{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 2462, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00040617384240454913, "grad_norm": 0.0, "learning_rate": 2.702702702702703e-07, "loss": 3.2487, "step": 1 }, { "epoch": 0.0008123476848090983, "grad_norm": 0.0, "learning_rate": 5.405405405405406e-07, "loss": 3.3287, "step": 2 }, { "epoch": 0.0012185215272136475, "grad_norm": 0.0, "learning_rate": 8.108108108108109e-07, "loss": 3.3526, "step": 3 }, { "epoch": 0.0016246953696181965, "grad_norm": 0.0, "learning_rate": 1.0810810810810812e-06, "loss": 3.1769, "step": 4 }, { "epoch": 0.0020308692120227455, "grad_norm": 0.0, "learning_rate": 1.3513513513513515e-06, "loss": 3.3049, "step": 5 }, { "epoch": 0.002437043054427295, "grad_norm": 0.0, "learning_rate": 1.6216216216216219e-06, "loss": 3.299, "step": 6 }, { "epoch": 0.002843216896831844, "grad_norm": 0.0, "learning_rate": 1.8918918918918922e-06, "loss": 3.3358, "step": 7 }, { "epoch": 0.003249390739236393, "grad_norm": 0.0, "learning_rate": 2.1621621621621623e-06, "loss": 3.1727, "step": 8 }, { "epoch": 0.0036555645816409425, "grad_norm": 0.0, "learning_rate": 2.432432432432433e-06, "loss": 3.155, "step": 9 }, { "epoch": 0.004061738424045491, "grad_norm": 0.0, "learning_rate": 2.702702702702703e-06, "loss": 3.2636, "step": 10 }, { "epoch": 0.004467912266450041, "grad_norm": 0.0, "learning_rate": 2.9729729729729736e-06, "loss": 3.2148, "step": 11 }, { "epoch": 0.00487408610885459, "grad_norm": 0.0, "learning_rate": 3.2432432432432437e-06, "loss": 3.1042, "step": 12 }, { "epoch": 0.005280259951259139, "grad_norm": 0.0, "learning_rate": 3.513513513513514e-06, "loss": 3.207, "step": 13 }, { "epoch": 0.005686433793663688, "grad_norm": 0.0, "learning_rate": 3.7837837837837844e-06, "loss": 3.2531, "step": 14 }, { "epoch": 0.006092607636068237, "grad_norm": 0.0, "learning_rate": 4.0540540540540545e-06, "loss": 3.2652, "step": 15 }, { "epoch": 0.006498781478472786, "grad_norm": 0.0, "learning_rate": 4.324324324324325e-06, "loss": 3.1545, "step": 16 }, { "epoch": 0.006904955320877335, "grad_norm": 0.0, "learning_rate": 4.594594594594596e-06, "loss": 3.1656, "step": 17 }, { "epoch": 0.007311129163281885, "grad_norm": 0.0, "learning_rate": 4.864864864864866e-06, "loss": 3.2177, "step": 18 }, { "epoch": 0.007717303005686434, "grad_norm": 0.0, "learning_rate": 5.135135135135135e-06, "loss": 3.1796, "step": 19 }, { "epoch": 0.008123476848090982, "grad_norm": 0.0, "learning_rate": 5.405405405405406e-06, "loss": 3.1439, "step": 20 }, { "epoch": 0.008529650690495532, "grad_norm": 0.0, "learning_rate": 5.675675675675676e-06, "loss": 3.1478, "step": 21 }, { "epoch": 0.008935824532900082, "grad_norm": 0.0, "learning_rate": 5.945945945945947e-06, "loss": 3.1481, "step": 22 }, { "epoch": 0.00934199837530463, "grad_norm": 0.0, "learning_rate": 6.2162162162162164e-06, "loss": 3.1122, "step": 23 }, { "epoch": 0.00974817221770918, "grad_norm": 0.0, "learning_rate": 6.486486486486487e-06, "loss": 3.1592, "step": 24 }, { "epoch": 0.010154346060113728, "grad_norm": 0.0, "learning_rate": 6.7567567567567575e-06, "loss": 3.1474, "step": 25 }, { "epoch": 0.010560519902518278, "grad_norm": 0.0, "learning_rate": 7.027027027027028e-06, "loss": 3.0779, "step": 26 }, { "epoch": 0.010966693744922826, "grad_norm": 0.0, "learning_rate": 7.297297297297298e-06, "loss": 3.15, "step": 27 }, { "epoch": 0.011372867587327376, "grad_norm": 0.0, "learning_rate": 7.567567567567569e-06, "loss": 3.1109, "step": 28 }, { "epoch": 0.011779041429731926, "grad_norm": 0.0, "learning_rate": 7.837837837837838e-06, "loss": 3.14, "step": 29 }, { "epoch": 0.012185215272136474, "grad_norm": 0.0, "learning_rate": 8.108108108108109e-06, "loss": 3.0277, "step": 30 }, { "epoch": 0.012591389114541024, "grad_norm": 0.0, "learning_rate": 8.378378378378378e-06, "loss": 3.0794, "step": 31 }, { "epoch": 0.012997562956945572, "grad_norm": 0.0, "learning_rate": 8.64864864864865e-06, "loss": 3.0829, "step": 32 }, { "epoch": 0.013403736799350122, "grad_norm": 0.0, "learning_rate": 8.91891891891892e-06, "loss": 3.0768, "step": 33 }, { "epoch": 0.01380991064175467, "grad_norm": 0.0, "learning_rate": 9.189189189189191e-06, "loss": 3.0548, "step": 34 }, { "epoch": 0.01421608448415922, "grad_norm": 0.0, "learning_rate": 9.45945945945946e-06, "loss": 3.0701, "step": 35 }, { "epoch": 0.01462225832656377, "grad_norm": 0.0, "learning_rate": 9.729729729729732e-06, "loss": 3.0187, "step": 36 }, { "epoch": 0.015028432168968318, "grad_norm": 0.0, "learning_rate": 1e-05, "loss": 3.0657, "step": 37 }, { "epoch": 0.015434606011372868, "grad_norm": 0.0, "learning_rate": 1.027027027027027e-05, "loss": 3.0582, "step": 38 }, { "epoch": 0.015840779853777416, "grad_norm": 0.0, "learning_rate": 1.0540540540540541e-05, "loss": 3.0392, "step": 39 }, { "epoch": 0.016246953696181964, "grad_norm": 0.0, "learning_rate": 1.0810810810810812e-05, "loss": 3.0482, "step": 40 }, { "epoch": 0.016653127538586516, "grad_norm": 0.0, "learning_rate": 1.1081081081081081e-05, "loss": 2.968, "step": 41 }, { "epoch": 0.017059301380991064, "grad_norm": 0.0, "learning_rate": 1.1351351351351352e-05, "loss": 2.9899, "step": 42 }, { "epoch": 0.017465475223395612, "grad_norm": 0.0, "learning_rate": 1.1621621621621622e-05, "loss": 3.0395, "step": 43 }, { "epoch": 0.017871649065800164, "grad_norm": 0.0, "learning_rate": 1.1891891891891894e-05, "loss": 2.9801, "step": 44 }, { "epoch": 0.018277822908204712, "grad_norm": 0.0, "learning_rate": 1.2162162162162164e-05, "loss": 2.9892, "step": 45 }, { "epoch": 0.01868399675060926, "grad_norm": 0.0, "learning_rate": 1.2432432432432433e-05, "loss": 3.0922, "step": 46 }, { "epoch": 0.019090170593013808, "grad_norm": 0.0, "learning_rate": 1.2702702702702702e-05, "loss": 3.0473, "step": 47 }, { "epoch": 0.01949634443541836, "grad_norm": 0.0, "learning_rate": 1.2972972972972975e-05, "loss": 2.9648, "step": 48 }, { "epoch": 0.019902518277822908, "grad_norm": 0.0, "learning_rate": 1.3243243243243244e-05, "loss": 3.0159, "step": 49 }, { "epoch": 0.020308692120227456, "grad_norm": 0.0, "learning_rate": 1.3513513513513515e-05, "loss": 3.0917, "step": 50 }, { "epoch": 0.020714865962632008, "grad_norm": 0.0, "learning_rate": 1.3783783783783784e-05, "loss": 3.0261, "step": 51 }, { "epoch": 0.021121039805036556, "grad_norm": 0.0, "learning_rate": 1.4054054054054055e-05, "loss": 3.0589, "step": 52 }, { "epoch": 0.021527213647441104, "grad_norm": 0.0, "learning_rate": 1.4324324324324326e-05, "loss": 3.0454, "step": 53 }, { "epoch": 0.021933387489845652, "grad_norm": 0.0, "learning_rate": 1.4594594594594596e-05, "loss": 3.0285, "step": 54 }, { "epoch": 0.022339561332250204, "grad_norm": 0.0, "learning_rate": 1.4864864864864865e-05, "loss": 3.0832, "step": 55 }, { "epoch": 0.022745735174654752, "grad_norm": 0.0, "learning_rate": 1.5135135135135138e-05, "loss": 2.9877, "step": 56 }, { "epoch": 0.0231519090170593, "grad_norm": 0.0, "learning_rate": 1.540540540540541e-05, "loss": 2.861, "step": 57 }, { "epoch": 0.023558082859463852, "grad_norm": 0.0, "learning_rate": 1.5675675675675676e-05, "loss": 3.0116, "step": 58 }, { "epoch": 0.0239642567018684, "grad_norm": 0.0, "learning_rate": 1.5945945945945947e-05, "loss": 2.997, "step": 59 }, { "epoch": 0.024370430544272948, "grad_norm": 0.0, "learning_rate": 1.6216216216216218e-05, "loss": 3.0151, "step": 60 }, { "epoch": 0.024776604386677496, "grad_norm": 0.0, "learning_rate": 1.648648648648649e-05, "loss": 2.9452, "step": 61 }, { "epoch": 0.025182778229082048, "grad_norm": 0.0, "learning_rate": 1.6756756756756757e-05, "loss": 3.0219, "step": 62 }, { "epoch": 0.025588952071486596, "grad_norm": 0.0, "learning_rate": 1.7027027027027028e-05, "loss": 2.9703, "step": 63 }, { "epoch": 0.025995125913891144, "grad_norm": 0.0, "learning_rate": 1.72972972972973e-05, "loss": 3.0034, "step": 64 }, { "epoch": 0.026401299756295696, "grad_norm": 0.0, "learning_rate": 1.756756756756757e-05, "loss": 2.7839, "step": 65 }, { "epoch": 0.026807473598700244, "grad_norm": 0.0, "learning_rate": 1.783783783783784e-05, "loss": 2.7971, "step": 66 }, { "epoch": 0.027213647441104792, "grad_norm": 0.0, "learning_rate": 1.8108108108108108e-05, "loss": 2.9821, "step": 67 }, { "epoch": 0.02761982128350934, "grad_norm": 0.0, "learning_rate": 1.8378378378378383e-05, "loss": 2.9834, "step": 68 }, { "epoch": 0.028025995125913892, "grad_norm": 0.0, "learning_rate": 1.864864864864865e-05, "loss": 2.8888, "step": 69 }, { "epoch": 0.02843216896831844, "grad_norm": 0.0, "learning_rate": 1.891891891891892e-05, "loss": 2.9156, "step": 70 }, { "epoch": 0.028838342810722988, "grad_norm": 0.0, "learning_rate": 1.918918918918919e-05, "loss": 2.8311, "step": 71 }, { "epoch": 0.02924451665312754, "grad_norm": 0.0, "learning_rate": 1.9459459459459463e-05, "loss": 2.9276, "step": 72 }, { "epoch": 0.029650690495532088, "grad_norm": 0.0, "learning_rate": 1.972972972972973e-05, "loss": 2.7668, "step": 73 }, { "epoch": 0.030056864337936636, "grad_norm": 0.0, "learning_rate": 2e-05, "loss": 2.904, "step": 74 }, { "epoch": 0.030463038180341188, "grad_norm": 0.0, "learning_rate": 1.9999991346315804e-05, "loss": 2.863, "step": 75 }, { "epoch": 0.030869212022745736, "grad_norm": 0.0, "learning_rate": 1.9999965385278197e-05, "loss": 2.7896, "step": 76 }, { "epoch": 0.03127538586515029, "grad_norm": 0.0, "learning_rate": 1.9999922116932104e-05, "loss": 2.8466, "step": 77 }, { "epoch": 0.03168155970755483, "grad_norm": 0.0, "learning_rate": 1.9999861541352416e-05, "loss": 2.7722, "step": 78 }, { "epoch": 0.032087733549959384, "grad_norm": 0.0, "learning_rate": 1.999978365864397e-05, "loss": 2.8221, "step": 79 }, { "epoch": 0.03249390739236393, "grad_norm": 0.0, "learning_rate": 1.9999688468941565e-05, "loss": 2.9196, "step": 80 }, { "epoch": 0.03290008123476848, "grad_norm": 0.0, "learning_rate": 1.9999575972409947e-05, "loss": 2.8137, "step": 81 }, { "epoch": 0.03330625507717303, "grad_norm": 0.0, "learning_rate": 1.9999446169243816e-05, "loss": 2.812, "step": 82 }, { "epoch": 0.033712428919577576, "grad_norm": 0.0, "learning_rate": 1.999929905966783e-05, "loss": 2.8855, "step": 83 }, { "epoch": 0.03411860276198213, "grad_norm": 0.0, "learning_rate": 1.9999134643936594e-05, "loss": 2.7789, "step": 84 }, { "epoch": 0.03452477660438668, "grad_norm": 0.0, "learning_rate": 1.9998952922334672e-05, "loss": 2.8942, "step": 85 }, { "epoch": 0.034930950446791224, "grad_norm": 0.0, "learning_rate": 1.9998753895176576e-05, "loss": 2.8191, "step": 86 }, { "epoch": 0.035337124289195776, "grad_norm": 0.0, "learning_rate": 1.9998537562806762e-05, "loss": 2.8464, "step": 87 }, { "epoch": 0.03574329813160033, "grad_norm": 0.0, "learning_rate": 1.9998303925599657e-05, "loss": 2.8738, "step": 88 }, { "epoch": 0.03614947197400487, "grad_norm": 0.0, "learning_rate": 1.9998052983959617e-05, "loss": 2.8404, "step": 89 }, { "epoch": 0.036555645816409424, "grad_norm": 0.0, "learning_rate": 1.999778473832096e-05, "loss": 2.784, "step": 90 }, { "epoch": 0.036961819658813976, "grad_norm": 0.0, "learning_rate": 1.9997499189147943e-05, "loss": 2.7153, "step": 91 }, { "epoch": 0.03736799350121852, "grad_norm": 0.0, "learning_rate": 1.999719633693478e-05, "loss": 2.9289, "step": 92 }, { "epoch": 0.03777416734362307, "grad_norm": 0.0, "learning_rate": 1.9996876182205633e-05, "loss": 2.8342, "step": 93 }, { "epoch": 0.038180341186027617, "grad_norm": 0.0, "learning_rate": 1.9996538725514597e-05, "loss": 2.9104, "step": 94 }, { "epoch": 0.03858651502843217, "grad_norm": 0.0, "learning_rate": 1.9996183967445725e-05, "loss": 2.7692, "step": 95 }, { "epoch": 0.03899268887083672, "grad_norm": 0.0, "learning_rate": 1.999581190861301e-05, "loss": 2.8681, "step": 96 }, { "epoch": 0.039398862713241264, "grad_norm": 0.0, "learning_rate": 1.999542254966039e-05, "loss": 2.8068, "step": 97 }, { "epoch": 0.039805036555645816, "grad_norm": 0.0, "learning_rate": 1.999501589126174e-05, "loss": 2.822, "step": 98 }, { "epoch": 0.04021121039805037, "grad_norm": 0.0, "learning_rate": 1.9994591934120876e-05, "loss": 2.6925, "step": 99 }, { "epoch": 0.04061738424045491, "grad_norm": 0.0, "learning_rate": 1.9994150678971564e-05, "loss": 2.6981, "step": 100 }, { "epoch": 0.041023558082859464, "grad_norm": 0.0, "learning_rate": 1.9993692126577493e-05, "loss": 2.7817, "step": 101 }, { "epoch": 0.041429731925264016, "grad_norm": 0.0, "learning_rate": 1.9993216277732302e-05, "loss": 2.7628, "step": 102 }, { "epoch": 0.04183590576766856, "grad_norm": 0.0, "learning_rate": 1.9992723133259555e-05, "loss": 2.8552, "step": 103 }, { "epoch": 0.04224207961007311, "grad_norm": 0.0, "learning_rate": 1.9992212694012757e-05, "loss": 2.7023, "step": 104 }, { "epoch": 0.042648253452477664, "grad_norm": 0.0, "learning_rate": 1.9991684960875348e-05, "loss": 2.7222, "step": 105 }, { "epoch": 0.04305442729488221, "grad_norm": 0.0, "learning_rate": 1.999113993476069e-05, "loss": 2.7878, "step": 106 }, { "epoch": 0.04346060113728676, "grad_norm": 0.0, "learning_rate": 1.9990577616612083e-05, "loss": 2.7559, "step": 107 }, { "epoch": 0.043866774979691305, "grad_norm": 0.0, "learning_rate": 1.998999800740275e-05, "loss": 2.6498, "step": 108 }, { "epoch": 0.044272948822095856, "grad_norm": 0.0, "learning_rate": 1.9989401108135843e-05, "loss": 2.6487, "step": 109 }, { "epoch": 0.04467912266450041, "grad_norm": 0.0, "learning_rate": 1.9988786919844437e-05, "loss": 2.7221, "step": 110 }, { "epoch": 0.04508529650690495, "grad_norm": 0.0, "learning_rate": 1.998815544359153e-05, "loss": 2.8401, "step": 111 }, { "epoch": 0.045491470349309504, "grad_norm": 0.0, "learning_rate": 1.9987506680470045e-05, "loss": 2.6593, "step": 112 }, { "epoch": 0.045897644191714056, "grad_norm": 0.0, "learning_rate": 1.998684063160281e-05, "loss": 2.5817, "step": 113 }, { "epoch": 0.0463038180341186, "grad_norm": 0.0, "learning_rate": 1.9986157298142595e-05, "loss": 2.7257, "step": 114 }, { "epoch": 0.04670999187652315, "grad_norm": 0.0, "learning_rate": 1.998545668127206e-05, "loss": 2.7774, "step": 115 }, { "epoch": 0.047116165718927704, "grad_norm": 0.0, "learning_rate": 1.998473878220379e-05, "loss": 2.6923, "step": 116 }, { "epoch": 0.04752233956133225, "grad_norm": 0.0, "learning_rate": 1.998400360218028e-05, "loss": 2.8439, "step": 117 }, { "epoch": 0.0479285134037368, "grad_norm": 0.0, "learning_rate": 1.9983251142473935e-05, "loss": 2.7283, "step": 118 }, { "epoch": 0.04833468724614135, "grad_norm": 0.0, "learning_rate": 1.9982481404387063e-05, "loss": 2.6864, "step": 119 }, { "epoch": 0.048740861088545896, "grad_norm": 0.0, "learning_rate": 1.998169438925188e-05, "loss": 2.5969, "step": 120 }, { "epoch": 0.04914703493095045, "grad_norm": 0.0, "learning_rate": 1.99808900984305e-05, "loss": 2.5429, "step": 121 }, { "epoch": 0.04955320877335499, "grad_norm": 0.0, "learning_rate": 1.9980068533314937e-05, "loss": 2.6451, "step": 122 }, { "epoch": 0.049959382615759544, "grad_norm": 0.0, "learning_rate": 1.9979229695327106e-05, "loss": 2.7139, "step": 123 }, { "epoch": 0.050365556458164096, "grad_norm": 0.0, "learning_rate": 1.9978373585918817e-05, "loss": 2.6816, "step": 124 }, { "epoch": 0.05077173030056864, "grad_norm": 0.0, "learning_rate": 1.9977500206571766e-05, "loss": 2.6698, "step": 125 }, { "epoch": 0.05117790414297319, "grad_norm": 0.0, "learning_rate": 1.9976609558797545e-05, "loss": 2.697, "step": 126 }, { "epoch": 0.051584077985377744, "grad_norm": 0.0, "learning_rate": 1.997570164413763e-05, "loss": 2.7854, "step": 127 }, { "epoch": 0.05199025182778229, "grad_norm": 0.0, "learning_rate": 1.9974776464163387e-05, "loss": 2.7656, "step": 128 }, { "epoch": 0.05239642567018684, "grad_norm": 0.0, "learning_rate": 1.9973834020476057e-05, "loss": 2.6282, "step": 129 }, { "epoch": 0.05280259951259139, "grad_norm": 0.0, "learning_rate": 1.9972874314706755e-05, "loss": 2.6657, "step": 130 }, { "epoch": 0.053208773354995936, "grad_norm": 0.0, "learning_rate": 1.9971897348516486e-05, "loss": 2.5828, "step": 131 }, { "epoch": 0.05361494719740049, "grad_norm": 0.0, "learning_rate": 1.997090312359612e-05, "loss": 2.7181, "step": 132 }, { "epoch": 0.05402112103980504, "grad_norm": 0.0, "learning_rate": 1.9969891641666403e-05, "loss": 2.6102, "step": 133 }, { "epoch": 0.054427294882209584, "grad_norm": 0.0, "learning_rate": 1.9968862904477936e-05, "loss": 2.7449, "step": 134 }, { "epoch": 0.054833468724614136, "grad_norm": 0.0, "learning_rate": 1.996781691381119e-05, "loss": 2.5124, "step": 135 }, { "epoch": 0.05523964256701868, "grad_norm": 0.0, "learning_rate": 1.9966753671476514e-05, "loss": 2.7268, "step": 136 }, { "epoch": 0.05564581640942323, "grad_norm": 0.0, "learning_rate": 1.9965673179314087e-05, "loss": 2.5129, "step": 137 }, { "epoch": 0.056051990251827784, "grad_norm": 0.0, "learning_rate": 1.9964575439193966e-05, "loss": 2.7181, "step": 138 }, { "epoch": 0.05645816409423233, "grad_norm": 0.0, "learning_rate": 1.9963460453016043e-05, "loss": 2.6967, "step": 139 }, { "epoch": 0.05686433793663688, "grad_norm": 0.0, "learning_rate": 1.996232822271007e-05, "loss": 2.8036, "step": 140 }, { "epoch": 0.05727051177904143, "grad_norm": 0.0, "learning_rate": 1.9961178750235636e-05, "loss": 2.618, "step": 141 }, { "epoch": 0.057676685621445976, "grad_norm": 0.0, "learning_rate": 1.996001203758218e-05, "loss": 2.5764, "step": 142 }, { "epoch": 0.05808285946385053, "grad_norm": 0.0, "learning_rate": 1.995882808676897e-05, "loss": 2.6804, "step": 143 }, { "epoch": 0.05848903330625508, "grad_norm": 0.0, "learning_rate": 1.995762689984512e-05, "loss": 2.5402, "step": 144 }, { "epoch": 0.058895207148659624, "grad_norm": 0.0, "learning_rate": 1.9956408478889558e-05, "loss": 2.6001, "step": 145 }, { "epoch": 0.059301380991064176, "grad_norm": 0.0, "learning_rate": 1.995517282601106e-05, "loss": 2.6716, "step": 146 }, { "epoch": 0.05970755483346873, "grad_norm": 0.0, "learning_rate": 1.995391994334821e-05, "loss": 2.6504, "step": 147 }, { "epoch": 0.06011372867587327, "grad_norm": 0.0, "learning_rate": 1.9952649833069423e-05, "loss": 2.6441, "step": 148 }, { "epoch": 0.060519902518277824, "grad_norm": 0.0, "learning_rate": 1.995136249737292e-05, "loss": 2.6496, "step": 149 }, { "epoch": 0.060926076360682375, "grad_norm": 0.0, "learning_rate": 1.9950057938486745e-05, "loss": 2.6008, "step": 150 }, { "epoch": 0.06133225020308692, "grad_norm": 0.0, "learning_rate": 1.9948736158668745e-05, "loss": 2.6165, "step": 151 }, { "epoch": 0.06173842404549147, "grad_norm": 0.0, "learning_rate": 1.994739716020657e-05, "loss": 2.7585, "step": 152 }, { "epoch": 0.062144597887896016, "grad_norm": 0.0, "learning_rate": 1.994604094541768e-05, "loss": 2.5329, "step": 153 }, { "epoch": 0.06255077173030058, "grad_norm": 0.0, "learning_rate": 1.994466751664932e-05, "loss": 2.7699, "step": 154 }, { "epoch": 0.06295694557270512, "grad_norm": 0.0, "learning_rate": 1.9943276876278532e-05, "loss": 2.6047, "step": 155 }, { "epoch": 0.06336311941510966, "grad_norm": 0.0, "learning_rate": 1.9941869026712158e-05, "loss": 2.4986, "step": 156 }, { "epoch": 0.06376929325751422, "grad_norm": 0.0, "learning_rate": 1.9940443970386807e-05, "loss": 2.6596, "step": 157 }, { "epoch": 0.06417546709991877, "grad_norm": 0.0, "learning_rate": 1.993900170976888e-05, "loss": 2.7416, "step": 158 }, { "epoch": 0.06458164094232331, "grad_norm": 0.0, "learning_rate": 1.9937542247354547e-05, "loss": 2.69, "step": 159 }, { "epoch": 0.06498781478472786, "grad_norm": 0.0, "learning_rate": 1.9936065585669762e-05, "loss": 2.5718, "step": 160 }, { "epoch": 0.06539398862713242, "grad_norm": 0.0, "learning_rate": 1.9934571727270225e-05, "loss": 2.6647, "step": 161 }, { "epoch": 0.06580016246953696, "grad_norm": 0.0, "learning_rate": 1.9933060674741422e-05, "loss": 2.6567, "step": 162 }, { "epoch": 0.0662063363119415, "grad_norm": 0.0, "learning_rate": 1.9931532430698583e-05, "loss": 2.6924, "step": 163 }, { "epoch": 0.06661251015434606, "grad_norm": 0.0, "learning_rate": 1.9929986997786698e-05, "loss": 2.6108, "step": 164 }, { "epoch": 0.06701868399675061, "grad_norm": 0.0, "learning_rate": 1.9928424378680505e-05, "loss": 2.5575, "step": 165 }, { "epoch": 0.06742485783915515, "grad_norm": 0.0, "learning_rate": 1.9926844576084483e-05, "loss": 2.6245, "step": 166 }, { "epoch": 0.06783103168155971, "grad_norm": 0.0, "learning_rate": 1.9925247592732856e-05, "loss": 2.6521, "step": 167 }, { "epoch": 0.06823720552396426, "grad_norm": 0.0, "learning_rate": 1.9923633431389584e-05, "loss": 2.5141, "step": 168 }, { "epoch": 0.0686433793663688, "grad_norm": 0.0, "learning_rate": 1.9922002094848353e-05, "loss": 2.658, "step": 169 }, { "epoch": 0.06904955320877336, "grad_norm": 0.0, "learning_rate": 1.992035358593258e-05, "loss": 2.5525, "step": 170 }, { "epoch": 0.0694557270511779, "grad_norm": 0.0, "learning_rate": 1.9918687907495396e-05, "loss": 2.6025, "step": 171 }, { "epoch": 0.06986190089358245, "grad_norm": 0.0, "learning_rate": 1.9917005062419657e-05, "loss": 2.572, "step": 172 }, { "epoch": 0.07026807473598701, "grad_norm": 0.0, "learning_rate": 1.991530505361792e-05, "loss": 2.6183, "step": 173 }, { "epoch": 0.07067424857839155, "grad_norm": 0.0, "learning_rate": 1.991358788403246e-05, "loss": 2.5517, "step": 174 }, { "epoch": 0.0710804224207961, "grad_norm": 0.0, "learning_rate": 1.991185355663524e-05, "loss": 2.6307, "step": 175 }, { "epoch": 0.07148659626320066, "grad_norm": 0.0, "learning_rate": 1.991010207442792e-05, "loss": 2.5407, "step": 176 }, { "epoch": 0.0718927701056052, "grad_norm": 0.0, "learning_rate": 1.9908333440441866e-05, "loss": 2.6134, "step": 177 }, { "epoch": 0.07229894394800974, "grad_norm": 0.0, "learning_rate": 1.990654765773811e-05, "loss": 2.6156, "step": 178 }, { "epoch": 0.0727051177904143, "grad_norm": 0.0, "learning_rate": 1.9904744729407376e-05, "loss": 2.6781, "step": 179 }, { "epoch": 0.07311129163281885, "grad_norm": 0.0, "learning_rate": 1.9902924658570052e-05, "loss": 2.4748, "step": 180 }, { "epoch": 0.07351746547522339, "grad_norm": 0.0, "learning_rate": 1.990108744837621e-05, "loss": 2.6295, "step": 181 }, { "epoch": 0.07392363931762795, "grad_norm": 0.0, "learning_rate": 1.9899233102005573e-05, "loss": 2.4331, "step": 182 }, { "epoch": 0.0743298131600325, "grad_norm": 0.0, "learning_rate": 1.9897361622667527e-05, "loss": 2.6205, "step": 183 }, { "epoch": 0.07473598700243704, "grad_norm": 0.0, "learning_rate": 1.989547301360111e-05, "loss": 2.5736, "step": 184 }, { "epoch": 0.0751421608448416, "grad_norm": 0.0, "learning_rate": 1.9893567278075006e-05, "loss": 2.6649, "step": 185 }, { "epoch": 0.07554833468724614, "grad_norm": 0.0, "learning_rate": 1.9891644419387545e-05, "loss": 2.5228, "step": 186 }, { "epoch": 0.07595450852965069, "grad_norm": 0.0, "learning_rate": 1.9889704440866688e-05, "loss": 2.6053, "step": 187 }, { "epoch": 0.07636068237205523, "grad_norm": 0.0, "learning_rate": 1.988774734587003e-05, "loss": 2.4187, "step": 188 }, { "epoch": 0.07676685621445979, "grad_norm": 0.0, "learning_rate": 1.9885773137784777e-05, "loss": 2.664, "step": 189 }, { "epoch": 0.07717303005686434, "grad_norm": 0.0, "learning_rate": 1.9883781820027777e-05, "loss": 2.5969, "step": 190 }, { "epoch": 0.07757920389926888, "grad_norm": 0.0, "learning_rate": 1.9881773396045467e-05, "loss": 2.6448, "step": 191 }, { "epoch": 0.07798537774167344, "grad_norm": 0.0, "learning_rate": 1.987974786931391e-05, "loss": 2.6517, "step": 192 }, { "epoch": 0.07839155158407798, "grad_norm": 0.0, "learning_rate": 1.987770524333875e-05, "loss": 2.7177, "step": 193 }, { "epoch": 0.07879772542648253, "grad_norm": 0.0, "learning_rate": 1.987564552165524e-05, "loss": 2.712, "step": 194 }, { "epoch": 0.07920389926888709, "grad_norm": 0.0, "learning_rate": 1.9873568707828213e-05, "loss": 2.6228, "step": 195 }, { "epoch": 0.07961007311129163, "grad_norm": 0.0, "learning_rate": 1.9871474805452093e-05, "loss": 2.5829, "step": 196 }, { "epoch": 0.08001624695369618, "grad_norm": 0.0, "learning_rate": 1.9869363818150867e-05, "loss": 2.6179, "step": 197 }, { "epoch": 0.08042242079610074, "grad_norm": 0.0, "learning_rate": 1.9867235749578108e-05, "loss": 2.4955, "step": 198 }, { "epoch": 0.08082859463850528, "grad_norm": 0.0, "learning_rate": 1.9865090603416934e-05, "loss": 2.61, "step": 199 }, { "epoch": 0.08123476848090982, "grad_norm": 0.0, "learning_rate": 1.986292838338003e-05, "loss": 2.6388, "step": 200 }, { "epoch": 0.08164094232331438, "grad_norm": 0.0, "learning_rate": 1.9860749093209632e-05, "loss": 2.5935, "step": 201 }, { "epoch": 0.08204711616571893, "grad_norm": 0.0, "learning_rate": 1.9858552736677516e-05, "loss": 2.5476, "step": 202 }, { "epoch": 0.08245329000812347, "grad_norm": 0.0, "learning_rate": 1.9856339317584997e-05, "loss": 2.5109, "step": 203 }, { "epoch": 0.08285946385052803, "grad_norm": 0.0, "learning_rate": 1.9854108839762923e-05, "loss": 2.4542, "step": 204 }, { "epoch": 0.08326563769293258, "grad_norm": 0.0, "learning_rate": 1.985186130707166e-05, "loss": 2.5878, "step": 205 }, { "epoch": 0.08367181153533712, "grad_norm": 0.0, "learning_rate": 1.984959672340111e-05, "loss": 2.6267, "step": 206 }, { "epoch": 0.08407798537774168, "grad_norm": 0.0, "learning_rate": 1.984731509267065e-05, "loss": 2.6271, "step": 207 }, { "epoch": 0.08448415922014622, "grad_norm": 0.0, "learning_rate": 1.9845016418829194e-05, "loss": 2.611, "step": 208 }, { "epoch": 0.08489033306255077, "grad_norm": 0.0, "learning_rate": 1.984270070585514e-05, "loss": 2.5934, "step": 209 }, { "epoch": 0.08529650690495533, "grad_norm": 0.0, "learning_rate": 1.984036795775638e-05, "loss": 2.5147, "step": 210 }, { "epoch": 0.08570268074735987, "grad_norm": 0.0, "learning_rate": 1.983801817857028e-05, "loss": 2.6927, "step": 211 }, { "epoch": 0.08610885458976442, "grad_norm": 0.0, "learning_rate": 1.98356513723637e-05, "loss": 2.4915, "step": 212 }, { "epoch": 0.08651502843216897, "grad_norm": 0.0, "learning_rate": 1.983326754323295e-05, "loss": 2.4961, "step": 213 }, { "epoch": 0.08692120227457352, "grad_norm": 0.0, "learning_rate": 1.9830866695303817e-05, "loss": 2.4596, "step": 214 }, { "epoch": 0.08732737611697806, "grad_norm": 0.0, "learning_rate": 1.982844883273153e-05, "loss": 2.5278, "step": 215 }, { "epoch": 0.08773354995938261, "grad_norm": 0.0, "learning_rate": 1.9826013959700776e-05, "loss": 2.6371, "step": 216 }, { "epoch": 0.08813972380178717, "grad_norm": 0.0, "learning_rate": 1.9823562080425684e-05, "loss": 2.573, "step": 217 }, { "epoch": 0.08854589764419171, "grad_norm": 0.0, "learning_rate": 1.9821093199149806e-05, "loss": 2.5947, "step": 218 }, { "epoch": 0.08895207148659626, "grad_norm": 0.0, "learning_rate": 1.9818607320146127e-05, "loss": 2.7128, "step": 219 }, { "epoch": 0.08935824532900082, "grad_norm": 0.0, "learning_rate": 1.9816104447717048e-05, "loss": 2.5598, "step": 220 }, { "epoch": 0.08976441917140536, "grad_norm": 0.0, "learning_rate": 1.9813584586194388e-05, "loss": 2.6209, "step": 221 }, { "epoch": 0.0901705930138099, "grad_norm": 0.0, "learning_rate": 1.981104773993936e-05, "loss": 2.5921, "step": 222 }, { "epoch": 0.09057676685621446, "grad_norm": 0.0, "learning_rate": 1.9808493913342575e-05, "loss": 2.5029, "step": 223 }, { "epoch": 0.09098294069861901, "grad_norm": 0.0, "learning_rate": 1.980592311082404e-05, "loss": 2.6118, "step": 224 }, { "epoch": 0.09138911454102355, "grad_norm": 0.0, "learning_rate": 1.9803335336833137e-05, "loss": 2.604, "step": 225 }, { "epoch": 0.09179528838342811, "grad_norm": 0.0, "learning_rate": 1.980073059584862e-05, "loss": 2.4417, "step": 226 }, { "epoch": 0.09220146222583266, "grad_norm": 0.0, "learning_rate": 1.979810889237861e-05, "loss": 2.6098, "step": 227 }, { "epoch": 0.0926076360682372, "grad_norm": 0.0, "learning_rate": 1.9795470230960584e-05, "loss": 2.529, "step": 228 }, { "epoch": 0.09301380991064176, "grad_norm": 0.0, "learning_rate": 1.9792814616161374e-05, "loss": 2.5323, "step": 229 }, { "epoch": 0.0934199837530463, "grad_norm": 0.0, "learning_rate": 1.9790142052577148e-05, "loss": 2.5105, "step": 230 }, { "epoch": 0.09382615759545085, "grad_norm": 0.0, "learning_rate": 1.978745254483341e-05, "loss": 2.5626, "step": 231 }, { "epoch": 0.09423233143785541, "grad_norm": 0.0, "learning_rate": 1.9784746097584992e-05, "loss": 2.6687, "step": 232 }, { "epoch": 0.09463850528025995, "grad_norm": 0.0, "learning_rate": 1.9782022715516044e-05, "loss": 2.5656, "step": 233 }, { "epoch": 0.0950446791226645, "grad_norm": 0.0, "learning_rate": 1.977928240334002e-05, "loss": 2.5871, "step": 234 }, { "epoch": 0.09545085296506906, "grad_norm": 0.0, "learning_rate": 1.9776525165799677e-05, "loss": 2.4851, "step": 235 }, { "epoch": 0.0958570268074736, "grad_norm": 0.0, "learning_rate": 1.9773751007667074e-05, "loss": 2.5824, "step": 236 }, { "epoch": 0.09626320064987814, "grad_norm": 0.0, "learning_rate": 1.9770959933743543e-05, "loss": 2.5024, "step": 237 }, { "epoch": 0.0966693744922827, "grad_norm": 0.0, "learning_rate": 1.9768151948859705e-05, "loss": 2.5092, "step": 238 }, { "epoch": 0.09707554833468725, "grad_norm": 0.0, "learning_rate": 1.9765327057875433e-05, "loss": 2.4837, "step": 239 }, { "epoch": 0.09748172217709179, "grad_norm": 0.0, "learning_rate": 1.976248526567988e-05, "loss": 2.4404, "step": 240 }, { "epoch": 0.09788789601949635, "grad_norm": 0.0, "learning_rate": 1.9759626577191436e-05, "loss": 2.6692, "step": 241 }, { "epoch": 0.0982940698619009, "grad_norm": 0.0, "learning_rate": 1.9756750997357738e-05, "loss": 2.5357, "step": 242 }, { "epoch": 0.09870024370430544, "grad_norm": 0.0, "learning_rate": 1.9753858531155658e-05, "loss": 2.4505, "step": 243 }, { "epoch": 0.09910641754670999, "grad_norm": 0.0, "learning_rate": 1.9750949183591296e-05, "loss": 2.3494, "step": 244 }, { "epoch": 0.09951259138911454, "grad_norm": 0.0, "learning_rate": 1.9748022959699962e-05, "loss": 2.5109, "step": 245 }, { "epoch": 0.09991876523151909, "grad_norm": 0.0, "learning_rate": 1.9745079864546184e-05, "loss": 2.5236, "step": 246 }, { "epoch": 0.10032493907392363, "grad_norm": 0.0, "learning_rate": 1.9742119903223687e-05, "loss": 2.4757, "step": 247 }, { "epoch": 0.10073111291632819, "grad_norm": 0.0, "learning_rate": 1.973914308085538e-05, "loss": 2.536, "step": 248 }, { "epoch": 0.10113728675873274, "grad_norm": 0.0, "learning_rate": 1.973614940259336e-05, "loss": 2.6013, "step": 249 }, { "epoch": 0.10154346060113728, "grad_norm": 0.0, "learning_rate": 1.97331388736189e-05, "loss": 2.4722, "step": 250 }, { "epoch": 0.10194963444354184, "grad_norm": 0.0, "learning_rate": 1.973011149914243e-05, "loss": 2.5483, "step": 251 }, { "epoch": 0.10235580828594638, "grad_norm": 0.0, "learning_rate": 1.972706728440354e-05, "loss": 2.6347, "step": 252 }, { "epoch": 0.10276198212835093, "grad_norm": 0.0, "learning_rate": 1.9724006234670962e-05, "loss": 2.5082, "step": 253 }, { "epoch": 0.10316815597075549, "grad_norm": 0.0, "learning_rate": 1.972092835524257e-05, "loss": 2.3805, "step": 254 }, { "epoch": 0.10357432981316003, "grad_norm": 0.0, "learning_rate": 1.9717833651445365e-05, "loss": 2.4964, "step": 255 }, { "epoch": 0.10398050365556458, "grad_norm": 0.0, "learning_rate": 1.9714722128635465e-05, "loss": 2.5169, "step": 256 }, { "epoch": 0.10438667749796914, "grad_norm": 0.0, "learning_rate": 1.9711593792198088e-05, "loss": 2.5164, "step": 257 }, { "epoch": 0.10479285134037368, "grad_norm": 0.0, "learning_rate": 1.9708448647547575e-05, "loss": 2.5955, "step": 258 }, { "epoch": 0.10519902518277822, "grad_norm": 0.0, "learning_rate": 1.970528670012733e-05, "loss": 2.4372, "step": 259 }, { "epoch": 0.10560519902518278, "grad_norm": 0.0, "learning_rate": 1.9702107955409862e-05, "loss": 2.4362, "step": 260 }, { "epoch": 0.10601137286758733, "grad_norm": 0.0, "learning_rate": 1.969891241889674e-05, "loss": 2.3709, "step": 261 }, { "epoch": 0.10641754670999187, "grad_norm": 0.0, "learning_rate": 1.9695700096118594e-05, "loss": 2.4398, "step": 262 }, { "epoch": 0.10682372055239643, "grad_norm": 0.0, "learning_rate": 1.9692470992635112e-05, "loss": 2.5488, "step": 263 }, { "epoch": 0.10722989439480098, "grad_norm": 0.0, "learning_rate": 1.968922511403502e-05, "loss": 2.4555, "step": 264 }, { "epoch": 0.10763606823720552, "grad_norm": 0.0, "learning_rate": 1.9685962465936077e-05, "loss": 2.4706, "step": 265 }, { "epoch": 0.10804224207961008, "grad_norm": 0.0, "learning_rate": 1.9682683053985073e-05, "loss": 2.5327, "step": 266 }, { "epoch": 0.10844841592201462, "grad_norm": 0.0, "learning_rate": 1.9679386883857808e-05, "loss": 2.4217, "step": 267 }, { "epoch": 0.10885458976441917, "grad_norm": 0.0, "learning_rate": 1.967607396125908e-05, "loss": 2.5391, "step": 268 }, { "epoch": 0.10926076360682373, "grad_norm": 0.0, "learning_rate": 1.967274429192269e-05, "loss": 2.5734, "step": 269 }, { "epoch": 0.10966693744922827, "grad_norm": 0.0, "learning_rate": 1.966939788161142e-05, "loss": 2.5788, "step": 270 }, { "epoch": 0.11007311129163282, "grad_norm": 0.0, "learning_rate": 1.9666034736117023e-05, "loss": 2.4827, "step": 271 }, { "epoch": 0.11047928513403736, "grad_norm": 0.0, "learning_rate": 1.966265486126022e-05, "loss": 2.4844, "step": 272 }, { "epoch": 0.11088545897644192, "grad_norm": 0.0, "learning_rate": 1.9659258262890683e-05, "loss": 2.3344, "step": 273 }, { "epoch": 0.11129163281884646, "grad_norm": 0.0, "learning_rate": 1.9655844946887035e-05, "loss": 2.4904, "step": 274 }, { "epoch": 0.11169780666125101, "grad_norm": 0.0, "learning_rate": 1.9652414919156824e-05, "loss": 2.4608, "step": 275 }, { "epoch": 0.11210398050365557, "grad_norm": 0.0, "learning_rate": 1.964896818563652e-05, "loss": 2.5146, "step": 276 }, { "epoch": 0.11251015434606011, "grad_norm": 0.0, "learning_rate": 1.9645504752291524e-05, "loss": 2.2942, "step": 277 }, { "epoch": 0.11291632818846466, "grad_norm": 0.0, "learning_rate": 1.9642024625116117e-05, "loss": 2.3531, "step": 278 }, { "epoch": 0.11332250203086922, "grad_norm": 0.0, "learning_rate": 1.963852781013349e-05, "loss": 2.3435, "step": 279 }, { "epoch": 0.11372867587327376, "grad_norm": 0.0, "learning_rate": 1.9635014313395708e-05, "loss": 2.6081, "step": 280 }, { "epoch": 0.1141348497156783, "grad_norm": 0.0, "learning_rate": 1.9631484140983704e-05, "loss": 2.5363, "step": 281 }, { "epoch": 0.11454102355808286, "grad_norm": 0.0, "learning_rate": 1.9627937299007286e-05, "loss": 2.5373, "step": 282 }, { "epoch": 0.11494719740048741, "grad_norm": 0.0, "learning_rate": 1.9624373793605098e-05, "loss": 2.5559, "step": 283 }, { "epoch": 0.11535337124289195, "grad_norm": 0.0, "learning_rate": 1.9620793630944632e-05, "loss": 2.6517, "step": 284 }, { "epoch": 0.11575954508529651, "grad_norm": 0.0, "learning_rate": 1.9617196817222207e-05, "loss": 2.4819, "step": 285 }, { "epoch": 0.11616571892770106, "grad_norm": 0.0, "learning_rate": 1.961358335866296e-05, "loss": 2.4354, "step": 286 }, { "epoch": 0.1165718927701056, "grad_norm": 0.0, "learning_rate": 1.9609953261520838e-05, "loss": 2.4315, "step": 287 }, { "epoch": 0.11697806661251016, "grad_norm": 0.0, "learning_rate": 1.960630653207858e-05, "loss": 2.4082, "step": 288 }, { "epoch": 0.1173842404549147, "grad_norm": 0.0, "learning_rate": 1.9602643176647727e-05, "loss": 2.3746, "step": 289 }, { "epoch": 0.11779041429731925, "grad_norm": 0.0, "learning_rate": 1.959896320156857e-05, "loss": 2.4664, "step": 290 }, { "epoch": 0.11819658813972381, "grad_norm": 0.0, "learning_rate": 1.9595266613210185e-05, "loss": 2.4704, "step": 291 }, { "epoch": 0.11860276198212835, "grad_norm": 0.0, "learning_rate": 1.9591553417970392e-05, "loss": 2.3398, "step": 292 }, { "epoch": 0.1190089358245329, "grad_norm": 0.0, "learning_rate": 1.958782362227575e-05, "loss": 2.55, "step": 293 }, { "epoch": 0.11941510966693746, "grad_norm": 0.0, "learning_rate": 1.958407723258156e-05, "loss": 2.3901, "step": 294 }, { "epoch": 0.119821283509342, "grad_norm": 0.0, "learning_rate": 1.9580314255371836e-05, "loss": 2.4696, "step": 295 }, { "epoch": 0.12022745735174654, "grad_norm": 0.0, "learning_rate": 1.9576534697159298e-05, "loss": 2.5475, "step": 296 }, { "epoch": 0.1206336311941511, "grad_norm": 0.0, "learning_rate": 1.957273856448537e-05, "loss": 2.4814, "step": 297 }, { "epoch": 0.12103980503655565, "grad_norm": 0.0, "learning_rate": 1.9568925863920155e-05, "loss": 2.4406, "step": 298 }, { "epoch": 0.12144597887896019, "grad_norm": 0.0, "learning_rate": 1.9565096602062436e-05, "loss": 2.4638, "step": 299 }, { "epoch": 0.12185215272136475, "grad_norm": 0.0, "learning_rate": 1.9561250785539655e-05, "loss": 2.2899, "step": 300 }, { "epoch": 0.1222583265637693, "grad_norm": 0.0, "learning_rate": 1.955738842100792e-05, "loss": 2.4481, "step": 301 }, { "epoch": 0.12266450040617384, "grad_norm": 0.0, "learning_rate": 1.955350951515195e-05, "loss": 2.5665, "step": 302 }, { "epoch": 0.12307067424857839, "grad_norm": 0.0, "learning_rate": 1.9549614074685123e-05, "loss": 2.395, "step": 303 }, { "epoch": 0.12347684809098294, "grad_norm": 0.0, "learning_rate": 1.9545702106349415e-05, "loss": 2.5631, "step": 304 }, { "epoch": 0.12388302193338749, "grad_norm": 0.0, "learning_rate": 1.954177361691542e-05, "loss": 2.5223, "step": 305 }, { "epoch": 0.12428919577579203, "grad_norm": 0.0, "learning_rate": 1.9537828613182314e-05, "loss": 2.4297, "step": 306 }, { "epoch": 0.12469536961819659, "grad_norm": 0.0, "learning_rate": 1.953386710197786e-05, "loss": 2.4152, "step": 307 }, { "epoch": 0.12510154346060115, "grad_norm": 0.0, "learning_rate": 1.9529889090158394e-05, "loss": 2.2819, "step": 308 }, { "epoch": 0.1255077173030057, "grad_norm": 0.0, "learning_rate": 1.9525894584608807e-05, "loss": 2.3583, "step": 309 }, { "epoch": 0.12591389114541024, "grad_norm": 0.0, "learning_rate": 1.9521883592242537e-05, "loss": 2.5911, "step": 310 }, { "epoch": 0.12632006498781478, "grad_norm": 0.0, "learning_rate": 1.9517856120001558e-05, "loss": 2.5261, "step": 311 }, { "epoch": 0.12672623883021933, "grad_norm": 0.0, "learning_rate": 1.9513812174856358e-05, "loss": 2.6172, "step": 312 }, { "epoch": 0.12713241267262387, "grad_norm": 0.0, "learning_rate": 1.950975176380595e-05, "loss": 2.3406, "step": 313 }, { "epoch": 0.12753858651502845, "grad_norm": 0.0, "learning_rate": 1.950567489387783e-05, "loss": 2.4539, "step": 314 }, { "epoch": 0.127944760357433, "grad_norm": 0.0, "learning_rate": 1.9501581572127992e-05, "loss": 2.443, "step": 315 }, { "epoch": 0.12835093419983754, "grad_norm": 0.0, "learning_rate": 1.94974718056409e-05, "loss": 2.5373, "step": 316 }, { "epoch": 0.12875710804224208, "grad_norm": 0.0, "learning_rate": 1.949334560152947e-05, "loss": 2.3649, "step": 317 }, { "epoch": 0.12916328188464662, "grad_norm": 0.0, "learning_rate": 1.9489202966935084e-05, "loss": 2.3887, "step": 318 }, { "epoch": 0.12956945572705117, "grad_norm": 0.0, "learning_rate": 1.9485043909027544e-05, "loss": 2.4963, "step": 319 }, { "epoch": 0.12997562956945571, "grad_norm": 0.0, "learning_rate": 1.9480868435005098e-05, "loss": 2.4733, "step": 320 }, { "epoch": 0.1303818034118603, "grad_norm": 0.0, "learning_rate": 1.947667655209438e-05, "loss": 2.4964, "step": 321 }, { "epoch": 0.13078797725426483, "grad_norm": 0.0, "learning_rate": 1.947246826755044e-05, "loss": 2.4526, "step": 322 }, { "epoch": 0.13119415109666938, "grad_norm": 0.0, "learning_rate": 1.9468243588656716e-05, "loss": 2.467, "step": 323 }, { "epoch": 0.13160032493907392, "grad_norm": 0.0, "learning_rate": 1.9464002522725005e-05, "loss": 2.4461, "step": 324 }, { "epoch": 0.13200649878147847, "grad_norm": 0.0, "learning_rate": 1.9459745077095484e-05, "loss": 2.3733, "step": 325 }, { "epoch": 0.132412672623883, "grad_norm": 0.0, "learning_rate": 1.945547125913667e-05, "loss": 2.322, "step": 326 }, { "epoch": 0.13281884646628758, "grad_norm": 0.0, "learning_rate": 1.945118107624542e-05, "loss": 2.478, "step": 327 }, { "epoch": 0.13322502030869213, "grad_norm": 0.0, "learning_rate": 1.9446874535846904e-05, "loss": 2.4545, "step": 328 }, { "epoch": 0.13363119415109667, "grad_norm": 0.0, "learning_rate": 1.9442551645394615e-05, "loss": 2.383, "step": 329 }, { "epoch": 0.13403736799350122, "grad_norm": 0.0, "learning_rate": 1.943821241237034e-05, "loss": 2.4319, "step": 330 }, { "epoch": 0.13444354183590576, "grad_norm": 0.0, "learning_rate": 1.9433856844284142e-05, "loss": 2.3597, "step": 331 }, { "epoch": 0.1348497156783103, "grad_norm": 0.0, "learning_rate": 1.9429484948674374e-05, "loss": 2.3389, "step": 332 }, { "epoch": 0.13525588952071488, "grad_norm": 0.0, "learning_rate": 1.9425096733107626e-05, "loss": 2.5248, "step": 333 }, { "epoch": 0.13566206336311942, "grad_norm": 0.0, "learning_rate": 1.9420692205178753e-05, "loss": 2.3457, "step": 334 }, { "epoch": 0.13606823720552397, "grad_norm": 0.0, "learning_rate": 1.941627137251083e-05, "loss": 2.4781, "step": 335 }, { "epoch": 0.1364744110479285, "grad_norm": 0.0, "learning_rate": 1.9411834242755156e-05, "loss": 2.332, "step": 336 }, { "epoch": 0.13688058489033306, "grad_norm": 0.0, "learning_rate": 1.9407380823591234e-05, "loss": 2.3431, "step": 337 }, { "epoch": 0.1372867587327376, "grad_norm": 0.0, "learning_rate": 1.9402911122726756e-05, "loss": 2.5436, "step": 338 }, { "epoch": 0.13769293257514217, "grad_norm": 0.0, "learning_rate": 1.939842514789761e-05, "loss": 2.3648, "step": 339 }, { "epoch": 0.13809910641754672, "grad_norm": 0.0, "learning_rate": 1.9393922906867827e-05, "loss": 2.349, "step": 340 }, { "epoch": 0.13850528025995126, "grad_norm": 0.0, "learning_rate": 1.9389404407429602e-05, "loss": 2.5552, "step": 341 }, { "epoch": 0.1389114541023558, "grad_norm": 0.0, "learning_rate": 1.9384869657403277e-05, "loss": 2.381, "step": 342 }, { "epoch": 0.13931762794476035, "grad_norm": 0.0, "learning_rate": 1.9380318664637305e-05, "loss": 2.4306, "step": 343 }, { "epoch": 0.1397238017871649, "grad_norm": 0.0, "learning_rate": 1.9375751437008253e-05, "loss": 2.4738, "step": 344 }, { "epoch": 0.14012997562956944, "grad_norm": 0.0, "learning_rate": 1.9371167982420794e-05, "loss": 2.6753, "step": 345 }, { "epoch": 0.14053614947197401, "grad_norm": 0.0, "learning_rate": 1.9366568308807685e-05, "loss": 2.3961, "step": 346 }, { "epoch": 0.14094232331437856, "grad_norm": 0.0, "learning_rate": 1.936195242412975e-05, "loss": 2.3725, "step": 347 }, { "epoch": 0.1413484971567831, "grad_norm": 0.0, "learning_rate": 1.9357320336375865e-05, "loss": 2.401, "step": 348 }, { "epoch": 0.14175467099918765, "grad_norm": 0.0, "learning_rate": 1.9352672053562957e-05, "loss": 2.4484, "step": 349 }, { "epoch": 0.1421608448415922, "grad_norm": 0.0, "learning_rate": 1.9348007583735985e-05, "loss": 2.4644, "step": 350 }, { "epoch": 0.14256701868399674, "grad_norm": 0.0, "learning_rate": 1.9343326934967913e-05, "loss": 2.3558, "step": 351 }, { "epoch": 0.1429731925264013, "grad_norm": 0.0, "learning_rate": 1.9338630115359714e-05, "loss": 2.4205, "step": 352 }, { "epoch": 0.14337936636880586, "grad_norm": 0.0, "learning_rate": 1.933391713304035e-05, "loss": 2.4525, "step": 353 }, { "epoch": 0.1437855402112104, "grad_norm": 0.0, "learning_rate": 1.9329187996166747e-05, "loss": 2.4579, "step": 354 }, { "epoch": 0.14419171405361494, "grad_norm": 0.0, "learning_rate": 1.9324442712923802e-05, "loss": 2.4017, "step": 355 }, { "epoch": 0.1445978878960195, "grad_norm": 0.0, "learning_rate": 1.931968129152435e-05, "loss": 2.4248, "step": 356 }, { "epoch": 0.14500406173842403, "grad_norm": 0.0, "learning_rate": 1.9314903740209155e-05, "loss": 2.3994, "step": 357 }, { "epoch": 0.1454102355808286, "grad_norm": 0.0, "learning_rate": 1.9310110067246905e-05, "loss": 2.4059, "step": 358 }, { "epoch": 0.14581640942323315, "grad_norm": 0.0, "learning_rate": 1.9305300280934187e-05, "loss": 2.3186, "step": 359 }, { "epoch": 0.1462225832656377, "grad_norm": 0.0, "learning_rate": 1.9300474389595472e-05, "loss": 2.3307, "step": 360 }, { "epoch": 0.14662875710804224, "grad_norm": 0.0, "learning_rate": 1.929563240158311e-05, "loss": 2.3616, "step": 361 }, { "epoch": 0.14703493095044678, "grad_norm": 0.0, "learning_rate": 1.9290774325277305e-05, "loss": 2.4012, "step": 362 }, { "epoch": 0.14744110479285133, "grad_norm": 0.0, "learning_rate": 1.9285900169086114e-05, "loss": 2.4284, "step": 363 }, { "epoch": 0.1478472786352559, "grad_norm": 0.0, "learning_rate": 1.9281009941445417e-05, "loss": 2.4652, "step": 364 }, { "epoch": 0.14825345247766045, "grad_norm": 0.0, "learning_rate": 1.927610365081891e-05, "loss": 2.4735, "step": 365 }, { "epoch": 0.148659626320065, "grad_norm": 0.0, "learning_rate": 1.9271181305698084e-05, "loss": 2.4312, "step": 366 }, { "epoch": 0.14906580016246954, "grad_norm": 0.0, "learning_rate": 1.9266242914602236e-05, "loss": 2.3812, "step": 367 }, { "epoch": 0.14947197400487408, "grad_norm": 0.0, "learning_rate": 1.9261288486078414e-05, "loss": 2.4233, "step": 368 }, { "epoch": 0.14987814784727863, "grad_norm": 0.0, "learning_rate": 1.925631802870143e-05, "loss": 2.4449, "step": 369 }, { "epoch": 0.1502843216896832, "grad_norm": 0.0, "learning_rate": 1.9251331551073843e-05, "loss": 2.4145, "step": 370 }, { "epoch": 0.15069049553208774, "grad_norm": 0.0, "learning_rate": 1.9246329061825926e-05, "loss": 2.4274, "step": 371 }, { "epoch": 0.1510966693744923, "grad_norm": 0.0, "learning_rate": 1.9241310569615673e-05, "loss": 2.4916, "step": 372 }, { "epoch": 0.15150284321689683, "grad_norm": 0.0, "learning_rate": 1.9236276083128778e-05, "loss": 2.3924, "step": 373 }, { "epoch": 0.15190901705930138, "grad_norm": 0.0, "learning_rate": 1.923122561107861e-05, "loss": 2.3765, "step": 374 }, { "epoch": 0.15231519090170592, "grad_norm": 0.0, "learning_rate": 1.9226159162206203e-05, "loss": 2.4051, "step": 375 }, { "epoch": 0.15272136474411047, "grad_norm": 0.0, "learning_rate": 1.9221076745280254e-05, "loss": 2.4943, "step": 376 }, { "epoch": 0.15312753858651504, "grad_norm": 0.0, "learning_rate": 1.9215978369097087e-05, "loss": 2.3159, "step": 377 }, { "epoch": 0.15353371242891958, "grad_norm": 0.0, "learning_rate": 1.9210864042480645e-05, "loss": 2.2422, "step": 378 }, { "epoch": 0.15393988627132413, "grad_norm": 0.0, "learning_rate": 1.9205733774282482e-05, "loss": 2.3656, "step": 379 }, { "epoch": 0.15434606011372867, "grad_norm": 0.0, "learning_rate": 1.9200587573381747e-05, "loss": 2.3676, "step": 380 }, { "epoch": 0.15475223395613322, "grad_norm": 0.0, "learning_rate": 1.9195425448685157e-05, "loss": 2.2896, "step": 381 }, { "epoch": 0.15515840779853776, "grad_norm": 0.0, "learning_rate": 1.9190247409126993e-05, "loss": 2.4015, "step": 382 }, { "epoch": 0.15556458164094233, "grad_norm": 0.0, "learning_rate": 1.9185053463669073e-05, "loss": 2.443, "step": 383 }, { "epoch": 0.15597075548334688, "grad_norm": 0.0, "learning_rate": 1.9179843621300757e-05, "loss": 2.5543, "step": 384 }, { "epoch": 0.15637692932575142, "grad_norm": 0.0, "learning_rate": 1.917461789103891e-05, "loss": 2.297, "step": 385 }, { "epoch": 0.15678310316815597, "grad_norm": 0.0, "learning_rate": 1.916937628192789e-05, "loss": 2.3869, "step": 386 }, { "epoch": 0.1571892770105605, "grad_norm": 0.0, "learning_rate": 1.9164118803039548e-05, "loss": 2.3473, "step": 387 }, { "epoch": 0.15759545085296506, "grad_norm": 0.0, "learning_rate": 1.9158845463473196e-05, "loss": 2.3561, "step": 388 }, { "epoch": 0.15800162469536963, "grad_norm": 0.0, "learning_rate": 1.9153556272355596e-05, "loss": 2.3772, "step": 389 }, { "epoch": 0.15840779853777417, "grad_norm": 0.0, "learning_rate": 1.9148251238840947e-05, "loss": 2.3861, "step": 390 }, { "epoch": 0.15881397238017872, "grad_norm": 0.0, "learning_rate": 1.9142930372110865e-05, "loss": 2.3979, "step": 391 }, { "epoch": 0.15922014622258326, "grad_norm": 0.0, "learning_rate": 1.913759368137437e-05, "loss": 2.3497, "step": 392 }, { "epoch": 0.1596263200649878, "grad_norm": 0.0, "learning_rate": 1.9132241175867867e-05, "loss": 2.3942, "step": 393 }, { "epoch": 0.16003249390739235, "grad_norm": 0.0, "learning_rate": 1.9126872864855142e-05, "loss": 2.3636, "step": 394 }, { "epoch": 0.16043866774979693, "grad_norm": 0.0, "learning_rate": 1.912148875762732e-05, "loss": 2.5085, "step": 395 }, { "epoch": 0.16084484159220147, "grad_norm": 0.0, "learning_rate": 1.9116088863502878e-05, "loss": 2.4302, "step": 396 }, { "epoch": 0.16125101543460602, "grad_norm": 0.0, "learning_rate": 1.9110673191827612e-05, "loss": 2.3496, "step": 397 }, { "epoch": 0.16165718927701056, "grad_norm": 0.0, "learning_rate": 1.9105241751974624e-05, "loss": 2.4599, "step": 398 }, { "epoch": 0.1620633631194151, "grad_norm": 0.0, "learning_rate": 1.9099794553344307e-05, "loss": 2.3191, "step": 399 }, { "epoch": 0.16246953696181965, "grad_norm": 0.0, "learning_rate": 1.9094331605364326e-05, "loss": 2.3583, "step": 400 }, { "epoch": 0.1628757108042242, "grad_norm": 0.0, "learning_rate": 1.908885291748961e-05, "loss": 2.2096, "step": 401 }, { "epoch": 0.16328188464662877, "grad_norm": 0.0, "learning_rate": 1.9083358499202323e-05, "loss": 2.493, "step": 402 }, { "epoch": 0.1636880584890333, "grad_norm": 0.0, "learning_rate": 1.9077848360011856e-05, "loss": 2.3626, "step": 403 }, { "epoch": 0.16409423233143786, "grad_norm": 0.0, "learning_rate": 1.9072322509454814e-05, "loss": 2.3202, "step": 404 }, { "epoch": 0.1645004061738424, "grad_norm": 0.0, "learning_rate": 1.906678095709499e-05, "loss": 2.4402, "step": 405 }, { "epoch": 0.16490658001624695, "grad_norm": 0.0, "learning_rate": 1.9061223712523352e-05, "loss": 2.3835, "step": 406 }, { "epoch": 0.1653127538586515, "grad_norm": 0.0, "learning_rate": 1.9055650785358023e-05, "loss": 2.2734, "step": 407 }, { "epoch": 0.16571892770105606, "grad_norm": 0.0, "learning_rate": 1.905006218524428e-05, "loss": 2.4199, "step": 408 }, { "epoch": 0.1661251015434606, "grad_norm": 0.0, "learning_rate": 1.9044457921854514e-05, "loss": 2.3536, "step": 409 }, { "epoch": 0.16653127538586515, "grad_norm": 0.0, "learning_rate": 1.903883800488824e-05, "loss": 2.3356, "step": 410 }, { "epoch": 0.1669374492282697, "grad_norm": 0.0, "learning_rate": 1.903320244407204e-05, "loss": 2.2875, "step": 411 }, { "epoch": 0.16734362307067424, "grad_norm": 0.0, "learning_rate": 1.9027551249159596e-05, "loss": 2.4932, "step": 412 }, { "epoch": 0.16774979691307879, "grad_norm": 0.0, "learning_rate": 1.9021884429931638e-05, "loss": 2.3694, "step": 413 }, { "epoch": 0.16815597075548336, "grad_norm": 0.0, "learning_rate": 1.9016201996195943e-05, "loss": 2.3417, "step": 414 }, { "epoch": 0.1685621445978879, "grad_norm": 0.0, "learning_rate": 1.90105039577873e-05, "loss": 2.442, "step": 415 }, { "epoch": 0.16896831844029245, "grad_norm": 0.0, "learning_rate": 1.900479032456752e-05, "loss": 2.3075, "step": 416 }, { "epoch": 0.169374492282697, "grad_norm": 0.0, "learning_rate": 1.8999061106425393e-05, "loss": 2.3224, "step": 417 }, { "epoch": 0.16978066612510154, "grad_norm": 0.0, "learning_rate": 1.8993316313276694e-05, "loss": 2.4144, "step": 418 }, { "epoch": 0.17018683996750608, "grad_norm": 0.0, "learning_rate": 1.8987555955064144e-05, "loss": 2.3506, "step": 419 }, { "epoch": 0.17059301380991065, "grad_norm": 0.0, "learning_rate": 1.898178004175741e-05, "loss": 2.3925, "step": 420 }, { "epoch": 0.1709991876523152, "grad_norm": 0.0, "learning_rate": 1.8975988583353078e-05, "loss": 2.4065, "step": 421 }, { "epoch": 0.17140536149471974, "grad_norm": 0.0, "learning_rate": 1.8970181589874637e-05, "loss": 2.404, "step": 422 }, { "epoch": 0.1718115353371243, "grad_norm": 0.0, "learning_rate": 1.8964359071372462e-05, "loss": 2.3813, "step": 423 }, { "epoch": 0.17221770917952883, "grad_norm": 0.0, "learning_rate": 1.895852103792381e-05, "loss": 2.2381, "step": 424 }, { "epoch": 0.17262388302193338, "grad_norm": 0.0, "learning_rate": 1.8952667499632764e-05, "loss": 2.3365, "step": 425 }, { "epoch": 0.17303005686433795, "grad_norm": 0.0, "learning_rate": 1.894679846663027e-05, "loss": 2.318, "step": 426 }, { "epoch": 0.1734362307067425, "grad_norm": 0.0, "learning_rate": 1.894091394907408e-05, "loss": 2.3967, "step": 427 }, { "epoch": 0.17384240454914704, "grad_norm": 0.0, "learning_rate": 1.893501395714874e-05, "loss": 2.3532, "step": 428 }, { "epoch": 0.17424857839155158, "grad_norm": 0.0, "learning_rate": 1.8929098501065592e-05, "loss": 2.2132, "step": 429 }, { "epoch": 0.17465475223395613, "grad_norm": 0.0, "learning_rate": 1.8923167591062723e-05, "loss": 2.3165, "step": 430 }, { "epoch": 0.17506092607636067, "grad_norm": 0.0, "learning_rate": 1.891722123740498e-05, "loss": 2.3694, "step": 431 }, { "epoch": 0.17546709991876522, "grad_norm": 0.0, "learning_rate": 1.8911259450383946e-05, "loss": 2.417, "step": 432 }, { "epoch": 0.1758732737611698, "grad_norm": 0.0, "learning_rate": 1.890528224031789e-05, "loss": 2.4756, "step": 433 }, { "epoch": 0.17627944760357434, "grad_norm": 0.0, "learning_rate": 1.8899289617551803e-05, "loss": 2.3689, "step": 434 }, { "epoch": 0.17668562144597888, "grad_norm": 0.0, "learning_rate": 1.8893281592457333e-05, "loss": 2.357, "step": 435 }, { "epoch": 0.17709179528838342, "grad_norm": 0.0, "learning_rate": 1.888725817543279e-05, "loss": 2.2996, "step": 436 }, { "epoch": 0.17749796913078797, "grad_norm": 0.0, "learning_rate": 1.8881219376903122e-05, "loss": 2.4149, "step": 437 }, { "epoch": 0.17790414297319251, "grad_norm": 0.0, "learning_rate": 1.8875165207319902e-05, "loss": 2.3769, "step": 438 }, { "epoch": 0.1783103168155971, "grad_norm": 0.0, "learning_rate": 1.8869095677161304e-05, "loss": 2.3688, "step": 439 }, { "epoch": 0.17871649065800163, "grad_norm": 0.0, "learning_rate": 1.886301079693209e-05, "loss": 2.4201, "step": 440 }, { "epoch": 0.17912266450040618, "grad_norm": 0.0, "learning_rate": 1.885691057716358e-05, "loss": 2.3479, "step": 441 }, { "epoch": 0.17952883834281072, "grad_norm": 0.0, "learning_rate": 1.8850795028413658e-05, "loss": 2.2269, "step": 442 }, { "epoch": 0.17993501218521527, "grad_norm": 0.0, "learning_rate": 1.8844664161266722e-05, "loss": 2.3099, "step": 443 }, { "epoch": 0.1803411860276198, "grad_norm": 0.0, "learning_rate": 1.883851798633369e-05, "loss": 2.2819, "step": 444 }, { "epoch": 0.18074735987002438, "grad_norm": 0.0, "learning_rate": 1.883235651425198e-05, "loss": 2.3802, "step": 445 }, { "epoch": 0.18115353371242893, "grad_norm": 0.0, "learning_rate": 1.882617975568547e-05, "loss": 2.2621, "step": 446 }, { "epoch": 0.18155970755483347, "grad_norm": 0.0, "learning_rate": 1.881998772132451e-05, "loss": 2.3538, "step": 447 }, { "epoch": 0.18196588139723802, "grad_norm": 0.0, "learning_rate": 1.8813780421885882e-05, "loss": 2.3, "step": 448 }, { "epoch": 0.18237205523964256, "grad_norm": 0.0, "learning_rate": 1.8807557868112783e-05, "loss": 2.3674, "step": 449 }, { "epoch": 0.1827782290820471, "grad_norm": 0.0, "learning_rate": 1.880132007077482e-05, "loss": 2.2754, "step": 450 }, { "epoch": 0.18318440292445168, "grad_norm": 0.0, "learning_rate": 1.879506704066798e-05, "loss": 2.4495, "step": 451 }, { "epoch": 0.18359057676685622, "grad_norm": 0.0, "learning_rate": 1.8788798788614606e-05, "loss": 2.2994, "step": 452 }, { "epoch": 0.18399675060926077, "grad_norm": 0.0, "learning_rate": 1.8782515325463397e-05, "loss": 2.3583, "step": 453 }, { "epoch": 0.1844029244516653, "grad_norm": 0.0, "learning_rate": 1.8776216662089373e-05, "loss": 2.3644, "step": 454 }, { "epoch": 0.18480909829406986, "grad_norm": 0.0, "learning_rate": 1.8769902809393866e-05, "loss": 2.3434, "step": 455 }, { "epoch": 0.1852152721364744, "grad_norm": 0.0, "learning_rate": 1.8763573778304495e-05, "loss": 2.3434, "step": 456 }, { "epoch": 0.18562144597887897, "grad_norm": 0.0, "learning_rate": 1.8757229579775136e-05, "loss": 2.2851, "step": 457 }, { "epoch": 0.18602761982128352, "grad_norm": 0.0, "learning_rate": 1.875087022478594e-05, "loss": 2.4705, "step": 458 }, { "epoch": 0.18643379366368806, "grad_norm": 0.0, "learning_rate": 1.8744495724343267e-05, "loss": 2.3118, "step": 459 }, { "epoch": 0.1868399675060926, "grad_norm": 0.0, "learning_rate": 1.8738106089479706e-05, "loss": 2.2906, "step": 460 }, { "epoch": 0.18724614134849715, "grad_norm": 0.0, "learning_rate": 1.873170133125403e-05, "loss": 2.2069, "step": 461 }, { "epoch": 0.1876523151909017, "grad_norm": 0.0, "learning_rate": 1.8725281460751198e-05, "loss": 2.2845, "step": 462 }, { "epoch": 0.18805848903330624, "grad_norm": 0.0, "learning_rate": 1.8718846489082303e-05, "loss": 2.5691, "step": 463 }, { "epoch": 0.18846466287571081, "grad_norm": 0.0, "learning_rate": 1.8712396427384595e-05, "loss": 2.4355, "step": 464 }, { "epoch": 0.18887083671811536, "grad_norm": 0.0, "learning_rate": 1.8705931286821437e-05, "loss": 2.3153, "step": 465 }, { "epoch": 0.1892770105605199, "grad_norm": 0.0, "learning_rate": 1.869945107858228e-05, "loss": 2.2724, "step": 466 }, { "epoch": 0.18968318440292445, "grad_norm": 0.0, "learning_rate": 1.8692955813882662e-05, "loss": 2.3407, "step": 467 }, { "epoch": 0.190089358245329, "grad_norm": 0.0, "learning_rate": 1.8686445503964174e-05, "loss": 2.2239, "step": 468 }, { "epoch": 0.19049553208773354, "grad_norm": 0.0, "learning_rate": 1.867992016009445e-05, "loss": 2.3319, "step": 469 }, { "epoch": 0.1909017059301381, "grad_norm": 0.0, "learning_rate": 1.867337979356715e-05, "loss": 2.3771, "step": 470 }, { "epoch": 0.19130787977254265, "grad_norm": 0.0, "learning_rate": 1.8666824415701915e-05, "loss": 2.2539, "step": 471 }, { "epoch": 0.1917140536149472, "grad_norm": 0.0, "learning_rate": 1.866025403784439e-05, "loss": 2.4249, "step": 472 }, { "epoch": 0.19212022745735174, "grad_norm": 0.0, "learning_rate": 1.865366867136616e-05, "loss": 2.4324, "step": 473 }, { "epoch": 0.1925264012997563, "grad_norm": 0.0, "learning_rate": 1.8647068327664774e-05, "loss": 2.3055, "step": 474 }, { "epoch": 0.19293257514216083, "grad_norm": 0.0, "learning_rate": 1.8640453018163677e-05, "loss": 2.4452, "step": 475 }, { "epoch": 0.1933387489845654, "grad_norm": 0.0, "learning_rate": 1.8633822754312233e-05, "loss": 2.2477, "step": 476 }, { "epoch": 0.19374492282696995, "grad_norm": 0.0, "learning_rate": 1.862717754758569e-05, "loss": 2.3411, "step": 477 }, { "epoch": 0.1941510966693745, "grad_norm": 0.0, "learning_rate": 1.8620517409485148e-05, "loss": 2.1814, "step": 478 }, { "epoch": 0.19455727051177904, "grad_norm": 0.0, "learning_rate": 1.861384235153755e-05, "loss": 2.4644, "step": 479 }, { "epoch": 0.19496344435418358, "grad_norm": 0.0, "learning_rate": 1.860715238529567e-05, "loss": 2.4534, "step": 480 }, { "epoch": 0.19536961819658813, "grad_norm": 0.0, "learning_rate": 1.8600447522338077e-05, "loss": 2.3062, "step": 481 }, { "epoch": 0.1957757920389927, "grad_norm": 0.0, "learning_rate": 1.8593727774269122e-05, "loss": 2.2894, "step": 482 }, { "epoch": 0.19618196588139725, "grad_norm": 0.0, "learning_rate": 1.8586993152718923e-05, "loss": 2.2769, "step": 483 }, { "epoch": 0.1965881397238018, "grad_norm": 0.0, "learning_rate": 1.8580243669343338e-05, "loss": 2.4535, "step": 484 }, { "epoch": 0.19699431356620634, "grad_norm": 0.0, "learning_rate": 1.8573479335823946e-05, "loss": 2.2412, "step": 485 }, { "epoch": 0.19740048740861088, "grad_norm": 0.0, "learning_rate": 1.8566700163868027e-05, "loss": 2.4249, "step": 486 }, { "epoch": 0.19780666125101543, "grad_norm": 0.0, "learning_rate": 1.8559906165208543e-05, "loss": 2.1975, "step": 487 }, { "epoch": 0.19821283509341997, "grad_norm": 0.0, "learning_rate": 1.855309735160412e-05, "loss": 2.4388, "step": 488 }, { "epoch": 0.19861900893582454, "grad_norm": 0.0, "learning_rate": 1.8546273734839024e-05, "loss": 2.336, "step": 489 }, { "epoch": 0.1990251827782291, "grad_norm": 0.0, "learning_rate": 1.8539435326723135e-05, "loss": 2.391, "step": 490 }, { "epoch": 0.19943135662063363, "grad_norm": 0.0, "learning_rate": 1.8532582139091945e-05, "loss": 2.3232, "step": 491 }, { "epoch": 0.19983753046303818, "grad_norm": 0.0, "learning_rate": 1.8525714183806508e-05, "loss": 2.4497, "step": 492 }, { "epoch": 0.20024370430544272, "grad_norm": 0.0, "learning_rate": 1.8518831472753456e-05, "loss": 2.2862, "step": 493 }, { "epoch": 0.20064987814784727, "grad_norm": 0.0, "learning_rate": 1.851193401784495e-05, "loss": 2.2218, "step": 494 }, { "epoch": 0.20105605199025184, "grad_norm": 0.0, "learning_rate": 1.8505021831018665e-05, "loss": 2.5144, "step": 495 }, { "epoch": 0.20146222583265638, "grad_norm": 0.0, "learning_rate": 1.8498094924237776e-05, "loss": 2.3951, "step": 496 }, { "epoch": 0.20186839967506093, "grad_norm": 0.0, "learning_rate": 1.8491153309490943e-05, "loss": 2.2684, "step": 497 }, { "epoch": 0.20227457351746547, "grad_norm": 0.0, "learning_rate": 1.848419699879227e-05, "loss": 2.2448, "step": 498 }, { "epoch": 0.20268074735987002, "grad_norm": 0.0, "learning_rate": 1.8477226004181298e-05, "loss": 2.1983, "step": 499 }, { "epoch": 0.20308692120227456, "grad_norm": 0.0, "learning_rate": 1.847024033772299e-05, "loss": 2.2771, "step": 500 }, { "epoch": 0.20349309504467913, "grad_norm": 0.0, "learning_rate": 1.8463240011507694e-05, "loss": 2.3556, "step": 501 }, { "epoch": 0.20389926888708368, "grad_norm": 0.0, "learning_rate": 1.845622503765113e-05, "loss": 2.2584, "step": 502 }, { "epoch": 0.20430544272948822, "grad_norm": 0.0, "learning_rate": 1.8449195428294372e-05, "loss": 2.2363, "step": 503 }, { "epoch": 0.20471161657189277, "grad_norm": 0.0, "learning_rate": 1.8442151195603825e-05, "loss": 2.3841, "step": 504 }, { "epoch": 0.2051177904142973, "grad_norm": 0.0, "learning_rate": 1.8435092351771205e-05, "loss": 2.2316, "step": 505 }, { "epoch": 0.20552396425670186, "grad_norm": 0.0, "learning_rate": 1.842801890901351e-05, "loss": 2.3825, "step": 506 }, { "epoch": 0.20593013809910643, "grad_norm": 0.0, "learning_rate": 1.8420930879573006e-05, "loss": 2.3271, "step": 507 }, { "epoch": 0.20633631194151097, "grad_norm": 0.0, "learning_rate": 1.841382827571721e-05, "loss": 2.4122, "step": 508 }, { "epoch": 0.20674248578391552, "grad_norm": 0.0, "learning_rate": 1.8406711109738856e-05, "loss": 2.3614, "step": 509 }, { "epoch": 0.20714865962632006, "grad_norm": 0.0, "learning_rate": 1.8399579393955893e-05, "loss": 2.3334, "step": 510 }, { "epoch": 0.2075548334687246, "grad_norm": 0.0, "learning_rate": 1.8392433140711438e-05, "loss": 2.3105, "step": 511 }, { "epoch": 0.20796100731112915, "grad_norm": 0.0, "learning_rate": 1.8385272362373775e-05, "loss": 2.3186, "step": 512 }, { "epoch": 0.20836718115353373, "grad_norm": 0.0, "learning_rate": 1.837809707133633e-05, "loss": 2.2958, "step": 513 }, { "epoch": 0.20877335499593827, "grad_norm": 0.0, "learning_rate": 1.837090728001764e-05, "loss": 2.4546, "step": 514 }, { "epoch": 0.20917952883834282, "grad_norm": 0.0, "learning_rate": 1.8363703000861344e-05, "loss": 2.3547, "step": 515 }, { "epoch": 0.20958570268074736, "grad_norm": 0.0, "learning_rate": 1.8356484246336154e-05, "loss": 2.397, "step": 516 }, { "epoch": 0.2099918765231519, "grad_norm": 0.0, "learning_rate": 1.834925102893583e-05, "loss": 2.3372, "step": 517 }, { "epoch": 0.21039805036555645, "grad_norm": 0.0, "learning_rate": 1.834200336117918e-05, "loss": 2.3076, "step": 518 }, { "epoch": 0.210804224207961, "grad_norm": 0.0, "learning_rate": 1.8334741255609988e-05, "loss": 2.3678, "step": 519 }, { "epoch": 0.21121039805036557, "grad_norm": 0.0, "learning_rate": 1.832746472479707e-05, "loss": 2.1341, "step": 520 }, { "epoch": 0.2116165718927701, "grad_norm": 0.0, "learning_rate": 1.8320173781334172e-05, "loss": 2.227, "step": 521 }, { "epoch": 0.21202274573517466, "grad_norm": 0.0, "learning_rate": 1.8312868437840002e-05, "loss": 2.2939, "step": 522 }, { "epoch": 0.2124289195775792, "grad_norm": 0.0, "learning_rate": 1.8305548706958188e-05, "loss": 2.3228, "step": 523 }, { "epoch": 0.21283509341998375, "grad_norm": 0.0, "learning_rate": 1.829821460135726e-05, "loss": 2.2234, "step": 524 }, { "epoch": 0.2132412672623883, "grad_norm": 0.0, "learning_rate": 1.829086613373062e-05, "loss": 2.3923, "step": 525 }, { "epoch": 0.21364744110479286, "grad_norm": 0.0, "learning_rate": 1.8283503316796536e-05, "loss": 2.465, "step": 526 }, { "epoch": 0.2140536149471974, "grad_norm": 0.0, "learning_rate": 1.8276126163298103e-05, "loss": 2.252, "step": 527 }, { "epoch": 0.21445978878960195, "grad_norm": 0.0, "learning_rate": 1.8268734686003232e-05, "loss": 2.3041, "step": 528 }, { "epoch": 0.2148659626320065, "grad_norm": 0.0, "learning_rate": 1.826132889770463e-05, "loss": 2.3616, "step": 529 }, { "epoch": 0.21527213647441104, "grad_norm": 0.0, "learning_rate": 1.8253908811219764e-05, "loss": 2.2784, "step": 530 }, { "epoch": 0.21567831031681559, "grad_norm": 0.0, "learning_rate": 1.8246474439390848e-05, "loss": 2.3518, "step": 531 }, { "epoch": 0.21608448415922016, "grad_norm": 0.0, "learning_rate": 1.823902579508483e-05, "loss": 2.2321, "step": 532 }, { "epoch": 0.2164906580016247, "grad_norm": 0.0, "learning_rate": 1.823156289119335e-05, "loss": 2.2933, "step": 533 }, { "epoch": 0.21689683184402925, "grad_norm": 0.0, "learning_rate": 1.822408574063273e-05, "loss": 2.2846, "step": 534 }, { "epoch": 0.2173030056864338, "grad_norm": 0.0, "learning_rate": 1.8216594356343953e-05, "loss": 2.3805, "step": 535 }, { "epoch": 0.21770917952883834, "grad_norm": 0.0, "learning_rate": 1.820908875129263e-05, "loss": 2.3217, "step": 536 }, { "epoch": 0.21811535337124288, "grad_norm": 0.0, "learning_rate": 1.8201568938468986e-05, "loss": 2.1517, "step": 537 }, { "epoch": 0.21852152721364745, "grad_norm": 0.0, "learning_rate": 1.8194034930887842e-05, "loss": 2.2157, "step": 538 }, { "epoch": 0.218927701056052, "grad_norm": 0.0, "learning_rate": 1.818648674158858e-05, "loss": 2.3376, "step": 539 }, { "epoch": 0.21933387489845654, "grad_norm": 0.0, "learning_rate": 1.8178924383635136e-05, "loss": 2.3447, "step": 540 }, { "epoch": 0.2197400487408611, "grad_norm": 0.0, "learning_rate": 1.817134787011596e-05, "loss": 2.1094, "step": 541 }, { "epoch": 0.22014622258326563, "grad_norm": 0.0, "learning_rate": 1.8163757214143993e-05, "loss": 2.4268, "step": 542 }, { "epoch": 0.22055239642567018, "grad_norm": 0.0, "learning_rate": 1.815615242885667e-05, "loss": 2.3089, "step": 543 }, { "epoch": 0.22095857026807472, "grad_norm": 0.0, "learning_rate": 1.8148533527415874e-05, "loss": 2.2621, "step": 544 }, { "epoch": 0.2213647441104793, "grad_norm": 0.0, "learning_rate": 1.814090052300792e-05, "loss": 2.3632, "step": 545 }, { "epoch": 0.22177091795288384, "grad_norm": 0.0, "learning_rate": 1.8133253428843524e-05, "loss": 2.4001, "step": 546 }, { "epoch": 0.22217709179528838, "grad_norm": 0.0, "learning_rate": 1.8125592258157797e-05, "loss": 2.3936, "step": 547 }, { "epoch": 0.22258326563769293, "grad_norm": 0.0, "learning_rate": 1.811791702421021e-05, "loss": 2.3061, "step": 548 }, { "epoch": 0.22298943948009747, "grad_norm": 0.0, "learning_rate": 1.811022774028457e-05, "loss": 2.2295, "step": 549 }, { "epoch": 0.22339561332250202, "grad_norm": 0.0, "learning_rate": 1.810252441968901e-05, "loss": 2.3328, "step": 550 }, { "epoch": 0.2238017871649066, "grad_norm": 0.0, "learning_rate": 1.809480707575594e-05, "loss": 2.3818, "step": 551 }, { "epoch": 0.22420796100731114, "grad_norm": 0.0, "learning_rate": 1.8087075721842064e-05, "loss": 2.4254, "step": 552 }, { "epoch": 0.22461413484971568, "grad_norm": 0.0, "learning_rate": 1.8079330371328314e-05, "loss": 2.3487, "step": 553 }, { "epoch": 0.22502030869212022, "grad_norm": 0.0, "learning_rate": 1.8071571037619856e-05, "loss": 2.3592, "step": 554 }, { "epoch": 0.22542648253452477, "grad_norm": 0.0, "learning_rate": 1.806379773414605e-05, "loss": 2.3026, "step": 555 }, { "epoch": 0.2258326563769293, "grad_norm": 0.0, "learning_rate": 1.8056010474360445e-05, "loss": 2.3976, "step": 556 }, { "epoch": 0.2262388302193339, "grad_norm": 0.0, "learning_rate": 1.8048209271740735e-05, "loss": 2.3345, "step": 557 }, { "epoch": 0.22664500406173843, "grad_norm": 0.0, "learning_rate": 1.804039413978875e-05, "loss": 2.2463, "step": 558 }, { "epoch": 0.22705117790414298, "grad_norm": 0.0, "learning_rate": 1.8032565092030423e-05, "loss": 2.2348, "step": 559 }, { "epoch": 0.22745735174654752, "grad_norm": 0.0, "learning_rate": 1.8024722142015784e-05, "loss": 2.4393, "step": 560 }, { "epoch": 0.22786352558895206, "grad_norm": 0.0, "learning_rate": 1.80168653033189e-05, "loss": 2.2826, "step": 561 }, { "epoch": 0.2282696994313566, "grad_norm": 0.0, "learning_rate": 1.8008994589537913e-05, "loss": 2.184, "step": 562 }, { "epoch": 0.22867587327376118, "grad_norm": 0.0, "learning_rate": 1.8001110014294938e-05, "loss": 2.159, "step": 563 }, { "epoch": 0.22908204711616573, "grad_norm": 0.0, "learning_rate": 1.7993211591236108e-05, "loss": 2.3164, "step": 564 }, { "epoch": 0.22948822095857027, "grad_norm": 0.0, "learning_rate": 1.7985299334031515e-05, "loss": 2.3322, "step": 565 }, { "epoch": 0.22989439480097482, "grad_norm": 0.0, "learning_rate": 1.7977373256375194e-05, "loss": 2.3496, "step": 566 }, { "epoch": 0.23030056864337936, "grad_norm": 0.0, "learning_rate": 1.7969433371985098e-05, "loss": 2.3686, "step": 567 }, { "epoch": 0.2307067424857839, "grad_norm": 0.0, "learning_rate": 1.796147969460308e-05, "loss": 2.3546, "step": 568 }, { "epoch": 0.23111291632818848, "grad_norm": 0.0, "learning_rate": 1.795351223799486e-05, "loss": 2.4048, "step": 569 }, { "epoch": 0.23151909017059302, "grad_norm": 0.0, "learning_rate": 1.7945531015950008e-05, "loss": 2.2933, "step": 570 }, { "epoch": 0.23192526401299757, "grad_norm": 0.0, "learning_rate": 1.7937536042281917e-05, "loss": 2.2626, "step": 571 }, { "epoch": 0.2323314378554021, "grad_norm": 0.0, "learning_rate": 1.7929527330827788e-05, "loss": 2.4657, "step": 572 }, { "epoch": 0.23273761169780666, "grad_norm": 0.0, "learning_rate": 1.792150489544859e-05, "loss": 2.2997, "step": 573 }, { "epoch": 0.2331437855402112, "grad_norm": 0.0, "learning_rate": 1.791346875002905e-05, "loss": 2.433, "step": 574 }, { "epoch": 0.23354995938261575, "grad_norm": 0.0, "learning_rate": 1.7905418908477617e-05, "loss": 2.3025, "step": 575 }, { "epoch": 0.23395613322502032, "grad_norm": 0.0, "learning_rate": 1.789735538472645e-05, "loss": 2.2842, "step": 576 }, { "epoch": 0.23436230706742486, "grad_norm": 0.0, "learning_rate": 1.7889278192731388e-05, "loss": 2.3679, "step": 577 }, { "epoch": 0.2347684809098294, "grad_norm": 0.0, "learning_rate": 1.7881187346471924e-05, "loss": 2.3807, "step": 578 }, { "epoch": 0.23517465475223395, "grad_norm": 0.0, "learning_rate": 1.7873082859951183e-05, "loss": 2.3975, "step": 579 }, { "epoch": 0.2355808285946385, "grad_norm": 0.0, "learning_rate": 1.7864964747195902e-05, "loss": 2.3499, "step": 580 }, { "epoch": 0.23598700243704304, "grad_norm": 0.0, "learning_rate": 1.785683302225639e-05, "loss": 2.2883, "step": 581 }, { "epoch": 0.23639317627944761, "grad_norm": 0.0, "learning_rate": 1.784868769920653e-05, "loss": 2.3568, "step": 582 }, { "epoch": 0.23679935012185216, "grad_norm": 0.0, "learning_rate": 1.7840528792143725e-05, "loss": 2.333, "step": 583 }, { "epoch": 0.2372055239642567, "grad_norm": 0.0, "learning_rate": 1.7832356315188907e-05, "loss": 2.4078, "step": 584 }, { "epoch": 0.23761169780666125, "grad_norm": 0.0, "learning_rate": 1.7824170282486473e-05, "loss": 2.1228, "step": 585 }, { "epoch": 0.2380178716490658, "grad_norm": 0.0, "learning_rate": 1.7815970708204296e-05, "loss": 2.237, "step": 586 }, { "epoch": 0.23842404549147034, "grad_norm": 0.0, "learning_rate": 1.780775760653368e-05, "loss": 2.3423, "step": 587 }, { "epoch": 0.2388302193338749, "grad_norm": 0.0, "learning_rate": 1.779953099168935e-05, "loss": 2.2765, "step": 588 }, { "epoch": 0.23923639317627945, "grad_norm": 0.0, "learning_rate": 1.77912908779094e-05, "loss": 2.3422, "step": 589 }, { "epoch": 0.239642567018684, "grad_norm": 0.0, "learning_rate": 1.77830372794553e-05, "loss": 2.2463, "step": 590 }, { "epoch": 0.24004874086108854, "grad_norm": 0.0, "learning_rate": 1.7774770210611865e-05, "loss": 2.3798, "step": 591 }, { "epoch": 0.2404549147034931, "grad_norm": 0.0, "learning_rate": 1.7766489685687208e-05, "loss": 2.3284, "step": 592 }, { "epoch": 0.24086108854589763, "grad_norm": 0.0, "learning_rate": 1.7758195719012742e-05, "loss": 2.0167, "step": 593 }, { "epoch": 0.2412672623883022, "grad_norm": 0.0, "learning_rate": 1.774988832494314e-05, "loss": 2.419, "step": 594 }, { "epoch": 0.24167343623070675, "grad_norm": 0.0, "learning_rate": 1.7741567517856313e-05, "loss": 2.3033, "step": 595 }, { "epoch": 0.2420796100731113, "grad_norm": 0.0, "learning_rate": 1.773323331215339e-05, "loss": 2.2696, "step": 596 }, { "epoch": 0.24248578391551584, "grad_norm": 0.0, "learning_rate": 1.772488572225869e-05, "loss": 2.3087, "step": 597 }, { "epoch": 0.24289195775792038, "grad_norm": 0.0, "learning_rate": 1.7716524762619695e-05, "loss": 2.3447, "step": 598 }, { "epoch": 0.24329813160032493, "grad_norm": 0.0, "learning_rate": 1.7708150447707017e-05, "loss": 2.2154, "step": 599 }, { "epoch": 0.2437043054427295, "grad_norm": 0.0, "learning_rate": 1.7699762792014402e-05, "loss": 2.3794, "step": 600 }, { "epoch": 0.24411047928513405, "grad_norm": 0.0, "learning_rate": 1.769136181005867e-05, "loss": 2.168, "step": 601 }, { "epoch": 0.2445166531275386, "grad_norm": 0.0, "learning_rate": 1.7682947516379706e-05, "loss": 2.3244, "step": 602 }, { "epoch": 0.24492282696994314, "grad_norm": 0.0, "learning_rate": 1.7674519925540447e-05, "loss": 2.4489, "step": 603 }, { "epoch": 0.24532900081234768, "grad_norm": 0.0, "learning_rate": 1.7666079052126827e-05, "loss": 2.1742, "step": 604 }, { "epoch": 0.24573517465475223, "grad_norm": 0.0, "learning_rate": 1.7657624910747783e-05, "loss": 2.4311, "step": 605 }, { "epoch": 0.24614134849715677, "grad_norm": 0.0, "learning_rate": 1.7649157516035205e-05, "loss": 2.3651, "step": 606 }, { "epoch": 0.24654752233956134, "grad_norm": 0.0, "learning_rate": 1.7640676882643926e-05, "loss": 2.3051, "step": 607 }, { "epoch": 0.2469536961819659, "grad_norm": 0.0, "learning_rate": 1.763218302525169e-05, "loss": 2.2251, "step": 608 }, { "epoch": 0.24735987002437043, "grad_norm": 0.0, "learning_rate": 1.7623675958559126e-05, "loss": 2.2732, "step": 609 }, { "epoch": 0.24776604386677498, "grad_norm": 0.0, "learning_rate": 1.7615155697289734e-05, "loss": 2.4586, "step": 610 }, { "epoch": 0.24817221770917952, "grad_norm": 0.0, "learning_rate": 1.7606622256189837e-05, "loss": 2.3971, "step": 611 }, { "epoch": 0.24857839155158407, "grad_norm": 0.0, "learning_rate": 1.7598075650028583e-05, "loss": 2.3014, "step": 612 }, { "epoch": 0.24898456539398864, "grad_norm": 0.0, "learning_rate": 1.758951589359789e-05, "loss": 2.4225, "step": 613 }, { "epoch": 0.24939073923639318, "grad_norm": 0.0, "learning_rate": 1.7580943001712457e-05, "loss": 2.5689, "step": 614 }, { "epoch": 0.24979691307879773, "grad_norm": 0.0, "learning_rate": 1.7572356989209692e-05, "loss": 2.3036, "step": 615 }, { "epoch": 0.2502030869212023, "grad_norm": 0.0, "learning_rate": 1.7563757870949724e-05, "loss": 2.1582, "step": 616 }, { "epoch": 0.25060926076360684, "grad_norm": 0.0, "learning_rate": 1.755514566181537e-05, "loss": 2.2498, "step": 617 }, { "epoch": 0.2510154346060114, "grad_norm": 0.0, "learning_rate": 1.7546520376712093e-05, "loss": 2.268, "step": 618 }, { "epoch": 0.25142160844841593, "grad_norm": 0.0, "learning_rate": 1.753788203056799e-05, "loss": 2.1596, "step": 619 }, { "epoch": 0.2518277822908205, "grad_norm": 0.0, "learning_rate": 1.752923063833377e-05, "loss": 2.3662, "step": 620 }, { "epoch": 0.252233956133225, "grad_norm": 0.0, "learning_rate": 1.7520566214982715e-05, "loss": 2.2781, "step": 621 }, { "epoch": 0.25264012997562957, "grad_norm": 0.0, "learning_rate": 1.7511888775510662e-05, "loss": 2.2516, "step": 622 }, { "epoch": 0.2530463038180341, "grad_norm": 0.0, "learning_rate": 1.750319833493597e-05, "loss": 2.2162, "step": 623 }, { "epoch": 0.25345247766043866, "grad_norm": 0.0, "learning_rate": 1.7494494908299508e-05, "loss": 2.4234, "step": 624 }, { "epoch": 0.2538586515028432, "grad_norm": 0.0, "learning_rate": 1.7485778510664618e-05, "loss": 2.181, "step": 625 }, { "epoch": 0.25426482534524775, "grad_norm": 0.0, "learning_rate": 1.7477049157117093e-05, "loss": 2.2596, "step": 626 }, { "epoch": 0.2546709991876523, "grad_norm": 0.0, "learning_rate": 1.746830686276514e-05, "loss": 2.2272, "step": 627 }, { "epoch": 0.2550771730300569, "grad_norm": 0.0, "learning_rate": 1.7459551642739375e-05, "loss": 2.3471, "step": 628 }, { "epoch": 0.25548334687246144, "grad_norm": 0.0, "learning_rate": 1.745078351219278e-05, "loss": 2.1418, "step": 629 }, { "epoch": 0.255889520714866, "grad_norm": 0.0, "learning_rate": 1.744200248630068e-05, "loss": 2.2758, "step": 630 }, { "epoch": 0.2562956945572705, "grad_norm": 0.0, "learning_rate": 1.7433208580260716e-05, "loss": 2.2983, "step": 631 }, { "epoch": 0.25670186839967507, "grad_norm": 0.0, "learning_rate": 1.7424401809292833e-05, "loss": 2.3826, "step": 632 }, { "epoch": 0.2571080422420796, "grad_norm": 0.0, "learning_rate": 1.741558218863923e-05, "loss": 2.2872, "step": 633 }, { "epoch": 0.25751421608448416, "grad_norm": 0.0, "learning_rate": 1.7406749733564344e-05, "loss": 2.1211, "step": 634 }, { "epoch": 0.2579203899268887, "grad_norm": 0.0, "learning_rate": 1.7397904459354843e-05, "loss": 2.2115, "step": 635 }, { "epoch": 0.25832656376929325, "grad_norm": 0.0, "learning_rate": 1.7389046381319557e-05, "loss": 2.436, "step": 636 }, { "epoch": 0.2587327376116978, "grad_norm": 0.0, "learning_rate": 1.7380175514789496e-05, "loss": 2.4339, "step": 637 }, { "epoch": 0.25913891145410234, "grad_norm": 0.0, "learning_rate": 1.737129187511779e-05, "loss": 2.2876, "step": 638 }, { "epoch": 0.2595450852965069, "grad_norm": 0.0, "learning_rate": 1.7362395477679685e-05, "loss": 2.2787, "step": 639 }, { "epoch": 0.25995125913891143, "grad_norm": 0.0, "learning_rate": 1.7353486337872502e-05, "loss": 2.166, "step": 640 }, { "epoch": 0.26035743298131603, "grad_norm": 0.0, "learning_rate": 1.734456447111562e-05, "loss": 2.3336, "step": 641 }, { "epoch": 0.2607636068237206, "grad_norm": 0.0, "learning_rate": 1.7335629892850436e-05, "loss": 2.4937, "step": 642 }, { "epoch": 0.2611697806661251, "grad_norm": 0.0, "learning_rate": 1.7326682618540364e-05, "loss": 2.1847, "step": 643 }, { "epoch": 0.26157595450852966, "grad_norm": 0.0, "learning_rate": 1.7317722663670767e-05, "loss": 2.3003, "step": 644 }, { "epoch": 0.2619821283509342, "grad_norm": 0.0, "learning_rate": 1.7308750043748987e-05, "loss": 2.3778, "step": 645 }, { "epoch": 0.26238830219333875, "grad_norm": 0.0, "learning_rate": 1.729976477430425e-05, "loss": 2.331, "step": 646 }, { "epoch": 0.2627944760357433, "grad_norm": 0.0, "learning_rate": 1.7290766870887707e-05, "loss": 2.1656, "step": 647 }, { "epoch": 0.26320064987814784, "grad_norm": 0.0, "learning_rate": 1.728175634907235e-05, "loss": 2.3379, "step": 648 }, { "epoch": 0.2636068237205524, "grad_norm": 0.0, "learning_rate": 1.727273322445303e-05, "loss": 2.3863, "step": 649 }, { "epoch": 0.26401299756295693, "grad_norm": 0.0, "learning_rate": 1.7263697512646397e-05, "loss": 2.3974, "step": 650 }, { "epoch": 0.2644191714053615, "grad_norm": 0.0, "learning_rate": 1.7254649229290887e-05, "loss": 2.3667, "step": 651 }, { "epoch": 0.264825345247766, "grad_norm": 0.0, "learning_rate": 1.7245588390046702e-05, "loss": 2.3077, "step": 652 }, { "epoch": 0.2652315190901706, "grad_norm": 0.0, "learning_rate": 1.7236515010595773e-05, "loss": 2.31, "step": 653 }, { "epoch": 0.26563769293257516, "grad_norm": 0.0, "learning_rate": 1.7227429106641726e-05, "loss": 2.1871, "step": 654 }, { "epoch": 0.2660438667749797, "grad_norm": 0.0, "learning_rate": 1.721833069390987e-05, "loss": 2.3447, "step": 655 }, { "epoch": 0.26645004061738425, "grad_norm": 0.0, "learning_rate": 1.7209219788147166e-05, "loss": 2.3719, "step": 656 }, { "epoch": 0.2668562144597888, "grad_norm": 0.0, "learning_rate": 1.720009640512219e-05, "loss": 2.3823, "step": 657 }, { "epoch": 0.26726238830219334, "grad_norm": 0.0, "learning_rate": 1.7190960560625127e-05, "loss": 2.2525, "step": 658 }, { "epoch": 0.2676685621445979, "grad_norm": 0.0, "learning_rate": 1.7181812270467708e-05, "loss": 2.3493, "step": 659 }, { "epoch": 0.26807473598700243, "grad_norm": 0.0, "learning_rate": 1.7172651550483224e-05, "loss": 2.3045, "step": 660 }, { "epoch": 0.268480909829407, "grad_norm": 0.0, "learning_rate": 1.7163478416526465e-05, "loss": 2.2381, "step": 661 }, { "epoch": 0.2688870836718115, "grad_norm": 0.0, "learning_rate": 1.7154292884473712e-05, "loss": 2.2322, "step": 662 }, { "epoch": 0.26929325751421607, "grad_norm": 0.0, "learning_rate": 1.7145094970222712e-05, "loss": 2.3794, "step": 663 }, { "epoch": 0.2696994313566206, "grad_norm": 0.0, "learning_rate": 1.7135884689692623e-05, "loss": 2.149, "step": 664 }, { "epoch": 0.27010560519902516, "grad_norm": 0.0, "learning_rate": 1.7126662058824025e-05, "loss": 2.4149, "step": 665 }, { "epoch": 0.27051177904142976, "grad_norm": 0.0, "learning_rate": 1.711742709357886e-05, "loss": 2.2043, "step": 666 }, { "epoch": 0.2709179528838343, "grad_norm": 0.0, "learning_rate": 1.710817980994042e-05, "loss": 2.2088, "step": 667 }, { "epoch": 0.27132412672623885, "grad_norm": 0.0, "learning_rate": 1.709892022391333e-05, "loss": 2.331, "step": 668 }, { "epoch": 0.2717303005686434, "grad_norm": 0.0, "learning_rate": 1.7089648351523485e-05, "loss": 2.1427, "step": 669 }, { "epoch": 0.27213647441104794, "grad_norm": 0.0, "learning_rate": 1.708036420881807e-05, "loss": 2.0988, "step": 670 }, { "epoch": 0.2725426482534525, "grad_norm": 0.0, "learning_rate": 1.7071067811865477e-05, "loss": 2.313, "step": 671 }, { "epoch": 0.272948822095857, "grad_norm": 0.0, "learning_rate": 1.706175917675533e-05, "loss": 2.1749, "step": 672 }, { "epoch": 0.27335499593826157, "grad_norm": 0.0, "learning_rate": 1.7052438319598426e-05, "loss": 2.2692, "step": 673 }, { "epoch": 0.2737611697806661, "grad_norm": 0.0, "learning_rate": 1.7043105256526723e-05, "loss": 2.357, "step": 674 }, { "epoch": 0.27416734362307066, "grad_norm": 0.0, "learning_rate": 1.7033760003693287e-05, "loss": 2.3545, "step": 675 }, { "epoch": 0.2745735174654752, "grad_norm": 0.0, "learning_rate": 1.7024402577272294e-05, "loss": 2.1584, "step": 676 }, { "epoch": 0.27497969130787975, "grad_norm": 0.0, "learning_rate": 1.7015032993458987e-05, "loss": 2.4054, "step": 677 }, { "epoch": 0.27538586515028435, "grad_norm": 0.0, "learning_rate": 1.7005651268469652e-05, "loss": 2.3507, "step": 678 }, { "epoch": 0.2757920389926889, "grad_norm": 0.0, "learning_rate": 1.6996257418541582e-05, "loss": 2.319, "step": 679 }, { "epoch": 0.27619821283509344, "grad_norm": 0.0, "learning_rate": 1.6986851459933067e-05, "loss": 2.1676, "step": 680 }, { "epoch": 0.276604386677498, "grad_norm": 0.0, "learning_rate": 1.6977433408923337e-05, "loss": 2.1611, "step": 681 }, { "epoch": 0.2770105605199025, "grad_norm": 0.0, "learning_rate": 1.6968003281812563e-05, "loss": 2.2256, "step": 682 }, { "epoch": 0.27741673436230707, "grad_norm": 0.0, "learning_rate": 1.6958561094921816e-05, "loss": 2.1479, "step": 683 }, { "epoch": 0.2778229082047116, "grad_norm": 0.0, "learning_rate": 1.6949106864593034e-05, "loss": 2.3652, "step": 684 }, { "epoch": 0.27822908204711616, "grad_norm": 0.0, "learning_rate": 1.6939640607189e-05, "loss": 2.3638, "step": 685 }, { "epoch": 0.2786352558895207, "grad_norm": 0.0, "learning_rate": 1.693016233909332e-05, "loss": 2.2184, "step": 686 }, { "epoch": 0.27904142973192525, "grad_norm": 0.0, "learning_rate": 1.6920672076710376e-05, "loss": 2.4067, "step": 687 }, { "epoch": 0.2794476035743298, "grad_norm": 0.0, "learning_rate": 1.6911169836465317e-05, "loss": 2.2919, "step": 688 }, { "epoch": 0.27985377741673434, "grad_norm": 0.0, "learning_rate": 1.6901655634804022e-05, "loss": 2.3061, "step": 689 }, { "epoch": 0.2802599512591389, "grad_norm": 0.0, "learning_rate": 1.689212948819307e-05, "loss": 2.2514, "step": 690 }, { "epoch": 0.2806661251015435, "grad_norm": 0.0, "learning_rate": 1.6882591413119712e-05, "loss": 2.1528, "step": 691 }, { "epoch": 0.28107229894394803, "grad_norm": 0.0, "learning_rate": 1.6873041426091845e-05, "loss": 2.3632, "step": 692 }, { "epoch": 0.2814784727863526, "grad_norm": 0.0, "learning_rate": 1.6863479543637985e-05, "loss": 2.1423, "step": 693 }, { "epoch": 0.2818846466287571, "grad_norm": 0.0, "learning_rate": 1.6853905782307235e-05, "loss": 2.1915, "step": 694 }, { "epoch": 0.28229082047116166, "grad_norm": 0.0, "learning_rate": 1.6844320158669256e-05, "loss": 2.4332, "step": 695 }, { "epoch": 0.2826969943135662, "grad_norm": 0.0, "learning_rate": 1.6834722689314237e-05, "loss": 2.2013, "step": 696 }, { "epoch": 0.28310316815597075, "grad_norm": 0.0, "learning_rate": 1.682511339085288e-05, "loss": 2.2823, "step": 697 }, { "epoch": 0.2835093419983753, "grad_norm": 0.0, "learning_rate": 1.681549227991634e-05, "loss": 2.2692, "step": 698 }, { "epoch": 0.28391551584077984, "grad_norm": 0.0, "learning_rate": 1.6805859373156237e-05, "loss": 2.2946, "step": 699 }, { "epoch": 0.2843216896831844, "grad_norm": 0.0, "learning_rate": 1.6796214687244595e-05, "loss": 2.3122, "step": 700 }, { "epoch": 0.28472786352558893, "grad_norm": 0.0, "learning_rate": 1.6786558238873826e-05, "loss": 2.2378, "step": 701 }, { "epoch": 0.2851340373679935, "grad_norm": 0.0, "learning_rate": 1.67768900447567e-05, "loss": 2.3654, "step": 702 }, { "epoch": 0.2855402112103981, "grad_norm": 0.0, "learning_rate": 1.6767210121626323e-05, "loss": 2.2735, "step": 703 }, { "epoch": 0.2859463850528026, "grad_norm": 0.0, "learning_rate": 1.6757518486236088e-05, "loss": 2.1789, "step": 704 }, { "epoch": 0.28635255889520717, "grad_norm": 0.0, "learning_rate": 1.6747815155359666e-05, "loss": 2.2796, "step": 705 }, { "epoch": 0.2867587327376117, "grad_norm": 0.0, "learning_rate": 1.6738100145790977e-05, "loss": 2.2621, "step": 706 }, { "epoch": 0.28716490658001625, "grad_norm": 0.0, "learning_rate": 1.6728373474344136e-05, "loss": 2.1962, "step": 707 }, { "epoch": 0.2875710804224208, "grad_norm": 0.0, "learning_rate": 1.6718635157853455e-05, "loss": 2.3858, "step": 708 }, { "epoch": 0.28797725426482534, "grad_norm": 0.0, "learning_rate": 1.67088852131734e-05, "loss": 2.232, "step": 709 }, { "epoch": 0.2883834281072299, "grad_norm": 0.0, "learning_rate": 1.6699123657178553e-05, "loss": 2.2416, "step": 710 }, { "epoch": 0.28878960194963443, "grad_norm": 0.0, "learning_rate": 1.6689350506763603e-05, "loss": 2.0869, "step": 711 }, { "epoch": 0.289195775792039, "grad_norm": 0.0, "learning_rate": 1.6679565778843305e-05, "loss": 2.3368, "step": 712 }, { "epoch": 0.2896019496344435, "grad_norm": 0.0, "learning_rate": 1.666976949035244e-05, "loss": 2.3393, "step": 713 }, { "epoch": 0.29000812347684807, "grad_norm": 0.0, "learning_rate": 1.6659961658245813e-05, "loss": 2.2703, "step": 714 }, { "epoch": 0.29041429731925267, "grad_norm": 0.0, "learning_rate": 1.6650142299498195e-05, "loss": 2.1874, "step": 715 }, { "epoch": 0.2908204711616572, "grad_norm": 0.0, "learning_rate": 1.6640311431104314e-05, "loss": 2.2878, "step": 716 }, { "epoch": 0.29122664500406176, "grad_norm": 0.0, "learning_rate": 1.6630469070078815e-05, "loss": 2.1634, "step": 717 }, { "epoch": 0.2916328188464663, "grad_norm": 0.0, "learning_rate": 1.6620615233456235e-05, "loss": 2.24, "step": 718 }, { "epoch": 0.29203899268887085, "grad_norm": 0.0, "learning_rate": 1.6610749938290977e-05, "loss": 2.154, "step": 719 }, { "epoch": 0.2924451665312754, "grad_norm": 0.0, "learning_rate": 1.6600873201657262e-05, "loss": 2.2913, "step": 720 }, { "epoch": 0.29285134037367994, "grad_norm": 0.0, "learning_rate": 1.6590985040649128e-05, "loss": 2.2738, "step": 721 }, { "epoch": 0.2932575142160845, "grad_norm": 0.0, "learning_rate": 1.658108547238038e-05, "loss": 2.2816, "step": 722 }, { "epoch": 0.293663688058489, "grad_norm": 0.0, "learning_rate": 1.657117451398456e-05, "loss": 2.1761, "step": 723 }, { "epoch": 0.29406986190089357, "grad_norm": 0.0, "learning_rate": 1.6561252182614932e-05, "loss": 2.3029, "step": 724 }, { "epoch": 0.2944760357432981, "grad_norm": 0.0, "learning_rate": 1.6551318495444443e-05, "loss": 2.322, "step": 725 }, { "epoch": 0.29488220958570266, "grad_norm": 0.0, "learning_rate": 1.6541373469665688e-05, "loss": 2.2716, "step": 726 }, { "epoch": 0.2952883834281072, "grad_norm": 0.0, "learning_rate": 1.653141712249089e-05, "loss": 2.2701, "step": 727 }, { "epoch": 0.2956945572705118, "grad_norm": 0.0, "learning_rate": 1.6521449471151867e-05, "loss": 2.2138, "step": 728 }, { "epoch": 0.29610073111291635, "grad_norm": 0.0, "learning_rate": 1.6511470532900002e-05, "loss": 2.3389, "step": 729 }, { "epoch": 0.2965069049553209, "grad_norm": 0.0, "learning_rate": 1.6501480325006206e-05, "loss": 2.2264, "step": 730 }, { "epoch": 0.29691307879772544, "grad_norm": 0.0, "learning_rate": 1.6491478864760902e-05, "loss": 2.144, "step": 731 }, { "epoch": 0.29731925264013, "grad_norm": 0.0, "learning_rate": 1.6481466169473988e-05, "loss": 2.146, "step": 732 }, { "epoch": 0.2977254264825345, "grad_norm": 0.0, "learning_rate": 1.64714422564748e-05, "loss": 2.1139, "step": 733 }, { "epoch": 0.2981316003249391, "grad_norm": 0.0, "learning_rate": 1.64614071431121e-05, "loss": 2.3376, "step": 734 }, { "epoch": 0.2985377741673436, "grad_norm": 0.0, "learning_rate": 1.6451360846754023e-05, "loss": 2.2112, "step": 735 }, { "epoch": 0.29894394800974816, "grad_norm": 0.0, "learning_rate": 1.6441303384788067e-05, "loss": 2.2749, "step": 736 }, { "epoch": 0.2993501218521527, "grad_norm": 0.0, "learning_rate": 1.643123477462105e-05, "loss": 2.2079, "step": 737 }, { "epoch": 0.29975629569455725, "grad_norm": 0.0, "learning_rate": 1.6421155033679085e-05, "loss": 2.3308, "step": 738 }, { "epoch": 0.3001624695369618, "grad_norm": 0.0, "learning_rate": 1.6411064179407556e-05, "loss": 2.1771, "step": 739 }, { "epoch": 0.3005686433793664, "grad_norm": 0.0, "learning_rate": 1.6400962229271074e-05, "loss": 2.1517, "step": 740 }, { "epoch": 0.30097481722177094, "grad_norm": 0.0, "learning_rate": 1.6390849200753452e-05, "loss": 2.247, "step": 741 }, { "epoch": 0.3013809910641755, "grad_norm": 0.0, "learning_rate": 1.6380725111357693e-05, "loss": 2.1538, "step": 742 }, { "epoch": 0.30178716490658003, "grad_norm": 0.0, "learning_rate": 1.6370589978605914e-05, "loss": 2.0464, "step": 743 }, { "epoch": 0.3021933387489846, "grad_norm": 0.0, "learning_rate": 1.636044382003938e-05, "loss": 2.2455, "step": 744 }, { "epoch": 0.3025995125913891, "grad_norm": 0.0, "learning_rate": 1.635028665321841e-05, "loss": 2.232, "step": 745 }, { "epoch": 0.30300568643379366, "grad_norm": 0.0, "learning_rate": 1.634011849572239e-05, "loss": 2.1484, "step": 746 }, { "epoch": 0.3034118602761982, "grad_norm": 0.0, "learning_rate": 1.632993936514972e-05, "loss": 2.2757, "step": 747 }, { "epoch": 0.30381803411860275, "grad_norm": 0.0, "learning_rate": 1.631974927911781e-05, "loss": 2.19, "step": 748 }, { "epoch": 0.3042242079610073, "grad_norm": 0.0, "learning_rate": 1.6309548255263003e-05, "loss": 2.2962, "step": 749 }, { "epoch": 0.30463038180341184, "grad_norm": 0.0, "learning_rate": 1.6299336311240593e-05, "loss": 2.3161, "step": 750 }, { "epoch": 0.3050365556458164, "grad_norm": 0.0, "learning_rate": 1.628911346472477e-05, "loss": 2.3098, "step": 751 }, { "epoch": 0.30544272948822093, "grad_norm": 0.0, "learning_rate": 1.6278879733408587e-05, "loss": 2.3443, "step": 752 }, { "epoch": 0.30584890333062553, "grad_norm": 0.0, "learning_rate": 1.6268635135003937e-05, "loss": 2.1582, "step": 753 }, { "epoch": 0.3062550771730301, "grad_norm": 0.0, "learning_rate": 1.6258379687241533e-05, "loss": 2.1179, "step": 754 }, { "epoch": 0.3066612510154346, "grad_norm": 0.0, "learning_rate": 1.6248113407870847e-05, "loss": 2.2984, "step": 755 }, { "epoch": 0.30706742485783917, "grad_norm": 0.0, "learning_rate": 1.6237836314660113e-05, "loss": 2.2647, "step": 756 }, { "epoch": 0.3074735987002437, "grad_norm": 0.0, "learning_rate": 1.622754842539627e-05, "loss": 2.1194, "step": 757 }, { "epoch": 0.30787977254264826, "grad_norm": 0.0, "learning_rate": 1.6217249757884954e-05, "loss": 2.2508, "step": 758 }, { "epoch": 0.3082859463850528, "grad_norm": 0.0, "learning_rate": 1.620694032995044e-05, "loss": 2.2725, "step": 759 }, { "epoch": 0.30869212022745735, "grad_norm": 0.0, "learning_rate": 1.6196620159435644e-05, "loss": 2.3877, "step": 760 }, { "epoch": 0.3090982940698619, "grad_norm": 0.0, "learning_rate": 1.6186289264202052e-05, "loss": 2.3427, "step": 761 }, { "epoch": 0.30950446791226643, "grad_norm": 0.0, "learning_rate": 1.6175947662129735e-05, "loss": 2.2695, "step": 762 }, { "epoch": 0.309910641754671, "grad_norm": 0.0, "learning_rate": 1.6165595371117287e-05, "loss": 2.2298, "step": 763 }, { "epoch": 0.3103168155970755, "grad_norm": 0.0, "learning_rate": 1.6155232409081794e-05, "loss": 2.2739, "step": 764 }, { "epoch": 0.3107229894394801, "grad_norm": 0.0, "learning_rate": 1.614485879395882e-05, "loss": 2.1387, "step": 765 }, { "epoch": 0.31112916328188467, "grad_norm": 0.0, "learning_rate": 1.6134474543702353e-05, "loss": 2.2687, "step": 766 }, { "epoch": 0.3115353371242892, "grad_norm": 0.0, "learning_rate": 1.6124079676284806e-05, "loss": 2.3746, "step": 767 }, { "epoch": 0.31194151096669376, "grad_norm": 0.0, "learning_rate": 1.6113674209696955e-05, "loss": 2.2171, "step": 768 }, { "epoch": 0.3123476848090983, "grad_norm": 0.0, "learning_rate": 1.6103258161947933e-05, "loss": 2.1757, "step": 769 }, { "epoch": 0.31275385865150285, "grad_norm": 0.0, "learning_rate": 1.609283155106517e-05, "loss": 2.2397, "step": 770 }, { "epoch": 0.3131600324939074, "grad_norm": 0.0, "learning_rate": 1.6082394395094385e-05, "loss": 2.1365, "step": 771 }, { "epoch": 0.31356620633631194, "grad_norm": 0.0, "learning_rate": 1.6071946712099552e-05, "loss": 2.1457, "step": 772 }, { "epoch": 0.3139723801787165, "grad_norm": 0.0, "learning_rate": 1.6061488520162854e-05, "loss": 2.0399, "step": 773 }, { "epoch": 0.314378554021121, "grad_norm": 0.0, "learning_rate": 1.605101983738468e-05, "loss": 2.3313, "step": 774 }, { "epoch": 0.31478472786352557, "grad_norm": 0.0, "learning_rate": 1.6040540681883555e-05, "loss": 2.2073, "step": 775 }, { "epoch": 0.3151909017059301, "grad_norm": 0.0, "learning_rate": 1.6030051071796146e-05, "loss": 2.3148, "step": 776 }, { "epoch": 0.31559707554833466, "grad_norm": 0.0, "learning_rate": 1.6019551025277206e-05, "loss": 2.2709, "step": 777 }, { "epoch": 0.31600324939073926, "grad_norm": 0.0, "learning_rate": 1.6009040560499548e-05, "loss": 2.1752, "step": 778 }, { "epoch": 0.3164094232331438, "grad_norm": 0.0, "learning_rate": 1.599851969565403e-05, "loss": 2.2317, "step": 779 }, { "epoch": 0.31681559707554835, "grad_norm": 0.0, "learning_rate": 1.598798844894949e-05, "loss": 2.2455, "step": 780 }, { "epoch": 0.3172217709179529, "grad_norm": 0.0, "learning_rate": 1.5977446838612748e-05, "loss": 2.3574, "step": 781 }, { "epoch": 0.31762794476035744, "grad_norm": 0.0, "learning_rate": 1.596689488288856e-05, "loss": 2.1362, "step": 782 }, { "epoch": 0.318034118602762, "grad_norm": 0.0, "learning_rate": 1.5956332600039582e-05, "loss": 2.2317, "step": 783 }, { "epoch": 0.31844029244516653, "grad_norm": 0.0, "learning_rate": 1.594576000834635e-05, "loss": 2.1919, "step": 784 }, { "epoch": 0.3188464662875711, "grad_norm": 0.0, "learning_rate": 1.5935177126107234e-05, "loss": 2.2036, "step": 785 }, { "epoch": 0.3192526401299756, "grad_norm": 0.0, "learning_rate": 1.5924583971638416e-05, "loss": 2.2215, "step": 786 }, { "epoch": 0.31965881397238016, "grad_norm": 0.0, "learning_rate": 1.5913980563273863e-05, "loss": 2.3342, "step": 787 }, { "epoch": 0.3200649878147847, "grad_norm": 0.0, "learning_rate": 1.5903366919365283e-05, "loss": 2.1495, "step": 788 }, { "epoch": 0.32047116165718925, "grad_norm": 0.0, "learning_rate": 1.5892743058282106e-05, "loss": 2.3158, "step": 789 }, { "epoch": 0.32087733549959385, "grad_norm": 0.0, "learning_rate": 1.5882108998411427e-05, "loss": 2.2487, "step": 790 }, { "epoch": 0.3212835093419984, "grad_norm": 0.0, "learning_rate": 1.587146475815802e-05, "loss": 2.1528, "step": 791 }, { "epoch": 0.32168968318440294, "grad_norm": 0.0, "learning_rate": 1.586081035594425e-05, "loss": 2.2317, "step": 792 }, { "epoch": 0.3220958570268075, "grad_norm": 0.0, "learning_rate": 1.585014581021009e-05, "loss": 2.1989, "step": 793 }, { "epoch": 0.32250203086921203, "grad_norm": 0.0, "learning_rate": 1.5839471139413065e-05, "loss": 2.203, "step": 794 }, { "epoch": 0.3229082047116166, "grad_norm": 0.0, "learning_rate": 1.5828786362028216e-05, "loss": 2.2866, "step": 795 }, { "epoch": 0.3233143785540211, "grad_norm": 0.0, "learning_rate": 1.5818091496548084e-05, "loss": 2.3737, "step": 796 }, { "epoch": 0.32372055239642566, "grad_norm": 0.0, "learning_rate": 1.5807386561482663e-05, "loss": 2.1959, "step": 797 }, { "epoch": 0.3241267262388302, "grad_norm": 0.0, "learning_rate": 1.5796671575359382e-05, "loss": 2.2726, "step": 798 }, { "epoch": 0.32453290008123475, "grad_norm": 0.0, "learning_rate": 1.5785946556723064e-05, "loss": 2.1687, "step": 799 }, { "epoch": 0.3249390739236393, "grad_norm": 0.0, "learning_rate": 1.5775211524135888e-05, "loss": 2.1893, "step": 800 }, { "epoch": 0.32534524776604384, "grad_norm": 0.0, "learning_rate": 1.5764466496177374e-05, "loss": 2.2465, "step": 801 }, { "epoch": 0.3257514216084484, "grad_norm": 0.0, "learning_rate": 1.5753711491444336e-05, "loss": 2.1829, "step": 802 }, { "epoch": 0.326157595450853, "grad_norm": 0.0, "learning_rate": 1.574294652855086e-05, "loss": 2.2347, "step": 803 }, { "epoch": 0.32656376929325753, "grad_norm": 0.0, "learning_rate": 1.573217162612826e-05, "loss": 2.2579, "step": 804 }, { "epoch": 0.3269699431356621, "grad_norm": 0.0, "learning_rate": 1.572138680282506e-05, "loss": 2.3573, "step": 805 }, { "epoch": 0.3273761169780666, "grad_norm": 0.0, "learning_rate": 1.571059207730695e-05, "loss": 2.2828, "step": 806 }, { "epoch": 0.32778229082047117, "grad_norm": 0.0, "learning_rate": 1.5699787468256757e-05, "loss": 2.1824, "step": 807 }, { "epoch": 0.3281884646628757, "grad_norm": 0.0, "learning_rate": 1.568897299437442e-05, "loss": 2.1359, "step": 808 }, { "epoch": 0.32859463850528026, "grad_norm": 0.0, "learning_rate": 1.567814867437694e-05, "loss": 2.1243, "step": 809 }, { "epoch": 0.3290008123476848, "grad_norm": 0.0, "learning_rate": 1.5667314526998373e-05, "loss": 2.2954, "step": 810 }, { "epoch": 0.32940698619008935, "grad_norm": 0.0, "learning_rate": 1.5656470570989778e-05, "loss": 2.168, "step": 811 }, { "epoch": 0.3298131600324939, "grad_norm": 0.0, "learning_rate": 1.564561682511918e-05, "loss": 2.1314, "step": 812 }, { "epoch": 0.33021933387489844, "grad_norm": 0.0, "learning_rate": 1.563475330817157e-05, "loss": 2.31, "step": 813 }, { "epoch": 0.330625507717303, "grad_norm": 0.0, "learning_rate": 1.5623880038948828e-05, "loss": 2.3119, "step": 814 }, { "epoch": 0.3310316815597076, "grad_norm": 0.0, "learning_rate": 1.561299703626972e-05, "loss": 2.1979, "step": 815 }, { "epoch": 0.3314378554021121, "grad_norm": 0.0, "learning_rate": 1.560210431896987e-05, "loss": 2.3119, "step": 816 }, { "epoch": 0.33184402924451667, "grad_norm": 0.0, "learning_rate": 1.5591201905901693e-05, "loss": 2.2017, "step": 817 }, { "epoch": 0.3322502030869212, "grad_norm": 0.0, "learning_rate": 1.55802898159344e-05, "loss": 2.2115, "step": 818 }, { "epoch": 0.33265637692932576, "grad_norm": 0.0, "learning_rate": 1.556936806795396e-05, "loss": 2.2302, "step": 819 }, { "epoch": 0.3330625507717303, "grad_norm": 0.0, "learning_rate": 1.5558436680863025e-05, "loss": 2.2278, "step": 820 }, { "epoch": 0.33346872461413485, "grad_norm": 0.0, "learning_rate": 1.5547495673580962e-05, "loss": 2.3017, "step": 821 }, { "epoch": 0.3338748984565394, "grad_norm": 0.0, "learning_rate": 1.553654506504377e-05, "loss": 2.1833, "step": 822 }, { "epoch": 0.33428107229894394, "grad_norm": 0.0, "learning_rate": 1.5525584874204072e-05, "loss": 2.3149, "step": 823 }, { "epoch": 0.3346872461413485, "grad_norm": 0.0, "learning_rate": 1.5514615120031077e-05, "loss": 2.3002, "step": 824 }, { "epoch": 0.335093419983753, "grad_norm": 0.0, "learning_rate": 1.5503635821510538e-05, "loss": 2.2364, "step": 825 }, { "epoch": 0.33549959382615757, "grad_norm": 0.0, "learning_rate": 1.5492646997644737e-05, "loss": 2.1446, "step": 826 }, { "epoch": 0.33590576766856217, "grad_norm": 0.0, "learning_rate": 1.5481648667452427e-05, "loss": 2.2477, "step": 827 }, { "epoch": 0.3363119415109667, "grad_norm": 0.0, "learning_rate": 1.547064084996883e-05, "loss": 2.2348, "step": 828 }, { "epoch": 0.33671811535337126, "grad_norm": 0.0, "learning_rate": 1.5459623564245583e-05, "loss": 2.3686, "step": 829 }, { "epoch": 0.3371242891957758, "grad_norm": 0.0, "learning_rate": 1.5448596829350706e-05, "loss": 2.1565, "step": 830 }, { "epoch": 0.33753046303818035, "grad_norm": 0.0, "learning_rate": 1.5437560664368574e-05, "loss": 2.1969, "step": 831 }, { "epoch": 0.3379366368805849, "grad_norm": 0.0, "learning_rate": 1.5426515088399882e-05, "loss": 2.215, "step": 832 }, { "epoch": 0.33834281072298944, "grad_norm": 0.0, "learning_rate": 1.5415460120561624e-05, "loss": 2.133, "step": 833 }, { "epoch": 0.338748984565394, "grad_norm": 0.0, "learning_rate": 1.540439577998703e-05, "loss": 2.0909, "step": 834 }, { "epoch": 0.33915515840779853, "grad_norm": 0.0, "learning_rate": 1.539332208582557e-05, "loss": 2.2679, "step": 835 }, { "epoch": 0.3395613322502031, "grad_norm": 0.0, "learning_rate": 1.538223905724289e-05, "loss": 2.2209, "step": 836 }, { "epoch": 0.3399675060926076, "grad_norm": 0.0, "learning_rate": 1.5371146713420793e-05, "loss": 2.2076, "step": 837 }, { "epoch": 0.34037367993501216, "grad_norm": 0.0, "learning_rate": 1.5360045073557214e-05, "loss": 2.2735, "step": 838 }, { "epoch": 0.3407798537774167, "grad_norm": 0.0, "learning_rate": 1.5348934156866163e-05, "loss": 2.2328, "step": 839 }, { "epoch": 0.3411860276198213, "grad_norm": 0.0, "learning_rate": 1.533781398257772e-05, "loss": 2.248, "step": 840 }, { "epoch": 0.34159220146222585, "grad_norm": 0.0, "learning_rate": 1.5326684569937977e-05, "loss": 2.2936, "step": 841 }, { "epoch": 0.3419983753046304, "grad_norm": 0.0, "learning_rate": 1.5315545938209016e-05, "loss": 2.1081, "step": 842 }, { "epoch": 0.34240454914703494, "grad_norm": 0.0, "learning_rate": 1.530439810666888e-05, "loss": 2.4121, "step": 843 }, { "epoch": 0.3428107229894395, "grad_norm": 0.0, "learning_rate": 1.529324109461153e-05, "loss": 2.1981, "step": 844 }, { "epoch": 0.34321689683184403, "grad_norm": 0.0, "learning_rate": 1.5282074921346827e-05, "loss": 2.3542, "step": 845 }, { "epoch": 0.3436230706742486, "grad_norm": 0.0, "learning_rate": 1.527089960620046e-05, "loss": 2.1933, "step": 846 }, { "epoch": 0.3440292445166531, "grad_norm": 0.0, "learning_rate": 1.5259715168513976e-05, "loss": 2.186, "step": 847 }, { "epoch": 0.34443541835905767, "grad_norm": 0.0, "learning_rate": 1.5248521627644684e-05, "loss": 2.2015, "step": 848 }, { "epoch": 0.3448415922014622, "grad_norm": 0.0, "learning_rate": 1.523731900296566e-05, "loss": 2.2112, "step": 849 }, { "epoch": 0.34524776604386676, "grad_norm": 0.0, "learning_rate": 1.5226107313865701e-05, "loss": 2.1368, "step": 850 }, { "epoch": 0.3456539398862713, "grad_norm": 0.0, "learning_rate": 1.5214886579749286e-05, "loss": 2.1085, "step": 851 }, { "epoch": 0.3460601137286759, "grad_norm": 0.0, "learning_rate": 1.5203656820036558e-05, "loss": 2.2378, "step": 852 }, { "epoch": 0.34646628757108044, "grad_norm": 0.0, "learning_rate": 1.519241805416327e-05, "loss": 2.2941, "step": 853 }, { "epoch": 0.346872461413485, "grad_norm": 0.0, "learning_rate": 1.5181170301580776e-05, "loss": 2.1019, "step": 854 }, { "epoch": 0.34727863525588953, "grad_norm": 0.0, "learning_rate": 1.5169913581755968e-05, "loss": 2.1857, "step": 855 }, { "epoch": 0.3476848090982941, "grad_norm": 0.0, "learning_rate": 1.515864791417127e-05, "loss": 2.1655, "step": 856 }, { "epoch": 0.3480909829406986, "grad_norm": 0.0, "learning_rate": 1.5147373318324588e-05, "loss": 2.3202, "step": 857 }, { "epoch": 0.34849715678310317, "grad_norm": 0.0, "learning_rate": 1.5136089813729276e-05, "loss": 2.205, "step": 858 }, { "epoch": 0.3489033306255077, "grad_norm": 0.0, "learning_rate": 1.5124797419914115e-05, "loss": 2.3771, "step": 859 }, { "epoch": 0.34930950446791226, "grad_norm": 0.0, "learning_rate": 1.5113496156423271e-05, "loss": 2.3059, "step": 860 }, { "epoch": 0.3497156783103168, "grad_norm": 0.0, "learning_rate": 1.5102186042816244e-05, "loss": 2.0896, "step": 861 }, { "epoch": 0.35012185215272135, "grad_norm": 0.0, "learning_rate": 1.509086709866788e-05, "loss": 2.2108, "step": 862 }, { "epoch": 0.3505280259951259, "grad_norm": 0.0, "learning_rate": 1.5079539343568281e-05, "loss": 2.2938, "step": 863 }, { "epoch": 0.35093419983753044, "grad_norm": 0.0, "learning_rate": 1.5068202797122813e-05, "loss": 2.2504, "step": 864 }, { "epoch": 0.35134037367993504, "grad_norm": 0.0, "learning_rate": 1.505685747895206e-05, "loss": 2.281, "step": 865 }, { "epoch": 0.3517465475223396, "grad_norm": 0.0, "learning_rate": 1.5045503408691776e-05, "loss": 2.2701, "step": 866 }, { "epoch": 0.3521527213647441, "grad_norm": 0.0, "learning_rate": 1.5034140605992871e-05, "loss": 2.1694, "step": 867 }, { "epoch": 0.35255889520714867, "grad_norm": 0.0, "learning_rate": 1.5022769090521366e-05, "loss": 2.2359, "step": 868 }, { "epoch": 0.3529650690495532, "grad_norm": 0.0, "learning_rate": 1.5011388881958357e-05, "loss": 2.3505, "step": 869 }, { "epoch": 0.35337124289195776, "grad_norm": 0.0, "learning_rate": 1.5000000000000002e-05, "loss": 2.2891, "step": 870 }, { "epoch": 0.3537774167343623, "grad_norm": 0.0, "learning_rate": 1.498860246435745e-05, "loss": 2.265, "step": 871 }, { "epoch": 0.35418359057676685, "grad_norm": 0.0, "learning_rate": 1.4977196294756832e-05, "loss": 2.0977, "step": 872 }, { "epoch": 0.3545897644191714, "grad_norm": 0.0, "learning_rate": 1.4965781510939235e-05, "loss": 2.1478, "step": 873 }, { "epoch": 0.35499593826157594, "grad_norm": 0.0, "learning_rate": 1.495435813266064e-05, "loss": 2.1593, "step": 874 }, { "epoch": 0.3554021121039805, "grad_norm": 0.0, "learning_rate": 1.4942926179691912e-05, "loss": 2.1062, "step": 875 }, { "epoch": 0.35580828594638503, "grad_norm": 0.0, "learning_rate": 1.493148567181875e-05, "loss": 2.1847, "step": 876 }, { "epoch": 0.35621445978878963, "grad_norm": 0.0, "learning_rate": 1.4920036628841666e-05, "loss": 2.4001, "step": 877 }, { "epoch": 0.3566206336311942, "grad_norm": 0.0, "learning_rate": 1.4908579070575936e-05, "loss": 2.2469, "step": 878 }, { "epoch": 0.3570268074735987, "grad_norm": 0.0, "learning_rate": 1.4897113016851584e-05, "loss": 2.2022, "step": 879 }, { "epoch": 0.35743298131600326, "grad_norm": 0.0, "learning_rate": 1.4885638487513327e-05, "loss": 2.1775, "step": 880 }, { "epoch": 0.3578391551584078, "grad_norm": 0.0, "learning_rate": 1.4874155502420558e-05, "loss": 2.2763, "step": 881 }, { "epoch": 0.35824532900081235, "grad_norm": 0.0, "learning_rate": 1.4862664081447297e-05, "loss": 2.3674, "step": 882 }, { "epoch": 0.3586515028432169, "grad_norm": 0.0, "learning_rate": 1.4851164244482182e-05, "loss": 2.2917, "step": 883 }, { "epoch": 0.35905767668562144, "grad_norm": 0.0, "learning_rate": 1.483965601142839e-05, "loss": 2.1617, "step": 884 }, { "epoch": 0.359463850528026, "grad_norm": 0.0, "learning_rate": 1.482813940220365e-05, "loss": 2.2967, "step": 885 }, { "epoch": 0.35987002437043053, "grad_norm": 0.0, "learning_rate": 1.4816614436740184e-05, "loss": 2.3223, "step": 886 }, { "epoch": 0.3602761982128351, "grad_norm": 0.0, "learning_rate": 1.4805081134984673e-05, "loss": 2.2353, "step": 887 }, { "epoch": 0.3606823720552396, "grad_norm": 0.0, "learning_rate": 1.4793539516898226e-05, "loss": 2.0492, "step": 888 }, { "epoch": 0.36108854589764416, "grad_norm": 0.0, "learning_rate": 1.4781989602456346e-05, "loss": 2.3003, "step": 889 }, { "epoch": 0.36149471974004876, "grad_norm": 0.0, "learning_rate": 1.4770431411648898e-05, "loss": 2.253, "step": 890 }, { "epoch": 0.3619008935824533, "grad_norm": 0.0, "learning_rate": 1.4758864964480066e-05, "loss": 2.2501, "step": 891 }, { "epoch": 0.36230706742485785, "grad_norm": 0.0, "learning_rate": 1.474729028096833e-05, "loss": 2.1898, "step": 892 }, { "epoch": 0.3627132412672624, "grad_norm": 0.0, "learning_rate": 1.4735707381146416e-05, "loss": 2.0529, "step": 893 }, { "epoch": 0.36311941510966694, "grad_norm": 0.0, "learning_rate": 1.4724116285061278e-05, "loss": 2.2086, "step": 894 }, { "epoch": 0.3635255889520715, "grad_norm": 0.0, "learning_rate": 1.4712517012774056e-05, "loss": 2.1241, "step": 895 }, { "epoch": 0.36393176279447603, "grad_norm": 0.0, "learning_rate": 1.4700909584360032e-05, "loss": 2.208, "step": 896 }, { "epoch": 0.3643379366368806, "grad_norm": 0.0, "learning_rate": 1.4689294019908614e-05, "loss": 2.044, "step": 897 }, { "epoch": 0.3647441104792851, "grad_norm": 0.0, "learning_rate": 1.4677670339523285e-05, "loss": 2.2585, "step": 898 }, { "epoch": 0.36515028432168967, "grad_norm": 0.0, "learning_rate": 1.4666038563321578e-05, "loss": 2.3411, "step": 899 }, { "epoch": 0.3655564581640942, "grad_norm": 0.0, "learning_rate": 1.465439871143504e-05, "loss": 2.2262, "step": 900 }, { "epoch": 0.36596263200649876, "grad_norm": 0.0, "learning_rate": 1.4642750804009184e-05, "loss": 2.1497, "step": 901 }, { "epoch": 0.36636880584890336, "grad_norm": 0.0, "learning_rate": 1.4631094861203478e-05, "loss": 2.1188, "step": 902 }, { "epoch": 0.3667749796913079, "grad_norm": 0.0, "learning_rate": 1.4619430903191292e-05, "loss": 2.0915, "step": 903 }, { "epoch": 0.36718115353371245, "grad_norm": 0.0, "learning_rate": 1.4607758950159862e-05, "loss": 2.1206, "step": 904 }, { "epoch": 0.367587327376117, "grad_norm": 0.0, "learning_rate": 1.4596079022310277e-05, "loss": 2.3112, "step": 905 }, { "epoch": 0.36799350121852153, "grad_norm": 0.0, "learning_rate": 1.4584391139857407e-05, "loss": 2.2055, "step": 906 }, { "epoch": 0.3683996750609261, "grad_norm": 0.0, "learning_rate": 1.4572695323029909e-05, "loss": 2.1936, "step": 907 }, { "epoch": 0.3688058489033306, "grad_norm": 0.0, "learning_rate": 1.4560991592070159e-05, "loss": 2.2516, "step": 908 }, { "epoch": 0.36921202274573517, "grad_norm": 0.0, "learning_rate": 1.454927996723424e-05, "loss": 2.2435, "step": 909 }, { "epoch": 0.3696181965881397, "grad_norm": 0.0, "learning_rate": 1.4537560468791889e-05, "loss": 2.2565, "step": 910 }, { "epoch": 0.37002437043054426, "grad_norm": 0.0, "learning_rate": 1.4525833117026476e-05, "loss": 2.1004, "step": 911 }, { "epoch": 0.3704305442729488, "grad_norm": 0.0, "learning_rate": 1.4514097932234959e-05, "loss": 2.2345, "step": 912 }, { "epoch": 0.37083671811535335, "grad_norm": 0.0, "learning_rate": 1.4502354934727855e-05, "loss": 2.2167, "step": 913 }, { "epoch": 0.37124289195775795, "grad_norm": 0.0, "learning_rate": 1.4490604144829204e-05, "loss": 2.3113, "step": 914 }, { "epoch": 0.3716490658001625, "grad_norm": 0.0, "learning_rate": 1.4478845582876528e-05, "loss": 2.1785, "step": 915 }, { "epoch": 0.37205523964256704, "grad_norm": 0.0, "learning_rate": 1.4467079269220805e-05, "loss": 2.4004, "step": 916 }, { "epoch": 0.3724614134849716, "grad_norm": 0.0, "learning_rate": 1.4455305224226429e-05, "loss": 2.1866, "step": 917 }, { "epoch": 0.3728675873273761, "grad_norm": 0.0, "learning_rate": 1.4443523468271168e-05, "loss": 2.3145, "step": 918 }, { "epoch": 0.37327376116978067, "grad_norm": 0.0, "learning_rate": 1.4431734021746147e-05, "loss": 2.3074, "step": 919 }, { "epoch": 0.3736799350121852, "grad_norm": 0.0, "learning_rate": 1.4419936905055794e-05, "loss": 2.2851, "step": 920 }, { "epoch": 0.37408610885458976, "grad_norm": 0.0, "learning_rate": 1.4408132138617812e-05, "loss": 2.1813, "step": 921 }, { "epoch": 0.3744922826969943, "grad_norm": 0.0, "learning_rate": 1.4396319742863145e-05, "loss": 2.1905, "step": 922 }, { "epoch": 0.37489845653939885, "grad_norm": 0.0, "learning_rate": 1.4384499738235942e-05, "loss": 2.2164, "step": 923 }, { "epoch": 0.3753046303818034, "grad_norm": 0.0, "learning_rate": 1.4372672145193523e-05, "loss": 2.1487, "step": 924 }, { "epoch": 0.37571080422420794, "grad_norm": 0.0, "learning_rate": 1.4360836984206336e-05, "loss": 2.1409, "step": 925 }, { "epoch": 0.3761169780666125, "grad_norm": 0.0, "learning_rate": 1.4348994275757933e-05, "loss": 2.0789, "step": 926 }, { "epoch": 0.3765231519090171, "grad_norm": 0.0, "learning_rate": 1.4337144040344919e-05, "loss": 2.1659, "step": 927 }, { "epoch": 0.37692932575142163, "grad_norm": 0.0, "learning_rate": 1.4325286298476941e-05, "loss": 2.1127, "step": 928 }, { "epoch": 0.3773354995938262, "grad_norm": 0.0, "learning_rate": 1.4313421070676626e-05, "loss": 2.0988, "step": 929 }, { "epoch": 0.3777416734362307, "grad_norm": 0.0, "learning_rate": 1.4301548377479562e-05, "loss": 2.1127, "step": 930 }, { "epoch": 0.37814784727863526, "grad_norm": 0.0, "learning_rate": 1.4289668239434256e-05, "loss": 2.1982, "step": 931 }, { "epoch": 0.3785540211210398, "grad_norm": 0.0, "learning_rate": 1.4277780677102098e-05, "loss": 2.2576, "step": 932 }, { "epoch": 0.37896019496344435, "grad_norm": 0.0, "learning_rate": 1.4265885711057333e-05, "loss": 2.3726, "step": 933 }, { "epoch": 0.3793663688058489, "grad_norm": 0.0, "learning_rate": 1.4253983361887017e-05, "loss": 2.1743, "step": 934 }, { "epoch": 0.37977254264825344, "grad_norm": 0.0, "learning_rate": 1.4242073650190984e-05, "loss": 2.2292, "step": 935 }, { "epoch": 0.380178716490658, "grad_norm": 0.0, "learning_rate": 1.423015659658181e-05, "loss": 2.2294, "step": 936 }, { "epoch": 0.38058489033306253, "grad_norm": 0.0, "learning_rate": 1.4218232221684778e-05, "loss": 2.1919, "step": 937 }, { "epoch": 0.3809910641754671, "grad_norm": 0.0, "learning_rate": 1.4206300546137844e-05, "loss": 2.2381, "step": 938 }, { "epoch": 0.3813972380178717, "grad_norm": 0.0, "learning_rate": 1.4194361590591599e-05, "loss": 2.1449, "step": 939 }, { "epoch": 0.3818034118602762, "grad_norm": 0.0, "learning_rate": 1.418241537570923e-05, "loss": 2.166, "step": 940 }, { "epoch": 0.38220958570268077, "grad_norm": 0.0, "learning_rate": 1.4170461922166499e-05, "loss": 2.2141, "step": 941 }, { "epoch": 0.3826157595450853, "grad_norm": 0.0, "learning_rate": 1.415850125065168e-05, "loss": 2.1729, "step": 942 }, { "epoch": 0.38302193338748985, "grad_norm": 0.0, "learning_rate": 1.414653338186555e-05, "loss": 2.1624, "step": 943 }, { "epoch": 0.3834281072298944, "grad_norm": 0.0, "learning_rate": 1.4134558336521342e-05, "loss": 2.1137, "step": 944 }, { "epoch": 0.38383428107229894, "grad_norm": 0.0, "learning_rate": 1.412257613534471e-05, "loss": 2.0904, "step": 945 }, { "epoch": 0.3842404549147035, "grad_norm": 0.0, "learning_rate": 1.4110586799073684e-05, "loss": 2.2695, "step": 946 }, { "epoch": 0.38464662875710803, "grad_norm": 0.0, "learning_rate": 1.4098590348458658e-05, "loss": 2.2839, "step": 947 }, { "epoch": 0.3850528025995126, "grad_norm": 0.0, "learning_rate": 1.4086586804262323e-05, "loss": 2.3075, "step": 948 }, { "epoch": 0.3854589764419171, "grad_norm": 0.0, "learning_rate": 1.4074576187259664e-05, "loss": 2.1906, "step": 949 }, { "epoch": 0.38586515028432167, "grad_norm": 0.0, "learning_rate": 1.4062558518237893e-05, "loss": 2.2454, "step": 950 }, { "epoch": 0.3862713241267262, "grad_norm": 0.0, "learning_rate": 1.4050533817996432e-05, "loss": 2.0874, "step": 951 }, { "epoch": 0.3866774979691308, "grad_norm": 0.0, "learning_rate": 1.4038502107346874e-05, "loss": 2.2704, "step": 952 }, { "epoch": 0.38708367181153536, "grad_norm": 0.0, "learning_rate": 1.4026463407112943e-05, "loss": 2.2519, "step": 953 }, { "epoch": 0.3874898456539399, "grad_norm": 0.0, "learning_rate": 1.4014417738130464e-05, "loss": 2.2242, "step": 954 }, { "epoch": 0.38789601949634445, "grad_norm": 0.0, "learning_rate": 1.4002365121247316e-05, "loss": 2.2161, "step": 955 }, { "epoch": 0.388302193338749, "grad_norm": 0.0, "learning_rate": 1.399030557732341e-05, "loss": 2.1168, "step": 956 }, { "epoch": 0.38870836718115354, "grad_norm": 0.0, "learning_rate": 1.3978239127230642e-05, "loss": 2.106, "step": 957 }, { "epoch": 0.3891145410235581, "grad_norm": 0.0, "learning_rate": 1.3966165791852862e-05, "loss": 2.0327, "step": 958 }, { "epoch": 0.3895207148659626, "grad_norm": 0.0, "learning_rate": 1.3954085592085835e-05, "loss": 2.2445, "step": 959 }, { "epoch": 0.38992688870836717, "grad_norm": 0.0, "learning_rate": 1.394199854883721e-05, "loss": 2.1192, "step": 960 }, { "epoch": 0.3903330625507717, "grad_norm": 0.0, "learning_rate": 1.3929904683026477e-05, "loss": 2.16, "step": 961 }, { "epoch": 0.39073923639317626, "grad_norm": 0.0, "learning_rate": 1.3917804015584932e-05, "loss": 2.2787, "step": 962 }, { "epoch": 0.3911454102355808, "grad_norm": 0.0, "learning_rate": 1.3905696567455649e-05, "loss": 2.0523, "step": 963 }, { "epoch": 0.3915515840779854, "grad_norm": 0.0, "learning_rate": 1.3893582359593436e-05, "loss": 2.326, "step": 964 }, { "epoch": 0.39195775792038995, "grad_norm": 0.0, "learning_rate": 1.3881461412964799e-05, "loss": 2.1414, "step": 965 }, { "epoch": 0.3923639317627945, "grad_norm": 0.0, "learning_rate": 1.3869333748547901e-05, "loss": 2.2256, "step": 966 }, { "epoch": 0.39277010560519904, "grad_norm": 0.0, "learning_rate": 1.3857199387332543e-05, "loss": 2.1903, "step": 967 }, { "epoch": 0.3931762794476036, "grad_norm": 0.0, "learning_rate": 1.3845058350320109e-05, "loss": 2.2899, "step": 968 }, { "epoch": 0.3935824532900081, "grad_norm": 0.0, "learning_rate": 1.3832910658523542e-05, "loss": 2.1873, "step": 969 }, { "epoch": 0.39398862713241267, "grad_norm": 0.0, "learning_rate": 1.3820756332967294e-05, "loss": 2.1731, "step": 970 }, { "epoch": 0.3943948009748172, "grad_norm": 0.0, "learning_rate": 1.3808595394687309e-05, "loss": 2.2098, "step": 971 }, { "epoch": 0.39480097481722176, "grad_norm": 0.0, "learning_rate": 1.3796427864730968e-05, "loss": 2.1245, "step": 972 }, { "epoch": 0.3952071486596263, "grad_norm": 0.0, "learning_rate": 1.3784253764157063e-05, "loss": 2.1876, "step": 973 }, { "epoch": 0.39561332250203085, "grad_norm": 0.0, "learning_rate": 1.3772073114035762e-05, "loss": 2.0982, "step": 974 }, { "epoch": 0.3960194963444354, "grad_norm": 0.0, "learning_rate": 1.3759885935448563e-05, "loss": 2.3171, "step": 975 }, { "epoch": 0.39642567018683994, "grad_norm": 0.0, "learning_rate": 1.3747692249488263e-05, "loss": 2.141, "step": 976 }, { "epoch": 0.39683184402924454, "grad_norm": 0.0, "learning_rate": 1.3735492077258924e-05, "loss": 2.2873, "step": 977 }, { "epoch": 0.3972380178716491, "grad_norm": 0.0, "learning_rate": 1.3723285439875836e-05, "loss": 2.1973, "step": 978 }, { "epoch": 0.39764419171405363, "grad_norm": 0.0, "learning_rate": 1.3711072358465473e-05, "loss": 2.2947, "step": 979 }, { "epoch": 0.3980503655564582, "grad_norm": 0.0, "learning_rate": 1.369885285416547e-05, "loss": 2.2924, "step": 980 }, { "epoch": 0.3984565393988627, "grad_norm": 0.0, "learning_rate": 1.368662694812457e-05, "loss": 2.2087, "step": 981 }, { "epoch": 0.39886271324126726, "grad_norm": 0.0, "learning_rate": 1.3674394661502595e-05, "loss": 2.282, "step": 982 }, { "epoch": 0.3992688870836718, "grad_norm": 0.0, "learning_rate": 1.3662156015470419e-05, "loss": 2.1999, "step": 983 }, { "epoch": 0.39967506092607635, "grad_norm": 0.0, "learning_rate": 1.3649911031209915e-05, "loss": 2.1088, "step": 984 }, { "epoch": 0.4000812347684809, "grad_norm": 0.0, "learning_rate": 1.3637659729913933e-05, "loss": 2.2563, "step": 985 }, { "epoch": 0.40048740861088544, "grad_norm": 0.0, "learning_rate": 1.3625402132786247e-05, "loss": 2.1971, "step": 986 }, { "epoch": 0.40089358245329, "grad_norm": 0.0, "learning_rate": 1.3613138261041532e-05, "loss": 2.2009, "step": 987 }, { "epoch": 0.40129975629569453, "grad_norm": 0.0, "learning_rate": 1.3600868135905323e-05, "loss": 2.1452, "step": 988 }, { "epoch": 0.40170593013809913, "grad_norm": 0.0, "learning_rate": 1.358859177861398e-05, "loss": 2.2133, "step": 989 }, { "epoch": 0.4021121039805037, "grad_norm": 0.0, "learning_rate": 1.3576309210414646e-05, "loss": 2.2156, "step": 990 }, { "epoch": 0.4025182778229082, "grad_norm": 0.0, "learning_rate": 1.3564020452565212e-05, "loss": 2.1613, "step": 991 }, { "epoch": 0.40292445166531277, "grad_norm": 0.0, "learning_rate": 1.3551725526334286e-05, "loss": 2.2145, "step": 992 }, { "epoch": 0.4033306255077173, "grad_norm": 0.0, "learning_rate": 1.3539424453001147e-05, "loss": 2.1498, "step": 993 }, { "epoch": 0.40373679935012186, "grad_norm": 0.0, "learning_rate": 1.352711725385572e-05, "loss": 2.3039, "step": 994 }, { "epoch": 0.4041429731925264, "grad_norm": 0.0, "learning_rate": 1.3514803950198526e-05, "loss": 2.1716, "step": 995 }, { "epoch": 0.40454914703493094, "grad_norm": 0.0, "learning_rate": 1.3502484563340652e-05, "loss": 2.2096, "step": 996 }, { "epoch": 0.4049553208773355, "grad_norm": 0.0, "learning_rate": 1.3490159114603713e-05, "loss": 2.2316, "step": 997 }, { "epoch": 0.40536149471974003, "grad_norm": 0.0, "learning_rate": 1.3477827625319826e-05, "loss": 2.212, "step": 998 }, { "epoch": 0.4057676685621446, "grad_norm": 0.0, "learning_rate": 1.3465490116831543e-05, "loss": 2.1684, "step": 999 }, { "epoch": 0.4061738424045491, "grad_norm": 0.0, "learning_rate": 1.3453146610491852e-05, "loss": 2.2691, "step": 1000 }, { "epoch": 0.4065800162469537, "grad_norm": 0.0, "learning_rate": 1.3440797127664112e-05, "loss": 2.1851, "step": 1001 }, { "epoch": 0.40698619008935827, "grad_norm": 0.0, "learning_rate": 1.3428441689722023e-05, "loss": 2.2404, "step": 1002 }, { "epoch": 0.4073923639317628, "grad_norm": 0.0, "learning_rate": 1.3416080318049603e-05, "loss": 2.1654, "step": 1003 }, { "epoch": 0.40779853777416736, "grad_norm": 0.0, "learning_rate": 1.340371303404113e-05, "loss": 2.0954, "step": 1004 }, { "epoch": 0.4082047116165719, "grad_norm": 0.0, "learning_rate": 1.3391339859101122e-05, "loss": 2.1074, "step": 1005 }, { "epoch": 0.40861088545897645, "grad_norm": 0.0, "learning_rate": 1.3378960814644283e-05, "loss": 2.2064, "step": 1006 }, { "epoch": 0.409017059301381, "grad_norm": 0.0, "learning_rate": 1.3366575922095484e-05, "loss": 2.123, "step": 1007 }, { "epoch": 0.40942323314378554, "grad_norm": 0.0, "learning_rate": 1.3354185202889716e-05, "loss": 2.0838, "step": 1008 }, { "epoch": 0.4098294069861901, "grad_norm": 0.0, "learning_rate": 1.334178867847205e-05, "loss": 2.2245, "step": 1009 }, { "epoch": 0.4102355808285946, "grad_norm": 0.0, "learning_rate": 1.3329386370297615e-05, "loss": 2.1052, "step": 1010 }, { "epoch": 0.41064175467099917, "grad_norm": 0.0, "learning_rate": 1.3316978299831531e-05, "loss": 2.0682, "step": 1011 }, { "epoch": 0.4110479285134037, "grad_norm": 0.0, "learning_rate": 1.3304564488548912e-05, "loss": 2.21, "step": 1012 }, { "epoch": 0.41145410235580826, "grad_norm": 0.0, "learning_rate": 1.3292144957934795e-05, "loss": 2.2127, "step": 1013 }, { "epoch": 0.41186027619821286, "grad_norm": 0.0, "learning_rate": 1.3279719729484117e-05, "loss": 2.175, "step": 1014 }, { "epoch": 0.4122664500406174, "grad_norm": 0.0, "learning_rate": 1.3267288824701686e-05, "loss": 2.2394, "step": 1015 }, { "epoch": 0.41267262388302195, "grad_norm": 0.0, "learning_rate": 1.3254852265102118e-05, "loss": 2.1575, "step": 1016 }, { "epoch": 0.4130787977254265, "grad_norm": 0.0, "learning_rate": 1.3242410072209831e-05, "loss": 2.2078, "step": 1017 }, { "epoch": 0.41348497156783104, "grad_norm": 0.0, "learning_rate": 1.3229962267558982e-05, "loss": 2.2983, "step": 1018 }, { "epoch": 0.4138911454102356, "grad_norm": 0.0, "learning_rate": 1.321750887269345e-05, "loss": 2.1222, "step": 1019 }, { "epoch": 0.41429731925264013, "grad_norm": 0.0, "learning_rate": 1.3205049909166782e-05, "loss": 2.1395, "step": 1020 }, { "epoch": 0.4147034930950447, "grad_norm": 0.0, "learning_rate": 1.3192585398542163e-05, "loss": 2.1075, "step": 1021 }, { "epoch": 0.4151096669374492, "grad_norm": 0.0, "learning_rate": 1.3180115362392383e-05, "loss": 2.0704, "step": 1022 }, { "epoch": 0.41551584077985376, "grad_norm": 0.0, "learning_rate": 1.3167639822299794e-05, "loss": 2.2332, "step": 1023 }, { "epoch": 0.4159220146222583, "grad_norm": 0.0, "learning_rate": 1.3155158799856269e-05, "loss": 2.169, "step": 1024 }, { "epoch": 0.41632818846466285, "grad_norm": 0.0, "learning_rate": 1.3142672316663178e-05, "loss": 2.1922, "step": 1025 }, { "epoch": 0.41673436230706745, "grad_norm": 0.0, "learning_rate": 1.3130180394331335e-05, "loss": 2.2715, "step": 1026 }, { "epoch": 0.417140536149472, "grad_norm": 0.0, "learning_rate": 1.3117683054480971e-05, "loss": 2.2056, "step": 1027 }, { "epoch": 0.41754670999187654, "grad_norm": 0.0, "learning_rate": 1.3105180318741691e-05, "loss": 2.1209, "step": 1028 }, { "epoch": 0.4179528838342811, "grad_norm": 0.0, "learning_rate": 1.309267220875244e-05, "loss": 2.0563, "step": 1029 }, { "epoch": 0.41835905767668563, "grad_norm": 0.0, "learning_rate": 1.3080158746161468e-05, "loss": 2.177, "step": 1030 }, { "epoch": 0.4187652315190902, "grad_norm": 0.0, "learning_rate": 1.306763995262628e-05, "loss": 2.232, "step": 1031 }, { "epoch": 0.4191714053614947, "grad_norm": 0.0, "learning_rate": 1.3055115849813622e-05, "loss": 2.1869, "step": 1032 }, { "epoch": 0.41957757920389926, "grad_norm": 0.0, "learning_rate": 1.3042586459399411e-05, "loss": 2.1057, "step": 1033 }, { "epoch": 0.4199837530463038, "grad_norm": 0.0, "learning_rate": 1.3030051803068729e-05, "loss": 2.2688, "step": 1034 }, { "epoch": 0.42038992688870835, "grad_norm": 0.0, "learning_rate": 1.3017511902515769e-05, "loss": 2.0909, "step": 1035 }, { "epoch": 0.4207961007311129, "grad_norm": 0.0, "learning_rate": 1.3004966779443794e-05, "loss": 2.127, "step": 1036 }, { "epoch": 0.42120227457351744, "grad_norm": 0.0, "learning_rate": 1.2992416455565114e-05, "loss": 2.1648, "step": 1037 }, { "epoch": 0.421608448415922, "grad_norm": 0.0, "learning_rate": 1.2979860952601038e-05, "loss": 2.2381, "step": 1038 }, { "epoch": 0.4220146222583266, "grad_norm": 0.0, "learning_rate": 1.2967300292281833e-05, "loss": 2.1502, "step": 1039 }, { "epoch": 0.42242079610073113, "grad_norm": 0.0, "learning_rate": 1.2954734496346704e-05, "loss": 2.112, "step": 1040 }, { "epoch": 0.4228269699431357, "grad_norm": 0.0, "learning_rate": 1.2942163586543728e-05, "loss": 2.1404, "step": 1041 }, { "epoch": 0.4232331437855402, "grad_norm": 0.0, "learning_rate": 1.2929587584629845e-05, "loss": 2.3397, "step": 1042 }, { "epoch": 0.42363931762794477, "grad_norm": 0.0, "learning_rate": 1.2917006512370809e-05, "loss": 2.1124, "step": 1043 }, { "epoch": 0.4240454914703493, "grad_norm": 0.0, "learning_rate": 1.2904420391541142e-05, "loss": 2.1463, "step": 1044 }, { "epoch": 0.42445166531275386, "grad_norm": 0.0, "learning_rate": 1.2891829243924108e-05, "loss": 2.2984, "step": 1045 }, { "epoch": 0.4248578391551584, "grad_norm": 0.0, "learning_rate": 1.2879233091311667e-05, "loss": 2.1716, "step": 1046 }, { "epoch": 0.42526401299756295, "grad_norm": 0.0, "learning_rate": 1.2866631955504445e-05, "loss": 2.2394, "step": 1047 }, { "epoch": 0.4256701868399675, "grad_norm": 0.0, "learning_rate": 1.2854025858311693e-05, "loss": 2.2095, "step": 1048 }, { "epoch": 0.42607636068237204, "grad_norm": 0.0, "learning_rate": 1.2841414821551252e-05, "loss": 2.1755, "step": 1049 }, { "epoch": 0.4264825345247766, "grad_norm": 0.0, "learning_rate": 1.2828798867049504e-05, "loss": 2.2072, "step": 1050 }, { "epoch": 0.4268887083671812, "grad_norm": 0.0, "learning_rate": 1.2816178016641344e-05, "loss": 2.2541, "step": 1051 }, { "epoch": 0.4272948822095857, "grad_norm": 0.0, "learning_rate": 1.2803552292170145e-05, "loss": 2.0908, "step": 1052 }, { "epoch": 0.42770105605199027, "grad_norm": 0.0, "learning_rate": 1.2790921715487715e-05, "loss": 2.3538, "step": 1053 }, { "epoch": 0.4281072298943948, "grad_norm": 0.0, "learning_rate": 1.2778286308454255e-05, "loss": 2.252, "step": 1054 }, { "epoch": 0.42851340373679936, "grad_norm": 0.0, "learning_rate": 1.2765646092938333e-05, "loss": 2.1349, "step": 1055 }, { "epoch": 0.4289195775792039, "grad_norm": 0.0, "learning_rate": 1.2753001090816832e-05, "loss": 2.2204, "step": 1056 }, { "epoch": 0.42932575142160845, "grad_norm": 0.0, "learning_rate": 1.2740351323974929e-05, "loss": 2.2027, "step": 1057 }, { "epoch": 0.429731925264013, "grad_norm": 0.0, "learning_rate": 1.2727696814306034e-05, "loss": 2.2562, "step": 1058 }, { "epoch": 0.43013809910641754, "grad_norm": 0.0, "learning_rate": 1.2715037583711775e-05, "loss": 2.2213, "step": 1059 }, { "epoch": 0.4305442729488221, "grad_norm": 0.0, "learning_rate": 1.2702373654101954e-05, "loss": 2.1633, "step": 1060 }, { "epoch": 0.4309504467912266, "grad_norm": 0.0, "learning_rate": 1.2689705047394494e-05, "loss": 2.2265, "step": 1061 }, { "epoch": 0.43135662063363117, "grad_norm": 0.0, "learning_rate": 1.2677031785515423e-05, "loss": 2.2089, "step": 1062 }, { "epoch": 0.4317627944760357, "grad_norm": 0.0, "learning_rate": 1.2664353890398822e-05, "loss": 2.1772, "step": 1063 }, { "epoch": 0.4321689683184403, "grad_norm": 0.0, "learning_rate": 1.2651671383986788e-05, "loss": 2.1418, "step": 1064 }, { "epoch": 0.43257514216084486, "grad_norm": 0.0, "learning_rate": 1.2638984288229408e-05, "loss": 2.051, "step": 1065 }, { "epoch": 0.4329813160032494, "grad_norm": 0.0, "learning_rate": 1.26262926250847e-05, "loss": 2.111, "step": 1066 }, { "epoch": 0.43338748984565395, "grad_norm": 0.0, "learning_rate": 1.2613596416518594e-05, "loss": 2.2185, "step": 1067 }, { "epoch": 0.4337936636880585, "grad_norm": 0.0, "learning_rate": 1.260089568450489e-05, "loss": 2.1404, "step": 1068 }, { "epoch": 0.43419983753046304, "grad_norm": 0.0, "learning_rate": 1.2588190451025209e-05, "loss": 1.9593, "step": 1069 }, { "epoch": 0.4346060113728676, "grad_norm": 0.0, "learning_rate": 1.2575480738068971e-05, "loss": 2.112, "step": 1070 }, { "epoch": 0.43501218521527213, "grad_norm": 0.0, "learning_rate": 1.2562766567633336e-05, "loss": 2.1948, "step": 1071 }, { "epoch": 0.4354183590576767, "grad_norm": 0.0, "learning_rate": 1.2550047961723192e-05, "loss": 2.3552, "step": 1072 }, { "epoch": 0.4358245329000812, "grad_norm": 0.0, "learning_rate": 1.2537324942351101e-05, "loss": 2.2378, "step": 1073 }, { "epoch": 0.43623070674248576, "grad_norm": 0.0, "learning_rate": 1.2524597531537261e-05, "loss": 2.2192, "step": 1074 }, { "epoch": 0.4366368805848903, "grad_norm": 0.0, "learning_rate": 1.2511865751309468e-05, "loss": 2.1775, "step": 1075 }, { "epoch": 0.4370430544272949, "grad_norm": 0.0, "learning_rate": 1.2499129623703086e-05, "loss": 2.2501, "step": 1076 }, { "epoch": 0.43744922826969945, "grad_norm": 0.0, "learning_rate": 1.2486389170760995e-05, "loss": 2.1625, "step": 1077 }, { "epoch": 0.437855402112104, "grad_norm": 0.0, "learning_rate": 1.2473644414533573e-05, "loss": 2.2327, "step": 1078 }, { "epoch": 0.43826157595450854, "grad_norm": 0.0, "learning_rate": 1.2460895377078632e-05, "loss": 2.1417, "step": 1079 }, { "epoch": 0.4386677497969131, "grad_norm": 0.0, "learning_rate": 1.244814208046141e-05, "loss": 2.2405, "step": 1080 }, { "epoch": 0.43907392363931763, "grad_norm": 0.0, "learning_rate": 1.2435384546754496e-05, "loss": 2.1885, "step": 1081 }, { "epoch": 0.4394800974817222, "grad_norm": 0.0, "learning_rate": 1.2422622798037833e-05, "loss": 2.0357, "step": 1082 }, { "epoch": 0.4398862713241267, "grad_norm": 0.0, "learning_rate": 1.2409856856398647e-05, "loss": 2.2452, "step": 1083 }, { "epoch": 0.44029244516653127, "grad_norm": 0.0, "learning_rate": 1.2397086743931421e-05, "loss": 2.1663, "step": 1084 }, { "epoch": 0.4406986190089358, "grad_norm": 0.0, "learning_rate": 1.2384312482737858e-05, "loss": 2.1294, "step": 1085 }, { "epoch": 0.44110479285134035, "grad_norm": 0.0, "learning_rate": 1.2371534094926852e-05, "loss": 2.2786, "step": 1086 }, { "epoch": 0.4415109666937449, "grad_norm": 0.0, "learning_rate": 1.2358751602614416e-05, "loss": 2.196, "step": 1087 }, { "epoch": 0.44191714053614944, "grad_norm": 0.0, "learning_rate": 1.234596502792369e-05, "loss": 2.3181, "step": 1088 }, { "epoch": 0.44232331437855404, "grad_norm": 0.0, "learning_rate": 1.2333174392984866e-05, "loss": 2.2173, "step": 1089 }, { "epoch": 0.4427294882209586, "grad_norm": 0.0, "learning_rate": 1.232037971993517e-05, "loss": 2.2705, "step": 1090 }, { "epoch": 0.44313566206336313, "grad_norm": 0.0, "learning_rate": 1.2307581030918807e-05, "loss": 2.1379, "step": 1091 }, { "epoch": 0.4435418359057677, "grad_norm": 0.0, "learning_rate": 1.2294778348086949e-05, "loss": 2.1896, "step": 1092 }, { "epoch": 0.4439480097481722, "grad_norm": 0.0, "learning_rate": 1.2281971693597663e-05, "loss": 2.2441, "step": 1093 }, { "epoch": 0.44435418359057677, "grad_norm": 0.0, "learning_rate": 1.2269161089615902e-05, "loss": 2.1991, "step": 1094 }, { "epoch": 0.4447603574329813, "grad_norm": 0.0, "learning_rate": 1.2256346558313452e-05, "loss": 2.1313, "step": 1095 }, { "epoch": 0.44516653127538586, "grad_norm": 0.0, "learning_rate": 1.224352812186889e-05, "loss": 2.0838, "step": 1096 }, { "epoch": 0.4455727051177904, "grad_norm": 0.0, "learning_rate": 1.2230705802467557e-05, "loss": 2.1562, "step": 1097 }, { "epoch": 0.44597887896019495, "grad_norm": 0.0, "learning_rate": 1.2217879622301514e-05, "loss": 2.262, "step": 1098 }, { "epoch": 0.4463850528025995, "grad_norm": 0.0, "learning_rate": 1.2205049603569505e-05, "loss": 2.2169, "step": 1099 }, { "epoch": 0.44679122664500404, "grad_norm": 0.0, "learning_rate": 1.2192215768476915e-05, "loss": 2.0763, "step": 1100 }, { "epoch": 0.44719740048740864, "grad_norm": 0.0, "learning_rate": 1.2179378139235735e-05, "loss": 2.0608, "step": 1101 }, { "epoch": 0.4476035743298132, "grad_norm": 0.0, "learning_rate": 1.2166536738064523e-05, "loss": 2.2179, "step": 1102 }, { "epoch": 0.4480097481722177, "grad_norm": 0.0, "learning_rate": 1.2153691587188363e-05, "loss": 2.2493, "step": 1103 }, { "epoch": 0.44841592201462227, "grad_norm": 0.0, "learning_rate": 1.2140842708838836e-05, "loss": 2.2148, "step": 1104 }, { "epoch": 0.4488220958570268, "grad_norm": 0.0, "learning_rate": 1.2127990125253966e-05, "loss": 2.2295, "step": 1105 }, { "epoch": 0.44922826969943136, "grad_norm": 0.0, "learning_rate": 1.2115133858678192e-05, "loss": 2.1762, "step": 1106 }, { "epoch": 0.4496344435418359, "grad_norm": 0.0, "learning_rate": 1.2102273931362327e-05, "loss": 2.0684, "step": 1107 }, { "epoch": 0.45004061738424045, "grad_norm": 0.0, "learning_rate": 1.2089410365563525e-05, "loss": 2.1168, "step": 1108 }, { "epoch": 0.450446791226645, "grad_norm": 0.0, "learning_rate": 1.207654318354523e-05, "loss": 2.1655, "step": 1109 }, { "epoch": 0.45085296506904954, "grad_norm": 0.0, "learning_rate": 1.2063672407577154e-05, "loss": 2.1142, "step": 1110 }, { "epoch": 0.4512591389114541, "grad_norm": 0.0, "learning_rate": 1.2050798059935214e-05, "loss": 2.2382, "step": 1111 }, { "epoch": 0.4516653127538586, "grad_norm": 0.0, "learning_rate": 1.2037920162901522e-05, "loss": 2.0719, "step": 1112 }, { "epoch": 0.45207148659626323, "grad_norm": 0.0, "learning_rate": 1.2025038738764332e-05, "loss": 2.125, "step": 1113 }, { "epoch": 0.4524776604386678, "grad_norm": 0.0, "learning_rate": 1.2012153809817992e-05, "loss": 2.2736, "step": 1114 }, { "epoch": 0.4528838342810723, "grad_norm": 0.0, "learning_rate": 1.1999265398362932e-05, "loss": 2.1533, "step": 1115 }, { "epoch": 0.45329000812347686, "grad_norm": 0.0, "learning_rate": 1.198637352670559e-05, "loss": 2.17, "step": 1116 }, { "epoch": 0.4536961819658814, "grad_norm": 0.0, "learning_rate": 1.1973478217158412e-05, "loss": 2.0277, "step": 1117 }, { "epoch": 0.45410235580828595, "grad_norm": 0.0, "learning_rate": 1.1960579492039783e-05, "loss": 2.1373, "step": 1118 }, { "epoch": 0.4545085296506905, "grad_norm": 0.0, "learning_rate": 1.1947677373673998e-05, "loss": 2.1482, "step": 1119 }, { "epoch": 0.45491470349309504, "grad_norm": 0.0, "learning_rate": 1.1934771884391234e-05, "loss": 2.1833, "step": 1120 }, { "epoch": 0.4553208773354996, "grad_norm": 0.0, "learning_rate": 1.1921863046527498e-05, "loss": 2.118, "step": 1121 }, { "epoch": 0.45572705117790413, "grad_norm": 0.0, "learning_rate": 1.1908950882424581e-05, "loss": 2.2352, "step": 1122 }, { "epoch": 0.4561332250203087, "grad_norm": 0.0, "learning_rate": 1.1896035414430051e-05, "loss": 2.1919, "step": 1123 }, { "epoch": 0.4565393988627132, "grad_norm": 0.0, "learning_rate": 1.1883116664897179e-05, "loss": 2.2806, "step": 1124 }, { "epoch": 0.45694557270511776, "grad_norm": 0.0, "learning_rate": 1.1870194656184924e-05, "loss": 2.2407, "step": 1125 }, { "epoch": 0.45735174654752236, "grad_norm": 0.0, "learning_rate": 1.1857269410657883e-05, "loss": 2.2488, "step": 1126 }, { "epoch": 0.4577579203899269, "grad_norm": 0.0, "learning_rate": 1.184434095068625e-05, "loss": 2.1106, "step": 1127 }, { "epoch": 0.45816409423233145, "grad_norm": 0.0, "learning_rate": 1.183140929864579e-05, "loss": 2.2157, "step": 1128 }, { "epoch": 0.458570268074736, "grad_norm": 0.0, "learning_rate": 1.1818474476917792e-05, "loss": 2.114, "step": 1129 }, { "epoch": 0.45897644191714054, "grad_norm": 0.0, "learning_rate": 1.1805536507889021e-05, "loss": 2.1844, "step": 1130 }, { "epoch": 0.4593826157595451, "grad_norm": 0.0, "learning_rate": 1.1792595413951706e-05, "loss": 2.0187, "step": 1131 }, { "epoch": 0.45978878960194963, "grad_norm": 0.0, "learning_rate": 1.177965121750347e-05, "loss": 2.2316, "step": 1132 }, { "epoch": 0.4601949634443542, "grad_norm": 0.0, "learning_rate": 1.1766703940947309e-05, "loss": 2.1647, "step": 1133 }, { "epoch": 0.4606011372867587, "grad_norm": 0.0, "learning_rate": 1.1753753606691554e-05, "loss": 2.0622, "step": 1134 }, { "epoch": 0.46100731112916327, "grad_norm": 0.0, "learning_rate": 1.1740800237149823e-05, "loss": 2.2113, "step": 1135 }, { "epoch": 0.4614134849715678, "grad_norm": 0.0, "learning_rate": 1.1727843854740997e-05, "loss": 2.1082, "step": 1136 }, { "epoch": 0.46181965881397236, "grad_norm": 0.0, "learning_rate": 1.1714884481889154e-05, "loss": 2.1074, "step": 1137 }, { "epoch": 0.46222583265637696, "grad_norm": 0.0, "learning_rate": 1.1701922141023566e-05, "loss": 2.0309, "step": 1138 }, { "epoch": 0.4626320064987815, "grad_norm": 0.0, "learning_rate": 1.1688956854578634e-05, "loss": 2.0905, "step": 1139 }, { "epoch": 0.46303818034118605, "grad_norm": 0.0, "learning_rate": 1.167598864499385e-05, "loss": 2.2399, "step": 1140 }, { "epoch": 0.4634443541835906, "grad_norm": 0.0, "learning_rate": 1.1663017534713778e-05, "loss": 2.1492, "step": 1141 }, { "epoch": 0.46385052802599513, "grad_norm": 0.0, "learning_rate": 1.1650043546187994e-05, "loss": 2.241, "step": 1142 }, { "epoch": 0.4642567018683997, "grad_norm": 0.0, "learning_rate": 1.1637066701871058e-05, "loss": 2.157, "step": 1143 }, { "epoch": 0.4646628757108042, "grad_norm": 0.0, "learning_rate": 1.1624087024222475e-05, "loss": 2.1526, "step": 1144 }, { "epoch": 0.46506904955320877, "grad_norm": 0.0, "learning_rate": 1.1611104535706645e-05, "loss": 2.1411, "step": 1145 }, { "epoch": 0.4654752233956133, "grad_norm": 0.0, "learning_rate": 1.1598119258792848e-05, "loss": 2.1684, "step": 1146 }, { "epoch": 0.46588139723801786, "grad_norm": 0.0, "learning_rate": 1.1585131215955175e-05, "loss": 2.1063, "step": 1147 }, { "epoch": 0.4662875710804224, "grad_norm": 0.0, "learning_rate": 1.157214042967251e-05, "loss": 2.0103, "step": 1148 }, { "epoch": 0.46669374492282695, "grad_norm": 0.0, "learning_rate": 1.1559146922428488e-05, "loss": 2.1876, "step": 1149 }, { "epoch": 0.4670999187652315, "grad_norm": 0.0, "learning_rate": 1.1546150716711448e-05, "loss": 2.192, "step": 1150 }, { "epoch": 0.4675060926076361, "grad_norm": 0.0, "learning_rate": 1.1533151835014406e-05, "loss": 2.0971, "step": 1151 }, { "epoch": 0.46791226645004064, "grad_norm": 0.0, "learning_rate": 1.1520150299835003e-05, "loss": 2.1185, "step": 1152 }, { "epoch": 0.4683184402924452, "grad_norm": 0.0, "learning_rate": 1.1507146133675475e-05, "loss": 2.1646, "step": 1153 }, { "epoch": 0.4687246141348497, "grad_norm": 0.0, "learning_rate": 1.1494139359042612e-05, "loss": 2.0858, "step": 1154 }, { "epoch": 0.46913078797725427, "grad_norm": 0.0, "learning_rate": 1.1481129998447715e-05, "loss": 2.1919, "step": 1155 }, { "epoch": 0.4695369618196588, "grad_norm": 0.0, "learning_rate": 1.146811807440657e-05, "loss": 2.1641, "step": 1156 }, { "epoch": 0.46994313566206336, "grad_norm": 0.0, "learning_rate": 1.1455103609439387e-05, "loss": 2.2056, "step": 1157 }, { "epoch": 0.4703493095044679, "grad_norm": 0.0, "learning_rate": 1.1442086626070781e-05, "loss": 2.0933, "step": 1158 }, { "epoch": 0.47075548334687245, "grad_norm": 0.0, "learning_rate": 1.1429067146829726e-05, "loss": 2.1324, "step": 1159 }, { "epoch": 0.471161657189277, "grad_norm": 0.0, "learning_rate": 1.1416045194249517e-05, "loss": 2.0757, "step": 1160 }, { "epoch": 0.47156783103168154, "grad_norm": 0.0, "learning_rate": 1.1403020790867724e-05, "loss": 2.1619, "step": 1161 }, { "epoch": 0.4719740048740861, "grad_norm": 0.0, "learning_rate": 1.1389993959226163e-05, "loss": 2.2536, "step": 1162 }, { "epoch": 0.4723801787164907, "grad_norm": 0.0, "learning_rate": 1.1376964721870847e-05, "loss": 2.2109, "step": 1163 }, { "epoch": 0.47278635255889523, "grad_norm": 0.0, "learning_rate": 1.1363933101351963e-05, "loss": 2.0944, "step": 1164 }, { "epoch": 0.4731925264012998, "grad_norm": 0.0, "learning_rate": 1.1350899120223813e-05, "loss": 2.1015, "step": 1165 }, { "epoch": 0.4735987002437043, "grad_norm": 0.0, "learning_rate": 1.1337862801044792e-05, "loss": 2.0539, "step": 1166 }, { "epoch": 0.47400487408610886, "grad_norm": 0.0, "learning_rate": 1.1324824166377335e-05, "loss": 2.2572, "step": 1167 }, { "epoch": 0.4744110479285134, "grad_norm": 0.0, "learning_rate": 1.1311783238787886e-05, "loss": 2.2219, "step": 1168 }, { "epoch": 0.47481722177091795, "grad_norm": 0.0, "learning_rate": 1.1298740040846863e-05, "loss": 2.0682, "step": 1169 }, { "epoch": 0.4752233956133225, "grad_norm": 0.0, "learning_rate": 1.1285694595128606e-05, "loss": 2.2505, "step": 1170 }, { "epoch": 0.47562956945572704, "grad_norm": 0.0, "learning_rate": 1.127264692421135e-05, "loss": 2.1445, "step": 1171 }, { "epoch": 0.4760357432981316, "grad_norm": 0.0, "learning_rate": 1.1259597050677178e-05, "loss": 2.1776, "step": 1172 }, { "epoch": 0.47644191714053613, "grad_norm": 0.0, "learning_rate": 1.1246544997111988e-05, "loss": 2.1466, "step": 1173 }, { "epoch": 0.4768480909829407, "grad_norm": 0.0, "learning_rate": 1.123349078610545e-05, "loss": 2.1459, "step": 1174 }, { "epoch": 0.4772542648253452, "grad_norm": 0.0, "learning_rate": 1.1220434440250966e-05, "loss": 2.1272, "step": 1175 }, { "epoch": 0.4776604386677498, "grad_norm": 0.0, "learning_rate": 1.1207375982145641e-05, "loss": 2.1906, "step": 1176 }, { "epoch": 0.47806661251015437, "grad_norm": 0.0, "learning_rate": 1.1194315434390222e-05, "loss": 2.0835, "step": 1177 }, { "epoch": 0.4784727863525589, "grad_norm": 0.0, "learning_rate": 1.1181252819589081e-05, "loss": 2.1153, "step": 1178 }, { "epoch": 0.47887896019496345, "grad_norm": 0.0, "learning_rate": 1.116818816035017e-05, "loss": 2.135, "step": 1179 }, { "epoch": 0.479285134037368, "grad_norm": 0.0, "learning_rate": 1.1155121479284976e-05, "loss": 2.3157, "step": 1180 }, { "epoch": 0.47969130787977254, "grad_norm": 0.0, "learning_rate": 1.1142052799008486e-05, "loss": 2.2787, "step": 1181 }, { "epoch": 0.4800974817221771, "grad_norm": 0.0, "learning_rate": 1.1128982142139142e-05, "loss": 2.3313, "step": 1182 }, { "epoch": 0.48050365556458163, "grad_norm": 0.0, "learning_rate": 1.1115909531298813e-05, "loss": 2.0153, "step": 1183 }, { "epoch": 0.4809098294069862, "grad_norm": 0.0, "learning_rate": 1.1102834989112752e-05, "loss": 2.1812, "step": 1184 }, { "epoch": 0.4813160032493907, "grad_norm": 0.0, "learning_rate": 1.1089758538209544e-05, "loss": 2.0404, "step": 1185 }, { "epoch": 0.48172217709179527, "grad_norm": 0.0, "learning_rate": 1.1076680201221093e-05, "loss": 2.1957, "step": 1186 }, { "epoch": 0.4821283509341998, "grad_norm": 0.0, "learning_rate": 1.106360000078255e-05, "loss": 2.1219, "step": 1187 }, { "epoch": 0.4825345247766044, "grad_norm": 0.0, "learning_rate": 1.1050517959532306e-05, "loss": 2.2441, "step": 1188 }, { "epoch": 0.48294069861900896, "grad_norm": 0.0, "learning_rate": 1.1037434100111925e-05, "loss": 2.0077, "step": 1189 }, { "epoch": 0.4833468724614135, "grad_norm": 0.0, "learning_rate": 1.1024348445166133e-05, "loss": 2.0731, "step": 1190 }, { "epoch": 0.48375304630381805, "grad_norm": 0.0, "learning_rate": 1.1011261017342751e-05, "loss": 2.231, "step": 1191 }, { "epoch": 0.4841592201462226, "grad_norm": 0.0, "learning_rate": 1.099817183929267e-05, "loss": 2.0983, "step": 1192 }, { "epoch": 0.48456539398862714, "grad_norm": 0.0, "learning_rate": 1.0985080933669816e-05, "loss": 2.162, "step": 1193 }, { "epoch": 0.4849715678310317, "grad_norm": 0.0, "learning_rate": 1.0971988323131099e-05, "loss": 2.0582, "step": 1194 }, { "epoch": 0.4853777416734362, "grad_norm": 0.0, "learning_rate": 1.0958894030336388e-05, "loss": 2.0499, "step": 1195 }, { "epoch": 0.48578391551584077, "grad_norm": 0.0, "learning_rate": 1.094579807794845e-05, "loss": 2.2981, "step": 1196 }, { "epoch": 0.4861900893582453, "grad_norm": 0.0, "learning_rate": 1.0932700488632937e-05, "loss": 2.1627, "step": 1197 }, { "epoch": 0.48659626320064986, "grad_norm": 0.0, "learning_rate": 1.091960128505833e-05, "loss": 2.3096, "step": 1198 }, { "epoch": 0.4870024370430544, "grad_norm": 0.0, "learning_rate": 1.0906500489895901e-05, "loss": 2.0529, "step": 1199 }, { "epoch": 0.487408610885459, "grad_norm": 0.0, "learning_rate": 1.089339812581968e-05, "loss": 2.2812, "step": 1200 }, { "epoch": 0.48781478472786355, "grad_norm": 0.0, "learning_rate": 1.0880294215506409e-05, "loss": 2.0306, "step": 1201 }, { "epoch": 0.4882209585702681, "grad_norm": 0.0, "learning_rate": 1.086718878163551e-05, "loss": 2.1281, "step": 1202 }, { "epoch": 0.48862713241267264, "grad_norm": 0.0, "learning_rate": 1.085408184688904e-05, "loss": 2.1612, "step": 1203 }, { "epoch": 0.4890333062550772, "grad_norm": 0.0, "learning_rate": 1.0840973433951653e-05, "loss": 2.0034, "step": 1204 }, { "epoch": 0.4894394800974817, "grad_norm": 0.0, "learning_rate": 1.0827863565510565e-05, "loss": 1.9795, "step": 1205 }, { "epoch": 0.48984565393988627, "grad_norm": 0.0, "learning_rate": 1.0814752264255508e-05, "loss": 2.0492, "step": 1206 }, { "epoch": 0.4902518277822908, "grad_norm": 0.0, "learning_rate": 1.0801639552878691e-05, "loss": 2.1032, "step": 1207 }, { "epoch": 0.49065800162469536, "grad_norm": 0.0, "learning_rate": 1.0788525454074765e-05, "loss": 2.2377, "step": 1208 }, { "epoch": 0.4910641754670999, "grad_norm": 0.0, "learning_rate": 1.0775409990540791e-05, "loss": 2.2063, "step": 1209 }, { "epoch": 0.49147034930950445, "grad_norm": 0.0, "learning_rate": 1.0762293184976178e-05, "loss": 2.0695, "step": 1210 }, { "epoch": 0.491876523151909, "grad_norm": 0.0, "learning_rate": 1.0749175060082671e-05, "loss": 1.951, "step": 1211 }, { "epoch": 0.49228269699431354, "grad_norm": 0.0, "learning_rate": 1.0736055638564287e-05, "loss": 2.0452, "step": 1212 }, { "epoch": 0.49268887083671814, "grad_norm": 0.0, "learning_rate": 1.0722934943127292e-05, "loss": 2.1725, "step": 1213 }, { "epoch": 0.4930950446791227, "grad_norm": 0.0, "learning_rate": 1.070981299648016e-05, "loss": 2.3675, "step": 1214 }, { "epoch": 0.49350121852152723, "grad_norm": 0.0, "learning_rate": 1.0696689821333526e-05, "loss": 2.0752, "step": 1215 }, { "epoch": 0.4939073923639318, "grad_norm": 0.0, "learning_rate": 1.0683565440400154e-05, "loss": 2.164, "step": 1216 }, { "epoch": 0.4943135662063363, "grad_norm": 0.0, "learning_rate": 1.067043987639489e-05, "loss": 2.0743, "step": 1217 }, { "epoch": 0.49471974004874086, "grad_norm": 0.0, "learning_rate": 1.0657313152034634e-05, "loss": 2.2017, "step": 1218 }, { "epoch": 0.4951259138911454, "grad_norm": 0.0, "learning_rate": 1.0644185290038292e-05, "loss": 2.0572, "step": 1219 }, { "epoch": 0.49553208773354995, "grad_norm": 0.0, "learning_rate": 1.0631056313126736e-05, "loss": 2.1713, "step": 1220 }, { "epoch": 0.4959382615759545, "grad_norm": 0.0, "learning_rate": 1.0617926244022774e-05, "loss": 2.263, "step": 1221 }, { "epoch": 0.49634443541835904, "grad_norm": 0.0, "learning_rate": 1.0604795105451096e-05, "loss": 2.1405, "step": 1222 }, { "epoch": 0.4967506092607636, "grad_norm": 0.0, "learning_rate": 1.0591662920138248e-05, "loss": 2.052, "step": 1223 }, { "epoch": 0.49715678310316813, "grad_norm": 0.0, "learning_rate": 1.057852971081259e-05, "loss": 2.2512, "step": 1224 }, { "epoch": 0.49756295694557273, "grad_norm": 0.0, "learning_rate": 1.0565395500204247e-05, "loss": 2.1173, "step": 1225 }, { "epoch": 0.4979691307879773, "grad_norm": 0.0, "learning_rate": 1.0552260311045082e-05, "loss": 2.078, "step": 1226 }, { "epoch": 0.4983753046303818, "grad_norm": 0.0, "learning_rate": 1.0539124166068656e-05, "loss": 2.1287, "step": 1227 }, { "epoch": 0.49878147847278637, "grad_norm": 0.0, "learning_rate": 1.0525987088010172e-05, "loss": 2.1638, "step": 1228 }, { "epoch": 0.4991876523151909, "grad_norm": 0.0, "learning_rate": 1.0512849099606458e-05, "loss": 2.0795, "step": 1229 }, { "epoch": 0.49959382615759546, "grad_norm": 0.0, "learning_rate": 1.0499710223595913e-05, "loss": 2.236, "step": 1230 }, { "epoch": 0.5, "grad_norm": 0.0, "learning_rate": 1.048657048271848e-05, "loss": 2.2051, "step": 1231 }, { "epoch": 0.5004061738424046, "grad_norm": 0.0, "learning_rate": 1.047342989971558e-05, "loss": 2.2705, "step": 1232 }, { "epoch": 0.5008123476848091, "grad_norm": 0.0, "learning_rate": 1.0460288497330117e-05, "loss": 2.1555, "step": 1233 }, { "epoch": 0.5012185215272137, "grad_norm": 0.0, "learning_rate": 1.0447146298306394e-05, "loss": 2.2011, "step": 1234 }, { "epoch": 0.5016246953696182, "grad_norm": 0.0, "learning_rate": 1.04340033253901e-05, "loss": 2.3176, "step": 1235 }, { "epoch": 0.5020308692120228, "grad_norm": 0.0, "learning_rate": 1.0420859601328266e-05, "loss": 2.0677, "step": 1236 }, { "epoch": 0.5024370430544273, "grad_norm": 0.0, "learning_rate": 1.0407715148869211e-05, "loss": 2.1847, "step": 1237 }, { "epoch": 0.5028432168968319, "grad_norm": 0.0, "learning_rate": 1.0394569990762528e-05, "loss": 2.1524, "step": 1238 }, { "epoch": 0.5032493907392364, "grad_norm": 0.0, "learning_rate": 1.0381424149759028e-05, "loss": 1.9388, "step": 1239 }, { "epoch": 0.503655564581641, "grad_norm": 0.0, "learning_rate": 1.0368277648610701e-05, "loss": 2.1015, "step": 1240 }, { "epoch": 0.5040617384240454, "grad_norm": 0.0, "learning_rate": 1.0355130510070682e-05, "loss": 2.1905, "step": 1241 }, { "epoch": 0.50446791226645, "grad_norm": 0.0, "learning_rate": 1.0341982756893203e-05, "loss": 2.155, "step": 1242 }, { "epoch": 0.5048740861088545, "grad_norm": 0.0, "learning_rate": 1.032883441183357e-05, "loss": 2.0491, "step": 1243 }, { "epoch": 0.5052802599512591, "grad_norm": 0.0, "learning_rate": 1.0315685497648107e-05, "loss": 2.2618, "step": 1244 }, { "epoch": 0.5056864337936637, "grad_norm": 0.0, "learning_rate": 1.0302536037094123e-05, "loss": 2.2017, "step": 1245 }, { "epoch": 0.5060926076360682, "grad_norm": 0.0, "learning_rate": 1.0289386052929874e-05, "loss": 2.0189, "step": 1246 }, { "epoch": 0.5064987814784728, "grad_norm": 0.0, "learning_rate": 1.0276235567914523e-05, "loss": 2.0108, "step": 1247 }, { "epoch": 0.5069049553208773, "grad_norm": 0.0, "learning_rate": 1.0263084604808097e-05, "loss": 2.0787, "step": 1248 }, { "epoch": 0.5073111291632819, "grad_norm": 0.0, "learning_rate": 1.0249933186371455e-05, "loss": 2.1227, "step": 1249 }, { "epoch": 0.5077173030056864, "grad_norm": 0.0, "learning_rate": 1.0236781335366239e-05, "loss": 2.2548, "step": 1250 }, { "epoch": 0.508123476848091, "grad_norm": 0.0, "learning_rate": 1.0223629074554846e-05, "loss": 2.2748, "step": 1251 }, { "epoch": 0.5085296506904955, "grad_norm": 0.0, "learning_rate": 1.0210476426700373e-05, "loss": 2.1332, "step": 1252 }, { "epoch": 0.5089358245329001, "grad_norm": 0.0, "learning_rate": 1.0197323414566596e-05, "loss": 2.0521, "step": 1253 }, { "epoch": 0.5093419983753046, "grad_norm": 0.0, "learning_rate": 1.0184170060917914e-05, "loss": 2.009, "step": 1254 }, { "epoch": 0.5097481722177092, "grad_norm": 0.0, "learning_rate": 1.0171016388519325e-05, "loss": 2.2128, "step": 1255 }, { "epoch": 0.5101543460601138, "grad_norm": 0.0, "learning_rate": 1.0157862420136371e-05, "loss": 2.1935, "step": 1256 }, { "epoch": 0.5105605199025183, "grad_norm": 0.0, "learning_rate": 1.0144708178535112e-05, "loss": 2.083, "step": 1257 }, { "epoch": 0.5109666937449229, "grad_norm": 0.0, "learning_rate": 1.0131553686482077e-05, "loss": 2.1653, "step": 1258 }, { "epoch": 0.5113728675873274, "grad_norm": 0.0, "learning_rate": 1.011839896674423e-05, "loss": 2.1431, "step": 1259 }, { "epoch": 0.511779041429732, "grad_norm": 0.0, "learning_rate": 1.0105244042088928e-05, "loss": 2.0776, "step": 1260 }, { "epoch": 0.5121852152721365, "grad_norm": 0.0, "learning_rate": 1.009208893528389e-05, "loss": 2.0631, "step": 1261 }, { "epoch": 0.512591389114541, "grad_norm": 0.0, "learning_rate": 1.0078933669097135e-05, "loss": 2.0369, "step": 1262 }, { "epoch": 0.5129975629569455, "grad_norm": 0.0, "learning_rate": 1.0065778266296973e-05, "loss": 2.1635, "step": 1263 }, { "epoch": 0.5134037367993501, "grad_norm": 0.0, "learning_rate": 1.0052622749651939e-05, "loss": 2.1372, "step": 1264 }, { "epoch": 0.5138099106417546, "grad_norm": 0.0, "learning_rate": 1.0039467141930776e-05, "loss": 2.0512, "step": 1265 }, { "epoch": 0.5142160844841592, "grad_norm": 0.0, "learning_rate": 1.002631146590238e-05, "loss": 2.2062, "step": 1266 }, { "epoch": 0.5146222583265637, "grad_norm": 0.0, "learning_rate": 1.0013155744335757e-05, "loss": 2.0638, "step": 1267 }, { "epoch": 0.5150284321689683, "grad_norm": 0.0, "learning_rate": 1e-05, "loss": 2.127, "step": 1268 }, { "epoch": 0.5154346060113729, "grad_norm": 0.0, "learning_rate": 9.986844255664248e-06, "loss": 2.0273, "step": 1269 }, { "epoch": 0.5158407798537774, "grad_norm": 0.0, "learning_rate": 9.973688534097624e-06, "loss": 2.062, "step": 1270 }, { "epoch": 0.516246953696182, "grad_norm": 0.0, "learning_rate": 9.960532858069226e-06, "loss": 2.191, "step": 1271 }, { "epoch": 0.5166531275385865, "grad_norm": 0.0, "learning_rate": 9.947377250348061e-06, "loss": 2.035, "step": 1272 }, { "epoch": 0.5170593013809911, "grad_norm": 0.0, "learning_rate": 9.934221733703032e-06, "loss": 2.1054, "step": 1273 }, { "epoch": 0.5174654752233956, "grad_norm": 0.0, "learning_rate": 9.92106633090287e-06, "loss": 2.1014, "step": 1274 }, { "epoch": 0.5178716490658002, "grad_norm": 0.0, "learning_rate": 9.907911064716113e-06, "loss": 2.0357, "step": 1275 }, { "epoch": 0.5182778229082047, "grad_norm": 0.0, "learning_rate": 9.894755957911074e-06, "loss": 2.0912, "step": 1276 }, { "epoch": 0.5186839967506093, "grad_norm": 0.0, "learning_rate": 9.88160103325577e-06, "loss": 2.0284, "step": 1277 }, { "epoch": 0.5190901705930138, "grad_norm": 0.0, "learning_rate": 9.868446313517927e-06, "loss": 2.1456, "step": 1278 }, { "epoch": 0.5194963444354184, "grad_norm": 0.0, "learning_rate": 9.855291821464893e-06, "loss": 2.0726, "step": 1279 }, { "epoch": 0.5199025182778229, "grad_norm": 0.0, "learning_rate": 9.842137579863632e-06, "loss": 2.0929, "step": 1280 }, { "epoch": 0.5203086921202275, "grad_norm": 0.0, "learning_rate": 9.828983611480678e-06, "loss": 2.1674, "step": 1281 }, { "epoch": 0.5207148659626321, "grad_norm": 0.0, "learning_rate": 9.815829939082087e-06, "loss": 2.2705, "step": 1282 }, { "epoch": 0.5211210398050365, "grad_norm": 0.0, "learning_rate": 9.802676585433408e-06, "loss": 2.2218, "step": 1283 }, { "epoch": 0.5215272136474411, "grad_norm": 0.0, "learning_rate": 9.78952357329963e-06, "loss": 2.009, "step": 1284 }, { "epoch": 0.5219333874898456, "grad_norm": 0.0, "learning_rate": 9.776370925445157e-06, "loss": 2.1574, "step": 1285 }, { "epoch": 0.5223395613322502, "grad_norm": 0.0, "learning_rate": 9.763218664633763e-06, "loss": 2.208, "step": 1286 }, { "epoch": 0.5227457351746547, "grad_norm": 0.0, "learning_rate": 9.750066813628547e-06, "loss": 2.096, "step": 1287 }, { "epoch": 0.5231519090170593, "grad_norm": 0.0, "learning_rate": 9.736915395191905e-06, "loss": 2.2305, "step": 1288 }, { "epoch": 0.5235580828594638, "grad_norm": 0.0, "learning_rate": 9.723764432085482e-06, "loss": 2.0651, "step": 1289 }, { "epoch": 0.5239642567018684, "grad_norm": 0.0, "learning_rate": 9.710613947070127e-06, "loss": 2.0742, "step": 1290 }, { "epoch": 0.5243704305442729, "grad_norm": 0.0, "learning_rate": 9.697463962905883e-06, "loss": 2.1389, "step": 1291 }, { "epoch": 0.5247766043866775, "grad_norm": 0.0, "learning_rate": 9.684314502351895e-06, "loss": 2.1537, "step": 1292 }, { "epoch": 0.525182778229082, "grad_norm": 0.0, "learning_rate": 9.671165588166432e-06, "loss": 2.1127, "step": 1293 }, { "epoch": 0.5255889520714866, "grad_norm": 0.0, "learning_rate": 9.658017243106802e-06, "loss": 2.1112, "step": 1294 }, { "epoch": 0.5259951259138912, "grad_norm": 0.0, "learning_rate": 9.644869489929322e-06, "loss": 2.2139, "step": 1295 }, { "epoch": 0.5264012997562957, "grad_norm": 0.0, "learning_rate": 9.631722351389302e-06, "loss": 2.3257, "step": 1296 }, { "epoch": 0.5268074735987003, "grad_norm": 0.0, "learning_rate": 9.618575850240972e-06, "loss": 2.2259, "step": 1297 }, { "epoch": 0.5272136474411048, "grad_norm": 0.0, "learning_rate": 9.605430009237474e-06, "loss": 2.1717, "step": 1298 }, { "epoch": 0.5276198212835094, "grad_norm": 0.0, "learning_rate": 9.592284851130794e-06, "loss": 2.1276, "step": 1299 }, { "epoch": 0.5280259951259139, "grad_norm": 0.0, "learning_rate": 9.579140398671737e-06, "loss": 2.0254, "step": 1300 }, { "epoch": 0.5284321689683185, "grad_norm": 0.0, "learning_rate": 9.565996674609903e-06, "loss": 2.1382, "step": 1301 }, { "epoch": 0.528838342810723, "grad_norm": 0.0, "learning_rate": 9.552853701693606e-06, "loss": 2.0563, "step": 1302 }, { "epoch": 0.5292445166531276, "grad_norm": 0.0, "learning_rate": 9.539711502669884e-06, "loss": 2.1082, "step": 1303 }, { "epoch": 0.529650690495532, "grad_norm": 0.0, "learning_rate": 9.526570100284423e-06, "loss": 2.0937, "step": 1304 }, { "epoch": 0.5300568643379366, "grad_norm": 0.0, "learning_rate": 9.513429517281525e-06, "loss": 2.042, "step": 1305 }, { "epoch": 0.5304630381803412, "grad_norm": 0.0, "learning_rate": 9.50028977640409e-06, "loss": 2.1646, "step": 1306 }, { "epoch": 0.5308692120227457, "grad_norm": 0.0, "learning_rate": 9.487150900393546e-06, "loss": 2.0178, "step": 1307 }, { "epoch": 0.5312753858651503, "grad_norm": 0.0, "learning_rate": 9.474012911989831e-06, "loss": 2.0801, "step": 1308 }, { "epoch": 0.5316815597075548, "grad_norm": 0.0, "learning_rate": 9.46087583393135e-06, "loss": 2.0296, "step": 1309 }, { "epoch": 0.5320877335499594, "grad_norm": 0.0, "learning_rate": 9.44773968895492e-06, "loss": 2.1394, "step": 1310 }, { "epoch": 0.5324939073923639, "grad_norm": 0.0, "learning_rate": 9.434604499795757e-06, "loss": 2.1581, "step": 1311 }, { "epoch": 0.5329000812347685, "grad_norm": 0.0, "learning_rate": 9.421470289187413e-06, "loss": 2.0935, "step": 1312 }, { "epoch": 0.533306255077173, "grad_norm": 0.0, "learning_rate": 9.408337079861755e-06, "loss": 2.1995, "step": 1313 }, { "epoch": 0.5337124289195776, "grad_norm": 0.0, "learning_rate": 9.395204894548907e-06, "loss": 2.2601, "step": 1314 }, { "epoch": 0.5341186027619821, "grad_norm": 0.0, "learning_rate": 9.38207375597723e-06, "loss": 2.238, "step": 1315 }, { "epoch": 0.5345247766043867, "grad_norm": 0.0, "learning_rate": 9.368943686873266e-06, "loss": 1.8965, "step": 1316 }, { "epoch": 0.5349309504467912, "grad_norm": 0.0, "learning_rate": 9.35581470996171e-06, "loss": 2.0592, "step": 1317 }, { "epoch": 0.5353371242891958, "grad_norm": 0.0, "learning_rate": 9.342686847965367e-06, "loss": 2.2001, "step": 1318 }, { "epoch": 0.5357432981316004, "grad_norm": 0.0, "learning_rate": 9.329560123605114e-06, "loss": 2.076, "step": 1319 }, { "epoch": 0.5361494719740049, "grad_norm": 0.0, "learning_rate": 9.31643455959985e-06, "loss": 2.0234, "step": 1320 }, { "epoch": 0.5365556458164095, "grad_norm": 0.0, "learning_rate": 9.303310178666478e-06, "loss": 2.1299, "step": 1321 }, { "epoch": 0.536961819658814, "grad_norm": 0.0, "learning_rate": 9.290187003519841e-06, "loss": 2.1533, "step": 1322 }, { "epoch": 0.5373679935012186, "grad_norm": 0.0, "learning_rate": 9.27706505687271e-06, "loss": 2.1013, "step": 1323 }, { "epoch": 0.537774167343623, "grad_norm": 0.0, "learning_rate": 9.263944361435718e-06, "loss": 2.009, "step": 1324 }, { "epoch": 0.5381803411860276, "grad_norm": 0.0, "learning_rate": 9.25082493991733e-06, "loss": 2.1214, "step": 1325 }, { "epoch": 0.5385865150284321, "grad_norm": 0.0, "learning_rate": 9.237706815023824e-06, "loss": 1.9846, "step": 1326 }, { "epoch": 0.5389926888708367, "grad_norm": 0.0, "learning_rate": 9.22459000945921e-06, "loss": 2.0915, "step": 1327 }, { "epoch": 0.5393988627132412, "grad_norm": 0.0, "learning_rate": 9.211474545925237e-06, "loss": 2.1396, "step": 1328 }, { "epoch": 0.5398050365556458, "grad_norm": 0.0, "learning_rate": 9.198360447121316e-06, "loss": 2.0042, "step": 1329 }, { "epoch": 0.5402112103980503, "grad_norm": 0.0, "learning_rate": 9.185247735744495e-06, "loss": 2.0506, "step": 1330 }, { "epoch": 0.5406173842404549, "grad_norm": 0.0, "learning_rate": 9.172136434489436e-06, "loss": 2.1639, "step": 1331 }, { "epoch": 0.5410235580828595, "grad_norm": 0.0, "learning_rate": 9.159026566048345e-06, "loss": 2.1566, "step": 1332 }, { "epoch": 0.541429731925264, "grad_norm": 0.0, "learning_rate": 9.145918153110962e-06, "loss": 2.1952, "step": 1333 }, { "epoch": 0.5418359057676686, "grad_norm": 0.0, "learning_rate": 9.132811218364494e-06, "loss": 2.2418, "step": 1334 }, { "epoch": 0.5422420796100731, "grad_norm": 0.0, "learning_rate": 9.119705784493593e-06, "loss": 2.1251, "step": 1335 }, { "epoch": 0.5426482534524777, "grad_norm": 0.0, "learning_rate": 9.106601874180324e-06, "loss": 2.1287, "step": 1336 }, { "epoch": 0.5430544272948822, "grad_norm": 0.0, "learning_rate": 9.093499510104102e-06, "loss": 2.0706, "step": 1337 }, { "epoch": 0.5434606011372868, "grad_norm": 0.0, "learning_rate": 9.080398714941672e-06, "loss": 2.1089, "step": 1338 }, { "epoch": 0.5438667749796913, "grad_norm": 0.0, "learning_rate": 9.067299511367064e-06, "loss": 1.9609, "step": 1339 }, { "epoch": 0.5442729488220959, "grad_norm": 0.0, "learning_rate": 9.054201922051552e-06, "loss": 2.164, "step": 1340 }, { "epoch": 0.5446791226645004, "grad_norm": 0.0, "learning_rate": 9.041105969663615e-06, "loss": 2.0635, "step": 1341 }, { "epoch": 0.545085296506905, "grad_norm": 0.0, "learning_rate": 9.028011676868901e-06, "loss": 2.0916, "step": 1342 }, { "epoch": 0.5454914703493096, "grad_norm": 0.0, "learning_rate": 9.014919066330187e-06, "loss": 2.0855, "step": 1343 }, { "epoch": 0.545897644191714, "grad_norm": 0.0, "learning_rate": 9.001828160707331e-06, "loss": 2.1625, "step": 1344 }, { "epoch": 0.5463038180341186, "grad_norm": 0.0, "learning_rate": 8.988738982657252e-06, "loss": 2.0672, "step": 1345 }, { "epoch": 0.5467099918765231, "grad_norm": 0.0, "learning_rate": 8.975651554833869e-06, "loss": 2.0992, "step": 1346 }, { "epoch": 0.5471161657189277, "grad_norm": 0.0, "learning_rate": 8.962565899888073e-06, "loss": 2.1581, "step": 1347 }, { "epoch": 0.5475223395613322, "grad_norm": 0.0, "learning_rate": 8.949482040467697e-06, "loss": 2.0065, "step": 1348 }, { "epoch": 0.5479285134037368, "grad_norm": 0.0, "learning_rate": 8.936399999217455e-06, "loss": 2.1379, "step": 1349 }, { "epoch": 0.5483346872461413, "grad_norm": 0.0, "learning_rate": 8.92331979877891e-06, "loss": 2.0939, "step": 1350 }, { "epoch": 0.5487408610885459, "grad_norm": 0.0, "learning_rate": 8.910241461790457e-06, "loss": 2.2297, "step": 1351 }, { "epoch": 0.5491470349309504, "grad_norm": 0.0, "learning_rate": 8.89716501088725e-06, "loss": 2.167, "step": 1352 }, { "epoch": 0.549553208773355, "grad_norm": 0.0, "learning_rate": 8.884090468701188e-06, "loss": 2.2216, "step": 1353 }, { "epoch": 0.5499593826157595, "grad_norm": 0.0, "learning_rate": 8.871017857860863e-06, "loss": 2.0596, "step": 1354 }, { "epoch": 0.5503655564581641, "grad_norm": 0.0, "learning_rate": 8.857947200991517e-06, "loss": 2.0186, "step": 1355 }, { "epoch": 0.5507717303005687, "grad_norm": 0.0, "learning_rate": 8.844878520715026e-06, "loss": 2.1902, "step": 1356 }, { "epoch": 0.5511779041429732, "grad_norm": 0.0, "learning_rate": 8.83181183964983e-06, "loss": 2.0569, "step": 1357 }, { "epoch": 0.5515840779853778, "grad_norm": 0.0, "learning_rate": 8.81874718041092e-06, "loss": 2.113, "step": 1358 }, { "epoch": 0.5519902518277823, "grad_norm": 0.0, "learning_rate": 8.805684565609785e-06, "loss": 2.1784, "step": 1359 }, { "epoch": 0.5523964256701869, "grad_norm": 0.0, "learning_rate": 8.792624017854364e-06, "loss": 2.1257, "step": 1360 }, { "epoch": 0.5528025995125914, "grad_norm": 0.0, "learning_rate": 8.779565559749037e-06, "loss": 2.1809, "step": 1361 }, { "epoch": 0.553208773354996, "grad_norm": 0.0, "learning_rate": 8.766509213894552e-06, "loss": 2.1125, "step": 1362 }, { "epoch": 0.5536149471974005, "grad_norm": 0.0, "learning_rate": 8.753455002888015e-06, "loss": 2.0791, "step": 1363 }, { "epoch": 0.554021121039805, "grad_norm": 0.0, "learning_rate": 8.740402949322827e-06, "loss": 2.0255, "step": 1364 }, { "epoch": 0.5544272948822095, "grad_norm": 0.0, "learning_rate": 8.727353075788653e-06, "loss": 2.1096, "step": 1365 }, { "epoch": 0.5548334687246141, "grad_norm": 0.0, "learning_rate": 8.714305404871397e-06, "loss": 2.1613, "step": 1366 }, { "epoch": 0.5552396425670186, "grad_norm": 0.0, "learning_rate": 8.701259959153139e-06, "loss": 2.0864, "step": 1367 }, { "epoch": 0.5556458164094232, "grad_norm": 0.0, "learning_rate": 8.688216761212118e-06, "loss": 2.1453, "step": 1368 }, { "epoch": 0.5560519902518278, "grad_norm": 0.0, "learning_rate": 8.675175833622669e-06, "loss": 2.0987, "step": 1369 }, { "epoch": 0.5564581640942323, "grad_norm": 0.0, "learning_rate": 8.662137198955211e-06, "loss": 2.1692, "step": 1370 }, { "epoch": 0.5568643379366369, "grad_norm": 0.0, "learning_rate": 8.64910087977619e-06, "loss": 1.8841, "step": 1371 }, { "epoch": 0.5572705117790414, "grad_norm": 0.0, "learning_rate": 8.636066898648039e-06, "loss": 2.1955, "step": 1372 }, { "epoch": 0.557676685621446, "grad_norm": 0.0, "learning_rate": 8.623035278129156e-06, "loss": 2.0237, "step": 1373 }, { "epoch": 0.5580828594638505, "grad_norm": 0.0, "learning_rate": 8.610006040773844e-06, "loss": 2.1402, "step": 1374 }, { "epoch": 0.5584890333062551, "grad_norm": 0.0, "learning_rate": 8.596979209132279e-06, "loss": 2.0498, "step": 1375 }, { "epoch": 0.5588952071486596, "grad_norm": 0.0, "learning_rate": 8.583954805750488e-06, "loss": 2.2756, "step": 1376 }, { "epoch": 0.5593013809910642, "grad_norm": 0.0, "learning_rate": 8.570932853170274e-06, "loss": 2.0918, "step": 1377 }, { "epoch": 0.5597075548334687, "grad_norm": 0.0, "learning_rate": 8.557913373929222e-06, "loss": 2.2289, "step": 1378 }, { "epoch": 0.5601137286758733, "grad_norm": 0.0, "learning_rate": 8.54489639056062e-06, "loss": 2.1917, "step": 1379 }, { "epoch": 0.5605199025182778, "grad_norm": 0.0, "learning_rate": 8.531881925593434e-06, "loss": 2.0762, "step": 1380 }, { "epoch": 0.5609260763606824, "grad_norm": 0.0, "learning_rate": 8.518870001552288e-06, "loss": 2.1482, "step": 1381 }, { "epoch": 0.561332250203087, "grad_norm": 0.0, "learning_rate": 8.50586064095739e-06, "loss": 2.0832, "step": 1382 }, { "epoch": 0.5617384240454915, "grad_norm": 0.0, "learning_rate": 8.492853866324529e-06, "loss": 2.1839, "step": 1383 }, { "epoch": 0.5621445978878961, "grad_norm": 0.0, "learning_rate": 8.479849700165002e-06, "loss": 2.2778, "step": 1384 }, { "epoch": 0.5625507717303005, "grad_norm": 0.0, "learning_rate": 8.466848164985595e-06, "loss": 2.0682, "step": 1385 }, { "epoch": 0.5629569455727051, "grad_norm": 0.0, "learning_rate": 8.453849283288554e-06, "loss": 2.1457, "step": 1386 }, { "epoch": 0.5633631194151096, "grad_norm": 0.0, "learning_rate": 8.440853077571513e-06, "loss": 1.9941, "step": 1387 }, { "epoch": 0.5637692932575142, "grad_norm": 0.0, "learning_rate": 8.427859570327494e-06, "loss": 2.2126, "step": 1388 }, { "epoch": 0.5641754670999187, "grad_norm": 0.0, "learning_rate": 8.41486878404483e-06, "loss": 2.1178, "step": 1389 }, { "epoch": 0.5645816409423233, "grad_norm": 0.0, "learning_rate": 8.401880741207155e-06, "loss": 2.2013, "step": 1390 }, { "epoch": 0.5649878147847278, "grad_norm": 0.0, "learning_rate": 8.388895464293358e-06, "loss": 2.1469, "step": 1391 }, { "epoch": 0.5653939886271324, "grad_norm": 0.0, "learning_rate": 8.375912975777529e-06, "loss": 1.944, "step": 1392 }, { "epoch": 0.565800162469537, "grad_norm": 0.0, "learning_rate": 8.362933298128944e-06, "loss": 2.0582, "step": 1393 }, { "epoch": 0.5662063363119415, "grad_norm": 0.0, "learning_rate": 8.349956453812009e-06, "loss": 2.2159, "step": 1394 }, { "epoch": 0.5666125101543461, "grad_norm": 0.0, "learning_rate": 8.336982465286225e-06, "loss": 2.1246, "step": 1395 }, { "epoch": 0.5670186839967506, "grad_norm": 0.0, "learning_rate": 8.324011355006152e-06, "loss": 2.1754, "step": 1396 }, { "epoch": 0.5674248578391552, "grad_norm": 0.0, "learning_rate": 8.311043145421369e-06, "loss": 2.0993, "step": 1397 }, { "epoch": 0.5678310316815597, "grad_norm": 0.0, "learning_rate": 8.298077858976435e-06, "loss": 2.1713, "step": 1398 }, { "epoch": 0.5682372055239643, "grad_norm": 0.0, "learning_rate": 8.285115518110847e-06, "loss": 1.982, "step": 1399 }, { "epoch": 0.5686433793663688, "grad_norm": 0.0, "learning_rate": 8.272156145259006e-06, "loss": 2.1729, "step": 1400 }, { "epoch": 0.5690495532087734, "grad_norm": 0.0, "learning_rate": 8.259199762850179e-06, "loss": 2.0354, "step": 1401 }, { "epoch": 0.5694557270511779, "grad_norm": 0.0, "learning_rate": 8.246246393308448e-06, "loss": 2.026, "step": 1402 }, { "epoch": 0.5698619008935825, "grad_norm": 0.0, "learning_rate": 8.233296059052695e-06, "loss": 2.1739, "step": 1403 }, { "epoch": 0.570268074735987, "grad_norm": 0.0, "learning_rate": 8.220348782496536e-06, "loss": 2.1526, "step": 1404 }, { "epoch": 0.5706742485783916, "grad_norm": 0.0, "learning_rate": 8.207404586048296e-06, "loss": 2.2567, "step": 1405 }, { "epoch": 0.5710804224207962, "grad_norm": 0.0, "learning_rate": 8.194463492110982e-06, "loss": 2.0147, "step": 1406 }, { "epoch": 0.5714865962632006, "grad_norm": 0.0, "learning_rate": 8.18152552308221e-06, "loss": 2.0679, "step": 1407 }, { "epoch": 0.5718927701056052, "grad_norm": 0.0, "learning_rate": 8.168590701354211e-06, "loss": 2.0239, "step": 1408 }, { "epoch": 0.5722989439480097, "grad_norm": 0.0, "learning_rate": 8.155659049313755e-06, "loss": 2.0991, "step": 1409 }, { "epoch": 0.5727051177904143, "grad_norm": 0.0, "learning_rate": 8.142730589342119e-06, "loss": 2.229, "step": 1410 }, { "epoch": 0.5731112916328188, "grad_norm": 0.0, "learning_rate": 8.129805343815078e-06, "loss": 2.1413, "step": 1411 }, { "epoch": 0.5735174654752234, "grad_norm": 0.0, "learning_rate": 8.116883335102821e-06, "loss": 1.9859, "step": 1412 }, { "epoch": 0.5739236393176279, "grad_norm": 0.0, "learning_rate": 8.103964585569952e-06, "loss": 2.1599, "step": 1413 }, { "epoch": 0.5743298131600325, "grad_norm": 0.0, "learning_rate": 8.091049117575424e-06, "loss": 2.0937, "step": 1414 }, { "epoch": 0.574735987002437, "grad_norm": 0.0, "learning_rate": 8.078136953472507e-06, "loss": 2.1231, "step": 1415 }, { "epoch": 0.5751421608448416, "grad_norm": 0.0, "learning_rate": 8.06522811560877e-06, "loss": 2.075, "step": 1416 }, { "epoch": 0.5755483346872461, "grad_norm": 0.0, "learning_rate": 8.052322626326001e-06, "loss": 2.1204, "step": 1417 }, { "epoch": 0.5759545085296507, "grad_norm": 0.0, "learning_rate": 8.03942050796022e-06, "loss": 2.0868, "step": 1418 }, { "epoch": 0.5763606823720553, "grad_norm": 0.0, "learning_rate": 8.026521782841591e-06, "loss": 2.0058, "step": 1419 }, { "epoch": 0.5767668562144598, "grad_norm": 0.0, "learning_rate": 8.013626473294411e-06, "loss": 2.1727, "step": 1420 }, { "epoch": 0.5771730300568644, "grad_norm": 0.0, "learning_rate": 8.000734601637075e-06, "loss": 2.177, "step": 1421 }, { "epoch": 0.5775792038992689, "grad_norm": 0.0, "learning_rate": 7.98784619018201e-06, "loss": 2.0605, "step": 1422 }, { "epoch": 0.5779853777416735, "grad_norm": 0.0, "learning_rate": 7.974961261235672e-06, "loss": 2.0598, "step": 1423 }, { "epoch": 0.578391551584078, "grad_norm": 0.0, "learning_rate": 7.962079837098481e-06, "loss": 1.9897, "step": 1424 }, { "epoch": 0.5787977254264826, "grad_norm": 0.0, "learning_rate": 7.94920194006479e-06, "loss": 2.2021, "step": 1425 }, { "epoch": 0.579203899268887, "grad_norm": 0.0, "learning_rate": 7.93632759242285e-06, "loss": 1.921, "step": 1426 }, { "epoch": 0.5796100731112916, "grad_norm": 0.0, "learning_rate": 7.923456816454769e-06, "loss": 2.2209, "step": 1427 }, { "epoch": 0.5800162469536961, "grad_norm": 0.0, "learning_rate": 7.910589634436478e-06, "loss": 2.0427, "step": 1428 }, { "epoch": 0.5804224207961007, "grad_norm": 0.0, "learning_rate": 7.897726068637675e-06, "loss": 2.1348, "step": 1429 }, { "epoch": 0.5808285946385053, "grad_norm": 0.0, "learning_rate": 7.884866141321811e-06, "loss": 1.9816, "step": 1430 }, { "epoch": 0.5812347684809098, "grad_norm": 0.0, "learning_rate": 7.872009874746039e-06, "loss": 2.1588, "step": 1431 }, { "epoch": 0.5816409423233144, "grad_norm": 0.0, "learning_rate": 7.859157291161165e-06, "loss": 2.1574, "step": 1432 }, { "epoch": 0.5820471161657189, "grad_norm": 0.0, "learning_rate": 7.846308412811639e-06, "loss": 2.1416, "step": 1433 }, { "epoch": 0.5824532900081235, "grad_norm": 0.0, "learning_rate": 7.833463261935482e-06, "loss": 1.8956, "step": 1434 }, { "epoch": 0.582859463850528, "grad_norm": 0.0, "learning_rate": 7.820621860764268e-06, "loss": 2.22, "step": 1435 }, { "epoch": 0.5832656376929326, "grad_norm": 0.0, "learning_rate": 7.80778423152309e-06, "loss": 2.1457, "step": 1436 }, { "epoch": 0.5836718115353371, "grad_norm": 0.0, "learning_rate": 7.794950396430497e-06, "loss": 1.9945, "step": 1437 }, { "epoch": 0.5840779853777417, "grad_norm": 0.0, "learning_rate": 7.782120377698489e-06, "loss": 2.056, "step": 1438 }, { "epoch": 0.5844841592201462, "grad_norm": 0.0, "learning_rate": 7.76929419753245e-06, "loss": 2.0349, "step": 1439 }, { "epoch": 0.5848903330625508, "grad_norm": 0.0, "learning_rate": 7.756471878131113e-06, "loss": 2.1978, "step": 1440 }, { "epoch": 0.5852965069049553, "grad_norm": 0.0, "learning_rate": 7.743653441686553e-06, "loss": 2.1875, "step": 1441 }, { "epoch": 0.5857026807473599, "grad_norm": 0.0, "learning_rate": 7.730838910384098e-06, "loss": 2.0372, "step": 1442 }, { "epoch": 0.5861088545897645, "grad_norm": 0.0, "learning_rate": 7.71802830640234e-06, "loss": 2.0692, "step": 1443 }, { "epoch": 0.586515028432169, "grad_norm": 0.0, "learning_rate": 7.705221651913058e-06, "loss": 1.982, "step": 1444 }, { "epoch": 0.5869212022745736, "grad_norm": 0.0, "learning_rate": 7.692418969081195e-06, "loss": 2.107, "step": 1445 }, { "epoch": 0.587327376116978, "grad_norm": 0.0, "learning_rate": 7.679620280064837e-06, "loss": 1.9769, "step": 1446 }, { "epoch": 0.5877335499593827, "grad_norm": 0.0, "learning_rate": 7.666825607015136e-06, "loss": 1.8649, "step": 1447 }, { "epoch": 0.5881397238017871, "grad_norm": 0.0, "learning_rate": 7.654034972076314e-06, "loss": 2.0281, "step": 1448 }, { "epoch": 0.5885458976441917, "grad_norm": 0.0, "learning_rate": 7.641248397385589e-06, "loss": 2.2794, "step": 1449 }, { "epoch": 0.5889520714865962, "grad_norm": 0.0, "learning_rate": 7.6284659050731525e-06, "loss": 2.1005, "step": 1450 }, { "epoch": 0.5893582453290008, "grad_norm": 0.0, "learning_rate": 7.615687517262143e-06, "loss": 2.1444, "step": 1451 }, { "epoch": 0.5897644191714053, "grad_norm": 0.0, "learning_rate": 7.602913256068583e-06, "loss": 2.2081, "step": 1452 }, { "epoch": 0.5901705930138099, "grad_norm": 0.0, "learning_rate": 7.590143143601356e-06, "loss": 2.031, "step": 1453 }, { "epoch": 0.5905767668562144, "grad_norm": 0.0, "learning_rate": 7.57737720196217e-06, "loss": 2.1199, "step": 1454 }, { "epoch": 0.590982940698619, "grad_norm": 0.0, "learning_rate": 7.564615453245506e-06, "loss": 2.087, "step": 1455 }, { "epoch": 0.5913891145410236, "grad_norm": 0.0, "learning_rate": 7.551857919538597e-06, "loss": 2.0854, "step": 1456 }, { "epoch": 0.5917952883834281, "grad_norm": 0.0, "learning_rate": 7.539104622921368e-06, "loss": 2.2261, "step": 1457 }, { "epoch": 0.5922014622258327, "grad_norm": 0.0, "learning_rate": 7.526355585466432e-06, "loss": 2.0819, "step": 1458 }, { "epoch": 0.5926076360682372, "grad_norm": 0.0, "learning_rate": 7.5136108292390096e-06, "loss": 2.0942, "step": 1459 }, { "epoch": 0.5930138099106418, "grad_norm": 0.0, "learning_rate": 7.500870376296918e-06, "loss": 2.0895, "step": 1460 }, { "epoch": 0.5934199837530463, "grad_norm": 0.0, "learning_rate": 7.488134248690535e-06, "loss": 1.9312, "step": 1461 }, { "epoch": 0.5938261575954509, "grad_norm": 0.0, "learning_rate": 7.4754024684627405e-06, "loss": 2.1533, "step": 1462 }, { "epoch": 0.5942323314378554, "grad_norm": 0.0, "learning_rate": 7.4626750576489e-06, "loss": 2.1081, "step": 1463 }, { "epoch": 0.59463850528026, "grad_norm": 0.0, "learning_rate": 7.449952038276811e-06, "loss": 2.0228, "step": 1464 }, { "epoch": 0.5950446791226645, "grad_norm": 0.0, "learning_rate": 7.437233432366667e-06, "loss": 2.0053, "step": 1465 }, { "epoch": 0.595450852965069, "grad_norm": 0.0, "learning_rate": 7.424519261931036e-06, "loss": 2.0616, "step": 1466 }, { "epoch": 0.5958570268074735, "grad_norm": 0.0, "learning_rate": 7.411809548974792e-06, "loss": 2.0614, "step": 1467 }, { "epoch": 0.5962632006498781, "grad_norm": 0.0, "learning_rate": 7.399104315495113e-06, "loss": 2.1736, "step": 1468 }, { "epoch": 0.5966693744922827, "grad_norm": 0.0, "learning_rate": 7.38640358348141e-06, "loss": 2.1545, "step": 1469 }, { "epoch": 0.5970755483346872, "grad_norm": 0.0, "learning_rate": 7.373707374915303e-06, "loss": 2.0499, "step": 1470 }, { "epoch": 0.5974817221770918, "grad_norm": 0.0, "learning_rate": 7.361015711770597e-06, "loss": 2.0947, "step": 1471 }, { "epoch": 0.5978878960194963, "grad_norm": 0.0, "learning_rate": 7.348328616013213e-06, "loss": 2.0263, "step": 1472 }, { "epoch": 0.5982940698619009, "grad_norm": 0.0, "learning_rate": 7.3356461096011825e-06, "loss": 1.8511, "step": 1473 }, { "epoch": 0.5987002437043054, "grad_norm": 0.0, "learning_rate": 7.322968214484583e-06, "loss": 2.0652, "step": 1474 }, { "epoch": 0.59910641754671, "grad_norm": 0.0, "learning_rate": 7.310294952605508e-06, "loss": 2.0724, "step": 1475 }, { "epoch": 0.5995125913891145, "grad_norm": 0.0, "learning_rate": 7.29762634589805e-06, "loss": 2.0848, "step": 1476 }, { "epoch": 0.5999187652315191, "grad_norm": 0.0, "learning_rate": 7.2849624162882256e-06, "loss": 2.0057, "step": 1477 }, { "epoch": 0.6003249390739236, "grad_norm": 0.0, "learning_rate": 7.27230318569397e-06, "loss": 2.2842, "step": 1478 }, { "epoch": 0.6007311129163282, "grad_norm": 0.0, "learning_rate": 7.259648676025077e-06, "loss": 1.9885, "step": 1479 }, { "epoch": 0.6011372867587328, "grad_norm": 0.0, "learning_rate": 7.2469989091831704e-06, "loss": 2.1768, "step": 1480 }, { "epoch": 0.6015434606011373, "grad_norm": 0.0, "learning_rate": 7.234353907061671e-06, "loss": 2.0462, "step": 1481 }, { "epoch": 0.6019496344435419, "grad_norm": 0.0, "learning_rate": 7.221713691545746e-06, "loss": 2.142, "step": 1482 }, { "epoch": 0.6023558082859464, "grad_norm": 0.0, "learning_rate": 7.2090782845122895e-06, "loss": 2.0281, "step": 1483 }, { "epoch": 0.602761982128351, "grad_norm": 0.0, "learning_rate": 7.1964477078298574e-06, "loss": 2.0204, "step": 1484 }, { "epoch": 0.6031681559707555, "grad_norm": 0.0, "learning_rate": 7.183821983358659e-06, "loss": 1.9932, "step": 1485 }, { "epoch": 0.6035743298131601, "grad_norm": 0.0, "learning_rate": 7.171201132950502e-06, "loss": 2.2471, "step": 1486 }, { "epoch": 0.6039805036555645, "grad_norm": 0.0, "learning_rate": 7.158585178448749e-06, "loss": 2.0932, "step": 1487 }, { "epoch": 0.6043866774979691, "grad_norm": 0.0, "learning_rate": 7.145974141688307e-06, "loss": 2.0869, "step": 1488 }, { "epoch": 0.6047928513403736, "grad_norm": 0.0, "learning_rate": 7.133368044495559e-06, "loss": 2.0065, "step": 1489 }, { "epoch": 0.6051990251827782, "grad_norm": 0.0, "learning_rate": 7.1207669086883366e-06, "loss": 2.031, "step": 1490 }, { "epoch": 0.6056051990251827, "grad_norm": 0.0, "learning_rate": 7.108170756075899e-06, "loss": 2.1585, "step": 1491 }, { "epoch": 0.6060113728675873, "grad_norm": 0.0, "learning_rate": 7.095579608458859e-06, "loss": 2.0402, "step": 1492 }, { "epoch": 0.6064175467099919, "grad_norm": 0.0, "learning_rate": 7.0829934876291926e-06, "loss": 2.1187, "step": 1493 }, { "epoch": 0.6068237205523964, "grad_norm": 0.0, "learning_rate": 7.070412415370158e-06, "loss": 2.2107, "step": 1494 }, { "epoch": 0.607229894394801, "grad_norm": 0.0, "learning_rate": 7.057836413456275e-06, "loss": 2.1331, "step": 1495 }, { "epoch": 0.6076360682372055, "grad_norm": 0.0, "learning_rate": 7.045265503653302e-06, "loss": 1.998, "step": 1496 }, { "epoch": 0.6080422420796101, "grad_norm": 0.0, "learning_rate": 7.0326997077181665e-06, "loss": 2.0854, "step": 1497 }, { "epoch": 0.6084484159220146, "grad_norm": 0.0, "learning_rate": 7.020139047398966e-06, "loss": 2.0701, "step": 1498 }, { "epoch": 0.6088545897644192, "grad_norm": 0.0, "learning_rate": 7.0075835444348905e-06, "loss": 1.9542, "step": 1499 }, { "epoch": 0.6092607636068237, "grad_norm": 0.0, "learning_rate": 6.995033220556209e-06, "loss": 2.0391, "step": 1500 }, { "epoch": 0.6096669374492283, "grad_norm": 0.0, "learning_rate": 6.982488097484237e-06, "loss": 2.1085, "step": 1501 }, { "epoch": 0.6100731112916328, "grad_norm": 0.0, "learning_rate": 6.969948196931272e-06, "loss": 1.9051, "step": 1502 }, { "epoch": 0.6104792851340374, "grad_norm": 0.0, "learning_rate": 6.957413540600593e-06, "loss": 2.0737, "step": 1503 }, { "epoch": 0.6108854589764419, "grad_norm": 0.0, "learning_rate": 6.944884150186384e-06, "loss": 2.2029, "step": 1504 }, { "epoch": 0.6112916328188465, "grad_norm": 0.0, "learning_rate": 6.932360047373721e-06, "loss": 2.0866, "step": 1505 }, { "epoch": 0.6116978066612511, "grad_norm": 0.0, "learning_rate": 6.919841253838537e-06, "loss": 2.0838, "step": 1506 }, { "epoch": 0.6121039805036556, "grad_norm": 0.0, "learning_rate": 6.907327791247562e-06, "loss": 2.1682, "step": 1507 }, { "epoch": 0.6125101543460602, "grad_norm": 0.0, "learning_rate": 6.894819681258312e-06, "loss": 2.1294, "step": 1508 }, { "epoch": 0.6129163281884646, "grad_norm": 0.0, "learning_rate": 6.882316945519034e-06, "loss": 2.1764, "step": 1509 }, { "epoch": 0.6133225020308692, "grad_norm": 0.0, "learning_rate": 6.869819605668669e-06, "loss": 2.0989, "step": 1510 }, { "epoch": 0.6137286758732737, "grad_norm": 0.0, "learning_rate": 6.8573276833368245e-06, "loss": 2.199, "step": 1511 }, { "epoch": 0.6141348497156783, "grad_norm": 0.0, "learning_rate": 6.844841200143733e-06, "loss": 2.1191, "step": 1512 }, { "epoch": 0.6145410235580828, "grad_norm": 0.0, "learning_rate": 6.83236017770021e-06, "loss": 1.9046, "step": 1513 }, { "epoch": 0.6149471974004874, "grad_norm": 0.0, "learning_rate": 6.819884637607619e-06, "loss": 2.167, "step": 1514 }, { "epoch": 0.6153533712428919, "grad_norm": 0.0, "learning_rate": 6.80741460145784e-06, "loss": 2.1701, "step": 1515 }, { "epoch": 0.6157595450852965, "grad_norm": 0.0, "learning_rate": 6.794950090833223e-06, "loss": 2.1708, "step": 1516 }, { "epoch": 0.6161657189277011, "grad_norm": 0.0, "learning_rate": 6.782491127306552e-06, "loss": 2.0167, "step": 1517 }, { "epoch": 0.6165718927701056, "grad_norm": 0.0, "learning_rate": 6.770037732441019e-06, "loss": 2.1323, "step": 1518 }, { "epoch": 0.6169780666125102, "grad_norm": 0.0, "learning_rate": 6.757589927790173e-06, "loss": 2.0919, "step": 1519 }, { "epoch": 0.6173842404549147, "grad_norm": 0.0, "learning_rate": 6.7451477348978835e-06, "loss": 2.1142, "step": 1520 }, { "epoch": 0.6177904142973193, "grad_norm": 0.0, "learning_rate": 6.7327111752983186e-06, "loss": 2.1022, "step": 1521 }, { "epoch": 0.6181965881397238, "grad_norm": 0.0, "learning_rate": 6.720280270515882e-06, "loss": 2.1306, "step": 1522 }, { "epoch": 0.6186027619821284, "grad_norm": 0.0, "learning_rate": 6.707855042065209e-06, "loss": 1.9427, "step": 1523 }, { "epoch": 0.6190089358245329, "grad_norm": 0.0, "learning_rate": 6.695435511451094e-06, "loss": 2.0002, "step": 1524 }, { "epoch": 0.6194151096669375, "grad_norm": 0.0, "learning_rate": 6.68302170016847e-06, "loss": 2.117, "step": 1525 }, { "epoch": 0.619821283509342, "grad_norm": 0.0, "learning_rate": 6.670613629702391e-06, "loss": 1.9938, "step": 1526 }, { "epoch": 0.6202274573517466, "grad_norm": 0.0, "learning_rate": 6.65821132152795e-06, "loss": 2.0941, "step": 1527 }, { "epoch": 0.620633631194151, "grad_norm": 0.0, "learning_rate": 6.645814797110287e-06, "loss": 1.9777, "step": 1528 }, { "epoch": 0.6210398050365556, "grad_norm": 0.0, "learning_rate": 6.63342407790452e-06, "loss": 2.1495, "step": 1529 }, { "epoch": 0.6214459788789602, "grad_norm": 0.0, "learning_rate": 6.62103918535572e-06, "loss": 1.9545, "step": 1530 }, { "epoch": 0.6218521527213647, "grad_norm": 0.0, "learning_rate": 6.608660140898884e-06, "loss": 1.9597, "step": 1531 }, { "epoch": 0.6222583265637693, "grad_norm": 0.0, "learning_rate": 6.596286965958872e-06, "loss": 2.0439, "step": 1532 }, { "epoch": 0.6226645004061738, "grad_norm": 0.0, "learning_rate": 6.5839196819504e-06, "loss": 2.018, "step": 1533 }, { "epoch": 0.6230706742485784, "grad_norm": 0.0, "learning_rate": 6.5715583102779815e-06, "loss": 2.1217, "step": 1534 }, { "epoch": 0.6234768480909829, "grad_norm": 0.0, "learning_rate": 6.559202872335893e-06, "loss": 1.9762, "step": 1535 }, { "epoch": 0.6238830219333875, "grad_norm": 0.0, "learning_rate": 6.546853389508152e-06, "loss": 2.0085, "step": 1536 }, { "epoch": 0.624289195775792, "grad_norm": 0.0, "learning_rate": 6.534509883168458e-06, "loss": 2.0326, "step": 1537 }, { "epoch": 0.6246953696181966, "grad_norm": 0.0, "learning_rate": 6.522172374680177e-06, "loss": 1.9932, "step": 1538 }, { "epoch": 0.6251015434606011, "grad_norm": 0.0, "learning_rate": 6.509840885396288e-06, "loss": 1.9958, "step": 1539 }, { "epoch": 0.6255077173030057, "grad_norm": 0.0, "learning_rate": 6.497515436659352e-06, "loss": 2.0479, "step": 1540 }, { "epoch": 0.6259138911454102, "grad_norm": 0.0, "learning_rate": 6.485196049801477e-06, "loss": 2.1116, "step": 1541 }, { "epoch": 0.6263200649878148, "grad_norm": 0.0, "learning_rate": 6.472882746144282e-06, "loss": 2.0949, "step": 1542 }, { "epoch": 0.6267262388302194, "grad_norm": 0.0, "learning_rate": 6.460575546998855e-06, "loss": 1.9934, "step": 1543 }, { "epoch": 0.6271324126726239, "grad_norm": 0.0, "learning_rate": 6.448274473665718e-06, "loss": 2.0993, "step": 1544 }, { "epoch": 0.6275385865150285, "grad_norm": 0.0, "learning_rate": 6.435979547434791e-06, "loss": 2.1076, "step": 1545 }, { "epoch": 0.627944760357433, "grad_norm": 0.0, "learning_rate": 6.423690789585359e-06, "loss": 2.1377, "step": 1546 }, { "epoch": 0.6283509341998376, "grad_norm": 0.0, "learning_rate": 6.411408221386021e-06, "loss": 2.0455, "step": 1547 }, { "epoch": 0.628757108042242, "grad_norm": 0.0, "learning_rate": 6.399131864094679e-06, "loss": 2.1002, "step": 1548 }, { "epoch": 0.6291632818846467, "grad_norm": 0.0, "learning_rate": 6.386861738958473e-06, "loss": 1.9772, "step": 1549 }, { "epoch": 0.6295694557270511, "grad_norm": 0.0, "learning_rate": 6.374597867213756e-06, "loss": 1.969, "step": 1550 }, { "epoch": 0.6299756295694557, "grad_norm": 0.0, "learning_rate": 6.362340270086071e-06, "loss": 2.1384, "step": 1551 }, { "epoch": 0.6303818034118602, "grad_norm": 0.0, "learning_rate": 6.350088968790085e-06, "loss": 2.1414, "step": 1552 }, { "epoch": 0.6307879772542648, "grad_norm": 0.0, "learning_rate": 6.3378439845295845e-06, "loss": 1.9076, "step": 1553 }, { "epoch": 0.6311941510966693, "grad_norm": 0.0, "learning_rate": 6.3256053384974105e-06, "loss": 2.1312, "step": 1554 }, { "epoch": 0.6316003249390739, "grad_norm": 0.0, "learning_rate": 6.313373051875434e-06, "loss": 1.9261, "step": 1555 }, { "epoch": 0.6320064987814785, "grad_norm": 0.0, "learning_rate": 6.301147145834534e-06, "loss": 1.9841, "step": 1556 }, { "epoch": 0.632412672623883, "grad_norm": 0.0, "learning_rate": 6.288927641534526e-06, "loss": 2.1701, "step": 1557 }, { "epoch": 0.6328188464662876, "grad_norm": 0.0, "learning_rate": 6.276714560124166e-06, "loss": 2.1476, "step": 1558 }, { "epoch": 0.6332250203086921, "grad_norm": 0.0, "learning_rate": 6.26450792274108e-06, "loss": 2.1146, "step": 1559 }, { "epoch": 0.6336311941510967, "grad_norm": 0.0, "learning_rate": 6.25230775051174e-06, "loss": 2.0341, "step": 1560 }, { "epoch": 0.6340373679935012, "grad_norm": 0.0, "learning_rate": 6.240114064551441e-06, "loss": 2.056, "step": 1561 }, { "epoch": 0.6344435418359058, "grad_norm": 0.0, "learning_rate": 6.2279268859642396e-06, "loss": 2.1324, "step": 1562 }, { "epoch": 0.6348497156783103, "grad_norm": 0.0, "learning_rate": 6.215746235842939e-06, "loss": 2.0261, "step": 1563 }, { "epoch": 0.6352558895207149, "grad_norm": 0.0, "learning_rate": 6.203572135269037e-06, "loss": 2.0637, "step": 1564 }, { "epoch": 0.6356620633631194, "grad_norm": 0.0, "learning_rate": 6.191404605312694e-06, "loss": 2.1048, "step": 1565 }, { "epoch": 0.636068237205524, "grad_norm": 0.0, "learning_rate": 6.179243667032709e-06, "loss": 2.1071, "step": 1566 }, { "epoch": 0.6364744110479286, "grad_norm": 0.0, "learning_rate": 6.167089341476462e-06, "loss": 2.1377, "step": 1567 }, { "epoch": 0.6368805848903331, "grad_norm": 0.0, "learning_rate": 6.154941649679894e-06, "loss": 2.1204, "step": 1568 }, { "epoch": 0.6372867587327377, "grad_norm": 0.0, "learning_rate": 6.1428006126674585e-06, "loss": 2.1702, "step": 1569 }, { "epoch": 0.6376929325751421, "grad_norm": 0.0, "learning_rate": 6.130666251452102e-06, "loss": 1.9974, "step": 1570 }, { "epoch": 0.6380991064175467, "grad_norm": 0.0, "learning_rate": 6.118538587035206e-06, "loss": 2.0669, "step": 1571 }, { "epoch": 0.6385052802599512, "grad_norm": 0.0, "learning_rate": 6.106417640406564e-06, "loss": 1.9267, "step": 1572 }, { "epoch": 0.6389114541023558, "grad_norm": 0.0, "learning_rate": 6.0943034325443525e-06, "loss": 2.0331, "step": 1573 }, { "epoch": 0.6393176279447603, "grad_norm": 0.0, "learning_rate": 6.082195984415069e-06, "loss": 2.1422, "step": 1574 }, { "epoch": 0.6397238017871649, "grad_norm": 0.0, "learning_rate": 6.070095316973526e-06, "loss": 2.0633, "step": 1575 }, { "epoch": 0.6401299756295694, "grad_norm": 0.0, "learning_rate": 6.0580014511627935e-06, "loss": 2.1259, "step": 1576 }, { "epoch": 0.640536149471974, "grad_norm": 0.0, "learning_rate": 6.0459144079141665e-06, "loss": 2.1838, "step": 1577 }, { "epoch": 0.6409423233143785, "grad_norm": 0.0, "learning_rate": 6.03383420814714e-06, "loss": 2.0005, "step": 1578 }, { "epoch": 0.6413484971567831, "grad_norm": 0.0, "learning_rate": 6.0217608727693575e-06, "loss": 2.182, "step": 1579 }, { "epoch": 0.6417546709991877, "grad_norm": 0.0, "learning_rate": 6.009694422676591e-06, "loss": 2.1163, "step": 1580 }, { "epoch": 0.6421608448415922, "grad_norm": 0.0, "learning_rate": 5.9976348787526875e-06, "loss": 2.1561, "step": 1581 }, { "epoch": 0.6425670186839968, "grad_norm": 0.0, "learning_rate": 5.9855822618695385e-06, "loss": 2.0477, "step": 1582 }, { "epoch": 0.6429731925264013, "grad_norm": 0.0, "learning_rate": 5.973536592887059e-06, "loss": 2.1352, "step": 1583 }, { "epoch": 0.6433793663688059, "grad_norm": 0.0, "learning_rate": 5.961497892653128e-06, "loss": 2.0825, "step": 1584 }, { "epoch": 0.6437855402112104, "grad_norm": 0.0, "learning_rate": 5.949466182003572e-06, "loss": 1.9768, "step": 1585 }, { "epoch": 0.644191714053615, "grad_norm": 0.0, "learning_rate": 5.937441481762112e-06, "loss": 2.1927, "step": 1586 }, { "epoch": 0.6445978878960195, "grad_norm": 0.0, "learning_rate": 5.925423812740338e-06, "loss": 1.9739, "step": 1587 }, { "epoch": 0.6450040617384241, "grad_norm": 0.0, "learning_rate": 5.913413195737678e-06, "loss": 2.1087, "step": 1588 }, { "epoch": 0.6454102355808286, "grad_norm": 0.0, "learning_rate": 5.901409651541346e-06, "loss": 2.1769, "step": 1589 }, { "epoch": 0.6458164094232332, "grad_norm": 0.0, "learning_rate": 5.889413200926317e-06, "loss": 2.151, "step": 1590 }, { "epoch": 0.6462225832656376, "grad_norm": 0.0, "learning_rate": 5.877423864655295e-06, "loss": 2.0459, "step": 1591 }, { "epoch": 0.6466287571080422, "grad_norm": 0.0, "learning_rate": 5.865441663478661e-06, "loss": 2.0575, "step": 1592 }, { "epoch": 0.6470349309504468, "grad_norm": 0.0, "learning_rate": 5.853466618134453e-06, "loss": 2.1, "step": 1593 }, { "epoch": 0.6474411047928513, "grad_norm": 0.0, "learning_rate": 5.841498749348322e-06, "loss": 2.0692, "step": 1594 }, { "epoch": 0.6478472786352559, "grad_norm": 0.0, "learning_rate": 5.829538077833503e-06, "loss": 2.1172, "step": 1595 }, { "epoch": 0.6482534524776604, "grad_norm": 0.0, "learning_rate": 5.817584624290773e-06, "loss": 2.0534, "step": 1596 }, { "epoch": 0.648659626320065, "grad_norm": 0.0, "learning_rate": 5.805638409408402e-06, "loss": 1.9584, "step": 1597 }, { "epoch": 0.6490658001624695, "grad_norm": 0.0, "learning_rate": 5.793699453862161e-06, "loss": 2.1548, "step": 1598 }, { "epoch": 0.6494719740048741, "grad_norm": 0.0, "learning_rate": 5.781767778315222e-06, "loss": 2.1617, "step": 1599 }, { "epoch": 0.6498781478472786, "grad_norm": 0.0, "learning_rate": 5.769843403418194e-06, "loss": 2.0745, "step": 1600 }, { "epoch": 0.6502843216896832, "grad_norm": 0.0, "learning_rate": 5.75792634980902e-06, "loss": 2.2049, "step": 1601 }, { "epoch": 0.6506904955320877, "grad_norm": 0.0, "learning_rate": 5.746016638112986e-06, "loss": 1.9851, "step": 1602 }, { "epoch": 0.6510966693744923, "grad_norm": 0.0, "learning_rate": 5.734114288942668e-06, "loss": 2.1381, "step": 1603 }, { "epoch": 0.6515028432168968, "grad_norm": 0.0, "learning_rate": 5.722219322897904e-06, "loss": 2.1199, "step": 1604 }, { "epoch": 0.6519090170593014, "grad_norm": 0.0, "learning_rate": 5.710331760565746e-06, "loss": 1.9653, "step": 1605 }, { "epoch": 0.652315190901706, "grad_norm": 0.0, "learning_rate": 5.698451622520442e-06, "loss": 2.1164, "step": 1606 }, { "epoch": 0.6527213647441105, "grad_norm": 0.0, "learning_rate": 5.686578929323378e-06, "loss": 2.0292, "step": 1607 }, { "epoch": 0.6531275385865151, "grad_norm": 0.0, "learning_rate": 5.674713701523063e-06, "loss": 2.0556, "step": 1608 }, { "epoch": 0.6535337124289196, "grad_norm": 0.0, "learning_rate": 5.662855959655082e-06, "loss": 2.0061, "step": 1609 }, { "epoch": 0.6539398862713242, "grad_norm": 0.0, "learning_rate": 5.651005724242072e-06, "loss": 2.0015, "step": 1610 }, { "epoch": 0.6543460601137286, "grad_norm": 0.0, "learning_rate": 5.639163015793665e-06, "loss": 1.9658, "step": 1611 }, { "epoch": 0.6547522339561332, "grad_norm": 0.0, "learning_rate": 5.627327854806478e-06, "loss": 2.1453, "step": 1612 }, { "epoch": 0.6551584077985377, "grad_norm": 0.0, "learning_rate": 5.615500261764062e-06, "loss": 1.9917, "step": 1613 }, { "epoch": 0.6555645816409423, "grad_norm": 0.0, "learning_rate": 5.603680257136857e-06, "loss": 1.9772, "step": 1614 }, { "epoch": 0.6559707554833468, "grad_norm": 0.0, "learning_rate": 5.591867861382193e-06, "loss": 2.0319, "step": 1615 }, { "epoch": 0.6563769293257514, "grad_norm": 0.0, "learning_rate": 5.58006309494421e-06, "loss": 2.048, "step": 1616 }, { "epoch": 0.656783103168156, "grad_norm": 0.0, "learning_rate": 5.568265978253857e-06, "loss": 2.1178, "step": 1617 }, { "epoch": 0.6571892770105605, "grad_norm": 0.0, "learning_rate": 5.556476531728836e-06, "loss": 2.059, "step": 1618 }, { "epoch": 0.6575954508529651, "grad_norm": 0.0, "learning_rate": 5.5446947757735755e-06, "loss": 1.9679, "step": 1619 }, { "epoch": 0.6580016246953696, "grad_norm": 0.0, "learning_rate": 5.532920730779198e-06, "loss": 1.8943, "step": 1620 }, { "epoch": 0.6584077985377742, "grad_norm": 0.0, "learning_rate": 5.521154417123478e-06, "loss": 2.157, "step": 1621 }, { "epoch": 0.6588139723801787, "grad_norm": 0.0, "learning_rate": 5.509395855170798e-06, "loss": 2.062, "step": 1622 }, { "epoch": 0.6592201462225833, "grad_norm": 0.0, "learning_rate": 5.497645065272149e-06, "loss": 2.0343, "step": 1623 }, { "epoch": 0.6596263200649878, "grad_norm": 0.0, "learning_rate": 5.485902067765041e-06, "loss": 2.088, "step": 1624 }, { "epoch": 0.6600324939073924, "grad_norm": 0.0, "learning_rate": 5.474166882973526e-06, "loss": 2.0258, "step": 1625 }, { "epoch": 0.6604386677497969, "grad_norm": 0.0, "learning_rate": 5.4624395312081125e-06, "loss": 2.0597, "step": 1626 }, { "epoch": 0.6608448415922015, "grad_norm": 0.0, "learning_rate": 5.450720032765762e-06, "loss": 2.1209, "step": 1627 }, { "epoch": 0.661251015434606, "grad_norm": 0.0, "learning_rate": 5.43900840792984e-06, "loss": 2.2295, "step": 1628 }, { "epoch": 0.6616571892770106, "grad_norm": 0.0, "learning_rate": 5.427304676970092e-06, "loss": 2.0556, "step": 1629 }, { "epoch": 0.6620633631194152, "grad_norm": 0.0, "learning_rate": 5.415608860142593e-06, "loss": 2.1701, "step": 1630 }, { "epoch": 0.6624695369618196, "grad_norm": 0.0, "learning_rate": 5.403920977689728e-06, "loss": 1.9631, "step": 1631 }, { "epoch": 0.6628757108042242, "grad_norm": 0.0, "learning_rate": 5.3922410498401355e-06, "loss": 2.1312, "step": 1632 }, { "epoch": 0.6632818846466287, "grad_norm": 0.0, "learning_rate": 5.380569096808712e-06, "loss": 2.0632, "step": 1633 }, { "epoch": 0.6636880584890333, "grad_norm": 0.0, "learning_rate": 5.368905138796523e-06, "loss": 1.9646, "step": 1634 }, { "epoch": 0.6640942323314378, "grad_norm": 0.0, "learning_rate": 5.357249195990817e-06, "loss": 2.0376, "step": 1635 }, { "epoch": 0.6645004061738424, "grad_norm": 0.0, "learning_rate": 5.345601288564963e-06, "loss": 2.0594, "step": 1636 }, { "epoch": 0.6649065800162469, "grad_norm": 0.0, "learning_rate": 5.333961436678422e-06, "loss": 2.0579, "step": 1637 }, { "epoch": 0.6653127538586515, "grad_norm": 0.0, "learning_rate": 5.322329660476715e-06, "loss": 2.1366, "step": 1638 }, { "epoch": 0.665718927701056, "grad_norm": 0.0, "learning_rate": 5.310705980091387e-06, "loss": 2.0695, "step": 1639 }, { "epoch": 0.6661251015434606, "grad_norm": 0.0, "learning_rate": 5.299090415639973e-06, "loss": 2.0223, "step": 1640 }, { "epoch": 0.6665312753858651, "grad_norm": 0.0, "learning_rate": 5.2874829872259494e-06, "loss": 2.1025, "step": 1641 }, { "epoch": 0.6669374492282697, "grad_norm": 0.0, "learning_rate": 5.275883714938726e-06, "loss": 2.0641, "step": 1642 }, { "epoch": 0.6673436230706743, "grad_norm": 0.0, "learning_rate": 5.264292618853587e-06, "loss": 2.2164, "step": 1643 }, { "epoch": 0.6677497969130788, "grad_norm": 0.0, "learning_rate": 5.252709719031675e-06, "loss": 2.1481, "step": 1644 }, { "epoch": 0.6681559707554834, "grad_norm": 0.0, "learning_rate": 5.241135035519935e-06, "loss": 2.0031, "step": 1645 }, { "epoch": 0.6685621445978879, "grad_norm": 0.0, "learning_rate": 5.2295685883511086e-06, "loss": 1.9852, "step": 1646 }, { "epoch": 0.6689683184402925, "grad_norm": 0.0, "learning_rate": 5.218010397543656e-06, "loss": 2.1054, "step": 1647 }, { "epoch": 0.669374492282697, "grad_norm": 0.0, "learning_rate": 5.2064604831017794e-06, "loss": 2.0635, "step": 1648 }, { "epoch": 0.6697806661251016, "grad_norm": 0.0, "learning_rate": 5.194918865015328e-06, "loss": 2.0783, "step": 1649 }, { "epoch": 0.670186839967506, "grad_norm": 0.0, "learning_rate": 5.183385563259819e-06, "loss": 2.1026, "step": 1650 }, { "epoch": 0.6705930138099107, "grad_norm": 0.0, "learning_rate": 5.171860597796352e-06, "loss": 2.0341, "step": 1651 }, { "epoch": 0.6709991876523151, "grad_norm": 0.0, "learning_rate": 5.160343988571614e-06, "loss": 2.0412, "step": 1652 }, { "epoch": 0.6714053614947197, "grad_norm": 0.0, "learning_rate": 5.1488357555178225e-06, "loss": 1.9604, "step": 1653 }, { "epoch": 0.6718115353371243, "grad_norm": 0.0, "learning_rate": 5.137335918552702e-06, "loss": 2.0956, "step": 1654 }, { "epoch": 0.6722177091795288, "grad_norm": 0.0, "learning_rate": 5.125844497579444e-06, "loss": 2.148, "step": 1655 }, { "epoch": 0.6726238830219334, "grad_norm": 0.0, "learning_rate": 5.114361512486676e-06, "loss": 1.969, "step": 1656 }, { "epoch": 0.6730300568643379, "grad_norm": 0.0, "learning_rate": 5.102886983148415e-06, "loss": 1.9613, "step": 1657 }, { "epoch": 0.6734362307067425, "grad_norm": 0.0, "learning_rate": 5.091420929424065e-06, "loss": 2.0375, "step": 1658 }, { "epoch": 0.673842404549147, "grad_norm": 0.0, "learning_rate": 5.079963371158333e-06, "loss": 2.1422, "step": 1659 }, { "epoch": 0.6742485783915516, "grad_norm": 0.0, "learning_rate": 5.068514328181251e-06, "loss": 1.9667, "step": 1660 }, { "epoch": 0.6746547522339561, "grad_norm": 0.0, "learning_rate": 5.057073820308089e-06, "loss": 2.1365, "step": 1661 }, { "epoch": 0.6750609260763607, "grad_norm": 0.0, "learning_rate": 5.045641867339361e-06, "loss": 2.0598, "step": 1662 }, { "epoch": 0.6754670999187652, "grad_norm": 0.0, "learning_rate": 5.034218489060767e-06, "loss": 2.1215, "step": 1663 }, { "epoch": 0.6758732737611698, "grad_norm": 0.0, "learning_rate": 5.0228037052431685e-06, "loss": 2.0742, "step": 1664 }, { "epoch": 0.6762794476035743, "grad_norm": 0.0, "learning_rate": 5.0113975356425525e-06, "loss": 2.1244, "step": 1665 }, { "epoch": 0.6766856214459789, "grad_norm": 0.0, "learning_rate": 5.000000000000003e-06, "loss": 1.9742, "step": 1666 }, { "epoch": 0.6770917952883835, "grad_norm": 0.0, "learning_rate": 4.988611118041644e-06, "loss": 2.0025, "step": 1667 }, { "epoch": 0.677497969130788, "grad_norm": 0.0, "learning_rate": 4.977230909478639e-06, "loss": 2.055, "step": 1668 }, { "epoch": 0.6779041429731926, "grad_norm": 0.0, "learning_rate": 4.965859394007132e-06, "loss": 1.9972, "step": 1669 }, { "epoch": 0.6783103168155971, "grad_norm": 0.0, "learning_rate": 4.954496591308227e-06, "loss": 1.8527, "step": 1670 }, { "epoch": 0.6787164906580017, "grad_norm": 0.0, "learning_rate": 4.943142521047945e-06, "loss": 1.9458, "step": 1671 }, { "epoch": 0.6791226645004061, "grad_norm": 0.0, "learning_rate": 4.931797202877186e-06, "loss": 2.1219, "step": 1672 }, { "epoch": 0.6795288383428107, "grad_norm": 0.0, "learning_rate": 4.9204606564317235e-06, "loss": 2.0577, "step": 1673 }, { "epoch": 0.6799350121852152, "grad_norm": 0.0, "learning_rate": 4.909132901332122e-06, "loss": 2.1396, "step": 1674 }, { "epoch": 0.6803411860276198, "grad_norm": 0.0, "learning_rate": 4.897813957183758e-06, "loss": 1.9921, "step": 1675 }, { "epoch": 0.6807473598700243, "grad_norm": 0.0, "learning_rate": 4.886503843576736e-06, "loss": 2.0457, "step": 1676 }, { "epoch": 0.6811535337124289, "grad_norm": 0.0, "learning_rate": 4.875202580085887e-06, "loss": 2.1205, "step": 1677 }, { "epoch": 0.6815597075548334, "grad_norm": 0.0, "learning_rate": 4.863910186270726e-06, "loss": 2.1265, "step": 1678 }, { "epoch": 0.681965881397238, "grad_norm": 0.0, "learning_rate": 4.852626681675415e-06, "loss": 2.0714, "step": 1679 }, { "epoch": 0.6823720552396426, "grad_norm": 0.0, "learning_rate": 4.841352085828731e-06, "loss": 2.0875, "step": 1680 }, { "epoch": 0.6827782290820471, "grad_norm": 0.0, "learning_rate": 4.830086418244036e-06, "loss": 2.039, "step": 1681 }, { "epoch": 0.6831844029244517, "grad_norm": 0.0, "learning_rate": 4.818829698419225e-06, "loss": 2.084, "step": 1682 }, { "epoch": 0.6835905767668562, "grad_norm": 0.0, "learning_rate": 4.807581945836733e-06, "loss": 2.1531, "step": 1683 }, { "epoch": 0.6839967506092608, "grad_norm": 0.0, "learning_rate": 4.796343179963442e-06, "loss": 2.1934, "step": 1684 }, { "epoch": 0.6844029244516653, "grad_norm": 0.0, "learning_rate": 4.785113420250715e-06, "loss": 2.0141, "step": 1685 }, { "epoch": 0.6848090982940699, "grad_norm": 0.0, "learning_rate": 4.773892686134301e-06, "loss": 2.0688, "step": 1686 }, { "epoch": 0.6852152721364744, "grad_norm": 0.0, "learning_rate": 4.762680997034341e-06, "loss": 2.0912, "step": 1687 }, { "epoch": 0.685621445978879, "grad_norm": 0.0, "learning_rate": 4.751478372355317e-06, "loss": 2.0568, "step": 1688 }, { "epoch": 0.6860276198212835, "grad_norm": 0.0, "learning_rate": 4.740284831486025e-06, "loss": 1.9608, "step": 1689 }, { "epoch": 0.6864337936636881, "grad_norm": 0.0, "learning_rate": 4.729100393799538e-06, "loss": 2.0189, "step": 1690 }, { "epoch": 0.6868399675060926, "grad_norm": 0.0, "learning_rate": 4.717925078653179e-06, "loss": 2.2655, "step": 1691 }, { "epoch": 0.6872461413484972, "grad_norm": 0.0, "learning_rate": 4.706758905388471e-06, "loss": 2.0378, "step": 1692 }, { "epoch": 0.6876523151909018, "grad_norm": 0.0, "learning_rate": 4.695601893331122e-06, "loss": 1.9853, "step": 1693 }, { "epoch": 0.6880584890333062, "grad_norm": 0.0, "learning_rate": 4.684454061790987e-06, "loss": 1.872, "step": 1694 }, { "epoch": 0.6884646628757108, "grad_norm": 0.0, "learning_rate": 4.673315430062027e-06, "loss": 1.9895, "step": 1695 }, { "epoch": 0.6888708367181153, "grad_norm": 0.0, "learning_rate": 4.662186017422282e-06, "loss": 1.9952, "step": 1696 }, { "epoch": 0.6892770105605199, "grad_norm": 0.0, "learning_rate": 4.651065843133837e-06, "loss": 2.0145, "step": 1697 }, { "epoch": 0.6896831844029244, "grad_norm": 0.0, "learning_rate": 4.639954926442792e-06, "loss": 2.0837, "step": 1698 }, { "epoch": 0.690089358245329, "grad_norm": 0.0, "learning_rate": 4.628853286579208e-06, "loss": 2.1278, "step": 1699 }, { "epoch": 0.6904955320877335, "grad_norm": 0.0, "learning_rate": 4.617760942757117e-06, "loss": 2.0971, "step": 1700 }, { "epoch": 0.6909017059301381, "grad_norm": 0.0, "learning_rate": 4.606677914174434e-06, "loss": 1.8796, "step": 1701 }, { "epoch": 0.6913078797725426, "grad_norm": 0.0, "learning_rate": 4.5956042200129725e-06, "loss": 2.0453, "step": 1702 }, { "epoch": 0.6917140536149472, "grad_norm": 0.0, "learning_rate": 4.584539879438379e-06, "loss": 1.9302, "step": 1703 }, { "epoch": 0.6921202274573518, "grad_norm": 0.0, "learning_rate": 4.573484911600118e-06, "loss": 1.9088, "step": 1704 }, { "epoch": 0.6925264012997563, "grad_norm": 0.0, "learning_rate": 4.562439335631428e-06, "loss": 2.0272, "step": 1705 }, { "epoch": 0.6929325751421609, "grad_norm": 0.0, "learning_rate": 4.551403170649299e-06, "loss": 1.8987, "step": 1706 }, { "epoch": 0.6933387489845654, "grad_norm": 0.0, "learning_rate": 4.540376435754417e-06, "loss": 2.03, "step": 1707 }, { "epoch": 0.69374492282697, "grad_norm": 0.0, "learning_rate": 4.529359150031172e-06, "loss": 2.0209, "step": 1708 }, { "epoch": 0.6941510966693745, "grad_norm": 0.0, "learning_rate": 4.518351332547573e-06, "loss": 2.1515, "step": 1709 }, { "epoch": 0.6945572705117791, "grad_norm": 0.0, "learning_rate": 4.507353002355269e-06, "loss": 2.0881, "step": 1710 }, { "epoch": 0.6949634443541836, "grad_norm": 0.0, "learning_rate": 4.496364178489465e-06, "loss": 1.9812, "step": 1711 }, { "epoch": 0.6953696181965882, "grad_norm": 0.0, "learning_rate": 4.485384879968926e-06, "loss": 2.1488, "step": 1712 }, { "epoch": 0.6957757920389926, "grad_norm": 0.0, "learning_rate": 4.474415125795929e-06, "loss": 2.1091, "step": 1713 }, { "epoch": 0.6961819658813972, "grad_norm": 0.0, "learning_rate": 4.4634549349562315e-06, "loss": 2.0442, "step": 1714 }, { "epoch": 0.6965881397238017, "grad_norm": 0.0, "learning_rate": 4.452504326419041e-06, "loss": 2.0152, "step": 1715 }, { "epoch": 0.6969943135662063, "grad_norm": 0.0, "learning_rate": 4.44156331913698e-06, "loss": 2.0759, "step": 1716 }, { "epoch": 0.6974004874086109, "grad_norm": 0.0, "learning_rate": 4.430631932046044e-06, "loss": 2.0234, "step": 1717 }, { "epoch": 0.6978066612510154, "grad_norm": 0.0, "learning_rate": 4.4197101840656e-06, "loss": 2.0348, "step": 1718 }, { "epoch": 0.69821283509342, "grad_norm": 0.0, "learning_rate": 4.408798094098311e-06, "loss": 2.069, "step": 1719 }, { "epoch": 0.6986190089358245, "grad_norm": 0.0, "learning_rate": 4.3978956810301355e-06, "loss": 1.9184, "step": 1720 }, { "epoch": 0.6990251827782291, "grad_norm": 0.0, "learning_rate": 4.387002963730282e-06, "loss": 1.9416, "step": 1721 }, { "epoch": 0.6994313566206336, "grad_norm": 0.0, "learning_rate": 4.376119961051175e-06, "loss": 2.1207, "step": 1722 }, { "epoch": 0.6998375304630382, "grad_norm": 0.0, "learning_rate": 4.365246691828432e-06, "loss": 2.0758, "step": 1723 }, { "epoch": 0.7002437043054427, "grad_norm": 0.0, "learning_rate": 4.354383174880819e-06, "loss": 1.8725, "step": 1724 }, { "epoch": 0.7006498781478473, "grad_norm": 0.0, "learning_rate": 4.343529429010228e-06, "loss": 2.1068, "step": 1725 }, { "epoch": 0.7010560519902518, "grad_norm": 0.0, "learning_rate": 4.33268547300163e-06, "loss": 2.0887, "step": 1726 }, { "epoch": 0.7014622258326564, "grad_norm": 0.0, "learning_rate": 4.3218513256230635e-06, "loss": 1.9625, "step": 1727 }, { "epoch": 0.7018683996750609, "grad_norm": 0.0, "learning_rate": 4.311027005625585e-06, "loss": 2.0433, "step": 1728 }, { "epoch": 0.7022745735174655, "grad_norm": 0.0, "learning_rate": 4.300212531743245e-06, "loss": 2.0534, "step": 1729 }, { "epoch": 0.7026807473598701, "grad_norm": 0.0, "learning_rate": 4.289407922693053e-06, "loss": 2.0215, "step": 1730 }, { "epoch": 0.7030869212022746, "grad_norm": 0.0, "learning_rate": 4.278613197174945e-06, "loss": 2.0377, "step": 1731 }, { "epoch": 0.7034930950446792, "grad_norm": 0.0, "learning_rate": 4.2678283738717396e-06, "loss": 2.0646, "step": 1732 }, { "epoch": 0.7038992688870837, "grad_norm": 0.0, "learning_rate": 4.257053471449144e-06, "loss": 1.9309, "step": 1733 }, { "epoch": 0.7043054427294883, "grad_norm": 0.0, "learning_rate": 4.2462885085556635e-06, "loss": 1.9566, "step": 1734 }, { "epoch": 0.7047116165718927, "grad_norm": 0.0, "learning_rate": 4.2355335038226295e-06, "loss": 2.0616, "step": 1735 }, { "epoch": 0.7051177904142973, "grad_norm": 0.0, "learning_rate": 4.2247884758641155e-06, "loss": 2.1168, "step": 1736 }, { "epoch": 0.7055239642567018, "grad_norm": 0.0, "learning_rate": 4.21405344327694e-06, "loss": 2.0227, "step": 1737 }, { "epoch": 0.7059301380991064, "grad_norm": 0.0, "learning_rate": 4.203328424640619e-06, "loss": 1.9847, "step": 1738 }, { "epoch": 0.7063363119415109, "grad_norm": 0.0, "learning_rate": 4.192613438517338e-06, "loss": 2.0114, "step": 1739 }, { "epoch": 0.7067424857839155, "grad_norm": 0.0, "learning_rate": 4.181908503451919e-06, "loss": 2.0769, "step": 1740 }, { "epoch": 0.7071486596263201, "grad_norm": 0.0, "learning_rate": 4.171213637971789e-06, "loss": 2.2288, "step": 1741 }, { "epoch": 0.7075548334687246, "grad_norm": 0.0, "learning_rate": 4.1605288605869365e-06, "loss": 2.0885, "step": 1742 }, { "epoch": 0.7079610073111292, "grad_norm": 0.0, "learning_rate": 4.149854189789914e-06, "loss": 2.1523, "step": 1743 }, { "epoch": 0.7083671811535337, "grad_norm": 0.0, "learning_rate": 4.139189644055751e-06, "loss": 2.1473, "step": 1744 }, { "epoch": 0.7087733549959383, "grad_norm": 0.0, "learning_rate": 4.128535241841987e-06, "loss": 2.0525, "step": 1745 }, { "epoch": 0.7091795288383428, "grad_norm": 0.0, "learning_rate": 4.117891001588574e-06, "loss": 1.9887, "step": 1746 }, { "epoch": 0.7095857026807474, "grad_norm": 0.0, "learning_rate": 4.107256941717898e-06, "loss": 2.1115, "step": 1747 }, { "epoch": 0.7099918765231519, "grad_norm": 0.0, "learning_rate": 4.096633080634717e-06, "loss": 1.978, "step": 1748 }, { "epoch": 0.7103980503655565, "grad_norm": 0.0, "learning_rate": 4.086019436726136e-06, "loss": 1.9176, "step": 1749 }, { "epoch": 0.710804224207961, "grad_norm": 0.0, "learning_rate": 4.075416028361584e-06, "loss": 2.0039, "step": 1750 }, { "epoch": 0.7112103980503656, "grad_norm": 0.0, "learning_rate": 4.064822873892771e-06, "loss": 2.1044, "step": 1751 }, { "epoch": 0.7116165718927701, "grad_norm": 0.0, "learning_rate": 4.054239991653654e-06, "loss": 2.159, "step": 1752 }, { "epoch": 0.7120227457351747, "grad_norm": 0.0, "learning_rate": 4.043667399960419e-06, "loss": 2.1217, "step": 1753 }, { "epoch": 0.7124289195775793, "grad_norm": 0.0, "learning_rate": 4.033105117111441e-06, "loss": 2.0775, "step": 1754 }, { "epoch": 0.7128350934199837, "grad_norm": 0.0, "learning_rate": 4.022553161387253e-06, "loss": 2.0207, "step": 1755 }, { "epoch": 0.7132412672623883, "grad_norm": 0.0, "learning_rate": 4.012011551050516e-06, "loss": 1.9978, "step": 1756 }, { "epoch": 0.7136474411047928, "grad_norm": 0.0, "learning_rate": 4.0014803043459726e-06, "loss": 1.9439, "step": 1757 }, { "epoch": 0.7140536149471974, "grad_norm": 0.0, "learning_rate": 3.9909594395004545e-06, "loss": 2.0567, "step": 1758 }, { "epoch": 0.7144597887896019, "grad_norm": 0.0, "learning_rate": 3.980448974722794e-06, "loss": 2.0277, "step": 1759 }, { "epoch": 0.7148659626320065, "grad_norm": 0.0, "learning_rate": 3.969948928203856e-06, "loss": 2.0964, "step": 1760 }, { "epoch": 0.715272136474411, "grad_norm": 0.0, "learning_rate": 3.959459318116447e-06, "loss": 2.2367, "step": 1761 }, { "epoch": 0.7156783103168156, "grad_norm": 0.0, "learning_rate": 3.948980162615323e-06, "loss": 1.9279, "step": 1762 }, { "epoch": 0.7160844841592201, "grad_norm": 0.0, "learning_rate": 3.938511479837147e-06, "loss": 2.0784, "step": 1763 }, { "epoch": 0.7164906580016247, "grad_norm": 0.0, "learning_rate": 3.928053287900452e-06, "loss": 1.9392, "step": 1764 }, { "epoch": 0.7168968318440292, "grad_norm": 0.0, "learning_rate": 3.917605604905618e-06, "loss": 2.1078, "step": 1765 }, { "epoch": 0.7173030056864338, "grad_norm": 0.0, "learning_rate": 3.907168448934836e-06, "loss": 2.0908, "step": 1766 }, { "epoch": 0.7177091795288384, "grad_norm": 0.0, "learning_rate": 3.896741838052069e-06, "loss": 2.1071, "step": 1767 }, { "epoch": 0.7181153533712429, "grad_norm": 0.0, "learning_rate": 3.886325790303046e-06, "loss": 2.1258, "step": 1768 }, { "epoch": 0.7185215272136475, "grad_norm": 0.0, "learning_rate": 3.875920323715195e-06, "loss": 1.9038, "step": 1769 }, { "epoch": 0.718927701056052, "grad_norm": 0.0, "learning_rate": 3.865525456297652e-06, "loss": 2.0354, "step": 1770 }, { "epoch": 0.7193338748984566, "grad_norm": 0.0, "learning_rate": 3.855141206041187e-06, "loss": 2.0428, "step": 1771 }, { "epoch": 0.7197400487408611, "grad_norm": 0.0, "learning_rate": 3.8447675909182095e-06, "loss": 2.0264, "step": 1772 }, { "epoch": 0.7201462225832657, "grad_norm": 0.0, "learning_rate": 3.8344046288827146e-06, "loss": 2.1437, "step": 1773 }, { "epoch": 0.7205523964256701, "grad_norm": 0.0, "learning_rate": 3.824052337870263e-06, "loss": 2.017, "step": 1774 }, { "epoch": 0.7209585702680747, "grad_norm": 0.0, "learning_rate": 3.8137107357979475e-06, "loss": 2.0581, "step": 1775 }, { "epoch": 0.7213647441104792, "grad_norm": 0.0, "learning_rate": 3.803379840564363e-06, "loss": 2.1007, "step": 1776 }, { "epoch": 0.7217709179528838, "grad_norm": 0.0, "learning_rate": 3.7930596700495626e-06, "loss": 2.1105, "step": 1777 }, { "epoch": 0.7221770917952883, "grad_norm": 0.0, "learning_rate": 3.7827502421150497e-06, "loss": 2.164, "step": 1778 }, { "epoch": 0.7225832656376929, "grad_norm": 0.0, "learning_rate": 3.7724515746037314e-06, "loss": 2.0472, "step": 1779 }, { "epoch": 0.7229894394800975, "grad_norm": 0.0, "learning_rate": 3.7621636853398902e-06, "loss": 1.9771, "step": 1780 }, { "epoch": 0.723395613322502, "grad_norm": 0.0, "learning_rate": 3.751886592129155e-06, "loss": 1.9079, "step": 1781 }, { "epoch": 0.7238017871649066, "grad_norm": 0.0, "learning_rate": 3.741620312758469e-06, "loss": 1.9131, "step": 1782 }, { "epoch": 0.7242079610073111, "grad_norm": 0.0, "learning_rate": 3.7313648649960667e-06, "loss": 2.0599, "step": 1783 }, { "epoch": 0.7246141348497157, "grad_norm": 0.0, "learning_rate": 3.721120266591416e-06, "loss": 1.9563, "step": 1784 }, { "epoch": 0.7250203086921202, "grad_norm": 0.0, "learning_rate": 3.7108865352752343e-06, "loss": 1.8759, "step": 1785 }, { "epoch": 0.7254264825345248, "grad_norm": 0.0, "learning_rate": 3.7006636887594095e-06, "loss": 1.8694, "step": 1786 }, { "epoch": 0.7258326563769293, "grad_norm": 0.0, "learning_rate": 3.690451744736999e-06, "loss": 2.0286, "step": 1787 }, { "epoch": 0.7262388302193339, "grad_norm": 0.0, "learning_rate": 3.6802507208821934e-06, "loss": 1.9929, "step": 1788 }, { "epoch": 0.7266450040617384, "grad_norm": 0.0, "learning_rate": 3.6700606348502787e-06, "loss": 1.9863, "step": 1789 }, { "epoch": 0.727051177904143, "grad_norm": 0.0, "learning_rate": 3.6598815042776135e-06, "loss": 2.0962, "step": 1790 }, { "epoch": 0.7274573517465476, "grad_norm": 0.0, "learning_rate": 3.649713346781597e-06, "loss": 2.0227, "step": 1791 }, { "epoch": 0.7278635255889521, "grad_norm": 0.0, "learning_rate": 3.639556179960623e-06, "loss": 2.0277, "step": 1792 }, { "epoch": 0.7282696994313567, "grad_norm": 0.0, "learning_rate": 3.6294100213940873e-06, "loss": 2.1085, "step": 1793 }, { "epoch": 0.7286758732737612, "grad_norm": 0.0, "learning_rate": 3.619274888642309e-06, "loss": 2.0296, "step": 1794 }, { "epoch": 0.7290820471161658, "grad_norm": 0.0, "learning_rate": 3.6091507992465493e-06, "loss": 2.0575, "step": 1795 }, { "epoch": 0.7294882209585702, "grad_norm": 0.0, "learning_rate": 3.5990377707289292e-06, "loss": 2.1535, "step": 1796 }, { "epoch": 0.7298943948009748, "grad_norm": 0.0, "learning_rate": 3.588935820592446e-06, "loss": 1.9535, "step": 1797 }, { "epoch": 0.7303005686433793, "grad_norm": 0.0, "learning_rate": 3.578844966320917e-06, "loss": 2.0684, "step": 1798 }, { "epoch": 0.7307067424857839, "grad_norm": 0.0, "learning_rate": 3.568765225378954e-06, "loss": 2.1852, "step": 1799 }, { "epoch": 0.7311129163281884, "grad_norm": 0.0, "learning_rate": 3.558696615211936e-06, "loss": 1.9688, "step": 1800 }, { "epoch": 0.731519090170593, "grad_norm": 0.0, "learning_rate": 3.548639153245982e-06, "loss": 2.0089, "step": 1801 }, { "epoch": 0.7319252640129975, "grad_norm": 0.0, "learning_rate": 3.5385928568879012e-06, "loss": 2.0862, "step": 1802 }, { "epoch": 0.7323314378554021, "grad_norm": 0.0, "learning_rate": 3.528557743525203e-06, "loss": 2.0029, "step": 1803 }, { "epoch": 0.7327376116978067, "grad_norm": 0.0, "learning_rate": 3.5185338305260163e-06, "loss": 2.0325, "step": 1804 }, { "epoch": 0.7331437855402112, "grad_norm": 0.0, "learning_rate": 3.5085211352391014e-06, "loss": 1.9006, "step": 1805 }, { "epoch": 0.7335499593826158, "grad_norm": 0.0, "learning_rate": 3.4985196749937976e-06, "loss": 2.06, "step": 1806 }, { "epoch": 0.7339561332250203, "grad_norm": 0.0, "learning_rate": 3.4885294671000015e-06, "loss": 1.9117, "step": 1807 }, { "epoch": 0.7343623070674249, "grad_norm": 0.0, "learning_rate": 3.478550528848134e-06, "loss": 2.075, "step": 1808 }, { "epoch": 0.7347684809098294, "grad_norm": 0.0, "learning_rate": 3.46858287750911e-06, "loss": 2.073, "step": 1809 }, { "epoch": 0.735174654752234, "grad_norm": 0.0, "learning_rate": 3.458626530334316e-06, "loss": 1.9292, "step": 1810 }, { "epoch": 0.7355808285946385, "grad_norm": 0.0, "learning_rate": 3.4486815045555612e-06, "loss": 2.0261, "step": 1811 }, { "epoch": 0.7359870024370431, "grad_norm": 0.0, "learning_rate": 3.438747817385071e-06, "loss": 2.1389, "step": 1812 }, { "epoch": 0.7363931762794476, "grad_norm": 0.0, "learning_rate": 3.4288254860154435e-06, "loss": 2.0675, "step": 1813 }, { "epoch": 0.7367993501218522, "grad_norm": 0.0, "learning_rate": 3.4189145276196244e-06, "loss": 2.0903, "step": 1814 }, { "epoch": 0.7372055239642566, "grad_norm": 0.0, "learning_rate": 3.4090149593508737e-06, "loss": 1.8568, "step": 1815 }, { "epoch": 0.7376116978066612, "grad_norm": 0.0, "learning_rate": 3.3991267983427424e-06, "loss": 1.9281, "step": 1816 }, { "epoch": 0.7380178716490658, "grad_norm": 0.0, "learning_rate": 3.3892500617090253e-06, "loss": 2.0964, "step": 1817 }, { "epoch": 0.7384240454914703, "grad_norm": 0.0, "learning_rate": 3.3793847665437674e-06, "loss": 2.1532, "step": 1818 }, { "epoch": 0.7388302193338749, "grad_norm": 0.0, "learning_rate": 3.3695309299211856e-06, "loss": 1.9562, "step": 1819 }, { "epoch": 0.7392363931762794, "grad_norm": 0.0, "learning_rate": 3.3596885688956893e-06, "loss": 2.0179, "step": 1820 }, { "epoch": 0.739642567018684, "grad_norm": 0.0, "learning_rate": 3.3498577005018083e-06, "loss": 2.0987, "step": 1821 }, { "epoch": 0.7400487408610885, "grad_norm": 0.0, "learning_rate": 3.340038341754189e-06, "loss": 2.0914, "step": 1822 }, { "epoch": 0.7404549147034931, "grad_norm": 0.0, "learning_rate": 3.330230509647561e-06, "loss": 2.2136, "step": 1823 }, { "epoch": 0.7408610885458976, "grad_norm": 0.0, "learning_rate": 3.3204342211566977e-06, "loss": 1.951, "step": 1824 }, { "epoch": 0.7412672623883022, "grad_norm": 0.0, "learning_rate": 3.3106494932363963e-06, "loss": 2.1297, "step": 1825 }, { "epoch": 0.7416734362307067, "grad_norm": 0.0, "learning_rate": 3.300876342821451e-06, "loss": 2.033, "step": 1826 }, { "epoch": 0.7420796100731113, "grad_norm": 0.0, "learning_rate": 3.2911147868266023e-06, "loss": 2.0135, "step": 1827 }, { "epoch": 0.7424857839155159, "grad_norm": 0.0, "learning_rate": 3.2813648421465483e-06, "loss": 2.1502, "step": 1828 }, { "epoch": 0.7428919577579204, "grad_norm": 0.0, "learning_rate": 3.2716265256558645e-06, "loss": 1.9298, "step": 1829 }, { "epoch": 0.743298131600325, "grad_norm": 0.0, "learning_rate": 3.2618998542090263e-06, "loss": 2.1073, "step": 1830 }, { "epoch": 0.7437043054427295, "grad_norm": 0.0, "learning_rate": 3.2521848446403337e-06, "loss": 1.9353, "step": 1831 }, { "epoch": 0.7441104792851341, "grad_norm": 0.0, "learning_rate": 3.2424815137639132e-06, "loss": 1.9977, "step": 1832 }, { "epoch": 0.7445166531275386, "grad_norm": 0.0, "learning_rate": 3.232789878373679e-06, "loss": 2.0785, "step": 1833 }, { "epoch": 0.7449228269699432, "grad_norm": 0.0, "learning_rate": 3.2231099552433e-06, "loss": 1.9604, "step": 1834 }, { "epoch": 0.7453290008123477, "grad_norm": 0.0, "learning_rate": 3.213441761126176e-06, "loss": 2.0326, "step": 1835 }, { "epoch": 0.7457351746547523, "grad_norm": 0.0, "learning_rate": 3.2037853127554096e-06, "loss": 2.0772, "step": 1836 }, { "epoch": 0.7461413484971567, "grad_norm": 0.0, "learning_rate": 3.194140626843767e-06, "loss": 2.0364, "step": 1837 }, { "epoch": 0.7465475223395613, "grad_norm": 0.0, "learning_rate": 3.1845077200836638e-06, "loss": 1.8897, "step": 1838 }, { "epoch": 0.7469536961819658, "grad_norm": 0.0, "learning_rate": 3.1748866091471253e-06, "loss": 2.0634, "step": 1839 }, { "epoch": 0.7473598700243704, "grad_norm": 0.0, "learning_rate": 3.165277310685764e-06, "loss": 1.9365, "step": 1840 }, { "epoch": 0.747766043866775, "grad_norm": 0.0, "learning_rate": 3.1556798413307466e-06, "loss": 1.8844, "step": 1841 }, { "epoch": 0.7481722177091795, "grad_norm": 0.0, "learning_rate": 3.1460942176927666e-06, "loss": 1.9921, "step": 1842 }, { "epoch": 0.7485783915515841, "grad_norm": 0.0, "learning_rate": 3.136520456362019e-06, "loss": 1.8865, "step": 1843 }, { "epoch": 0.7489845653939886, "grad_norm": 0.0, "learning_rate": 3.1269585739081564e-06, "loss": 1.8609, "step": 1844 }, { "epoch": 0.7493907392363932, "grad_norm": 0.0, "learning_rate": 3.117408586880293e-06, "loss": 2.061, "step": 1845 }, { "epoch": 0.7497969130787977, "grad_norm": 0.0, "learning_rate": 3.107870511806934e-06, "loss": 2.2184, "step": 1846 }, { "epoch": 0.7502030869212023, "grad_norm": 0.0, "learning_rate": 3.09834436519598e-06, "loss": 2.1547, "step": 1847 }, { "epoch": 0.7506092607636068, "grad_norm": 0.0, "learning_rate": 3.088830163534684e-06, "loss": 2.013, "step": 1848 }, { "epoch": 0.7510154346060114, "grad_norm": 0.0, "learning_rate": 3.0793279232896256e-06, "loss": 1.9212, "step": 1849 }, { "epoch": 0.7514216084484159, "grad_norm": 0.0, "learning_rate": 3.0698376609066828e-06, "loss": 1.909, "step": 1850 }, { "epoch": 0.7518277822908205, "grad_norm": 0.0, "learning_rate": 3.0603593928110044e-06, "loss": 2.1294, "step": 1851 }, { "epoch": 0.752233956133225, "grad_norm": 0.0, "learning_rate": 3.050893135406968e-06, "loss": 1.9651, "step": 1852 }, { "epoch": 0.7526401299756296, "grad_norm": 0.0, "learning_rate": 3.0414389050781877e-06, "loss": 1.9685, "step": 1853 }, { "epoch": 0.7530463038180342, "grad_norm": 0.0, "learning_rate": 3.0319967181874366e-06, "loss": 2.0938, "step": 1854 }, { "epoch": 0.7534524776604387, "grad_norm": 0.0, "learning_rate": 3.0225665910766668e-06, "loss": 2.012, "step": 1855 }, { "epoch": 0.7538586515028433, "grad_norm": 0.0, "learning_rate": 3.013148540066936e-06, "loss": 1.8942, "step": 1856 }, { "epoch": 0.7542648253452477, "grad_norm": 0.0, "learning_rate": 3.0037425814584175e-06, "loss": 1.8787, "step": 1857 }, { "epoch": 0.7546709991876523, "grad_norm": 0.0, "learning_rate": 2.9943487315303486e-06, "loss": 2.0224, "step": 1858 }, { "epoch": 0.7550771730300568, "grad_norm": 0.0, "learning_rate": 2.984967006541013e-06, "loss": 2.0522, "step": 1859 }, { "epoch": 0.7554833468724614, "grad_norm": 0.0, "learning_rate": 2.9755974227277073e-06, "loss": 2.1036, "step": 1860 }, { "epoch": 0.7558895207148659, "grad_norm": 0.0, "learning_rate": 2.966239996306717e-06, "loss": 1.8909, "step": 1861 }, { "epoch": 0.7562956945572705, "grad_norm": 0.0, "learning_rate": 2.9568947434732777e-06, "loss": 2.1235, "step": 1862 }, { "epoch": 0.756701868399675, "grad_norm": 0.0, "learning_rate": 2.9475616804015727e-06, "loss": 2.022, "step": 1863 }, { "epoch": 0.7571080422420796, "grad_norm": 0.0, "learning_rate": 2.9382408232446714e-06, "loss": 1.9403, "step": 1864 }, { "epoch": 0.7575142160844841, "grad_norm": 0.0, "learning_rate": 2.9289321881345257e-06, "loss": 2.0098, "step": 1865 }, { "epoch": 0.7579203899268887, "grad_norm": 0.0, "learning_rate": 2.919635791181934e-06, "loss": 1.851, "step": 1866 }, { "epoch": 0.7583265637692933, "grad_norm": 0.0, "learning_rate": 2.9103516484765137e-06, "loss": 2.176, "step": 1867 }, { "epoch": 0.7587327376116978, "grad_norm": 0.0, "learning_rate": 2.901079776086674e-06, "loss": 2.0907, "step": 1868 }, { "epoch": 0.7591389114541024, "grad_norm": 0.0, "learning_rate": 2.8918201900595788e-06, "loss": 2.0415, "step": 1869 }, { "epoch": 0.7595450852965069, "grad_norm": 0.0, "learning_rate": 2.882572906421145e-06, "loss": 2.0217, "step": 1870 }, { "epoch": 0.7599512591389115, "grad_norm": 0.0, "learning_rate": 2.87333794117598e-06, "loss": 2.001, "step": 1871 }, { "epoch": 0.760357432981316, "grad_norm": 0.0, "learning_rate": 2.8641153103073794e-06, "loss": 1.8916, "step": 1872 }, { "epoch": 0.7607636068237206, "grad_norm": 0.0, "learning_rate": 2.854905029777292e-06, "loss": 2.1893, "step": 1873 }, { "epoch": 0.7611697806661251, "grad_norm": 0.0, "learning_rate": 2.8457071155262885e-06, "loss": 2.2135, "step": 1874 }, { "epoch": 0.7615759545085297, "grad_norm": 0.0, "learning_rate": 2.836521583473537e-06, "loss": 1.9829, "step": 1875 }, { "epoch": 0.7619821283509342, "grad_norm": 0.0, "learning_rate": 2.8273484495167815e-06, "loss": 1.8687, "step": 1876 }, { "epoch": 0.7623883021933388, "grad_norm": 0.0, "learning_rate": 2.8181877295322923e-06, "loss": 2.1216, "step": 1877 }, { "epoch": 0.7627944760357434, "grad_norm": 0.0, "learning_rate": 2.809039439374878e-06, "loss": 2.0339, "step": 1878 }, { "epoch": 0.7632006498781478, "grad_norm": 0.0, "learning_rate": 2.7999035948778075e-06, "loss": 1.9225, "step": 1879 }, { "epoch": 0.7636068237205524, "grad_norm": 0.0, "learning_rate": 2.790780211852838e-06, "loss": 1.8471, "step": 1880 }, { "epoch": 0.7640129975629569, "grad_norm": 0.0, "learning_rate": 2.7816693060901334e-06, "loss": 2.1409, "step": 1881 }, { "epoch": 0.7644191714053615, "grad_norm": 0.0, "learning_rate": 2.7725708933582785e-06, "loss": 2.1637, "step": 1882 }, { "epoch": 0.764825345247766, "grad_norm": 0.0, "learning_rate": 2.7634849894042303e-06, "loss": 2.1391, "step": 1883 }, { "epoch": 0.7652315190901706, "grad_norm": 0.0, "learning_rate": 2.7544116099532982e-06, "loss": 1.9973, "step": 1884 }, { "epoch": 0.7656376929325751, "grad_norm": 0.0, "learning_rate": 2.7453507707091143e-06, "loss": 1.988, "step": 1885 }, { "epoch": 0.7660438667749797, "grad_norm": 0.0, "learning_rate": 2.7363024873536093e-06, "loss": 2.0629, "step": 1886 }, { "epoch": 0.7664500406173842, "grad_norm": 0.0, "learning_rate": 2.7272667755469707e-06, "loss": 2.0769, "step": 1887 }, { "epoch": 0.7668562144597888, "grad_norm": 0.0, "learning_rate": 2.718243650927652e-06, "loss": 1.9993, "step": 1888 }, { "epoch": 0.7672623883021933, "grad_norm": 0.0, "learning_rate": 2.7092331291122976e-06, "loss": 1.9491, "step": 1889 }, { "epoch": 0.7676685621445979, "grad_norm": 0.0, "learning_rate": 2.700235225695752e-06, "loss": 2.0339, "step": 1890 }, { "epoch": 0.7680747359870025, "grad_norm": 0.0, "learning_rate": 2.6912499562510175e-06, "loss": 2.045, "step": 1891 }, { "epoch": 0.768480909829407, "grad_norm": 0.0, "learning_rate": 2.6822773363292333e-06, "loss": 2.1218, "step": 1892 }, { "epoch": 0.7688870836718116, "grad_norm": 0.0, "learning_rate": 2.6733173814596405e-06, "loss": 2.1183, "step": 1893 }, { "epoch": 0.7692932575142161, "grad_norm": 0.0, "learning_rate": 2.6643701071495644e-06, "loss": 2.0136, "step": 1894 }, { "epoch": 0.7696994313566207, "grad_norm": 0.0, "learning_rate": 2.6554355288843847e-06, "loss": 2.1094, "step": 1895 }, { "epoch": 0.7701056051990252, "grad_norm": 0.0, "learning_rate": 2.646513662127501e-06, "loss": 2.0686, "step": 1896 }, { "epoch": 0.7705117790414298, "grad_norm": 0.0, "learning_rate": 2.6376045223203174e-06, "loss": 1.9496, "step": 1897 }, { "epoch": 0.7709179528838342, "grad_norm": 0.0, "learning_rate": 2.628708124882212e-06, "loss": 2.1232, "step": 1898 }, { "epoch": 0.7713241267262388, "grad_norm": 0.0, "learning_rate": 2.619824485210506e-06, "loss": 2.0487, "step": 1899 }, { "epoch": 0.7717303005686433, "grad_norm": 0.0, "learning_rate": 2.610953618680443e-06, "loss": 1.9861, "step": 1900 }, { "epoch": 0.7721364744110479, "grad_norm": 0.0, "learning_rate": 2.602095540645162e-06, "loss": 2.0174, "step": 1901 }, { "epoch": 0.7725426482534524, "grad_norm": 0.0, "learning_rate": 2.5932502664356553e-06, "loss": 2.0839, "step": 1902 }, { "epoch": 0.772948822095857, "grad_norm": 0.0, "learning_rate": 2.5844178113607756e-06, "loss": 2.0518, "step": 1903 }, { "epoch": 0.7733549959382616, "grad_norm": 0.0, "learning_rate": 2.5755981907071683e-06, "loss": 2.0429, "step": 1904 }, { "epoch": 0.7737611697806661, "grad_norm": 0.0, "learning_rate": 2.566791419739286e-06, "loss": 2.0421, "step": 1905 }, { "epoch": 0.7741673436230707, "grad_norm": 0.0, "learning_rate": 2.5579975136993253e-06, "loss": 2.0387, "step": 1906 }, { "epoch": 0.7745735174654752, "grad_norm": 0.0, "learning_rate": 2.5492164878072235e-06, "loss": 1.9755, "step": 1907 }, { "epoch": 0.7749796913078798, "grad_norm": 0.0, "learning_rate": 2.5404483572606264e-06, "loss": 1.9431, "step": 1908 }, { "epoch": 0.7753858651502843, "grad_norm": 0.0, "learning_rate": 2.531693137234861e-06, "loss": 2.05, "step": 1909 }, { "epoch": 0.7757920389926889, "grad_norm": 0.0, "learning_rate": 2.52295084288291e-06, "loss": 1.943, "step": 1910 }, { "epoch": 0.7761982128350934, "grad_norm": 0.0, "learning_rate": 2.5142214893353843e-06, "loss": 2.0487, "step": 1911 }, { "epoch": 0.776604386677498, "grad_norm": 0.0, "learning_rate": 2.505505091700492e-06, "loss": 2.1152, "step": 1912 }, { "epoch": 0.7770105605199025, "grad_norm": 0.0, "learning_rate": 2.496801665064035e-06, "loss": 2.1084, "step": 1913 }, { "epoch": 0.7774167343623071, "grad_norm": 0.0, "learning_rate": 2.4881112244893403e-06, "loss": 2.0125, "step": 1914 }, { "epoch": 0.7778229082047117, "grad_norm": 0.0, "learning_rate": 2.4794337850172867e-06, "loss": 1.9885, "step": 1915 }, { "epoch": 0.7782290820471162, "grad_norm": 0.0, "learning_rate": 2.470769361666231e-06, "loss": 1.9708, "step": 1916 }, { "epoch": 0.7786352558895208, "grad_norm": 0.0, "learning_rate": 2.4621179694320096e-06, "loss": 1.9482, "step": 1917 }, { "epoch": 0.7790414297319253, "grad_norm": 0.0, "learning_rate": 2.453479623287909e-06, "loss": 1.9961, "step": 1918 }, { "epoch": 0.7794476035743299, "grad_norm": 0.0, "learning_rate": 2.4448543381846314e-06, "loss": 2.1035, "step": 1919 }, { "epoch": 0.7798537774167343, "grad_norm": 0.0, "learning_rate": 2.436242129050276e-06, "loss": 1.9453, "step": 1920 }, { "epoch": 0.7802599512591389, "grad_norm": 0.0, "learning_rate": 2.4276430107903125e-06, "loss": 2.0646, "step": 1921 }, { "epoch": 0.7806661251015434, "grad_norm": 0.0, "learning_rate": 2.419056998287547e-06, "loss": 2.1336, "step": 1922 }, { "epoch": 0.781072298943948, "grad_norm": 0.0, "learning_rate": 2.4104841064021088e-06, "loss": 1.9627, "step": 1923 }, { "epoch": 0.7814784727863525, "grad_norm": 0.0, "learning_rate": 2.4019243499714205e-06, "loss": 2.1149, "step": 1924 }, { "epoch": 0.7818846466287571, "grad_norm": 0.0, "learning_rate": 2.393377743810166e-06, "loss": 2.0128, "step": 1925 }, { "epoch": 0.7822908204711616, "grad_norm": 0.0, "learning_rate": 2.3848443027102706e-06, "loss": 1.9839, "step": 1926 }, { "epoch": 0.7826969943135662, "grad_norm": 0.0, "learning_rate": 2.376324041440877e-06, "loss": 2.0042, "step": 1927 }, { "epoch": 0.7831031681559708, "grad_norm": 0.0, "learning_rate": 2.367816974748317e-06, "loss": 1.9932, "step": 1928 }, { "epoch": 0.7835093419983753, "grad_norm": 0.0, "learning_rate": 2.3593231173560773e-06, "loss": 1.8798, "step": 1929 }, { "epoch": 0.7839155158407799, "grad_norm": 0.0, "learning_rate": 2.3508424839647994e-06, "loss": 2.0245, "step": 1930 }, { "epoch": 0.7843216896831844, "grad_norm": 0.0, "learning_rate": 2.342375089252219e-06, "loss": 2.011, "step": 1931 }, { "epoch": 0.784727863525589, "grad_norm": 0.0, "learning_rate": 2.333920947873173e-06, "loss": 2.0029, "step": 1932 }, { "epoch": 0.7851340373679935, "grad_norm": 0.0, "learning_rate": 2.325480074459554e-06, "loss": 2.1062, "step": 1933 }, { "epoch": 0.7855402112103981, "grad_norm": 0.0, "learning_rate": 2.3170524836202936e-06, "loss": 2.0559, "step": 1934 }, { "epoch": 0.7859463850528026, "grad_norm": 0.0, "learning_rate": 2.3086381899413324e-06, "loss": 1.9752, "step": 1935 }, { "epoch": 0.7863525588952072, "grad_norm": 0.0, "learning_rate": 2.3002372079856016e-06, "loss": 2.0912, "step": 1936 }, { "epoch": 0.7867587327376117, "grad_norm": 0.0, "learning_rate": 2.291849552292982e-06, "loss": 2.0547, "step": 1937 }, { "epoch": 0.7871649065800163, "grad_norm": 0.0, "learning_rate": 2.2834752373803094e-06, "loss": 1.9807, "step": 1938 }, { "epoch": 0.7875710804224207, "grad_norm": 0.0, "learning_rate": 2.275114277741308e-06, "loss": 2.0859, "step": 1939 }, { "epoch": 0.7879772542648253, "grad_norm": 0.0, "learning_rate": 2.266766687846611e-06, "loss": 1.9668, "step": 1940 }, { "epoch": 0.7883834281072299, "grad_norm": 0.0, "learning_rate": 2.2584324821436887e-06, "loss": 2.0017, "step": 1941 }, { "epoch": 0.7887896019496344, "grad_norm": 0.0, "learning_rate": 2.250111675056863e-06, "loss": 1.9974, "step": 1942 }, { "epoch": 0.789195775792039, "grad_norm": 0.0, "learning_rate": 2.241804280987261e-06, "loss": 2.0402, "step": 1943 }, { "epoch": 0.7896019496344435, "grad_norm": 0.0, "learning_rate": 2.2335103143127934e-06, "loss": 2.0626, "step": 1944 }, { "epoch": 0.7900081234768481, "grad_norm": 0.0, "learning_rate": 2.225229789388137e-06, "loss": 2.0187, "step": 1945 }, { "epoch": 0.7904142973192526, "grad_norm": 0.0, "learning_rate": 2.216962720544703e-06, "loss": 1.9081, "step": 1946 }, { "epoch": 0.7908204711616572, "grad_norm": 0.0, "learning_rate": 2.208709122090602e-06, "loss": 2.0023, "step": 1947 }, { "epoch": 0.7912266450040617, "grad_norm": 0.0, "learning_rate": 2.200469008310655e-06, "loss": 2.0339, "step": 1948 }, { "epoch": 0.7916328188464663, "grad_norm": 0.0, "learning_rate": 2.1922423934663195e-06, "loss": 2.0432, "step": 1949 }, { "epoch": 0.7920389926888708, "grad_norm": 0.0, "learning_rate": 2.184029291795705e-06, "loss": 2.0507, "step": 1950 }, { "epoch": 0.7924451665312754, "grad_norm": 0.0, "learning_rate": 2.175829717513529e-06, "loss": 2.107, "step": 1951 }, { "epoch": 0.7928513403736799, "grad_norm": 0.0, "learning_rate": 2.167643684811096e-06, "loss": 1.9782, "step": 1952 }, { "epoch": 0.7932575142160845, "grad_norm": 0.0, "learning_rate": 2.159471207856275e-06, "loss": 1.9978, "step": 1953 }, { "epoch": 0.7936636880584891, "grad_norm": 0.0, "learning_rate": 2.151312300793473e-06, "loss": 2.0593, "step": 1954 }, { "epoch": 0.7940698619008936, "grad_norm": 0.0, "learning_rate": 2.143166977743615e-06, "loss": 2.0055, "step": 1955 }, { "epoch": 0.7944760357432982, "grad_norm": 0.0, "learning_rate": 2.135035252804104e-06, "loss": 1.9963, "step": 1956 }, { "epoch": 0.7948822095857027, "grad_norm": 0.0, "learning_rate": 2.1269171400488186e-06, "loss": 1.9866, "step": 1957 }, { "epoch": 0.7952883834281073, "grad_norm": 0.0, "learning_rate": 2.118812653528077e-06, "loss": 1.9388, "step": 1958 }, { "epoch": 0.7956945572705117, "grad_norm": 0.0, "learning_rate": 2.110721807268613e-06, "loss": 1.9919, "step": 1959 }, { "epoch": 0.7961007311129163, "grad_norm": 0.0, "learning_rate": 2.1026446152735503e-06, "loss": 2.0758, "step": 1960 }, { "epoch": 0.7965069049553208, "grad_norm": 0.0, "learning_rate": 2.0945810915223876e-06, "loss": 2.1538, "step": 1961 }, { "epoch": 0.7969130787977254, "grad_norm": 0.0, "learning_rate": 2.086531249970952e-06, "loss": 1.9382, "step": 1962 }, { "epoch": 0.7973192526401299, "grad_norm": 0.0, "learning_rate": 2.0784951045514135e-06, "loss": 2.0679, "step": 1963 }, { "epoch": 0.7977254264825345, "grad_norm": 0.0, "learning_rate": 2.0704726691722134e-06, "loss": 2.1086, "step": 1964 }, { "epoch": 0.7981316003249391, "grad_norm": 0.0, "learning_rate": 2.062463957718087e-06, "loss": 1.9981, "step": 1965 }, { "epoch": 0.7985377741673436, "grad_norm": 0.0, "learning_rate": 2.0544689840499988e-06, "loss": 1.9198, "step": 1966 }, { "epoch": 0.7989439480097482, "grad_norm": 0.0, "learning_rate": 2.046487762005146e-06, "loss": 1.9197, "step": 1967 }, { "epoch": 0.7993501218521527, "grad_norm": 0.0, "learning_rate": 2.0385203053969238e-06, "loss": 2.0124, "step": 1968 }, { "epoch": 0.7997562956945573, "grad_norm": 0.0, "learning_rate": 2.030566628014903e-06, "loss": 2.1314, "step": 1969 }, { "epoch": 0.8001624695369618, "grad_norm": 0.0, "learning_rate": 2.022626743624807e-06, "loss": 2.1097, "step": 1970 }, { "epoch": 0.8005686433793664, "grad_norm": 0.0, "learning_rate": 2.0147006659684875e-06, "loss": 1.9252, "step": 1971 }, { "epoch": 0.8009748172217709, "grad_norm": 0.0, "learning_rate": 2.006788408763891e-06, "loss": 1.9386, "step": 1972 }, { "epoch": 0.8013809910641755, "grad_norm": 0.0, "learning_rate": 1.9988899857050647e-06, "loss": 1.9693, "step": 1973 }, { "epoch": 0.80178716490658, "grad_norm": 0.0, "learning_rate": 1.991005410462089e-06, "loss": 2.0542, "step": 1974 }, { "epoch": 0.8021933387489846, "grad_norm": 0.0, "learning_rate": 1.9831346966810994e-06, "loss": 2.0425, "step": 1975 }, { "epoch": 0.8025995125913891, "grad_norm": 0.0, "learning_rate": 1.9752778579842214e-06, "loss": 2.0546, "step": 1976 }, { "epoch": 0.8030056864337937, "grad_norm": 0.0, "learning_rate": 1.967434907969579e-06, "loss": 1.9884, "step": 1977 }, { "epoch": 0.8034118602761983, "grad_norm": 0.0, "learning_rate": 1.9596058602112533e-06, "loss": 2.0716, "step": 1978 }, { "epoch": 0.8038180341186028, "grad_norm": 0.0, "learning_rate": 1.951790728259266e-06, "loss": 1.8453, "step": 1979 }, { "epoch": 0.8042242079610074, "grad_norm": 0.0, "learning_rate": 1.943989525639559e-06, "loss": 2.0662, "step": 1980 }, { "epoch": 0.8046303818034118, "grad_norm": 0.0, "learning_rate": 1.9362022658539526e-06, "loss": 1.7474, "step": 1981 }, { "epoch": 0.8050365556458164, "grad_norm": 0.0, "learning_rate": 1.928428962380148e-06, "loss": 1.9577, "step": 1982 }, { "epoch": 0.8054427294882209, "grad_norm": 0.0, "learning_rate": 1.9206696286716887e-06, "loss": 2.0333, "step": 1983 }, { "epoch": 0.8058489033306255, "grad_norm": 0.0, "learning_rate": 1.9129242781579383e-06, "loss": 1.985, "step": 1984 }, { "epoch": 0.80625507717303, "grad_norm": 0.0, "learning_rate": 1.9051929242440604e-06, "loss": 2.0687, "step": 1985 }, { "epoch": 0.8066612510154346, "grad_norm": 0.0, "learning_rate": 1.8974755803109968e-06, "loss": 1.9951, "step": 1986 }, { "epoch": 0.8070674248578391, "grad_norm": 0.0, "learning_rate": 1.889772259715431e-06, "loss": 1.9467, "step": 1987 }, { "epoch": 0.8074735987002437, "grad_norm": 0.0, "learning_rate": 1.8820829757897952e-06, "loss": 2.0612, "step": 1988 }, { "epoch": 0.8078797725426482, "grad_norm": 0.0, "learning_rate": 1.8744077418422036e-06, "loss": 1.9147, "step": 1989 }, { "epoch": 0.8082859463850528, "grad_norm": 0.0, "learning_rate": 1.866746571156479e-06, "loss": 1.9189, "step": 1990 }, { "epoch": 0.8086921202274574, "grad_norm": 0.0, "learning_rate": 1.8590994769920834e-06, "loss": 1.9322, "step": 1991 }, { "epoch": 0.8090982940698619, "grad_norm": 0.0, "learning_rate": 1.851466472584127e-06, "loss": 2.0667, "step": 1992 }, { "epoch": 0.8095044679122665, "grad_norm": 0.0, "learning_rate": 1.8438475711433312e-06, "loss": 2.0485, "step": 1993 }, { "epoch": 0.809910641754671, "grad_norm": 0.0, "learning_rate": 1.8362427858560094e-06, "loss": 2.0057, "step": 1994 }, { "epoch": 0.8103168155970756, "grad_norm": 0.0, "learning_rate": 1.8286521298840443e-06, "loss": 1.9323, "step": 1995 }, { "epoch": 0.8107229894394801, "grad_norm": 0.0, "learning_rate": 1.8210756163648657e-06, "loss": 2.0258, "step": 1996 }, { "epoch": 0.8111291632818847, "grad_norm": 0.0, "learning_rate": 1.8135132584114167e-06, "loss": 1.9728, "step": 1997 }, { "epoch": 0.8115353371242892, "grad_norm": 0.0, "learning_rate": 1.8059650691121611e-06, "loss": 2.0476, "step": 1998 }, { "epoch": 0.8119415109666938, "grad_norm": 0.0, "learning_rate": 1.7984310615310153e-06, "loss": 1.9584, "step": 1999 }, { "epoch": 0.8123476848090982, "grad_norm": 0.0, "learning_rate": 1.7909112487073754e-06, "loss": 2.0482, "step": 2000 }, { "epoch": 0.8127538586515028, "grad_norm": 0.0, "learning_rate": 1.783405643656051e-06, "loss": 2.0743, "step": 2001 }, { "epoch": 0.8131600324939074, "grad_norm": 0.0, "learning_rate": 1.7759142593672707e-06, "loss": 2.0408, "step": 2002 }, { "epoch": 0.8135662063363119, "grad_norm": 0.0, "learning_rate": 1.7684371088066509e-06, "loss": 1.827, "step": 2003 }, { "epoch": 0.8139723801787165, "grad_norm": 0.0, "learning_rate": 1.7609742049151702e-06, "loss": 1.9483, "step": 2004 }, { "epoch": 0.814378554021121, "grad_norm": 0.0, "learning_rate": 1.7535255606091505e-06, "loss": 2.0477, "step": 2005 }, { "epoch": 0.8147847278635256, "grad_norm": 0.0, "learning_rate": 1.74609118878024e-06, "loss": 2.0616, "step": 2006 }, { "epoch": 0.8151909017059301, "grad_norm": 0.0, "learning_rate": 1.7386711022953728e-06, "loss": 2.0788, "step": 2007 }, { "epoch": 0.8155970755483347, "grad_norm": 0.0, "learning_rate": 1.7312653139967694e-06, "loss": 2.1194, "step": 2008 }, { "epoch": 0.8160032493907392, "grad_norm": 0.0, "learning_rate": 1.7238738367019003e-06, "loss": 2.0241, "step": 2009 }, { "epoch": 0.8164094232331438, "grad_norm": 0.0, "learning_rate": 1.7164966832034668e-06, "loss": 1.931, "step": 2010 }, { "epoch": 0.8168155970755483, "grad_norm": 0.0, "learning_rate": 1.7091338662693814e-06, "loss": 2.0263, "step": 2011 }, { "epoch": 0.8172217709179529, "grad_norm": 0.0, "learning_rate": 1.7017853986427423e-06, "loss": 2.0955, "step": 2012 }, { "epoch": 0.8176279447603574, "grad_norm": 0.0, "learning_rate": 1.6944512930418145e-06, "loss": 1.9992, "step": 2013 }, { "epoch": 0.818034118602762, "grad_norm": 0.0, "learning_rate": 1.6871315621599982e-06, "loss": 1.8629, "step": 2014 }, { "epoch": 0.8184402924451666, "grad_norm": 0.0, "learning_rate": 1.679826218665832e-06, "loss": 1.8735, "step": 2015 }, { "epoch": 0.8188464662875711, "grad_norm": 0.0, "learning_rate": 1.6725352752029332e-06, "loss": 1.9735, "step": 2016 }, { "epoch": 0.8192526401299757, "grad_norm": 0.0, "learning_rate": 1.6652587443900115e-06, "loss": 2.0017, "step": 2017 }, { "epoch": 0.8196588139723802, "grad_norm": 0.0, "learning_rate": 1.6579966388208257e-06, "loss": 1.9551, "step": 2018 }, { "epoch": 0.8200649878147848, "grad_norm": 0.0, "learning_rate": 1.650748971064169e-06, "loss": 2.0187, "step": 2019 }, { "epoch": 0.8204711616571893, "grad_norm": 0.0, "learning_rate": 1.643515753663847e-06, "loss": 2.1338, "step": 2020 }, { "epoch": 0.8208773354995939, "grad_norm": 0.0, "learning_rate": 1.636296999138659e-06, "loss": 1.9588, "step": 2021 }, { "epoch": 0.8212835093419983, "grad_norm": 0.0, "learning_rate": 1.6290927199823604e-06, "loss": 2.0189, "step": 2022 }, { "epoch": 0.8216896831844029, "grad_norm": 0.0, "learning_rate": 1.6219029286636733e-06, "loss": 2.0828, "step": 2023 }, { "epoch": 0.8220958570268074, "grad_norm": 0.0, "learning_rate": 1.6147276376262256e-06, "loss": 1.9983, "step": 2024 }, { "epoch": 0.822502030869212, "grad_norm": 0.0, "learning_rate": 1.607566859288565e-06, "loss": 2.0031, "step": 2025 }, { "epoch": 0.8229082047116165, "grad_norm": 0.0, "learning_rate": 1.6004206060441096e-06, "loss": 1.9908, "step": 2026 }, { "epoch": 0.8233143785540211, "grad_norm": 0.0, "learning_rate": 1.5932888902611453e-06, "loss": 1.8913, "step": 2027 }, { "epoch": 0.8237205523964257, "grad_norm": 0.0, "learning_rate": 1.5861717242827935e-06, "loss": 1.9554, "step": 2028 }, { "epoch": 0.8241267262388302, "grad_norm": 0.0, "learning_rate": 1.579069120426997e-06, "loss": 2.0935, "step": 2029 }, { "epoch": 0.8245329000812348, "grad_norm": 0.0, "learning_rate": 1.5719810909864941e-06, "loss": 2.0035, "step": 2030 }, { "epoch": 0.8249390739236393, "grad_norm": 0.0, "learning_rate": 1.5649076482287995e-06, "loss": 1.9869, "step": 2031 }, { "epoch": 0.8253452477660439, "grad_norm": 0.0, "learning_rate": 1.5578488043961748e-06, "loss": 1.9527, "step": 2032 }, { "epoch": 0.8257514216084484, "grad_norm": 0.0, "learning_rate": 1.550804571705632e-06, "loss": 1.9945, "step": 2033 }, { "epoch": 0.826157595450853, "grad_norm": 0.0, "learning_rate": 1.543774962348874e-06, "loss": 1.9436, "step": 2034 }, { "epoch": 0.8265637692932575, "grad_norm": 0.0, "learning_rate": 1.5367599884923101e-06, "loss": 2.1102, "step": 2035 }, { "epoch": 0.8269699431356621, "grad_norm": 0.0, "learning_rate": 1.5297596622770117e-06, "loss": 1.9169, "step": 2036 }, { "epoch": 0.8273761169780666, "grad_norm": 0.0, "learning_rate": 1.5227739958187027e-06, "loss": 2.0167, "step": 2037 }, { "epoch": 0.8277822908204712, "grad_norm": 0.0, "learning_rate": 1.5158030012077329e-06, "loss": 2.0314, "step": 2038 }, { "epoch": 0.8281884646628757, "grad_norm": 0.0, "learning_rate": 1.5088466905090592e-06, "loss": 1.9159, "step": 2039 }, { "epoch": 0.8285946385052803, "grad_norm": 0.0, "learning_rate": 1.5019050757622266e-06, "loss": 1.9208, "step": 2040 }, { "epoch": 0.8290008123476849, "grad_norm": 0.0, "learning_rate": 1.4949781689813403e-06, "loss": 2.02, "step": 2041 }, { "epoch": 0.8294069861900893, "grad_norm": 0.0, "learning_rate": 1.4880659821550547e-06, "loss": 1.8278, "step": 2042 }, { "epoch": 0.829813160032494, "grad_norm": 0.0, "learning_rate": 1.4811685272465471e-06, "loss": 1.8654, "step": 2043 }, { "epoch": 0.8302193338748984, "grad_norm": 0.0, "learning_rate": 1.4742858161934937e-06, "loss": 1.9327, "step": 2044 }, { "epoch": 0.830625507717303, "grad_norm": 0.0, "learning_rate": 1.4674178609080603e-06, "loss": 1.8045, "step": 2045 }, { "epoch": 0.8310316815597075, "grad_norm": 0.0, "learning_rate": 1.4605646732768685e-06, "loss": 2.0083, "step": 2046 }, { "epoch": 0.8314378554021121, "grad_norm": 0.0, "learning_rate": 1.4537262651609784e-06, "loss": 2.0928, "step": 2047 }, { "epoch": 0.8318440292445166, "grad_norm": 0.0, "learning_rate": 1.4469026483958837e-06, "loss": 2.021, "step": 2048 }, { "epoch": 0.8322502030869212, "grad_norm": 0.0, "learning_rate": 1.4400938347914583e-06, "loss": 2.0684, "step": 2049 }, { "epoch": 0.8326563769293257, "grad_norm": 0.0, "learning_rate": 1.4332998361319783e-06, "loss": 2.0315, "step": 2050 }, { "epoch": 0.8330625507717303, "grad_norm": 0.0, "learning_rate": 1.4265206641760588e-06, "loss": 1.9685, "step": 2051 }, { "epoch": 0.8334687246141349, "grad_norm": 0.0, "learning_rate": 1.4197563306566654e-06, "loss": 2.111, "step": 2052 }, { "epoch": 0.8338748984565394, "grad_norm": 0.0, "learning_rate": 1.4130068472810799e-06, "loss": 1.9925, "step": 2053 }, { "epoch": 0.834281072298944, "grad_norm": 0.0, "learning_rate": 1.4062722257308803e-06, "loss": 1.9667, "step": 2054 }, { "epoch": 0.8346872461413485, "grad_norm": 0.0, "learning_rate": 1.3995524776619262e-06, "loss": 1.8639, "step": 2055 }, { "epoch": 0.8350934199837531, "grad_norm": 0.0, "learning_rate": 1.3928476147043335e-06, "loss": 1.9788, "step": 2056 }, { "epoch": 0.8354995938261576, "grad_norm": 0.0, "learning_rate": 1.3861576484624506e-06, "loss": 2.0145, "step": 2057 }, { "epoch": 0.8359057676685622, "grad_norm": 0.0, "learning_rate": 1.3794825905148557e-06, "loss": 1.924, "step": 2058 }, { "epoch": 0.8363119415109667, "grad_norm": 0.0, "learning_rate": 1.3728224524143096e-06, "loss": 2.033, "step": 2059 }, { "epoch": 0.8367181153533713, "grad_norm": 0.0, "learning_rate": 1.3661772456877675e-06, "loss": 1.9989, "step": 2060 }, { "epoch": 0.8371242891957758, "grad_norm": 0.0, "learning_rate": 1.359546981836326e-06, "loss": 2.0376, "step": 2061 }, { "epoch": 0.8375304630381804, "grad_norm": 0.0, "learning_rate": 1.3529316723352303e-06, "loss": 1.9486, "step": 2062 }, { "epoch": 0.8379366368805848, "grad_norm": 0.0, "learning_rate": 1.346331328633841e-06, "loss": 1.903, "step": 2063 }, { "epoch": 0.8383428107229894, "grad_norm": 0.0, "learning_rate": 1.339745962155613e-06, "loss": 1.8416, "step": 2064 }, { "epoch": 0.838748984565394, "grad_norm": 0.0, "learning_rate": 1.3331755842980886e-06, "loss": 2.033, "step": 2065 }, { "epoch": 0.8391551584077985, "grad_norm": 0.0, "learning_rate": 1.3266202064328548e-06, "loss": 1.9796, "step": 2066 }, { "epoch": 0.8395613322502031, "grad_norm": 0.0, "learning_rate": 1.3200798399055514e-06, "loss": 1.9744, "step": 2067 }, { "epoch": 0.8399675060926076, "grad_norm": 0.0, "learning_rate": 1.3135544960358281e-06, "loss": 1.9621, "step": 2068 }, { "epoch": 0.8403736799350122, "grad_norm": 0.0, "learning_rate": 1.307044186117341e-06, "loss": 2.0617, "step": 2069 }, { "epoch": 0.8407798537774167, "grad_norm": 0.0, "learning_rate": 1.3005489214177213e-06, "loss": 2.045, "step": 2070 }, { "epoch": 0.8411860276198213, "grad_norm": 0.0, "learning_rate": 1.2940687131785668e-06, "loss": 1.9981, "step": 2071 }, { "epoch": 0.8415922014622258, "grad_norm": 0.0, "learning_rate": 1.2876035726154046e-06, "loss": 1.9532, "step": 2072 }, { "epoch": 0.8419983753046304, "grad_norm": 0.0, "learning_rate": 1.2811535109176999e-06, "loss": 2.0372, "step": 2073 }, { "epoch": 0.8424045491470349, "grad_norm": 0.0, "learning_rate": 1.2747185392488048e-06, "loss": 1.9951, "step": 2074 }, { "epoch": 0.8428107229894395, "grad_norm": 0.0, "learning_rate": 1.268298668745971e-06, "loss": 1.996, "step": 2075 }, { "epoch": 0.843216896831844, "grad_norm": 0.0, "learning_rate": 1.2618939105202966e-06, "loss": 1.898, "step": 2076 }, { "epoch": 0.8436230706742486, "grad_norm": 0.0, "learning_rate": 1.2555042756567347e-06, "loss": 2.0223, "step": 2077 }, { "epoch": 0.8440292445166532, "grad_norm": 0.0, "learning_rate": 1.249129775214064e-06, "loss": 2.0379, "step": 2078 }, { "epoch": 0.8444354183590577, "grad_norm": 0.0, "learning_rate": 1.2427704202248646e-06, "loss": 2.0679, "step": 2079 }, { "epoch": 0.8448415922014623, "grad_norm": 0.0, "learning_rate": 1.2364262216955092e-06, "loss": 2.076, "step": 2080 }, { "epoch": 0.8452477660438668, "grad_norm": 0.0, "learning_rate": 1.2300971906061354e-06, "loss": 2.0448, "step": 2081 }, { "epoch": 0.8456539398862714, "grad_norm": 0.0, "learning_rate": 1.2237833379106257e-06, "loss": 1.769, "step": 2082 }, { "epoch": 0.8460601137286758, "grad_norm": 0.0, "learning_rate": 1.2174846745366065e-06, "loss": 1.8179, "step": 2083 }, { "epoch": 0.8464662875710804, "grad_norm": 0.0, "learning_rate": 1.2112012113853955e-06, "loss": 2.1034, "step": 2084 }, { "epoch": 0.8468724614134849, "grad_norm": 0.0, "learning_rate": 1.2049329593320248e-06, "loss": 2.0479, "step": 2085 }, { "epoch": 0.8472786352558895, "grad_norm": 0.0, "learning_rate": 1.1986799292251816e-06, "loss": 1.9967, "step": 2086 }, { "epoch": 0.847684809098294, "grad_norm": 0.0, "learning_rate": 1.1924421318872182e-06, "loss": 1.887, "step": 2087 }, { "epoch": 0.8480909829406986, "grad_norm": 0.0, "learning_rate": 1.1862195781141206e-06, "loss": 2.0084, "step": 2088 }, { "epoch": 0.8484971567831032, "grad_norm": 0.0, "learning_rate": 1.1800122786754908e-06, "loss": 2.0228, "step": 2089 }, { "epoch": 0.8489033306255077, "grad_norm": 0.0, "learning_rate": 1.1738202443145307e-06, "loss": 1.9175, "step": 2090 }, { "epoch": 0.8493095044679123, "grad_norm": 0.0, "learning_rate": 1.1676434857480245e-06, "loss": 2.1184, "step": 2091 }, { "epoch": 0.8497156783103168, "grad_norm": 0.0, "learning_rate": 1.161482013666313e-06, "loss": 2.1404, "step": 2092 }, { "epoch": 0.8501218521527214, "grad_norm": 0.0, "learning_rate": 1.1553358387332825e-06, "loss": 1.9236, "step": 2093 }, { "epoch": 0.8505280259951259, "grad_norm": 0.0, "learning_rate": 1.1492049715863464e-06, "loss": 1.9098, "step": 2094 }, { "epoch": 0.8509341998375305, "grad_norm": 0.0, "learning_rate": 1.143089422836422e-06, "loss": 1.977, "step": 2095 }, { "epoch": 0.851340373679935, "grad_norm": 0.0, "learning_rate": 1.1369892030679142e-06, "loss": 1.9678, "step": 2096 }, { "epoch": 0.8517465475223396, "grad_norm": 0.0, "learning_rate": 1.1309043228386984e-06, "loss": 1.9612, "step": 2097 }, { "epoch": 0.8521527213647441, "grad_norm": 0.0, "learning_rate": 1.1248347926801029e-06, "loss": 2.0604, "step": 2098 }, { "epoch": 0.8525588952071487, "grad_norm": 0.0, "learning_rate": 1.1187806230968801e-06, "loss": 1.9636, "step": 2099 }, { "epoch": 0.8529650690495532, "grad_norm": 0.0, "learning_rate": 1.112741824567215e-06, "loss": 2.0582, "step": 2100 }, { "epoch": 0.8533712428919578, "grad_norm": 0.0, "learning_rate": 1.1067184075426695e-06, "loss": 1.9882, "step": 2101 }, { "epoch": 0.8537774167343624, "grad_norm": 0.0, "learning_rate": 1.100710382448198e-06, "loss": 2.2015, "step": 2102 }, { "epoch": 0.8541835905767668, "grad_norm": 0.0, "learning_rate": 1.0947177596821101e-06, "loss": 1.894, "step": 2103 }, { "epoch": 0.8545897644191714, "grad_norm": 0.0, "learning_rate": 1.0887405496160575e-06, "loss": 1.9542, "step": 2104 }, { "epoch": 0.8549959382615759, "grad_norm": 0.0, "learning_rate": 1.0827787625950193e-06, "loss": 1.8706, "step": 2105 }, { "epoch": 0.8554021121039805, "grad_norm": 0.0, "learning_rate": 1.0768324089372816e-06, "loss": 1.8183, "step": 2106 }, { "epoch": 0.855808285946385, "grad_norm": 0.0, "learning_rate": 1.0709014989344113e-06, "loss": 2.0053, "step": 2107 }, { "epoch": 0.8562144597887896, "grad_norm": 0.0, "learning_rate": 1.0649860428512604e-06, "loss": 2.1309, "step": 2108 }, { "epoch": 0.8566206336311941, "grad_norm": 0.0, "learning_rate": 1.0590860509259205e-06, "loss": 1.8927, "step": 2109 }, { "epoch": 0.8570268074735987, "grad_norm": 0.0, "learning_rate": 1.053201533369731e-06, "loss": 2.1209, "step": 2110 }, { "epoch": 0.8574329813160032, "grad_norm": 0.0, "learning_rate": 1.0473325003672386e-06, "loss": 1.8436, "step": 2111 }, { "epoch": 0.8578391551584078, "grad_norm": 0.0, "learning_rate": 1.0414789620761968e-06, "loss": 2.0495, "step": 2112 }, { "epoch": 0.8582453290008123, "grad_norm": 0.0, "learning_rate": 1.0356409286275393e-06, "loss": 1.8718, "step": 2113 }, { "epoch": 0.8586515028432169, "grad_norm": 0.0, "learning_rate": 1.029818410125365e-06, "loss": 1.9463, "step": 2114 }, { "epoch": 0.8590576766856215, "grad_norm": 0.0, "learning_rate": 1.0240114166469229e-06, "loss": 1.9503, "step": 2115 }, { "epoch": 0.859463850528026, "grad_norm": 0.0, "learning_rate": 1.0182199582425911e-06, "loss": 1.9, "step": 2116 }, { "epoch": 0.8598700243704306, "grad_norm": 0.0, "learning_rate": 1.0124440449358553e-06, "loss": 2.0225, "step": 2117 }, { "epoch": 0.8602761982128351, "grad_norm": 0.0, "learning_rate": 1.0066836867233087e-06, "loss": 2.1188, "step": 2118 }, { "epoch": 0.8606823720552397, "grad_norm": 0.0, "learning_rate": 1.000938893574609e-06, "loss": 2.0477, "step": 2119 }, { "epoch": 0.8610885458976442, "grad_norm": 0.0, "learning_rate": 9.952096754324847e-07, "loss": 1.9922, "step": 2120 }, { "epoch": 0.8614947197400488, "grad_norm": 0.0, "learning_rate": 9.894960422127032e-07, "loss": 2.006, "step": 2121 }, { "epoch": 0.8619008935824533, "grad_norm": 0.0, "learning_rate": 9.837980038040607e-07, "loss": 1.9491, "step": 2122 }, { "epoch": 0.8623070674248579, "grad_norm": 0.0, "learning_rate": 9.78115570068362e-07, "loss": 1.9933, "step": 2123 }, { "epoch": 0.8627132412672623, "grad_norm": 0.0, "learning_rate": 9.724487508404045e-07, "loss": 1.9658, "step": 2124 }, { "epoch": 0.8631194151096669, "grad_norm": 0.0, "learning_rate": 9.667975559279641e-07, "loss": 1.8924, "step": 2125 }, { "epoch": 0.8635255889520714, "grad_norm": 0.0, "learning_rate": 9.611619951117657e-07, "loss": 2.0453, "step": 2126 }, { "epoch": 0.863931762794476, "grad_norm": 0.0, "learning_rate": 9.555420781454861e-07, "loss": 2.0803, "step": 2127 }, { "epoch": 0.8643379366368806, "grad_norm": 0.0, "learning_rate": 9.499378147557225e-07, "loss": 1.913, "step": 2128 }, { "epoch": 0.8647441104792851, "grad_norm": 0.0, "learning_rate": 9.443492146419786e-07, "loss": 1.9068, "step": 2129 }, { "epoch": 0.8651502843216897, "grad_norm": 0.0, "learning_rate": 9.387762874766515e-07, "loss": 2.0384, "step": 2130 }, { "epoch": 0.8655564581640942, "grad_norm": 0.0, "learning_rate": 9.332190429050137e-07, "loss": 2.0282, "step": 2131 }, { "epoch": 0.8659626320064988, "grad_norm": 0.0, "learning_rate": 9.276774905451868e-07, "loss": 1.898, "step": 2132 }, { "epoch": 0.8663688058489033, "grad_norm": 0.0, "learning_rate": 9.221516399881469e-07, "loss": 1.897, "step": 2133 }, { "epoch": 0.8667749796913079, "grad_norm": 0.0, "learning_rate": 9.166415007976803e-07, "loss": 2.0147, "step": 2134 }, { "epoch": 0.8671811535337124, "grad_norm": 0.0, "learning_rate": 9.111470825103952e-07, "loss": 2.0314, "step": 2135 }, { "epoch": 0.867587327376117, "grad_norm": 0.0, "learning_rate": 9.05668394635677e-07, "loss": 2.0919, "step": 2136 }, { "epoch": 0.8679935012185215, "grad_norm": 0.0, "learning_rate": 9.002054466556964e-07, "loss": 1.9033, "step": 2137 }, { "epoch": 0.8683996750609261, "grad_norm": 0.0, "learning_rate": 8.94758248025378e-07, "loss": 2.0417, "step": 2138 }, { "epoch": 0.8688058489033307, "grad_norm": 0.0, "learning_rate": 8.893268081723894e-07, "loss": 1.9907, "step": 2139 }, { "epoch": 0.8692120227457352, "grad_norm": 0.0, "learning_rate": 8.839111364971231e-07, "loss": 2.0324, "step": 2140 }, { "epoch": 0.8696181965881398, "grad_norm": 0.0, "learning_rate": 8.785112423726827e-07, "loss": 1.9322, "step": 2141 }, { "epoch": 0.8700243704305443, "grad_norm": 0.0, "learning_rate": 8.7312713514486e-07, "loss": 2.0949, "step": 2142 }, { "epoch": 0.8704305442729489, "grad_norm": 0.0, "learning_rate": 8.677588241321344e-07, "loss": 1.8217, "step": 2143 }, { "epoch": 0.8708367181153533, "grad_norm": 0.0, "learning_rate": 8.624063186256327e-07, "loss": 2.0106, "step": 2144 }, { "epoch": 0.871242891957758, "grad_norm": 0.0, "learning_rate": 8.570696278891388e-07, "loss": 2.0793, "step": 2145 }, { "epoch": 0.8716490658001624, "grad_norm": 0.0, "learning_rate": 8.517487611590558e-07, "loss": 2.0193, "step": 2146 }, { "epoch": 0.872055239642567, "grad_norm": 0.0, "learning_rate": 8.46443727644406e-07, "loss": 2.0104, "step": 2147 }, { "epoch": 0.8724614134849715, "grad_norm": 0.0, "learning_rate": 8.411545365268059e-07, "loss": 1.8193, "step": 2148 }, { "epoch": 0.8728675873273761, "grad_norm": 0.0, "learning_rate": 8.358811969604541e-07, "loss": 1.9529, "step": 2149 }, { "epoch": 0.8732737611697806, "grad_norm": 0.0, "learning_rate": 8.306237180721121e-07, "loss": 2.0446, "step": 2150 }, { "epoch": 0.8736799350121852, "grad_norm": 0.0, "learning_rate": 8.253821089610959e-07, "loss": 1.9356, "step": 2151 }, { "epoch": 0.8740861088545898, "grad_norm": 0.0, "learning_rate": 8.201563786992462e-07, "loss": 2.1414, "step": 2152 }, { "epoch": 0.8744922826969943, "grad_norm": 0.0, "learning_rate": 8.149465363309295e-07, "loss": 1.9174, "step": 2153 }, { "epoch": 0.8748984565393989, "grad_norm": 0.0, "learning_rate": 8.097525908730108e-07, "loss": 1.8254, "step": 2154 }, { "epoch": 0.8753046303818034, "grad_norm": 0.0, "learning_rate": 8.045745513148451e-07, "loss": 1.9834, "step": 2155 }, { "epoch": 0.875710804224208, "grad_norm": 0.0, "learning_rate": 7.994124266182568e-07, "loss": 2.036, "step": 2156 }, { "epoch": 0.8761169780666125, "grad_norm": 0.0, "learning_rate": 7.942662257175204e-07, "loss": 1.7861, "step": 2157 }, { "epoch": 0.8765231519090171, "grad_norm": 0.0, "learning_rate": 7.891359575193613e-07, "loss": 2.0725, "step": 2158 }, { "epoch": 0.8769293257514216, "grad_norm": 0.0, "learning_rate": 7.840216309029169e-07, "loss": 2.0342, "step": 2159 }, { "epoch": 0.8773354995938262, "grad_norm": 0.0, "learning_rate": 7.789232547197489e-07, "loss": 1.9945, "step": 2160 }, { "epoch": 0.8777416734362307, "grad_norm": 0.0, "learning_rate": 7.738408377937978e-07, "loss": 2.0734, "step": 2161 }, { "epoch": 0.8781478472786353, "grad_norm": 0.0, "learning_rate": 7.687743889213939e-07, "loss": 2.0107, "step": 2162 }, { "epoch": 0.8785540211210398, "grad_norm": 0.0, "learning_rate": 7.63723916871224e-07, "loss": 1.9751, "step": 2163 }, { "epoch": 0.8789601949634444, "grad_norm": 0.0, "learning_rate": 7.586894303843284e-07, "loss": 2.0618, "step": 2164 }, { "epoch": 0.879366368805849, "grad_norm": 0.0, "learning_rate": 7.536709381740781e-07, "loss": 1.9379, "step": 2165 }, { "epoch": 0.8797725426482534, "grad_norm": 0.0, "learning_rate": 7.486684489261609e-07, "loss": 2.013, "step": 2166 }, { "epoch": 0.880178716490658, "grad_norm": 0.0, "learning_rate": 7.43681971298571e-07, "loss": 2.0308, "step": 2167 }, { "epoch": 0.8805848903330625, "grad_norm": 0.0, "learning_rate": 7.387115139215895e-07, "loss": 1.906, "step": 2168 }, { "epoch": 0.8809910641754671, "grad_norm": 0.0, "learning_rate": 7.337570853977649e-07, "loss": 1.9261, "step": 2169 }, { "epoch": 0.8813972380178716, "grad_norm": 0.0, "learning_rate": 7.288186943019171e-07, "loss": 1.9921, "step": 2170 }, { "epoch": 0.8818034118602762, "grad_norm": 0.0, "learning_rate": 7.238963491810935e-07, "loss": 2.0762, "step": 2171 }, { "epoch": 0.8822095857026807, "grad_norm": 0.0, "learning_rate": 7.189900585545861e-07, "loss": 1.9944, "step": 2172 }, { "epoch": 0.8826157595450853, "grad_norm": 0.0, "learning_rate": 7.140998309138869e-07, "loss": 2.0129, "step": 2173 }, { "epoch": 0.8830219333874898, "grad_norm": 0.0, "learning_rate": 7.092256747226944e-07, "loss": 2.0708, "step": 2174 }, { "epoch": 0.8834281072298944, "grad_norm": 0.0, "learning_rate": 7.043675984168918e-07, "loss": 1.9468, "step": 2175 }, { "epoch": 0.8838342810722989, "grad_norm": 0.0, "learning_rate": 6.995256104045289e-07, "loss": 1.9879, "step": 2176 }, { "epoch": 0.8842404549147035, "grad_norm": 0.0, "learning_rate": 6.946997190658156e-07, "loss": 2.0197, "step": 2177 }, { "epoch": 0.8846466287571081, "grad_norm": 0.0, "learning_rate": 6.89889932753095e-07, "loss": 2.0034, "step": 2178 }, { "epoch": 0.8850528025995126, "grad_norm": 0.0, "learning_rate": 6.850962597908462e-07, "loss": 2.0649, "step": 2179 }, { "epoch": 0.8854589764419172, "grad_norm": 0.0, "learning_rate": 6.803187084756524e-07, "loss": 1.966, "step": 2180 }, { "epoch": 0.8858651502843217, "grad_norm": 0.0, "learning_rate": 6.755572870761983e-07, "loss": 2.1015, "step": 2181 }, { "epoch": 0.8862713241267263, "grad_norm": 0.0, "learning_rate": 6.708120038332533e-07, "loss": 1.9672, "step": 2182 }, { "epoch": 0.8866774979691308, "grad_norm": 0.0, "learning_rate": 6.66082866959653e-07, "loss": 2.1031, "step": 2183 }, { "epoch": 0.8870836718115354, "grad_norm": 0.0, "learning_rate": 6.61369884640286e-07, "loss": 1.9654, "step": 2184 }, { "epoch": 0.8874898456539398, "grad_norm": 0.0, "learning_rate": 6.566730650320896e-07, "loss": 1.9763, "step": 2185 }, { "epoch": 0.8878960194963444, "grad_norm": 0.0, "learning_rate": 6.519924162640168e-07, "loss": 2.0478, "step": 2186 }, { "epoch": 0.8883021933387489, "grad_norm": 0.0, "learning_rate": 6.473279464370441e-07, "loss": 2.0701, "step": 2187 }, { "epoch": 0.8887083671811535, "grad_norm": 0.0, "learning_rate": 6.426796636241372e-07, "loss": 2.1371, "step": 2188 }, { "epoch": 0.8891145410235581, "grad_norm": 0.0, "learning_rate": 6.380475758702532e-07, "loss": 2.0179, "step": 2189 }, { "epoch": 0.8895207148659626, "grad_norm": 0.0, "learning_rate": 6.334316911923155e-07, "loss": 1.9756, "step": 2190 }, { "epoch": 0.8899268887083672, "grad_norm": 0.0, "learning_rate": 6.288320175792062e-07, "loss": 1.9853, "step": 2191 }, { "epoch": 0.8903330625507717, "grad_norm": 0.0, "learning_rate": 6.242485629917494e-07, "loss": 1.973, "step": 2192 }, { "epoch": 0.8907392363931763, "grad_norm": 0.0, "learning_rate": 6.196813353627007e-07, "loss": 1.9181, "step": 2193 }, { "epoch": 0.8911454102355808, "grad_norm": 0.0, "learning_rate": 6.151303425967259e-07, "loss": 1.9252, "step": 2194 }, { "epoch": 0.8915515840779854, "grad_norm": 0.0, "learning_rate": 6.105955925704e-07, "loss": 2.0554, "step": 2195 }, { "epoch": 0.8919577579203899, "grad_norm": 0.0, "learning_rate": 6.060770931321758e-07, "loss": 2.0207, "step": 2196 }, { "epoch": 0.8923639317627945, "grad_norm": 0.0, "learning_rate": 6.015748521023946e-07, "loss": 1.9275, "step": 2197 }, { "epoch": 0.892770105605199, "grad_norm": 0.0, "learning_rate": 5.970888772732453e-07, "loss": 1.9397, "step": 2198 }, { "epoch": 0.8931762794476036, "grad_norm": 0.0, "learning_rate": 5.926191764087707e-07, "loss": 2.018, "step": 2199 }, { "epoch": 0.8935824532900081, "grad_norm": 0.0, "learning_rate": 5.881657572448474e-07, "loss": 2.0782, "step": 2200 }, { "epoch": 0.8939886271324127, "grad_norm": 0.0, "learning_rate": 5.837286274891718e-07, "loss": 1.8667, "step": 2201 }, { "epoch": 0.8943948009748173, "grad_norm": 0.0, "learning_rate": 5.793077948212478e-07, "loss": 2.034, "step": 2202 }, { "epoch": 0.8948009748172218, "grad_norm": 0.0, "learning_rate": 5.74903266892376e-07, "loss": 1.9759, "step": 2203 }, { "epoch": 0.8952071486596264, "grad_norm": 0.0, "learning_rate": 5.705150513256297e-07, "loss": 1.9783, "step": 2204 }, { "epoch": 0.8956133225020309, "grad_norm": 0.0, "learning_rate": 5.661431557158592e-07, "loss": 2.0661, "step": 2205 }, { "epoch": 0.8960194963444355, "grad_norm": 0.0, "learning_rate": 5.617875876296641e-07, "loss": 2.0099, "step": 2206 }, { "epoch": 0.8964256701868399, "grad_norm": 0.0, "learning_rate": 5.574483546053866e-07, "loss": 1.9842, "step": 2207 }, { "epoch": 0.8968318440292445, "grad_norm": 0.0, "learning_rate": 5.531254641530981e-07, "loss": 1.9419, "step": 2208 }, { "epoch": 0.897238017871649, "grad_norm": 0.0, "learning_rate": 5.488189237545827e-07, "loss": 2.0462, "step": 2209 }, { "epoch": 0.8976441917140536, "grad_norm": 0.0, "learning_rate": 5.445287408633304e-07, "loss": 2.0243, "step": 2210 }, { "epoch": 0.8980503655564581, "grad_norm": 0.0, "learning_rate": 5.402549229045151e-07, "loss": 1.9367, "step": 2211 }, { "epoch": 0.8984565393988627, "grad_norm": 0.0, "learning_rate": 5.359974772749965e-07, "loss": 1.9233, "step": 2212 }, { "epoch": 0.8988627132412672, "grad_norm": 0.0, "learning_rate": 5.317564113432883e-07, "loss": 2.22, "step": 2213 }, { "epoch": 0.8992688870836718, "grad_norm": 0.0, "learning_rate": 5.27531732449561e-07, "loss": 1.9437, "step": 2214 }, { "epoch": 0.8996750609260764, "grad_norm": 0.0, "learning_rate": 5.233234479056226e-07, "loss": 1.9679, "step": 2215 }, { "epoch": 0.9000812347684809, "grad_norm": 0.0, "learning_rate": 5.191315649949046e-07, "loss": 1.9925, "step": 2216 }, { "epoch": 0.9004874086108855, "grad_norm": 0.0, "learning_rate": 5.149560909724549e-07, "loss": 1.9493, "step": 2217 }, { "epoch": 0.90089358245329, "grad_norm": 0.0, "learning_rate": 5.107970330649204e-07, "loss": 2.0336, "step": 2218 }, { "epoch": 0.9012997562956946, "grad_norm": 0.0, "learning_rate": 5.066543984705319e-07, "loss": 1.9505, "step": 2219 }, { "epoch": 0.9017059301380991, "grad_norm": 0.0, "learning_rate": 5.025281943591054e-07, "loss": 1.9355, "step": 2220 }, { "epoch": 0.9021121039805037, "grad_norm": 0.0, "learning_rate": 4.984184278720083e-07, "loss": 1.9167, "step": 2221 }, { "epoch": 0.9025182778229082, "grad_norm": 0.0, "learning_rate": 4.943251061221721e-07, "loss": 1.9744, "step": 2222 }, { "epoch": 0.9029244516653128, "grad_norm": 0.0, "learning_rate": 4.902482361940541e-07, "loss": 1.8155, "step": 2223 }, { "epoch": 0.9033306255077173, "grad_norm": 0.0, "learning_rate": 4.861878251436436e-07, "loss": 2.0078, "step": 2224 }, { "epoch": 0.9037367993501219, "grad_norm": 0.0, "learning_rate": 4.821438799984457e-07, "loss": 2.1208, "step": 2225 }, { "epoch": 0.9041429731925265, "grad_norm": 0.0, "learning_rate": 4.78116407757464e-07, "loss": 1.9739, "step": 2226 }, { "epoch": 0.904549147034931, "grad_norm": 0.0, "learning_rate": 4.741054153911939e-07, "loss": 1.9327, "step": 2227 }, { "epoch": 0.9049553208773355, "grad_norm": 0.0, "learning_rate": 4.701109098416079e-07, "loss": 1.9687, "step": 2228 }, { "epoch": 0.90536149471974, "grad_norm": 0.0, "learning_rate": 4.661328980221413e-07, "loss": 1.9434, "step": 2229 }, { "epoch": 0.9057676685621446, "grad_norm": 0.0, "learning_rate": 4.6217138681769026e-07, "loss": 1.8943, "step": 2230 }, { "epoch": 0.9061738424045491, "grad_norm": 0.0, "learning_rate": 4.5822638308458344e-07, "loss": 2.0225, "step": 2231 }, { "epoch": 0.9065800162469537, "grad_norm": 0.0, "learning_rate": 4.5429789365058596e-07, "loss": 1.9893, "step": 2232 }, { "epoch": 0.9069861900893582, "grad_norm": 0.0, "learning_rate": 4.5038592531488014e-07, "loss": 1.9511, "step": 2233 }, { "epoch": 0.9073923639317628, "grad_norm": 0.0, "learning_rate": 4.464904848480522e-07, "loss": 2.0774, "step": 2234 }, { "epoch": 0.9077985377741673, "grad_norm": 0.0, "learning_rate": 4.426115789920848e-07, "loss": 1.952, "step": 2235 }, { "epoch": 0.9082047116165719, "grad_norm": 0.0, "learning_rate": 4.3874921446034445e-07, "loss": 1.9467, "step": 2236 }, { "epoch": 0.9086108854589764, "grad_norm": 0.0, "learning_rate": 4.349033979375683e-07, "loss": 1.9334, "step": 2237 }, { "epoch": 0.909017059301381, "grad_norm": 0.0, "learning_rate": 4.310741360798498e-07, "loss": 1.9832, "step": 2238 }, { "epoch": 0.9094232331437856, "grad_norm": 0.0, "learning_rate": 4.272614355146354e-07, "loss": 2.0676, "step": 2239 }, { "epoch": 0.9098294069861901, "grad_norm": 0.0, "learning_rate": 4.234653028407054e-07, "loss": 1.9796, "step": 2240 }, { "epoch": 0.9102355808285947, "grad_norm": 0.0, "learning_rate": 4.1968574462816656e-07, "loss": 1.9263, "step": 2241 }, { "epoch": 0.9106417546709992, "grad_norm": 0.0, "learning_rate": 4.1592276741844075e-07, "loss": 1.8886, "step": 2242 }, { "epoch": 0.9110479285134038, "grad_norm": 0.0, "learning_rate": 4.1217637772425157e-07, "loss": 1.9253, "step": 2243 }, { "epoch": 0.9114541023558083, "grad_norm": 0.0, "learning_rate": 4.084465820296102e-07, "loss": 1.917, "step": 2244 }, { "epoch": 0.9118602761982129, "grad_norm": 0.0, "learning_rate": 4.0473338678981735e-07, "loss": 1.9758, "step": 2245 }, { "epoch": 0.9122664500406173, "grad_norm": 0.0, "learning_rate": 4.0103679843142895e-07, "loss": 1.8313, "step": 2246 }, { "epoch": 0.912672623883022, "grad_norm": 0.0, "learning_rate": 3.973568233522751e-07, "loss": 2.0597, "step": 2247 }, { "epoch": 0.9130787977254264, "grad_norm": 0.0, "learning_rate": 3.936934679214188e-07, "loss": 2.0158, "step": 2248 }, { "epoch": 0.913484971567831, "grad_norm": 0.0, "learning_rate": 3.9004673847916506e-07, "loss": 1.9233, "step": 2249 }, { "epoch": 0.9138911454102355, "grad_norm": 0.0, "learning_rate": 3.864166413370429e-07, "loss": 2.018, "step": 2250 }, { "epoch": 0.9142973192526401, "grad_norm": 0.0, "learning_rate": 3.828031827777967e-07, "loss": 2.092, "step": 2251 }, { "epoch": 0.9147034930950447, "grad_norm": 0.0, "learning_rate": 3.792063690553716e-07, "loss": 2.1489, "step": 2252 }, { "epoch": 0.9151096669374492, "grad_norm": 0.0, "learning_rate": 3.7562620639490564e-07, "loss": 1.915, "step": 2253 }, { "epoch": 0.9155158407798538, "grad_norm": 0.0, "learning_rate": 3.720627009927158e-07, "loss": 1.9782, "step": 2254 }, { "epoch": 0.9159220146222583, "grad_norm": 0.0, "learning_rate": 3.6851585901629737e-07, "loss": 2.0381, "step": 2255 }, { "epoch": 0.9163281884646629, "grad_norm": 0.0, "learning_rate": 3.649856866042933e-07, "loss": 2.0425, "step": 2256 }, { "epoch": 0.9167343623070674, "grad_norm": 0.0, "learning_rate": 3.614721898665108e-07, "loss": 1.9785, "step": 2257 }, { "epoch": 0.917140536149472, "grad_norm": 0.0, "learning_rate": 3.5797537488388326e-07, "loss": 1.9196, "step": 2258 }, { "epoch": 0.9175467099918765, "grad_norm": 0.0, "learning_rate": 3.5449524770847733e-07, "loss": 2.0303, "step": 2259 }, { "epoch": 0.9179528838342811, "grad_norm": 0.0, "learning_rate": 3.510318143634794e-07, "loss": 1.9487, "step": 2260 }, { "epoch": 0.9183590576766856, "grad_norm": 0.0, "learning_rate": 3.475850808431791e-07, "loss": 1.9716, "step": 2261 }, { "epoch": 0.9187652315190902, "grad_norm": 0.0, "learning_rate": 3.441550531129667e-07, "loss": 2.0555, "step": 2262 }, { "epoch": 0.9191714053614947, "grad_norm": 0.0, "learning_rate": 3.4074173710931804e-07, "loss": 2.0764, "step": 2263 }, { "epoch": 0.9195775792038993, "grad_norm": 0.0, "learning_rate": 3.3734513873978193e-07, "loss": 1.8729, "step": 2264 }, { "epoch": 0.9199837530463039, "grad_norm": 0.0, "learning_rate": 3.339652638829782e-07, "loss": 2.0684, "step": 2265 }, { "epoch": 0.9203899268887084, "grad_norm": 0.0, "learning_rate": 3.3060211838858104e-07, "loss": 1.9708, "step": 2266 }, { "epoch": 0.920796100731113, "grad_norm": 0.0, "learning_rate": 3.272557080773098e-07, "loss": 1.8649, "step": 2267 }, { "epoch": 0.9212022745735174, "grad_norm": 0.0, "learning_rate": 3.239260387409215e-07, "loss": 1.9282, "step": 2268 }, { "epoch": 0.921608448415922, "grad_norm": 0.0, "learning_rate": 3.2061311614219437e-07, "loss": 1.8468, "step": 2269 }, { "epoch": 0.9220146222583265, "grad_norm": 0.0, "learning_rate": 3.1731694601492834e-07, "loss": 1.9627, "step": 2270 }, { "epoch": 0.9224207961007311, "grad_norm": 0.0, "learning_rate": 3.1403753406392454e-07, "loss": 1.9611, "step": 2271 }, { "epoch": 0.9228269699431356, "grad_norm": 0.0, "learning_rate": 3.1077488596498505e-07, "loss": 2.0448, "step": 2272 }, { "epoch": 0.9232331437855402, "grad_norm": 0.0, "learning_rate": 3.075290073648918e-07, "loss": 2.085, "step": 2273 }, { "epoch": 0.9236393176279447, "grad_norm": 0.0, "learning_rate": 3.042999038814076e-07, "loss": 1.9537, "step": 2274 }, { "epoch": 0.9240454914703493, "grad_norm": 0.0, "learning_rate": 3.0108758110326096e-07, "loss": 1.9528, "step": 2275 }, { "epoch": 0.9244516653127539, "grad_norm": 0.0, "learning_rate": 2.978920445901379e-07, "loss": 1.9437, "step": 2276 }, { "epoch": 0.9248578391551584, "grad_norm": 0.0, "learning_rate": 2.9471329987266985e-07, "loss": 1.9655, "step": 2277 }, { "epoch": 0.925264012997563, "grad_norm": 0.0, "learning_rate": 2.915513524524294e-07, "loss": 2.1049, "step": 2278 }, { "epoch": 0.9256701868399675, "grad_norm": 0.0, "learning_rate": 2.8840620780191233e-07, "loss": 1.9516, "step": 2279 }, { "epoch": 0.9260763606823721, "grad_norm": 0.0, "learning_rate": 2.852778713645399e-07, "loss": 1.8598, "step": 2280 }, { "epoch": 0.9264825345247766, "grad_norm": 0.0, "learning_rate": 2.8216634855463556e-07, "loss": 2.0367, "step": 2281 }, { "epoch": 0.9268887083671812, "grad_norm": 0.0, "learning_rate": 2.790716447574304e-07, "loss": 1.9733, "step": 2282 }, { "epoch": 0.9272948822095857, "grad_norm": 0.0, "learning_rate": 2.759937653290401e-07, "loss": 2.024, "step": 2283 }, { "epoch": 0.9277010560519903, "grad_norm": 0.0, "learning_rate": 2.729327155964634e-07, "loss": 1.933, "step": 2284 }, { "epoch": 0.9281072298943948, "grad_norm": 0.0, "learning_rate": 2.698885008575725e-07, "loss": 1.9114, "step": 2285 }, { "epoch": 0.9285134037367994, "grad_norm": 0.0, "learning_rate": 2.668611263811016e-07, "loss": 1.9945, "step": 2286 }, { "epoch": 0.9289195775792038, "grad_norm": 0.0, "learning_rate": 2.6385059740664074e-07, "loss": 1.9631, "step": 2287 }, { "epoch": 0.9293257514216084, "grad_norm": 0.0, "learning_rate": 2.6085691914462306e-07, "loss": 2.1133, "step": 2288 }, { "epoch": 0.929731925264013, "grad_norm": 0.0, "learning_rate": 2.578800967763162e-07, "loss": 1.9909, "step": 2289 }, { "epoch": 0.9301380991064175, "grad_norm": 0.0, "learning_rate": 2.5492013545381666e-07, "loss": 1.8895, "step": 2290 }, { "epoch": 0.9305442729488221, "grad_norm": 0.0, "learning_rate": 2.519770403000399e-07, "loss": 1.8376, "step": 2291 }, { "epoch": 0.9309504467912266, "grad_norm": 0.0, "learning_rate": 2.4905081640870686e-07, "loss": 1.9927, "step": 2292 }, { "epoch": 0.9313566206336312, "grad_norm": 0.0, "learning_rate": 2.461414688443442e-07, "loss": 2.0609, "step": 2293 }, { "epoch": 0.9317627944760357, "grad_norm": 0.0, "learning_rate": 2.4324900264226405e-07, "loss": 2.0594, "step": 2294 }, { "epoch": 0.9321689683184403, "grad_norm": 0.0, "learning_rate": 2.403734228085663e-07, "loss": 1.969, "step": 2295 }, { "epoch": 0.9325751421608448, "grad_norm": 0.0, "learning_rate": 2.3751473432012007e-07, "loss": 1.9778, "step": 2296 }, { "epoch": 0.9329813160032494, "grad_norm": 0.0, "learning_rate": 2.3467294212456747e-07, "loss": 1.956, "step": 2297 }, { "epoch": 0.9333874898456539, "grad_norm": 0.0, "learning_rate": 2.3184805114029872e-07, "loss": 2.1207, "step": 2298 }, { "epoch": 0.9337936636880585, "grad_norm": 0.0, "learning_rate": 2.2904006625645735e-07, "loss": 1.9848, "step": 2299 }, { "epoch": 0.934199837530463, "grad_norm": 0.0, "learning_rate": 2.2624899233292807e-07, "loss": 2.0149, "step": 2300 }, { "epoch": 0.9346060113728676, "grad_norm": 0.0, "learning_rate": 2.2347483420032347e-07, "loss": 1.9895, "step": 2301 }, { "epoch": 0.9350121852152722, "grad_norm": 0.0, "learning_rate": 2.2071759665998282e-07, "loss": 2.0018, "step": 2302 }, { "epoch": 0.9354183590576767, "grad_norm": 0.0, "learning_rate": 2.1797728448395894e-07, "loss": 1.9066, "step": 2303 }, { "epoch": 0.9358245329000813, "grad_norm": 0.0, "learning_rate": 2.15253902415008e-07, "loss": 1.8844, "step": 2304 }, { "epoch": 0.9362307067424858, "grad_norm": 0.0, "learning_rate": 2.125474551665918e-07, "loss": 2.0586, "step": 2305 }, { "epoch": 0.9366368805848904, "grad_norm": 0.0, "learning_rate": 2.098579474228546e-07, "loss": 1.9121, "step": 2306 }, { "epoch": 0.9370430544272949, "grad_norm": 0.0, "learning_rate": 2.071853838386295e-07, "loss": 2.0305, "step": 2307 }, { "epoch": 0.9374492282696995, "grad_norm": 0.0, "learning_rate": 2.045297690394188e-07, "loss": 1.9564, "step": 2308 }, { "epoch": 0.9378554021121039, "grad_norm": 0.0, "learning_rate": 2.0189110762139362e-07, "loss": 2.0806, "step": 2309 }, { "epoch": 0.9382615759545085, "grad_norm": 0.0, "learning_rate": 1.9926940415138206e-07, "loss": 1.8526, "step": 2310 }, { "epoch": 0.938667749796913, "grad_norm": 0.0, "learning_rate": 1.966646631668634e-07, "loss": 2.0774, "step": 2311 }, { "epoch": 0.9390739236393176, "grad_norm": 0.0, "learning_rate": 1.9407688917595923e-07, "loss": 2.0189, "step": 2312 }, { "epoch": 0.9394800974817222, "grad_norm": 0.0, "learning_rate": 1.9150608665742477e-07, "loss": 1.9929, "step": 2313 }, { "epoch": 0.9398862713241267, "grad_norm": 0.0, "learning_rate": 1.8895226006064084e-07, "loss": 1.9296, "step": 2314 }, { "epoch": 0.9402924451665313, "grad_norm": 0.0, "learning_rate": 1.864154138056129e-07, "loss": 1.8951, "step": 2315 }, { "epoch": 0.9406986190089358, "grad_norm": 0.0, "learning_rate": 1.838955522829522e-07, "loss": 1.9526, "step": 2316 }, { "epoch": 0.9411047928513404, "grad_norm": 0.0, "learning_rate": 1.8139267985387567e-07, "loss": 2.0136, "step": 2317 }, { "epoch": 0.9415109666937449, "grad_norm": 0.0, "learning_rate": 1.7890680085019597e-07, "loss": 1.81, "step": 2318 }, { "epoch": 0.9419171405361495, "grad_norm": 0.0, "learning_rate": 1.7643791957431823e-07, "loss": 1.938, "step": 2319 }, { "epoch": 0.942323314378554, "grad_norm": 0.0, "learning_rate": 1.7398604029922328e-07, "loss": 1.9923, "step": 2320 }, { "epoch": 0.9427294882209586, "grad_norm": 0.0, "learning_rate": 1.7155116726847109e-07, "loss": 2.0574, "step": 2321 }, { "epoch": 0.9431356620633631, "grad_norm": 0.0, "learning_rate": 1.6913330469618628e-07, "loss": 1.9931, "step": 2322 }, { "epoch": 0.9435418359057677, "grad_norm": 0.0, "learning_rate": 1.667324567670503e-07, "loss": 1.9952, "step": 2323 }, { "epoch": 0.9439480097481722, "grad_norm": 0.0, "learning_rate": 1.6434862763630156e-07, "loss": 2.1351, "step": 2324 }, { "epoch": 0.9443541835905768, "grad_norm": 0.0, "learning_rate": 1.6198182142971862e-07, "loss": 2.09, "step": 2325 }, { "epoch": 0.9447603574329814, "grad_norm": 0.0, "learning_rate": 1.5963204224362261e-07, "loss": 1.9103, "step": 2326 }, { "epoch": 0.9451665312753859, "grad_norm": 0.0, "learning_rate": 1.5729929414486145e-07, "loss": 2.0384, "step": 2327 }, { "epoch": 0.9455727051177905, "grad_norm": 0.0, "learning_rate": 1.5498358117080893e-07, "loss": 2.0107, "step": 2328 }, { "epoch": 0.945978878960195, "grad_norm": 0.0, "learning_rate": 1.5268490732935238e-07, "loss": 2.0008, "step": 2329 }, { "epoch": 0.9463850528025995, "grad_norm": 0.0, "learning_rate": 1.504032765988961e-07, "loss": 1.9905, "step": 2330 }, { "epoch": 0.946791226645004, "grad_norm": 0.0, "learning_rate": 1.4813869292833683e-07, "loss": 1.9504, "step": 2331 }, { "epoch": 0.9471974004874086, "grad_norm": 0.0, "learning_rate": 1.4589116023707827e-07, "loss": 1.9302, "step": 2332 }, { "epoch": 0.9476035743298131, "grad_norm": 0.0, "learning_rate": 1.4366068241500442e-07, "loss": 1.9135, "step": 2333 }, { "epoch": 0.9480097481722177, "grad_norm": 0.0, "learning_rate": 1.4144726332248726e-07, "loss": 2.0581, "step": 2334 }, { "epoch": 0.9484159220146222, "grad_norm": 0.0, "learning_rate": 1.3925090679037135e-07, "loss": 2.0342, "step": 2335 }, { "epoch": 0.9488220958570268, "grad_norm": 0.0, "learning_rate": 1.370716166199726e-07, "loss": 1.9769, "step": 2336 }, { "epoch": 0.9492282696994313, "grad_norm": 0.0, "learning_rate": 1.3490939658306946e-07, "loss": 1.893, "step": 2337 }, { "epoch": 0.9496344435418359, "grad_norm": 0.0, "learning_rate": 1.327642504218951e-07, "loss": 1.9815, "step": 2338 }, { "epoch": 0.9500406173842405, "grad_norm": 0.0, "learning_rate": 1.3063618184913195e-07, "loss": 2.0068, "step": 2339 }, { "epoch": 0.950446791226645, "grad_norm": 0.0, "learning_rate": 1.2852519454790936e-07, "loss": 2.0844, "step": 2340 }, { "epoch": 0.9508529650690496, "grad_norm": 0.0, "learning_rate": 1.2643129217178697e-07, "loss": 1.8867, "step": 2341 }, { "epoch": 0.9512591389114541, "grad_norm": 0.0, "learning_rate": 1.2435447834476254e-07, "loss": 1.9537, "step": 2342 }, { "epoch": 0.9516653127538587, "grad_norm": 0.0, "learning_rate": 1.2229475666125312e-07, "loss": 2.0395, "step": 2343 }, { "epoch": 0.9520714865962632, "grad_norm": 0.0, "learning_rate": 1.2025213068609375e-07, "loss": 1.9234, "step": 2344 }, { "epoch": 0.9524776604386678, "grad_norm": 0.0, "learning_rate": 1.1822660395453323e-07, "loss": 1.9289, "step": 2345 }, { "epoch": 0.9528838342810723, "grad_norm": 0.0, "learning_rate": 1.1621817997222507e-07, "loss": 2.0861, "step": 2346 }, { "epoch": 0.9532900081234769, "grad_norm": 0.0, "learning_rate": 1.1422686221522317e-07, "loss": 2.0012, "step": 2347 }, { "epoch": 0.9536961819658814, "grad_norm": 0.0, "learning_rate": 1.122526541299751e-07, "loss": 1.9673, "step": 2348 }, { "epoch": 0.954102355808286, "grad_norm": 0.0, "learning_rate": 1.1029555913331324e-07, "loss": 2.0194, "step": 2349 }, { "epoch": 0.9545085296506904, "grad_norm": 0.0, "learning_rate": 1.0835558061245587e-07, "loss": 2.0225, "step": 2350 }, { "epoch": 0.954914703493095, "grad_norm": 0.0, "learning_rate": 1.0643272192499499e-07, "loss": 2.0114, "step": 2351 }, { "epoch": 0.9553208773354996, "grad_norm": 0.0, "learning_rate": 1.045269863988918e-07, "loss": 1.8157, "step": 2352 }, { "epoch": 0.9557270511779041, "grad_norm": 0.0, "learning_rate": 1.0263837733247573e-07, "loss": 1.9377, "step": 2353 }, { "epoch": 0.9561332250203087, "grad_norm": 0.0, "learning_rate": 1.0076689799442874e-07, "loss": 1.9137, "step": 2354 }, { "epoch": 0.9565393988627132, "grad_norm": 0.0, "learning_rate": 9.891255162379099e-08, "loss": 2.0387, "step": 2355 }, { "epoch": 0.9569455727051178, "grad_norm": 0.0, "learning_rate": 9.707534142994746e-08, "loss": 1.9868, "step": 2356 }, { "epoch": 0.9573517465475223, "grad_norm": 0.0, "learning_rate": 9.525527059262684e-08, "loss": 1.9906, "step": 2357 }, { "epoch": 0.9577579203899269, "grad_norm": 0.0, "learning_rate": 9.34523422618916e-08, "loss": 1.8936, "step": 2358 }, { "epoch": 0.9581640942323314, "grad_norm": 0.0, "learning_rate": 9.166655955813564e-08, "loss": 1.9964, "step": 2359 }, { "epoch": 0.958570268074736, "grad_norm": 0.0, "learning_rate": 8.989792557207889e-08, "loss": 1.9598, "step": 2360 }, { "epoch": 0.9589764419171405, "grad_norm": 0.0, "learning_rate": 8.814644336476274e-08, "loss": 1.9436, "step": 2361 }, { "epoch": 0.9593826157595451, "grad_norm": 0.0, "learning_rate": 8.641211596754129e-08, "loss": 2.0392, "step": 2362 }, { "epoch": 0.9597887896019497, "grad_norm": 0.0, "learning_rate": 8.469494638207898e-08, "loss": 1.9291, "step": 2363 }, { "epoch": 0.9601949634443542, "grad_norm": 0.0, "learning_rate": 8.299493758034294e-08, "loss": 2.0019, "step": 2364 }, { "epoch": 0.9606011372867588, "grad_norm": 0.0, "learning_rate": 8.131209250460403e-08, "loss": 1.9764, "step": 2365 }, { "epoch": 0.9610073111291633, "grad_norm": 0.0, "learning_rate": 7.964641406742135e-08, "loss": 2.0736, "step": 2366 }, { "epoch": 0.9614134849715679, "grad_norm": 0.0, "learning_rate": 7.799790515164884e-08, "loss": 1.776, "step": 2367 }, { "epoch": 0.9618196588139724, "grad_norm": 0.0, "learning_rate": 7.63665686104187e-08, "loss": 1.9797, "step": 2368 }, { "epoch": 0.962225832656377, "grad_norm": 0.0, "learning_rate": 7.475240726714572e-08, "loss": 1.9517, "step": 2369 }, { "epoch": 0.9626320064987814, "grad_norm": 0.0, "learning_rate": 7.315542391551966e-08, "loss": 1.8873, "step": 2370 }, { "epoch": 0.963038180341186, "grad_norm": 0.0, "learning_rate": 7.157562131949847e-08, "loss": 1.9845, "step": 2371 }, { "epoch": 0.9634443541835905, "grad_norm": 0.0, "learning_rate": 7.001300221330387e-08, "loss": 2.0557, "step": 2372 }, { "epoch": 0.9638505280259951, "grad_norm": 0.0, "learning_rate": 6.846756930141806e-08, "loss": 1.9752, "step": 2373 }, { "epoch": 0.9642567018683996, "grad_norm": 0.0, "learning_rate": 6.693932525857927e-08, "loss": 1.876, "step": 2374 }, { "epoch": 0.9646628757108042, "grad_norm": 0.0, "learning_rate": 6.542827272977614e-08, "loss": 2.0394, "step": 2375 }, { "epoch": 0.9650690495532088, "grad_norm": 0.0, "learning_rate": 6.393441433024228e-08, "loss": 2.0881, "step": 2376 }, { "epoch": 0.9654752233956133, "grad_norm": 0.0, "learning_rate": 6.245775264545285e-08, "loss": 2.0782, "step": 2377 }, { "epoch": 0.9658813972380179, "grad_norm": 0.0, "learning_rate": 6.099829023112236e-08, "loss": 1.9187, "step": 2378 }, { "epoch": 0.9662875710804224, "grad_norm": 0.0, "learning_rate": 5.955602961319473e-08, "loss": 1.8699, "step": 2379 }, { "epoch": 0.966693744922827, "grad_norm": 0.0, "learning_rate": 5.813097328784434e-08, "loss": 1.983, "step": 2380 }, { "epoch": 0.9670999187652315, "grad_norm": 0.0, "learning_rate": 5.672312372146827e-08, "loss": 1.8955, "step": 2381 }, { "epoch": 0.9675060926076361, "grad_norm": 0.0, "learning_rate": 5.533248335068409e-08, "loss": 2.0964, "step": 2382 }, { "epoch": 0.9679122664500406, "grad_norm": 0.0, "learning_rate": 5.39590545823232e-08, "loss": 1.9526, "step": 2383 }, { "epoch": 0.9683184402924452, "grad_norm": 0.0, "learning_rate": 5.260283979343084e-08, "loss": 2.0021, "step": 2384 }, { "epoch": 0.9687246141348497, "grad_norm": 0.0, "learning_rate": 5.126384133125606e-08, "loss": 1.9355, "step": 2385 }, { "epoch": 0.9691307879772543, "grad_norm": 0.0, "learning_rate": 4.994206151325509e-08, "loss": 2.0465, "step": 2386 }, { "epoch": 0.9695369618196588, "grad_norm": 0.0, "learning_rate": 4.8637502627080224e-08, "loss": 2.0229, "step": 2387 }, { "epoch": 0.9699431356620634, "grad_norm": 0.0, "learning_rate": 4.73501669305787e-08, "loss": 1.944, "step": 2388 }, { "epoch": 0.970349309504468, "grad_norm": 0.0, "learning_rate": 4.608005665178938e-08, "loss": 1.9556, "step": 2389 }, { "epoch": 0.9707554833468724, "grad_norm": 0.0, "learning_rate": 4.482717398894165e-08, "loss": 2.0306, "step": 2390 }, { "epoch": 0.971161657189277, "grad_norm": 0.0, "learning_rate": 4.359152111044207e-08, "loss": 2.0438, "step": 2391 }, { "epoch": 0.9715678310316815, "grad_norm": 0.0, "learning_rate": 4.2373100154883274e-08, "loss": 1.9476, "step": 2392 }, { "epoch": 0.9719740048740861, "grad_norm": 0.0, "learning_rate": 4.117191323103065e-08, "loss": 2.0444, "step": 2393 }, { "epoch": 0.9723801787164906, "grad_norm": 0.0, "learning_rate": 3.998796241782232e-08, "loss": 2.0052, "step": 2394 }, { "epoch": 0.9727863525588952, "grad_norm": 0.0, "learning_rate": 3.882124976436585e-08, "loss": 1.9659, "step": 2395 }, { "epoch": 0.9731925264012997, "grad_norm": 0.0, "learning_rate": 3.7671777289932654e-08, "loss": 2.1011, "step": 2396 }, { "epoch": 0.9735987002437043, "grad_norm": 0.0, "learning_rate": 3.6539546983959115e-08, "loss": 2.0373, "step": 2397 }, { "epoch": 0.9740048740861088, "grad_norm": 0.0, "learning_rate": 3.5424560806036625e-08, "loss": 1.9998, "step": 2398 }, { "epoch": 0.9744110479285134, "grad_norm": 0.0, "learning_rate": 3.432682068591375e-08, "loss": 1.8799, "step": 2399 }, { "epoch": 0.974817221770918, "grad_norm": 0.0, "learning_rate": 3.324632852348853e-08, "loss": 1.9425, "step": 2400 }, { "epoch": 0.9752233956133225, "grad_norm": 0.0, "learning_rate": 3.21830861888095e-08, "loss": 2.0589, "step": 2401 }, { "epoch": 0.9756295694557271, "grad_norm": 0.0, "learning_rate": 3.1137095522068006e-08, "loss": 1.9937, "step": 2402 }, { "epoch": 0.9760357432981316, "grad_norm": 0.0, "learning_rate": 3.0108358333600375e-08, "loss": 1.9836, "step": 2403 }, { "epoch": 0.9764419171405362, "grad_norm": 0.0, "learning_rate": 2.9096876403879036e-08, "loss": 1.9529, "step": 2404 }, { "epoch": 0.9768480909829407, "grad_norm": 0.0, "learning_rate": 2.8102651483513655e-08, "loss": 2.0527, "step": 2405 }, { "epoch": 0.9772542648253453, "grad_norm": 0.0, "learning_rate": 2.7125685293245552e-08, "loss": 1.9627, "step": 2406 }, { "epoch": 0.9776604386677498, "grad_norm": 0.0, "learning_rate": 2.6165979523945505e-08, "loss": 1.9697, "step": 2407 }, { "epoch": 0.9780666125101544, "grad_norm": 0.0, "learning_rate": 2.5223535836612634e-08, "loss": 1.9469, "step": 2408 }, { "epoch": 0.9784727863525589, "grad_norm": 0.0, "learning_rate": 2.429835586236773e-08, "loss": 1.8708, "step": 2409 }, { "epoch": 0.9788789601949635, "grad_norm": 0.0, "learning_rate": 2.3390441202455484e-08, "loss": 1.8585, "step": 2410 }, { "epoch": 0.9792851340373679, "grad_norm": 0.0, "learning_rate": 2.2499793428235605e-08, "loss": 1.9765, "step": 2411 }, { "epoch": 0.9796913078797725, "grad_norm": 0.0, "learning_rate": 2.1626414081186153e-08, "loss": 1.9214, "step": 2412 }, { "epoch": 0.9800974817221771, "grad_norm": 0.0, "learning_rate": 2.0770304672896868e-08, "loss": 1.9792, "step": 2413 }, { "epoch": 0.9805036555645816, "grad_norm": 0.0, "learning_rate": 1.993146668506585e-08, "loss": 2.0423, "step": 2414 }, { "epoch": 0.9809098294069862, "grad_norm": 0.0, "learning_rate": 1.9109901569505095e-08, "loss": 1.9412, "step": 2415 }, { "epoch": 0.9813160032493907, "grad_norm": 0.0, "learning_rate": 1.830561074812276e-08, "loss": 1.9753, "step": 2416 }, { "epoch": 0.9817221770917953, "grad_norm": 0.0, "learning_rate": 1.7518595612938673e-08, "loss": 1.8814, "step": 2417 }, { "epoch": 0.9821283509341998, "grad_norm": 0.0, "learning_rate": 1.6748857526066588e-08, "loss": 2.0546, "step": 2418 }, { "epoch": 0.9825345247766044, "grad_norm": 0.0, "learning_rate": 1.5996397819720844e-08, "loss": 2.0215, "step": 2419 }, { "epoch": 0.9829406986190089, "grad_norm": 0.0, "learning_rate": 1.5261217796211923e-08, "loss": 2.053, "step": 2420 }, { "epoch": 0.9833468724614135, "grad_norm": 0.0, "learning_rate": 1.4543318727942013e-08, "loss": 2.0265, "step": 2421 }, { "epoch": 0.983753046303818, "grad_norm": 0.0, "learning_rate": 1.3842701857406104e-08, "loss": 1.8434, "step": 2422 }, { "epoch": 0.9841592201462226, "grad_norm": 0.0, "learning_rate": 1.3159368397188677e-08, "loss": 1.9364, "step": 2423 }, { "epoch": 0.9845653939886271, "grad_norm": 0.0, "learning_rate": 1.2493319529958137e-08, "loss": 1.9876, "step": 2424 }, { "epoch": 0.9849715678310317, "grad_norm": 0.0, "learning_rate": 1.1844556408470153e-08, "loss": 2.0148, "step": 2425 }, { "epoch": 0.9853777416734363, "grad_norm": 0.0, "learning_rate": 1.1213080155564327e-08, "loss": 1.9353, "step": 2426 }, { "epoch": 0.9857839155158408, "grad_norm": 0.0, "learning_rate": 1.0598891864158634e-08, "loss": 1.8777, "step": 2427 }, { "epoch": 0.9861900893582454, "grad_norm": 0.0, "learning_rate": 1.0001992597251653e-08, "loss": 1.9467, "step": 2428 }, { "epoch": 0.9865962632006499, "grad_norm": 0.0, "learning_rate": 9.422383387919232e-09, "loss": 1.8844, "step": 2429 }, { "epoch": 0.9870024370430545, "grad_norm": 0.0, "learning_rate": 8.860065239311155e-09, "loss": 1.9274, "step": 2430 }, { "epoch": 0.987408610885459, "grad_norm": 0.0, "learning_rate": 8.315039124654478e-09, "loss": 2.1342, "step": 2431 }, { "epoch": 0.9878147847278635, "grad_norm": 0.0, "learning_rate": 7.787305987243532e-09, "loss": 1.8757, "step": 2432 }, { "epoch": 0.988220958570268, "grad_norm": 0.0, "learning_rate": 7.276866740447697e-09, "loss": 1.9953, "step": 2433 }, { "epoch": 0.9886271324126726, "grad_norm": 0.0, "learning_rate": 6.783722267701409e-09, "loss": 2.071, "step": 2434 }, { "epoch": 0.9890333062550771, "grad_norm": 0.0, "learning_rate": 6.307873422508604e-09, "loss": 1.9389, "step": 2435 }, { "epoch": 0.9894394800974817, "grad_norm": 0.0, "learning_rate": 5.849321028438271e-09, "loss": 1.8921, "step": 2436 }, { "epoch": 0.9898456539398862, "grad_norm": 0.0, "learning_rate": 5.408065879124458e-09, "loss": 2.0526, "step": 2437 }, { "epoch": 0.9902518277822908, "grad_norm": 0.0, "learning_rate": 4.984108738261828e-09, "loss": 2.0183, "step": 2438 }, { "epoch": 0.9906580016246954, "grad_norm": 0.0, "learning_rate": 4.577450339611211e-09, "loss": 1.9212, "step": 2439 }, { "epoch": 0.9910641754670999, "grad_norm": 0.0, "learning_rate": 4.188091386989612e-09, "loss": 1.944, "step": 2440 }, { "epoch": 0.9914703493095045, "grad_norm": 0.0, "learning_rate": 3.816032554275762e-09, "loss": 2.0006, "step": 2441 }, { "epoch": 0.991876523151909, "grad_norm": 0.0, "learning_rate": 3.4612744854045645e-09, "loss": 2.098, "step": 2442 }, { "epoch": 0.9922826969943136, "grad_norm": 0.0, "learning_rate": 3.123817794370432e-09, "loss": 2.0428, "step": 2443 }, { "epoch": 0.9926888708367181, "grad_norm": 0.0, "learning_rate": 2.8036630652206187e-09, "loss": 2.0121, "step": 2444 }, { "epoch": 0.9930950446791227, "grad_norm": 0.0, "learning_rate": 2.5008108520585552e-09, "loss": 1.9739, "step": 2445 }, { "epoch": 0.9935012185215272, "grad_norm": 0.0, "learning_rate": 2.215261679042735e-09, "loss": 1.8375, "step": 2446 }, { "epoch": 0.9939073923639318, "grad_norm": 0.0, "learning_rate": 1.9470160403844974e-09, "loss": 1.9766, "step": 2447 }, { "epoch": 0.9943135662063363, "grad_norm": 0.0, "learning_rate": 1.6960744003435836e-09, "loss": 1.8805, "step": 2448 }, { "epoch": 0.9947197400487409, "grad_norm": 0.0, "learning_rate": 1.4624371932370206e-09, "loss": 1.9837, "step": 2449 }, { "epoch": 0.9951259138911455, "grad_norm": 0.0, "learning_rate": 1.246104823426908e-09, "loss": 1.9101, "step": 2450 }, { "epoch": 0.99553208773355, "grad_norm": 0.0, "learning_rate": 1.047077665329299e-09, "loss": 2.0056, "step": 2451 }, { "epoch": 0.9959382615759546, "grad_norm": 0.0, "learning_rate": 8.653560634075409e-10, "loss": 1.9863, "step": 2452 }, { "epoch": 0.996344435418359, "grad_norm": 0.0, "learning_rate": 7.009403321722729e-10, "loss": 1.9342, "step": 2453 }, { "epoch": 0.9967506092607636, "grad_norm": 0.0, "learning_rate": 5.538307561858691e-10, "loss": 2.0287, "step": 2454 }, { "epoch": 0.9971567831031681, "grad_norm": 0.0, "learning_rate": 4.2402759005577555e-10, "loss": 2.0698, "step": 2455 }, { "epoch": 0.9975629569455727, "grad_norm": 0.0, "learning_rate": 3.115310584367315e-10, "loss": 1.9531, "step": 2456 }, { "epoch": 0.9979691307879772, "grad_norm": 0.0, "learning_rate": 2.163413560307692e-10, "loss": 1.8703, "step": 2457 }, { "epoch": 0.9983753046303818, "grad_norm": 0.0, "learning_rate": 1.3845864758610384e-10, "loss": 2.0286, "step": 2458 }, { "epoch": 0.9987814784727863, "grad_norm": 0.0, "learning_rate": 7.78830678982434e-11, "loss": 2.0303, "step": 2459 }, { "epoch": 0.9991876523151909, "grad_norm": 0.0, "learning_rate": 3.461472180554815e-11, "loss": 1.9552, "step": 2460 }, { "epoch": 0.9995938261575954, "grad_norm": 0.0, "learning_rate": 8.653684195891742e-12, "loss": 2.0066, "step": 2461 }, { "epoch": 1.0, "grad_norm": 0.0, "learning_rate": 0.0, "loss": 2.063, "step": 2462 }, { "epoch": 1.0, "step": 2462, "total_flos": 1.5137828909141197e+18, "train_loss": 2.2033056199889565, "train_runtime": 160077.9377, "train_samples_per_second": 1.969, "train_steps_per_second": 0.015 } ], "logging_steps": 1.0, "max_steps": 2462, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 1.5137828909141197e+18, "train_batch_size": 1, "trial_name": null, "trial_params": null }