": 0.0715, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 4.737215189873417, | |
| "grad_norm": 0.28546837073269926, | |
| "learning_rate": 3.087927871059804e-07, | |
| "loss": 0.0737, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 4.74126582278481, | |
| "grad_norm": 0.27874639963890896, | |
| "learning_rate": 2.989367499569418e-07, | |
| "loss": 0.0662, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 4.745316455696202, | |
| "grad_norm": 0.28308672256323414, | |
| "learning_rate": 2.8923938246290917e-07, | |
| "loss": 0.0741, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.749367088607595, | |
| "grad_norm": 0.301729767578112, | |
| "learning_rate": 2.7970076272514804e-07, | |
| "loss": 0.084, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 4.7534177215189874, | |
| "grad_norm": 0.28212100368975296, | |
| "learning_rate": 2.703209675663887e-07, | |
| "loss": 0.0798, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 4.7574683544303795, | |
| "grad_norm": 0.2926324966662583, | |
| "learning_rate": 2.6110007253021374e-07, | |
| "loss": 0.0677, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 4.7615189873417725, | |
| "grad_norm": 0.28980906025680236, | |
| "learning_rate": 2.520381518804471e-07, | |
| "loss": 0.0696, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 4.7655696202531646, | |
| "grad_norm": 0.29577855335484277, | |
| "learning_rate": 2.4313527860054585e-07, | |
| "loss": 0.0729, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.769620253164557, | |
| "grad_norm": 0.29128181419340027, | |
| "learning_rate": 2.343915243930317e-07, | |
| "loss": 0.0767, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 4.77367088607595, | |
| "grad_norm": 0.3013774041330498, | |
| "learning_rate": 2.2580695967889367e-07, | |
| "loss": 0.0699, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 4.777721518987342, | |
| "grad_norm": 0.3082947764121362, | |
| "learning_rate": 2.1738165359704189e-07, | |
| "loss": 0.082, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 4.781772151898734, | |
| "grad_norm": 0.30209048852638565, | |
| "learning_rate": 2.0911567400373257e-07, | |
| "loss": 0.096, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 4.785822784810127, | |
| "grad_norm": 0.2784539804816474, | |
| "learning_rate": 2.0100908747202607e-07, | |
| "loss": 0.0687, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.789873417721519, | |
| "grad_norm": 0.2811784129458952, | |
| "learning_rate": 1.9306195929125638e-07, | |
| "loss": 0.0673, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 4.793924050632912, | |
| "grad_norm": 0.2858731538348693, | |
| "learning_rate": 1.8527435346650247e-07, | |
| "loss": 0.0774, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 4.797974683544304, | |
| "grad_norm": 0.28695354008395374, | |
| "learning_rate": 1.7764633271807108e-07, | |
| "loss": 0.0648, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 4.802025316455696, | |
| "grad_norm": 0.2911749586643828, | |
| "learning_rate": 1.7017795848099262e-07, | |
| "loss": 0.0714, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 4.806075949367089, | |
| "grad_norm": 0.30382371281865367, | |
| "learning_rate": 1.6286929090452596e-07, | |
| "loss": 0.0693, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.810126582278481, | |
| "grad_norm": 0.2795344081922359, | |
| "learning_rate": 1.557203888516745e-07, | |
| "loss": 0.0633, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 4.814177215189874, | |
| "grad_norm": 0.2924687098212547, | |
| "learning_rate": 1.487313098987131e-07, | |
| "loss": 0.0764, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 4.818227848101266, | |
| "grad_norm": 0.28494772947354224, | |
| "learning_rate": 1.4190211033472402e-07, | |
| "loss": 0.0737, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 4.822278481012658, | |
| "grad_norm": 0.2991454077013667, | |
| "learning_rate": 1.3523284516113955e-07, | |
| "loss": 0.0712, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 4.826329113924051, | |
| "grad_norm": 0.2956485729120339, | |
| "learning_rate": 1.2872356809130682e-07, | |
| "loss": 0.0746, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.830379746835443, | |
| "grad_norm": 0.3035474674340979, | |
| "learning_rate": 1.2237433155004807e-07, | |
| "loss": 0.0851, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 4.834430379746835, | |
| "grad_norm": 0.2989657570122506, | |
| "learning_rate": 1.1618518667323886e-07, | |
| "loss": 0.0671, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 4.838481012658228, | |
| "grad_norm": 0.28651074427962725, | |
| "learning_rate": 1.1015618330740385e-07, | |
| "loss": 0.069, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 4.84253164556962, | |
| "grad_norm": 0.29140780346011835, | |
| "learning_rate": 1.042873700093061e-07, | |
| "loss": 0.0758, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 4.846582278481012, | |
| "grad_norm": 0.29401571678392746, | |
| "learning_rate": 9.857879404556291e-08, | |
| "loss": 0.0824, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.850632911392405, | |
| "grad_norm": 0.28672840842088204, | |
| "learning_rate": 9.303050139225722e-08, | |
| "loss": 0.0692, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 4.854683544303797, | |
| "grad_norm": 0.2881078613363558, | |
| "learning_rate": 8.76425367345779e-08, | |
| "loss": 0.071, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 4.85873417721519, | |
| "grad_norm": 0.2873927225632679, | |
| "learning_rate": 8.241494346644897e-08, | |
| "loss": 0.0734, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 4.862784810126582, | |
| "grad_norm": 0.2828418006272296, | |
| "learning_rate": 7.734776369019204e-08, | |
| "loss": 0.0737, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 4.866835443037974, | |
| "grad_norm": 0.29528096971769136, | |
| "learning_rate": 7.244103821617332e-08, | |
| "loss": 0.0807, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.870886075949367, | |
| "grad_norm": 0.3036885234961058, | |
| "learning_rate": 6.769480656248606e-08, | |
| "loss": 0.0744, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 4.874936708860759, | |
| "grad_norm": 0.3024701859608035, | |
| "learning_rate": 6.310910695462635e-08, | |
| "loss": 0.0794, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 4.878987341772152, | |
| "grad_norm": 0.2992783889731382, | |
| "learning_rate": 5.8683976325191185e-08, | |
| "loss": 0.0835, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 4.883037974683544, | |
| "grad_norm": 0.2979650267791592, | |
| "learning_rate": 5.4419450313571984e-08, | |
| "loss": 0.0701, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 4.8870886075949365, | |
| "grad_norm": 0.28840078098625826, | |
| "learning_rate": 5.031556326567488e-08, | |
| "loss": 0.0772, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.891139240506329, | |
| "grad_norm": 0.2946389074002119, | |
| "learning_rate": 4.637234823364312e-08, | |
| "loss": 0.0806, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 4.8951898734177215, | |
| "grad_norm": 0.2977772418467221, | |
| "learning_rate": 4.258983697558838e-08, | |
| "loss": 0.0769, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 4.899240506329114, | |
| "grad_norm": 0.2853870920918102, | |
| "learning_rate": 3.896805995533548e-08, | |
| "loss": 0.0745, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 4.9032911392405065, | |
| "grad_norm": 0.28701110091798887, | |
| "learning_rate": 3.550704634218028e-08, | |
| "loss": 0.0744, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 4.907341772151899, | |
| "grad_norm": 0.27740330250699163, | |
| "learning_rate": 3.2206824010647676e-08, | |
| "loss": 0.0687, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.911392405063291, | |
| "grad_norm": 0.283413954738808, | |
| "learning_rate": 2.9067419540278476e-08, | |
| "loss": 0.0837, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 4.915443037974684, | |
| "grad_norm": 0.2948604672576155, | |
| "learning_rate": 2.6088858215400638e-08, | |
| "loss": 0.0816, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 4.919493670886076, | |
| "grad_norm": 0.2968359817989385, | |
| "learning_rate": 2.3271164024940564e-08, | |
| "loss": 0.0821, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 4.923544303797469, | |
| "grad_norm": 0.28982806661763233, | |
| "learning_rate": 2.061435966221881e-08, | |
| "loss": 0.0789, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 4.927594936708861, | |
| "grad_norm": 0.2980911867392471, | |
| "learning_rate": 1.811846652477245e-08, | |
| "loss": 0.0846, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.931645569620253, | |
| "grad_norm": 0.28566453201765923, | |
| "learning_rate": 1.5783504714184106e-08, | |
| "loss": 0.0791, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 4.935696202531646, | |
| "grad_norm": 0.2780364223509791, | |
| "learning_rate": 1.360949303591097e-08, | |
| "loss": 0.0623, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 4.939746835443038, | |
| "grad_norm": 0.2994569621281494, | |
| "learning_rate": 1.1596448999144916e-08, | |
| "loss": 0.0817, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 4.943797468354431, | |
| "grad_norm": 0.29676057549950285, | |
| "learning_rate": 9.744388816668172e-09, | |
| "loss": 0.0636, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 4.947848101265823, | |
| "grad_norm": 0.28683605248952426, | |
| "learning_rate": 8.05332740472009e-09, | |
| "loss": 0.0809, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.951898734177215, | |
| "grad_norm": 0.2830786945435166, | |
| "learning_rate": 6.523278382872811e-09, | |
| "loss": 0.0697, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 4.955949367088608, | |
| "grad_norm": 0.29557318421670636, | |
| "learning_rate": 5.15425407393133e-09, | |
| "loss": 0.0852, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "grad_norm": 0.28399950405251057, | |
| "learning_rate": 3.94626550383137e-09, | |
| "loss": 0.0631, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 4.964050632911392, | |
| "grad_norm": 0.28377358542773673, | |
| "learning_rate": 2.899322401546112e-09, | |
| "loss": 0.0606, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 4.968101265822785, | |
| "grad_norm": 0.3038573547059663, | |
| "learning_rate": 2.013433199010706e-09, | |
| "loss": 0.0803, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.972151898734177, | |
| "grad_norm": 0.2863955703046627, | |
| "learning_rate": 1.2886050310556563e-09, | |
| "loss": 0.0768, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 4.976202531645569, | |
| "grad_norm": 0.28530390476821577, | |
| "learning_rate": 7.248437353468695e-10, | |
| "loss": 0.0722, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 4.980253164556962, | |
| "grad_norm": 0.3020147611176079, | |
| "learning_rate": 3.221538523412449e-10, | |
| "loss": 0.0864, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 4.984303797468354, | |
| "grad_norm": 0.3124686707703266, | |
| "learning_rate": 8.053862524670663e-11, | |
| "loss": 0.088, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 4.988354430379747, | |
| "grad_norm": 0.2912564641316393, | |
| "learning_rate": 0.0, | |
| "loss": 0.0952, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.988354430379747, | |
| "step": 1230, | |
| "total_flos": 9.737215825160438e+17, | |
| "train_loss": 0.40829433875718735, | |
| "train_runtime": 27742.0258, | |
| "train_samples_per_second": 5.695, | |
| "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": 9.737215825160438e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |