{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 379.746835443038, "eval_steps": 500, "global_step": 30000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.12658227848101267, "grad_norm": 0.5513319373130798, "learning_rate": 6.000000000000001e-07, "loss": 1.3018, "step": 10 }, { "epoch": 0.25316455696202533, "grad_norm": 0.5655359625816345, "learning_rate": 1.2666666666666667e-06, "loss": 1.3009, "step": 20 }, { "epoch": 0.379746835443038, "grad_norm": 0.5913138389587402, "learning_rate": 1.9333333333333336e-06, "loss": 1.2945, "step": 30 }, { "epoch": 0.5063291139240507, "grad_norm": 0.5336909890174866, "learning_rate": 2.6e-06, "loss": 1.2857, "step": 40 }, { "epoch": 0.6329113924050633, "grad_norm": 0.7177284359931946, "learning_rate": 3.2666666666666666e-06, "loss": 1.2734, "step": 50 }, { "epoch": 0.759493670886076, "grad_norm": 0.7843838930130005, "learning_rate": 3.9333333333333335e-06, "loss": 1.2557, "step": 60 }, { "epoch": 0.8860759493670886, "grad_norm": 0.8489007353782654, "learning_rate": 4.6e-06, "loss": 1.2375, "step": 70 }, { "epoch": 1.0126582278481013, "grad_norm": 0.7142075896263123, "learning_rate": 5.266666666666667e-06, "loss": 1.2146, "step": 80 }, { "epoch": 1.139240506329114, "grad_norm": 0.4863379895687103, "learning_rate": 5.933333333333334e-06, "loss": 1.2025, "step": 90 }, { "epoch": 1.2658227848101267, "grad_norm": 0.38184744119644165, "learning_rate": 6.6e-06, "loss": 1.1876, "step": 100 }, { "epoch": 1.3924050632911391, "grad_norm": 0.2662145793437958, "learning_rate": 7.266666666666668e-06, "loss": 1.1816, "step": 110 }, { "epoch": 1.518987341772152, "grad_norm": 0.3294070363044739, "learning_rate": 7.933333333333334e-06, "loss": 1.1805, "step": 120 }, { "epoch": 1.6455696202531644, "grad_norm": 0.3084878921508789, "learning_rate": 8.599999999999999e-06, "loss": 1.1751, "step": 130 }, { "epoch": 1.7721518987341773, "grad_norm": 0.3515961468219757, "learning_rate": 9.266666666666667e-06, "loss": 1.1695, "step": 140 }, { "epoch": 1.8987341772151898, "grad_norm": 0.3893003761768341, "learning_rate": 9.933333333333334e-06, "loss": 1.1637, "step": 150 }, { "epoch": 2.0253164556962027, "grad_norm": 0.36362600326538086, "learning_rate": 1.06e-05, "loss": 1.1553, "step": 160 }, { "epoch": 2.151898734177215, "grad_norm": 0.4011550843715668, "learning_rate": 1.1266666666666667e-05, "loss": 1.1508, "step": 170 }, { "epoch": 2.278481012658228, "grad_norm": 0.40479904413223267, "learning_rate": 1.1933333333333333e-05, "loss": 1.1408, "step": 180 }, { "epoch": 2.4050632911392404, "grad_norm": 0.4014892876148224, "learning_rate": 1.2600000000000001e-05, "loss": 1.1385, "step": 190 }, { "epoch": 2.5316455696202533, "grad_norm": 0.4534638524055481, "learning_rate": 1.3266666666666666e-05, "loss": 1.1327, "step": 200 }, { "epoch": 2.6582278481012658, "grad_norm": 0.5342440605163574, "learning_rate": 1.3933333333333334e-05, "loss": 1.1233, "step": 210 }, { "epoch": 2.7848101265822782, "grad_norm": 0.45370814204216003, "learning_rate": 1.4599999999999999e-05, "loss": 1.1154, "step": 220 }, { "epoch": 2.911392405063291, "grad_norm": 0.5013312697410583, "learning_rate": 1.5266666666666667e-05, "loss": 1.1099, "step": 230 }, { "epoch": 3.037974683544304, "grad_norm": 0.5222480893135071, "learning_rate": 1.5933333333333332e-05, "loss": 1.1001, "step": 240 }, { "epoch": 3.1645569620253164, "grad_norm": 0.852782130241394, "learning_rate": 1.66e-05, "loss": 1.0861, "step": 250 }, { "epoch": 3.291139240506329, "grad_norm": 0.9118990898132324, "learning_rate": 1.726666666666667e-05, "loss": 1.0574, "step": 260 }, { "epoch": 3.4177215189873418, "grad_norm": 1.348787784576416, "learning_rate": 1.7933333333333337e-05, "loss": 0.9985, "step": 270 }, { "epoch": 3.5443037974683547, "grad_norm": 1.2913002967834473, "learning_rate": 1.86e-05, "loss": 0.9203, "step": 280 }, { "epoch": 3.670886075949367, "grad_norm": 1.7578246593475342, "learning_rate": 1.926666666666667e-05, "loss": 0.8464, "step": 290 }, { "epoch": 3.7974683544303796, "grad_norm": 1.4012209177017212, "learning_rate": 1.9933333333333334e-05, "loss": 0.7976, "step": 300 }, { "epoch": 3.9240506329113924, "grad_norm": 1.7191272974014282, "learning_rate": 2.06e-05, "loss": 0.7476, "step": 310 }, { "epoch": 4.050632911392405, "grad_norm": 1.739491581916809, "learning_rate": 2.1266666666666667e-05, "loss": 0.6921, "step": 320 }, { "epoch": 4.177215189873418, "grad_norm": 2.007615327835083, "learning_rate": 2.1933333333333332e-05, "loss": 0.6685, "step": 330 }, { "epoch": 4.30379746835443, "grad_norm": 2.0354361534118652, "learning_rate": 2.26e-05, "loss": 0.6461, "step": 340 }, { "epoch": 4.430379746835443, "grad_norm": 2.403346538543701, "learning_rate": 2.326666666666667e-05, "loss": 0.618, "step": 350 }, { "epoch": 4.556962025316456, "grad_norm": 2.1794471740722656, "learning_rate": 2.3933333333333337e-05, "loss": 0.6075, "step": 360 }, { "epoch": 4.6835443037974684, "grad_norm": 2.588928461074829, "learning_rate": 2.46e-05, "loss": 0.5814, "step": 370 }, { "epoch": 4.810126582278481, "grad_norm": 2.4363818168640137, "learning_rate": 2.5266666666666666e-05, "loss": 0.5794, "step": 380 }, { "epoch": 4.936708860759493, "grad_norm": 2.2946314811706543, "learning_rate": 2.5933333333333338e-05, "loss": 0.5712, "step": 390 }, { "epoch": 5.063291139240507, "grad_norm": 2.1703038215637207, "learning_rate": 2.6600000000000003e-05, "loss": 0.5632, "step": 400 }, { "epoch": 5.189873417721519, "grad_norm": 2.399822235107422, "learning_rate": 2.7266666666666668e-05, "loss": 0.5653, "step": 410 }, { "epoch": 5.3164556962025316, "grad_norm": 2.5700266361236572, "learning_rate": 2.7933333333333332e-05, "loss": 0.5569, "step": 420 }, { "epoch": 5.443037974683544, "grad_norm": 2.1505465507507324, "learning_rate": 2.86e-05, "loss": 0.5576, "step": 430 }, { "epoch": 5.569620253164557, "grad_norm": 2.101299524307251, "learning_rate": 2.926666666666667e-05, "loss": 0.5505, "step": 440 }, { "epoch": 5.69620253164557, "grad_norm": 1.7083948850631714, "learning_rate": 2.9933333333333337e-05, "loss": 0.5445, "step": 450 }, { "epoch": 5.822784810126582, "grad_norm": 2.1228315830230713, "learning_rate": 3.06e-05, "loss": 0.5422, "step": 460 }, { "epoch": 5.949367088607595, "grad_norm": 2.2119195461273193, "learning_rate": 3.126666666666666e-05, "loss": 0.5372, "step": 470 }, { "epoch": 6.075949367088608, "grad_norm": 2.2164721488952637, "learning_rate": 3.1933333333333335e-05, "loss": 0.5378, "step": 480 }, { "epoch": 6.2025316455696204, "grad_norm": 1.8868210315704346, "learning_rate": 3.26e-05, "loss": 0.5349, "step": 490 }, { "epoch": 6.329113924050633, "grad_norm": 2.381728410720825, "learning_rate": 3.326666666666667e-05, "loss": 0.5344, "step": 500 }, { "epoch": 6.455696202531645, "grad_norm": 2.3358376026153564, "learning_rate": 3.3933333333333336e-05, "loss": 0.5281, "step": 510 }, { "epoch": 6.582278481012658, "grad_norm": 2.338210105895996, "learning_rate": 3.46e-05, "loss": 0.5297, "step": 520 }, { "epoch": 6.708860759493671, "grad_norm": 2.7216763496398926, "learning_rate": 3.526666666666667e-05, "loss": 0.5268, "step": 530 }, { "epoch": 6.8354430379746836, "grad_norm": 1.975274682044983, "learning_rate": 3.593333333333334e-05, "loss": 0.5286, "step": 540 }, { "epoch": 6.962025316455696, "grad_norm": 2.6043806076049805, "learning_rate": 3.66e-05, "loss": 0.5157, "step": 550 }, { "epoch": 7.0886075949367084, "grad_norm": 2.249964714050293, "learning_rate": 3.726666666666667e-05, "loss": 0.52, "step": 560 }, { "epoch": 7.215189873417722, "grad_norm": 2.4465601444244385, "learning_rate": 3.793333333333334e-05, "loss": 0.5215, "step": 570 }, { "epoch": 7.341772151898734, "grad_norm": 2.083763837814331, "learning_rate": 3.86e-05, "loss": 0.5172, "step": 580 }, { "epoch": 7.468354430379747, "grad_norm": 2.6223790645599365, "learning_rate": 3.926666666666667e-05, "loss": 0.5137, "step": 590 }, { "epoch": 7.594936708860759, "grad_norm": 2.1126513481140137, "learning_rate": 3.993333333333333e-05, "loss": 0.513, "step": 600 }, { "epoch": 7.7215189873417724, "grad_norm": 2.5394580364227295, "learning_rate": 4.0600000000000004e-05, "loss": 0.5074, "step": 610 }, { "epoch": 7.848101265822785, "grad_norm": 2.394052743911743, "learning_rate": 4.126666666666667e-05, "loss": 0.5127, "step": 620 }, { "epoch": 7.974683544303797, "grad_norm": 2.447455644607544, "learning_rate": 4.1933333333333334e-05, "loss": 0.5164, "step": 630 }, { "epoch": 8.10126582278481, "grad_norm": 2.2050960063934326, "learning_rate": 4.26e-05, "loss": 0.5138, "step": 640 }, { "epoch": 8.227848101265822, "grad_norm": 2.3660080432891846, "learning_rate": 4.3266666666666664e-05, "loss": 0.5089, "step": 650 }, { "epoch": 8.354430379746836, "grad_norm": 2.424452066421509, "learning_rate": 4.3933333333333335e-05, "loss": 0.5055, "step": 660 }, { "epoch": 8.481012658227849, "grad_norm": 2.2885427474975586, "learning_rate": 4.46e-05, "loss": 0.4975, "step": 670 }, { "epoch": 8.60759493670886, "grad_norm": 2.2519845962524414, "learning_rate": 4.526666666666667e-05, "loss": 0.4935, "step": 680 }, { "epoch": 8.734177215189874, "grad_norm": 2.1786277294158936, "learning_rate": 4.5933333333333336e-05, "loss": 0.4885, "step": 690 }, { "epoch": 8.860759493670885, "grad_norm": 2.121755361557007, "learning_rate": 4.660000000000001e-05, "loss": 0.4861, "step": 700 }, { "epoch": 8.987341772151899, "grad_norm": 2.968592643737793, "learning_rate": 4.726666666666667e-05, "loss": 0.4841, "step": 710 }, { "epoch": 9.113924050632912, "grad_norm": 2.660475730895996, "learning_rate": 4.793333333333334e-05, "loss": 0.4838, "step": 720 }, { "epoch": 9.240506329113924, "grad_norm": 2.069873094558716, "learning_rate": 4.86e-05, "loss": 0.4824, "step": 730 }, { "epoch": 9.367088607594937, "grad_norm": 2.1236796379089355, "learning_rate": 4.926666666666667e-05, "loss": 0.4816, "step": 740 }, { "epoch": 9.49367088607595, "grad_norm": 1.9090505838394165, "learning_rate": 4.993333333333334e-05, "loss": 0.4793, "step": 750 }, { "epoch": 9.620253164556962, "grad_norm": 2.7258217334747314, "learning_rate": 5.0600000000000003e-05, "loss": 0.4833, "step": 760 }, { "epoch": 9.746835443037975, "grad_norm": 2.4134349822998047, "learning_rate": 5.1266666666666675e-05, "loss": 0.4802, "step": 770 }, { "epoch": 9.873417721518987, "grad_norm": 2.3482956886291504, "learning_rate": 5.193333333333333e-05, "loss": 0.4738, "step": 780 }, { "epoch": 10.0, "grad_norm": 2.469525098800659, "learning_rate": 5.2600000000000005e-05, "loss": 0.4671, "step": 790 }, { "epoch": 10.126582278481013, "grad_norm": 2.5011403560638428, "learning_rate": 5.326666666666666e-05, "loss": 0.4669, "step": 800 }, { "epoch": 10.253164556962025, "grad_norm": 2.552440643310547, "learning_rate": 5.3933333333333334e-05, "loss": 0.4622, "step": 810 }, { "epoch": 10.379746835443038, "grad_norm": 2.4903154373168945, "learning_rate": 5.4600000000000006e-05, "loss": 0.4593, "step": 820 }, { "epoch": 10.50632911392405, "grad_norm": 2.6741697788238525, "learning_rate": 5.5266666666666664e-05, "loss": 0.451, "step": 830 }, { "epoch": 10.632911392405063, "grad_norm": 2.6026761531829834, "learning_rate": 5.5933333333333335e-05, "loss": 0.4482, "step": 840 }, { "epoch": 10.759493670886076, "grad_norm": 2.863369941711426, "learning_rate": 5.66e-05, "loss": 0.4439, "step": 850 }, { "epoch": 10.886075949367088, "grad_norm": 2.700861692428589, "learning_rate": 5.726666666666667e-05, "loss": 0.4424, "step": 860 }, { "epoch": 11.012658227848101, "grad_norm": 2.6111884117126465, "learning_rate": 5.7933333333333337e-05, "loss": 0.4357, "step": 870 }, { "epoch": 11.139240506329115, "grad_norm": 2.111107587814331, "learning_rate": 5.86e-05, "loss": 0.4332, "step": 880 }, { "epoch": 11.265822784810126, "grad_norm": 2.6753222942352295, "learning_rate": 5.926666666666667e-05, "loss": 0.4233, "step": 890 }, { "epoch": 11.39240506329114, "grad_norm": 2.912189245223999, "learning_rate": 5.9933333333333345e-05, "loss": 0.4211, "step": 900 }, { "epoch": 11.518987341772151, "grad_norm": 2.967435598373413, "learning_rate": 6.06e-05, "loss": 0.4105, "step": 910 }, { "epoch": 11.645569620253164, "grad_norm": 3.524245262145996, "learning_rate": 6.126666666666667e-05, "loss": 0.4062, "step": 920 }, { "epoch": 11.772151898734178, "grad_norm": 4.115475654602051, "learning_rate": 6.193333333333333e-05, "loss": 0.4046, "step": 930 }, { "epoch": 11.89873417721519, "grad_norm": 3.537309408187866, "learning_rate": 6.26e-05, "loss": 0.4062, "step": 940 }, { "epoch": 12.025316455696203, "grad_norm": 3.6304872035980225, "learning_rate": 6.326666666666667e-05, "loss": 0.3835, "step": 950 }, { "epoch": 12.151898734177216, "grad_norm": 4.854063987731934, "learning_rate": 6.393333333333333e-05, "loss": 0.3793, "step": 960 }, { "epoch": 12.278481012658228, "grad_norm": 3.9066970348358154, "learning_rate": 6.460000000000001e-05, "loss": 0.3753, "step": 970 }, { "epoch": 12.405063291139241, "grad_norm": 3.026914119720459, "learning_rate": 6.526666666666666e-05, "loss": 0.355, "step": 980 }, { "epoch": 12.531645569620252, "grad_norm": 3.0555994510650635, "learning_rate": 6.593333333333334e-05, "loss": 0.3576, "step": 990 }, { "epoch": 12.658227848101266, "grad_norm": 2.791698932647705, "learning_rate": 6.66e-05, "loss": 0.338, "step": 1000 }, { "epoch": 12.784810126582279, "grad_norm": 6.145380020141602, "learning_rate": 6.726666666666667e-05, "loss": 0.3356, "step": 1010 }, { "epoch": 12.91139240506329, "grad_norm": 4.256630897521973, "learning_rate": 6.793333333333334e-05, "loss": 0.3338, "step": 1020 }, { "epoch": 13.037974683544304, "grad_norm": 3.8724238872528076, "learning_rate": 6.860000000000001e-05, "loss": 0.3092, "step": 1030 }, { "epoch": 13.164556962025316, "grad_norm": 3.8069210052490234, "learning_rate": 6.926666666666667e-05, "loss": 0.2958, "step": 1040 }, { "epoch": 13.291139240506329, "grad_norm": 6.6001152992248535, "learning_rate": 6.993333333333334e-05, "loss": 0.2821, "step": 1050 }, { "epoch": 13.417721518987342, "grad_norm": 5.3725738525390625, "learning_rate": 7.06e-05, "loss": 0.2815, "step": 1060 }, { "epoch": 13.544303797468354, "grad_norm": 4.718721866607666, "learning_rate": 7.126666666666667e-05, "loss": 0.2746, "step": 1070 }, { "epoch": 13.670886075949367, "grad_norm": 5.398708820343018, "learning_rate": 7.193333333333334e-05, "loss": 0.2639, "step": 1080 }, { "epoch": 13.79746835443038, "grad_norm": 5.898968696594238, "learning_rate": 7.26e-05, "loss": 0.254, "step": 1090 }, { "epoch": 13.924050632911392, "grad_norm": 5.093671798706055, "learning_rate": 7.326666666666667e-05, "loss": 0.2384, "step": 1100 }, { "epoch": 14.050632911392405, "grad_norm": 4.636958599090576, "learning_rate": 7.393333333333333e-05, "loss": 0.2326, "step": 1110 }, { "epoch": 14.177215189873417, "grad_norm": 3.994814872741699, "learning_rate": 7.46e-05, "loss": 0.2237, "step": 1120 }, { "epoch": 14.30379746835443, "grad_norm": 5.495810508728027, "learning_rate": 7.526666666666668e-05, "loss": 0.2167, "step": 1130 }, { "epoch": 14.430379746835444, "grad_norm": 4.490682601928711, "learning_rate": 7.593333333333334e-05, "loss": 0.2024, "step": 1140 }, { "epoch": 14.556962025316455, "grad_norm": 4.138688564300537, "learning_rate": 7.66e-05, "loss": 0.1852, "step": 1150 }, { "epoch": 14.683544303797468, "grad_norm": 3.6954426765441895, "learning_rate": 7.726666666666667e-05, "loss": 0.1724, "step": 1160 }, { "epoch": 14.810126582278482, "grad_norm": 6.45929479598999, "learning_rate": 7.793333333333333e-05, "loss": 0.1758, "step": 1170 }, { "epoch": 14.936708860759493, "grad_norm": 4.519597053527832, "learning_rate": 7.860000000000001e-05, "loss": 0.1717, "step": 1180 }, { "epoch": 15.063291139240507, "grad_norm": 4.992429733276367, "learning_rate": 7.926666666666666e-05, "loss": 0.1529, "step": 1190 }, { "epoch": 15.189873417721518, "grad_norm": 5.37672233581543, "learning_rate": 7.993333333333334e-05, "loss": 0.176, "step": 1200 }, { "epoch": 15.316455696202532, "grad_norm": 3.232316255569458, "learning_rate": 8.060000000000001e-05, "loss": 0.143, "step": 1210 }, { "epoch": 15.443037974683545, "grad_norm": 5.803571701049805, "learning_rate": 8.126666666666667e-05, "loss": 0.1458, "step": 1220 }, { "epoch": 15.569620253164556, "grad_norm": 6.328423023223877, "learning_rate": 8.193333333333334e-05, "loss": 0.1521, "step": 1230 }, { "epoch": 15.69620253164557, "grad_norm": 3.698668956756592, "learning_rate": 8.26e-05, "loss": 0.1309, "step": 1240 }, { "epoch": 15.822784810126583, "grad_norm": 3.283508539199829, "learning_rate": 8.326666666666667e-05, "loss": 0.1209, "step": 1250 }, { "epoch": 15.949367088607595, "grad_norm": 6.439619064331055, "learning_rate": 8.393333333333335e-05, "loss": 0.118, "step": 1260 }, { "epoch": 16.075949367088608, "grad_norm": 5.516022205352783, "learning_rate": 8.46e-05, "loss": 0.123, "step": 1270 }, { "epoch": 16.20253164556962, "grad_norm": 6.012554168701172, "learning_rate": 8.526666666666667e-05, "loss": 0.1265, "step": 1280 }, { "epoch": 16.32911392405063, "grad_norm": 3.0568230152130127, "learning_rate": 8.593333333333333e-05, "loss": 0.1196, "step": 1290 }, { "epoch": 16.455696202531644, "grad_norm": 2.9236395359039307, "learning_rate": 8.66e-05, "loss": 0.1102, "step": 1300 }, { "epoch": 16.582278481012658, "grad_norm": 2.830207109451294, "learning_rate": 8.726666666666667e-05, "loss": 0.0993, "step": 1310 }, { "epoch": 16.70886075949367, "grad_norm": 2.768869400024414, "learning_rate": 8.793333333333333e-05, "loss": 0.0905, "step": 1320 }, { "epoch": 16.835443037974684, "grad_norm": 2.8727898597717285, "learning_rate": 8.86e-05, "loss": 0.1076, "step": 1330 }, { "epoch": 16.962025316455698, "grad_norm": 2.73938250541687, "learning_rate": 8.926666666666668e-05, "loss": 0.1026, "step": 1340 }, { "epoch": 17.088607594936708, "grad_norm": 2.700979709625244, "learning_rate": 8.993333333333334e-05, "loss": 0.0999, "step": 1350 }, { "epoch": 17.21518987341772, "grad_norm": 2.8123114109039307, "learning_rate": 9.06e-05, "loss": 0.0972, "step": 1360 }, { "epoch": 17.341772151898734, "grad_norm": 3.2849531173706055, "learning_rate": 9.126666666666667e-05, "loss": 0.1017, "step": 1370 }, { "epoch": 17.468354430379748, "grad_norm": 2.9631850719451904, "learning_rate": 9.193333333333334e-05, "loss": 0.0899, "step": 1380 }, { "epoch": 17.59493670886076, "grad_norm": 2.329077959060669, "learning_rate": 9.260000000000001e-05, "loss": 0.0899, "step": 1390 }, { "epoch": 17.72151898734177, "grad_norm": 2.7561373710632324, "learning_rate": 9.326666666666667e-05, "loss": 0.0895, "step": 1400 }, { "epoch": 17.848101265822784, "grad_norm": 2.938918352127075, "learning_rate": 9.393333333333334e-05, "loss": 0.0885, "step": 1410 }, { "epoch": 17.974683544303797, "grad_norm": 2.9288649559020996, "learning_rate": 9.46e-05, "loss": 0.0908, "step": 1420 }, { "epoch": 18.10126582278481, "grad_norm": 3.197842597961426, "learning_rate": 9.526666666666667e-05, "loss": 0.0888, "step": 1430 }, { "epoch": 18.227848101265824, "grad_norm": 2.640136480331421, "learning_rate": 9.593333333333334e-05, "loss": 0.087, "step": 1440 }, { "epoch": 18.354430379746834, "grad_norm": 2.3861618041992188, "learning_rate": 9.66e-05, "loss": 0.0876, "step": 1450 }, { "epoch": 18.481012658227847, "grad_norm": 2.4457623958587646, "learning_rate": 9.726666666666667e-05, "loss": 0.0959, "step": 1460 }, { "epoch": 18.60759493670886, "grad_norm": 2.173258066177368, "learning_rate": 9.793333333333333e-05, "loss": 0.0825, "step": 1470 }, { "epoch": 18.734177215189874, "grad_norm": 2.8218343257904053, "learning_rate": 9.86e-05, "loss": 0.0873, "step": 1480 }, { "epoch": 18.860759493670887, "grad_norm": 2.4145612716674805, "learning_rate": 9.926666666666668e-05, "loss": 0.0837, "step": 1490 }, { "epoch": 18.9873417721519, "grad_norm": 2.6727967262268066, "learning_rate": 9.993333333333334e-05, "loss": 0.0858, "step": 1500 }, { "epoch": 19.11392405063291, "grad_norm": 1.9554531574249268, "learning_rate": 9.999997539434007e-05, "loss": 0.0855, "step": 1510 }, { "epoch": 19.240506329113924, "grad_norm": 2.456997871398926, "learning_rate": 9.999989033776898e-05, "loss": 0.0765, "step": 1520 }, { "epoch": 19.367088607594937, "grad_norm": 2.282543659210205, "learning_rate": 9.999974452661641e-05, "loss": 0.0833, "step": 1530 }, { "epoch": 19.49367088607595, "grad_norm": 2.0353522300720215, "learning_rate": 9.999953796105959e-05, "loss": 0.082, "step": 1540 }, { "epoch": 19.620253164556964, "grad_norm": 3.002187967300415, "learning_rate": 9.999927064134949e-05, "loss": 0.0836, "step": 1550 }, { "epoch": 19.746835443037973, "grad_norm": 2.8009581565856934, "learning_rate": 9.999894256781095e-05, "loss": 0.0832, "step": 1560 }, { "epoch": 19.873417721518987, "grad_norm": 2.8053760528564453, "learning_rate": 9.99985537408426e-05, "loss": 0.0821, "step": 1570 }, { "epoch": 20.0, "grad_norm": 2.507418155670166, "learning_rate": 9.999810416091688e-05, "loss": 0.077, "step": 1580 }, { "epoch": 20.126582278481013, "grad_norm": 2.506423234939575, "learning_rate": 9.99975938285801e-05, "loss": 0.0772, "step": 1590 }, { "epoch": 20.253164556962027, "grad_norm": 2.460216760635376, "learning_rate": 9.999702274445236e-05, "loss": 0.0831, "step": 1600 }, { "epoch": 20.379746835443036, "grad_norm": 2.22212815284729, "learning_rate": 9.999639090922756e-05, "loss": 0.0756, "step": 1610 }, { "epoch": 20.50632911392405, "grad_norm": 2.5354509353637695, "learning_rate": 9.999569832367346e-05, "loss": 0.0705, "step": 1620 }, { "epoch": 20.632911392405063, "grad_norm": 2.264050245285034, "learning_rate": 9.999494498863162e-05, "loss": 0.0746, "step": 1630 }, { "epoch": 20.759493670886076, "grad_norm": 2.0292508602142334, "learning_rate": 9.99941309050174e-05, "loss": 0.0692, "step": 1640 }, { "epoch": 20.88607594936709, "grad_norm": 2.3730621337890625, "learning_rate": 9.999325607381999e-05, "loss": 0.07, "step": 1650 }, { "epoch": 21.0126582278481, "grad_norm": 2.155402898788452, "learning_rate": 9.999232049610238e-05, "loss": 0.0672, "step": 1660 }, { "epoch": 21.139240506329113, "grad_norm": 2.2973403930664062, "learning_rate": 9.999132417300142e-05, "loss": 0.0707, "step": 1670 }, { "epoch": 21.265822784810126, "grad_norm": 2.2573273181915283, "learning_rate": 9.99902671057277e-05, "loss": 0.0724, "step": 1680 }, { "epoch": 21.39240506329114, "grad_norm": 2.054298162460327, "learning_rate": 9.998914929556569e-05, "loss": 0.0792, "step": 1690 }, { "epoch": 21.518987341772153, "grad_norm": 2.271177053451538, "learning_rate": 9.998797074387361e-05, "loss": 0.0761, "step": 1700 }, { "epoch": 21.645569620253166, "grad_norm": 2.307734251022339, "learning_rate": 9.99867314520835e-05, "loss": 0.0724, "step": 1710 }, { "epoch": 21.772151898734176, "grad_norm": 2.126559257507324, "learning_rate": 9.998543142170126e-05, "loss": 0.0679, "step": 1720 }, { "epoch": 21.89873417721519, "grad_norm": 1.8648324012756348, "learning_rate": 9.99840706543065e-05, "loss": 0.0693, "step": 1730 }, { "epoch": 22.025316455696203, "grad_norm": 2.261537790298462, "learning_rate": 9.998264915155274e-05, "loss": 0.0662, "step": 1740 }, { "epoch": 22.151898734177216, "grad_norm": 2.2258408069610596, "learning_rate": 9.998116691516718e-05, "loss": 0.0744, "step": 1750 }, { "epoch": 22.27848101265823, "grad_norm": 2.2407755851745605, "learning_rate": 9.997962394695091e-05, "loss": 0.0684, "step": 1760 }, { "epoch": 22.40506329113924, "grad_norm": 2.066622495651245, "learning_rate": 9.997802024877875e-05, "loss": 0.0692, "step": 1770 }, { "epoch": 22.531645569620252, "grad_norm": 2.0904812812805176, "learning_rate": 9.99763558225994e-05, "loss": 0.0684, "step": 1780 }, { "epoch": 22.658227848101266, "grad_norm": 1.7152646780014038, "learning_rate": 9.997463067043526e-05, "loss": 0.0701, "step": 1790 }, { "epoch": 22.78481012658228, "grad_norm": 1.9799325466156006, "learning_rate": 9.997284479438253e-05, "loss": 0.0677, "step": 1800 }, { "epoch": 22.911392405063292, "grad_norm": 2.3889966011047363, "learning_rate": 9.997099819661127e-05, "loss": 0.0645, "step": 1810 }, { "epoch": 23.037974683544302, "grad_norm": 2.1051926612854004, "learning_rate": 9.996909087936524e-05, "loss": 0.067, "step": 1820 }, { "epoch": 23.164556962025316, "grad_norm": 2.300703763961792, "learning_rate": 9.996712284496201e-05, "loss": 0.063, "step": 1830 }, { "epoch": 23.29113924050633, "grad_norm": 1.9028583765029907, "learning_rate": 9.996509409579293e-05, "loss": 0.069, "step": 1840 }, { "epoch": 23.417721518987342, "grad_norm": 2.290872097015381, "learning_rate": 9.996300463432312e-05, "loss": 0.0655, "step": 1850 }, { "epoch": 23.544303797468356, "grad_norm": 2.120954751968384, "learning_rate": 9.996085446309148e-05, "loss": 0.0648, "step": 1860 }, { "epoch": 23.67088607594937, "grad_norm": 2.0536272525787354, "learning_rate": 9.995864358471066e-05, "loss": 0.0604, "step": 1870 }, { "epoch": 23.79746835443038, "grad_norm": 2.0836892127990723, "learning_rate": 9.99563720018671e-05, "loss": 0.0607, "step": 1880 }, { "epoch": 23.924050632911392, "grad_norm": 2.038742780685425, "learning_rate": 9.995403971732098e-05, "loss": 0.0675, "step": 1890 }, { "epoch": 24.050632911392405, "grad_norm": 2.1215431690216064, "learning_rate": 9.995164673390625e-05, "loss": 0.0588, "step": 1900 }, { "epoch": 24.17721518987342, "grad_norm": 2.1127493381500244, "learning_rate": 9.994919305453059e-05, "loss": 0.0613, "step": 1910 }, { "epoch": 24.303797468354432, "grad_norm": 1.8694216012954712, "learning_rate": 9.994667868217548e-05, "loss": 0.0631, "step": 1920 }, { "epoch": 24.430379746835442, "grad_norm": 1.8166369199752808, "learning_rate": 9.99441036198961e-05, "loss": 0.0618, "step": 1930 }, { "epoch": 24.556962025316455, "grad_norm": 1.7402926683425903, "learning_rate": 9.99414678708214e-05, "loss": 0.0617, "step": 1940 }, { "epoch": 24.68354430379747, "grad_norm": 1.7546372413635254, "learning_rate": 9.993877143815407e-05, "loss": 0.06, "step": 1950 }, { "epoch": 24.810126582278482, "grad_norm": 1.9886579513549805, "learning_rate": 9.993601432517053e-05, "loss": 0.06, "step": 1960 }, { "epoch": 24.936708860759495, "grad_norm": 1.640436053276062, "learning_rate": 9.993319653522091e-05, "loss": 0.0637, "step": 1970 }, { "epoch": 25.063291139240505, "grad_norm": 2.0011415481567383, "learning_rate": 9.993031807172911e-05, "loss": 0.0611, "step": 1980 }, { "epoch": 25.189873417721518, "grad_norm": 1.7352300882339478, "learning_rate": 9.992737893819273e-05, "loss": 0.0606, "step": 1990 }, { "epoch": 25.31645569620253, "grad_norm": 1.8793390989303589, "learning_rate": 9.992437913818312e-05, "loss": 0.0613, "step": 2000 }, { "epoch": 25.443037974683545, "grad_norm": 1.9070268869400024, "learning_rate": 9.992131867534526e-05, "loss": 0.0572, "step": 2010 }, { "epoch": 25.569620253164558, "grad_norm": 1.888676404953003, "learning_rate": 9.991819755339796e-05, "loss": 0.0623, "step": 2020 }, { "epoch": 25.696202531645568, "grad_norm": 1.8511658906936646, "learning_rate": 9.991501577613365e-05, "loss": 0.0593, "step": 2030 }, { "epoch": 25.82278481012658, "grad_norm": 1.934926986694336, "learning_rate": 9.99117733474185e-05, "loss": 0.0568, "step": 2040 }, { "epoch": 25.949367088607595, "grad_norm": 2.0611813068389893, "learning_rate": 9.990847027119234e-05, "loss": 0.0591, "step": 2050 }, { "epoch": 26.075949367088608, "grad_norm": 2.075688123703003, "learning_rate": 9.990510655146877e-05, "loss": 0.0596, "step": 2060 }, { "epoch": 26.20253164556962, "grad_norm": 1.7787792682647705, "learning_rate": 9.990168219233496e-05, "loss": 0.0599, "step": 2070 }, { "epoch": 26.32911392405063, "grad_norm": 2.1297836303710938, "learning_rate": 9.989819719795188e-05, "loss": 0.0508, "step": 2080 }, { "epoch": 26.455696202531644, "grad_norm": 1.9286338090896606, "learning_rate": 9.989465157255412e-05, "loss": 0.0595, "step": 2090 }, { "epoch": 26.582278481012658, "grad_norm": 1.7370514869689941, "learning_rate": 9.989104532044994e-05, "loss": 0.0547, "step": 2100 }, { "epoch": 26.70886075949367, "grad_norm": 1.5902724266052246, "learning_rate": 9.988737844602128e-05, "loss": 0.05, "step": 2110 }, { "epoch": 26.835443037974684, "grad_norm": 1.9551624059677124, "learning_rate": 9.988365095372372e-05, "loss": 0.0534, "step": 2120 }, { "epoch": 26.962025316455698, "grad_norm": 1.8140512704849243, "learning_rate": 9.987986284808654e-05, "loss": 0.06, "step": 2130 }, { "epoch": 27.088607594936708, "grad_norm": 1.7308670282363892, "learning_rate": 9.987601413371264e-05, "loss": 0.0527, "step": 2140 }, { "epoch": 27.21518987341772, "grad_norm": 1.6692501306533813, "learning_rate": 9.987210481527855e-05, "loss": 0.0546, "step": 2150 }, { "epoch": 27.341772151898734, "grad_norm": 1.6720436811447144, "learning_rate": 9.98681348975345e-05, "loss": 0.0513, "step": 2160 }, { "epoch": 27.468354430379748, "grad_norm": 1.5918365716934204, "learning_rate": 9.986410438530427e-05, "loss": 0.0505, "step": 2170 }, { "epoch": 27.59493670886076, "grad_norm": 2.125399589538574, "learning_rate": 9.986001328348534e-05, "loss": 0.0514, "step": 2180 }, { "epoch": 27.72151898734177, "grad_norm": 1.9963953495025635, "learning_rate": 9.985586159704878e-05, "loss": 0.0498, "step": 2190 }, { "epoch": 27.848101265822784, "grad_norm": 1.4963257312774658, "learning_rate": 9.985164933103929e-05, "loss": 0.0617, "step": 2200 }, { "epoch": 27.974683544303797, "grad_norm": 1.9560141563415527, "learning_rate": 9.984737649057513e-05, "loss": 0.055, "step": 2210 }, { "epoch": 28.10126582278481, "grad_norm": 1.6828334331512451, "learning_rate": 9.984304308084827e-05, "loss": 0.0567, "step": 2220 }, { "epoch": 28.227848101265824, "grad_norm": 1.646760106086731, "learning_rate": 9.983864910712416e-05, "loss": 0.0512, "step": 2230 }, { "epoch": 28.354430379746834, "grad_norm": 1.528198003768921, "learning_rate": 9.98341945747419e-05, "loss": 0.0524, "step": 2240 }, { "epoch": 28.481012658227847, "grad_norm": 1.7126085758209229, "learning_rate": 9.98296794891142e-05, "loss": 0.0517, "step": 2250 }, { "epoch": 28.60759493670886, "grad_norm": 1.6763677597045898, "learning_rate": 9.982510385572725e-05, "loss": 0.0575, "step": 2260 }, { "epoch": 28.734177215189874, "grad_norm": 1.8366398811340332, "learning_rate": 9.982046768014094e-05, "loss": 0.0524, "step": 2270 }, { "epoch": 28.860759493670887, "grad_norm": 1.7732946872711182, "learning_rate": 9.981577096798863e-05, "loss": 0.0476, "step": 2280 }, { "epoch": 28.9873417721519, "grad_norm": 1.6984881162643433, "learning_rate": 9.981101372497727e-05, "loss": 0.0533, "step": 2290 }, { "epoch": 29.11392405063291, "grad_norm": 2.0939345359802246, "learning_rate": 9.980619595688737e-05, "loss": 0.0503, "step": 2300 }, { "epoch": 29.240506329113924, "grad_norm": 1.5777472257614136, "learning_rate": 9.980131766957295e-05, "loss": 0.0553, "step": 2310 }, { "epoch": 29.367088607594937, "grad_norm": 1.6525834798812866, "learning_rate": 9.979637886896163e-05, "loss": 0.0524, "step": 2320 }, { "epoch": 29.49367088607595, "grad_norm": 1.867486596107483, "learning_rate": 9.979137956105447e-05, "loss": 0.0563, "step": 2330 }, { "epoch": 29.620253164556964, "grad_norm": 1.6072065830230713, "learning_rate": 9.978631975192613e-05, "loss": 0.0482, "step": 2340 }, { "epoch": 29.746835443037973, "grad_norm": 1.2181282043457031, "learning_rate": 9.978119944772475e-05, "loss": 0.0551, "step": 2350 }, { "epoch": 29.873417721518987, "grad_norm": 1.4790440797805786, "learning_rate": 9.977601865467197e-05, "loss": 0.0511, "step": 2360 }, { "epoch": 30.0, "grad_norm": 1.600636601448059, "learning_rate": 9.977077737906297e-05, "loss": 0.0479, "step": 2370 }, { "epoch": 30.126582278481013, "grad_norm": 1.5224621295928955, "learning_rate": 9.976547562726636e-05, "loss": 0.0515, "step": 2380 }, { "epoch": 30.253164556962027, "grad_norm": 1.664806604385376, "learning_rate": 9.976011340572429e-05, "loss": 0.0517, "step": 2390 }, { "epoch": 30.379746835443036, "grad_norm": 1.416125774383545, "learning_rate": 9.975469072095237e-05, "loss": 0.0486, "step": 2400 }, { "epoch": 30.50632911392405, "grad_norm": 1.5895745754241943, "learning_rate": 9.974920757953965e-05, "loss": 0.0442, "step": 2410 }, { "epoch": 30.632911392405063, "grad_norm": 1.7233142852783203, "learning_rate": 9.97436639881487e-05, "loss": 0.0568, "step": 2420 }, { "epoch": 30.759493670886076, "grad_norm": 1.5579676628112793, "learning_rate": 9.973805995351545e-05, "loss": 0.0492, "step": 2430 }, { "epoch": 30.88607594936709, "grad_norm": 2.280303716659546, "learning_rate": 9.973239548244939e-05, "loss": 0.056, "step": 2440 }, { "epoch": 31.0126582278481, "grad_norm": 1.4474360942840576, "learning_rate": 9.972667058183333e-05, "loss": 0.0531, "step": 2450 }, { "epoch": 31.139240506329113, "grad_norm": 1.4837006330490112, "learning_rate": 9.972088525862362e-05, "loss": 0.0465, "step": 2460 }, { "epoch": 31.265822784810126, "grad_norm": 1.6732004880905151, "learning_rate": 9.971503951984995e-05, "loss": 0.0502, "step": 2470 }, { "epoch": 31.39240506329114, "grad_norm": 1.420230746269226, "learning_rate": 9.970913337261543e-05, "loss": 0.0466, "step": 2480 }, { "epoch": 31.518987341772153, "grad_norm": 1.7215665578842163, "learning_rate": 9.97031668240966e-05, "loss": 0.0542, "step": 2490 }, { "epoch": 31.645569620253166, "grad_norm": 1.5131992101669312, "learning_rate": 9.969713988154339e-05, "loss": 0.0488, "step": 2500 }, { "epoch": 31.772151898734176, "grad_norm": 1.3093454837799072, "learning_rate": 9.969105255227906e-05, "loss": 0.049, "step": 2510 }, { "epoch": 31.89873417721519, "grad_norm": 1.4509352445602417, "learning_rate": 9.968490484370035e-05, "loss": 0.0482, "step": 2520 }, { "epoch": 32.0253164556962, "grad_norm": 1.7185330390930176, "learning_rate": 9.967869676327726e-05, "loss": 0.048, "step": 2530 }, { "epoch": 32.151898734177216, "grad_norm": 1.4482355117797852, "learning_rate": 9.96724283185532e-05, "loss": 0.0456, "step": 2540 }, { "epoch": 32.278481012658226, "grad_norm": 1.5459964275360107, "learning_rate": 9.966609951714494e-05, "loss": 0.0494, "step": 2550 }, { "epoch": 32.40506329113924, "grad_norm": 1.5911484956741333, "learning_rate": 9.965971036674255e-05, "loss": 0.0541, "step": 2560 }, { "epoch": 32.53164556962025, "grad_norm": 1.8603941202163696, "learning_rate": 9.965326087510947e-05, "loss": 0.0445, "step": 2570 }, { "epoch": 32.65822784810126, "grad_norm": 1.5258524417877197, "learning_rate": 9.964675105008244e-05, "loss": 0.0477, "step": 2580 }, { "epoch": 32.78481012658228, "grad_norm": 1.4535887241363525, "learning_rate": 9.964018089957147e-05, "loss": 0.0414, "step": 2590 }, { "epoch": 32.91139240506329, "grad_norm": 1.245955467224121, "learning_rate": 9.963355043155996e-05, "loss": 0.0511, "step": 2600 }, { "epoch": 33.037974683544306, "grad_norm": 1.5431768894195557, "learning_rate": 9.962685965410455e-05, "loss": 0.0469, "step": 2610 }, { "epoch": 33.164556962025316, "grad_norm": 1.1873111724853516, "learning_rate": 9.962010857533514e-05, "loss": 0.0463, "step": 2620 }, { "epoch": 33.29113924050633, "grad_norm": 1.4043241739273071, "learning_rate": 9.961329720345493e-05, "loss": 0.045, "step": 2630 }, { "epoch": 33.41772151898734, "grad_norm": 1.2250679731369019, "learning_rate": 9.96064255467404e-05, "loss": 0.048, "step": 2640 }, { "epoch": 33.54430379746835, "grad_norm": 1.3751749992370605, "learning_rate": 9.959949361354126e-05, "loss": 0.0468, "step": 2650 }, { "epoch": 33.67088607594937, "grad_norm": 1.556969165802002, "learning_rate": 9.959250141228045e-05, "loss": 0.0472, "step": 2660 }, { "epoch": 33.79746835443038, "grad_norm": 1.6381007432937622, "learning_rate": 9.958544895145414e-05, "loss": 0.0474, "step": 2670 }, { "epoch": 33.924050632911396, "grad_norm": 1.6285345554351807, "learning_rate": 9.957833623963177e-05, "loss": 0.049, "step": 2680 }, { "epoch": 34.050632911392405, "grad_norm": 1.3586374521255493, "learning_rate": 9.957116328545593e-05, "loss": 0.0472, "step": 2690 }, { "epoch": 34.177215189873415, "grad_norm": 1.387245535850525, "learning_rate": 9.956393009764244e-05, "loss": 0.044, "step": 2700 }, { "epoch": 34.30379746835443, "grad_norm": 1.3078941106796265, "learning_rate": 9.955663668498032e-05, "loss": 0.0431, "step": 2710 }, { "epoch": 34.43037974683544, "grad_norm": 1.2781531810760498, "learning_rate": 9.954928305633173e-05, "loss": 0.0441, "step": 2720 }, { "epoch": 34.55696202531646, "grad_norm": 1.4285024404525757, "learning_rate": 9.954186922063204e-05, "loss": 0.0452, "step": 2730 }, { "epoch": 34.68354430379747, "grad_norm": 1.3509777784347534, "learning_rate": 9.953439518688974e-05, "loss": 0.0414, "step": 2740 }, { "epoch": 34.81012658227848, "grad_norm": 1.4490294456481934, "learning_rate": 9.952686096418652e-05, "loss": 0.0399, "step": 2750 }, { "epoch": 34.936708860759495, "grad_norm": 1.4212442636489868, "learning_rate": 9.951926656167714e-05, "loss": 0.0409, "step": 2760 }, { "epoch": 35.063291139240505, "grad_norm": 1.597275972366333, "learning_rate": 9.951161198858953e-05, "loss": 0.046, "step": 2770 }, { "epoch": 35.18987341772152, "grad_norm": 1.4766473770141602, "learning_rate": 9.950389725422471e-05, "loss": 0.0408, "step": 2780 }, { "epoch": 35.31645569620253, "grad_norm": 1.4127964973449707, "learning_rate": 9.949612236795682e-05, "loss": 0.0422, "step": 2790 }, { "epoch": 35.44303797468354, "grad_norm": 1.1118073463439941, "learning_rate": 9.948828733923305e-05, "loss": 0.0438, "step": 2800 }, { "epoch": 35.56962025316456, "grad_norm": 1.4752812385559082, "learning_rate": 9.948039217757374e-05, "loss": 0.0457, "step": 2810 }, { "epoch": 35.69620253164557, "grad_norm": 1.3044487237930298, "learning_rate": 9.947243689257225e-05, "loss": 0.043, "step": 2820 }, { "epoch": 35.822784810126585, "grad_norm": 1.2891746759414673, "learning_rate": 9.946442149389497e-05, "loss": 0.0399, "step": 2830 }, { "epoch": 35.949367088607595, "grad_norm": 1.1242961883544922, "learning_rate": 9.945634599128139e-05, "loss": 0.0421, "step": 2840 }, { "epoch": 36.075949367088604, "grad_norm": 1.2997679710388184, "learning_rate": 9.944821039454402e-05, "loss": 0.0423, "step": 2850 }, { "epoch": 36.20253164556962, "grad_norm": 1.3160228729248047, "learning_rate": 9.944001471356835e-05, "loss": 0.0499, "step": 2860 }, { "epoch": 36.32911392405063, "grad_norm": 1.4972009658813477, "learning_rate": 9.94317589583129e-05, "loss": 0.0455, "step": 2870 }, { "epoch": 36.45569620253165, "grad_norm": 1.5283229351043701, "learning_rate": 9.942344313880922e-05, "loss": 0.0418, "step": 2880 }, { "epoch": 36.58227848101266, "grad_norm": 1.4833893775939941, "learning_rate": 9.941506726516179e-05, "loss": 0.0456, "step": 2890 }, { "epoch": 36.70886075949367, "grad_norm": 1.4040194749832153, "learning_rate": 9.94066313475481e-05, "loss": 0.0397, "step": 2900 }, { "epoch": 36.835443037974684, "grad_norm": 1.2494182586669922, "learning_rate": 9.939813539621857e-05, "loss": 0.0389, "step": 2910 }, { "epoch": 36.962025316455694, "grad_norm": 1.3371855020523071, "learning_rate": 9.93895794214966e-05, "loss": 0.0441, "step": 2920 }, { "epoch": 37.08860759493671, "grad_norm": 1.2104462385177612, "learning_rate": 9.938096343377852e-05, "loss": 0.0454, "step": 2930 }, { "epoch": 37.21518987341772, "grad_norm": 0.9761514663696289, "learning_rate": 9.937228744353353e-05, "loss": 0.0486, "step": 2940 }, { "epoch": 37.34177215189873, "grad_norm": 1.3859795331954956, "learning_rate": 9.936355146130379e-05, "loss": 0.0406, "step": 2950 }, { "epoch": 37.46835443037975, "grad_norm": 1.19649076461792, "learning_rate": 9.935475549770435e-05, "loss": 0.0428, "step": 2960 }, { "epoch": 37.59493670886076, "grad_norm": 1.267019271850586, "learning_rate": 9.934589956342315e-05, "loss": 0.0401, "step": 2970 }, { "epoch": 37.721518987341774, "grad_norm": 1.3906549215316772, "learning_rate": 9.933698366922093e-05, "loss": 0.0383, "step": 2980 }, { "epoch": 37.848101265822784, "grad_norm": 1.4417158365249634, "learning_rate": 9.93280078259314e-05, "loss": 0.0395, "step": 2990 }, { "epoch": 37.9746835443038, "grad_norm": 1.3787323236465454, "learning_rate": 9.931897204446104e-05, "loss": 0.0404, "step": 3000 }, { "epoch": 38.10126582278481, "grad_norm": 1.3218048810958862, "learning_rate": 9.930987633578915e-05, "loss": 0.0385, "step": 3010 }, { "epoch": 38.22784810126582, "grad_norm": 1.5283139944076538, "learning_rate": 9.93007207109679e-05, "loss": 0.0402, "step": 3020 }, { "epoch": 38.35443037974684, "grad_norm": 1.3579915761947632, "learning_rate": 9.929150518112224e-05, "loss": 0.0374, "step": 3030 }, { "epoch": 38.48101265822785, "grad_norm": 1.3059357404708862, "learning_rate": 9.928222975744991e-05, "loss": 0.0406, "step": 3040 }, { "epoch": 38.607594936708864, "grad_norm": 1.324473261833191, "learning_rate": 9.92728944512214e-05, "loss": 0.0372, "step": 3050 }, { "epoch": 38.734177215189874, "grad_norm": 1.4157681465148926, "learning_rate": 9.926349927378001e-05, "loss": 0.0397, "step": 3060 }, { "epoch": 38.860759493670884, "grad_norm": 1.013658046722412, "learning_rate": 9.925404423654174e-05, "loss": 0.0373, "step": 3070 }, { "epoch": 38.9873417721519, "grad_norm": 1.1138534545898438, "learning_rate": 9.924452935099537e-05, "loss": 0.0392, "step": 3080 }, { "epoch": 39.11392405063291, "grad_norm": 1.521822214126587, "learning_rate": 9.92349546287024e-05, "loss": 0.0435, "step": 3090 }, { "epoch": 39.24050632911393, "grad_norm": 1.5652226209640503, "learning_rate": 9.9225320081297e-05, "loss": 0.046, "step": 3100 }, { "epoch": 39.36708860759494, "grad_norm": 1.5923287868499756, "learning_rate": 9.921562572048606e-05, "loss": 0.0413, "step": 3110 }, { "epoch": 39.49367088607595, "grad_norm": 1.4563900232315063, "learning_rate": 9.920587155804913e-05, "loss": 0.04, "step": 3120 }, { "epoch": 39.620253164556964, "grad_norm": 0.9785522222518921, "learning_rate": 9.919605760583845e-05, "loss": 0.0402, "step": 3130 }, { "epoch": 39.74683544303797, "grad_norm": 1.4076848030090332, "learning_rate": 9.91861838757789e-05, "loss": 0.0401, "step": 3140 }, { "epoch": 39.87341772151899, "grad_norm": 1.4128766059875488, "learning_rate": 9.917625037986798e-05, "loss": 0.0394, "step": 3150 }, { "epoch": 40.0, "grad_norm": 1.1887348890304565, "learning_rate": 9.916625713017583e-05, "loss": 0.0382, "step": 3160 }, { "epoch": 40.12658227848101, "grad_norm": 1.21052086353302, "learning_rate": 9.915620413884519e-05, "loss": 0.0369, "step": 3170 }, { "epoch": 40.25316455696203, "grad_norm": 1.1511962413787842, "learning_rate": 9.914609141809139e-05, "loss": 0.0365, "step": 3180 }, { "epoch": 40.379746835443036, "grad_norm": 1.4313950538635254, "learning_rate": 9.913591898020235e-05, "loss": 0.0401, "step": 3190 }, { "epoch": 40.50632911392405, "grad_norm": 1.5324219465255737, "learning_rate": 9.912568683753853e-05, "loss": 0.0405, "step": 3200 }, { "epoch": 40.63291139240506, "grad_norm": 1.2750862836837769, "learning_rate": 9.911539500253295e-05, "loss": 0.039, "step": 3210 }, { "epoch": 40.75949367088607, "grad_norm": 1.2275537252426147, "learning_rate": 9.910504348769118e-05, "loss": 0.0366, "step": 3220 }, { "epoch": 40.88607594936709, "grad_norm": 1.1674925088882446, "learning_rate": 9.909463230559127e-05, "loss": 0.0356, "step": 3230 }, { "epoch": 41.0126582278481, "grad_norm": 1.239423394203186, "learning_rate": 9.908416146888376e-05, "loss": 0.0392, "step": 3240 }, { "epoch": 41.139240506329116, "grad_norm": 1.2494473457336426, "learning_rate": 9.907363099029175e-05, "loss": 0.0371, "step": 3250 }, { "epoch": 41.265822784810126, "grad_norm": 1.2085329294204712, "learning_rate": 9.906304088261073e-05, "loss": 0.042, "step": 3260 }, { "epoch": 41.392405063291136, "grad_norm": 1.3277506828308105, "learning_rate": 9.905239115870872e-05, "loss": 0.037, "step": 3270 }, { "epoch": 41.51898734177215, "grad_norm": 1.2171432971954346, "learning_rate": 9.90416818315261e-05, "loss": 0.0381, "step": 3280 }, { "epoch": 41.64556962025316, "grad_norm": 1.1145105361938477, "learning_rate": 9.903091291407573e-05, "loss": 0.0452, "step": 3290 }, { "epoch": 41.77215189873418, "grad_norm": 1.2240761518478394, "learning_rate": 9.902008441944286e-05, "loss": 0.0367, "step": 3300 }, { "epoch": 41.89873417721519, "grad_norm": 1.2811228036880493, "learning_rate": 9.900919636078512e-05, "loss": 0.0384, "step": 3310 }, { "epoch": 42.0253164556962, "grad_norm": 1.2527257204055786, "learning_rate": 9.899824875133255e-05, "loss": 0.0433, "step": 3320 }, { "epoch": 42.151898734177216, "grad_norm": 1.2986490726470947, "learning_rate": 9.898724160438749e-05, "loss": 0.0372, "step": 3330 }, { "epoch": 42.278481012658226, "grad_norm": 0.9297201633453369, "learning_rate": 9.89761749333247e-05, "loss": 0.0388, "step": 3340 }, { "epoch": 42.40506329113924, "grad_norm": 1.063752293586731, "learning_rate": 9.896504875159122e-05, "loss": 0.0363, "step": 3350 }, { "epoch": 42.53164556962025, "grad_norm": 1.127341866493225, "learning_rate": 9.89538630727064e-05, "loss": 0.0367, "step": 3360 }, { "epoch": 42.65822784810126, "grad_norm": 1.2588235139846802, "learning_rate": 9.894261791026189e-05, "loss": 0.0412, "step": 3370 }, { "epoch": 42.78481012658228, "grad_norm": 1.1755564212799072, "learning_rate": 9.893131327792165e-05, "loss": 0.0355, "step": 3380 }, { "epoch": 42.91139240506329, "grad_norm": 1.2799650430679321, "learning_rate": 9.891994918942182e-05, "loss": 0.0341, "step": 3390 }, { "epoch": 43.037974683544306, "grad_norm": 1.15727961063385, "learning_rate": 9.890852565857092e-05, "loss": 0.0425, "step": 3400 }, { "epoch": 43.164556962025316, "grad_norm": 1.231336236000061, "learning_rate": 9.889704269924954e-05, "loss": 0.0327, "step": 3410 }, { "epoch": 43.29113924050633, "grad_norm": 1.1485258340835571, "learning_rate": 9.888550032541059e-05, "loss": 0.033, "step": 3420 }, { "epoch": 43.41772151898734, "grad_norm": 1.2472343444824219, "learning_rate": 9.887389855107916e-05, "loss": 0.0355, "step": 3430 }, { "epoch": 43.54430379746835, "grad_norm": 1.0023990869522095, "learning_rate": 9.886223739035248e-05, "loss": 0.0411, "step": 3440 }, { "epoch": 43.67088607594937, "grad_norm": 1.2424426078796387, "learning_rate": 9.885051685739997e-05, "loss": 0.0379, "step": 3450 }, { "epoch": 43.79746835443038, "grad_norm": 1.1531319618225098, "learning_rate": 9.883873696646316e-05, "loss": 0.0419, "step": 3460 }, { "epoch": 43.924050632911396, "grad_norm": 1.1577609777450562, "learning_rate": 9.882689773185575e-05, "loss": 0.0324, "step": 3470 }, { "epoch": 44.050632911392405, "grad_norm": 1.1094541549682617, "learning_rate": 9.881499916796353e-05, "loss": 0.0387, "step": 3480 }, { "epoch": 44.177215189873415, "grad_norm": 1.2745198011398315, "learning_rate": 9.880304128924434e-05, "loss": 0.0349, "step": 3490 }, { "epoch": 44.30379746835443, "grad_norm": 1.2117058038711548, "learning_rate": 9.879102411022817e-05, "loss": 0.0364, "step": 3500 }, { "epoch": 44.43037974683544, "grad_norm": 1.0685536861419678, "learning_rate": 9.877894764551703e-05, "loss": 0.0328, "step": 3510 }, { "epoch": 44.55696202531646, "grad_norm": 1.071201205253601, "learning_rate": 9.876681190978494e-05, "loss": 0.0367, "step": 3520 }, { "epoch": 44.68354430379747, "grad_norm": 1.2042531967163086, "learning_rate": 9.875461691777797e-05, "loss": 0.0373, "step": 3530 }, { "epoch": 44.81012658227848, "grad_norm": 1.3120919466018677, "learning_rate": 9.874236268431417e-05, "loss": 0.0351, "step": 3540 }, { "epoch": 44.936708860759495, "grad_norm": 1.592692255973816, "learning_rate": 9.873004922428361e-05, "loss": 0.0358, "step": 3550 }, { "epoch": 45.063291139240505, "grad_norm": 1.0386817455291748, "learning_rate": 9.871767655264829e-05, "loss": 0.0378, "step": 3560 }, { "epoch": 45.18987341772152, "grad_norm": 1.1642653942108154, "learning_rate": 9.87052446844422e-05, "loss": 0.0353, "step": 3570 }, { "epoch": 45.31645569620253, "grad_norm": 1.16920006275177, "learning_rate": 9.869275363477122e-05, "loss": 0.0379, "step": 3580 }, { "epoch": 45.44303797468354, "grad_norm": 1.000443458557129, "learning_rate": 9.868020341881312e-05, "loss": 0.0353, "step": 3590 }, { "epoch": 45.56962025316456, "grad_norm": 1.044503927230835, "learning_rate": 9.866759405181765e-05, "loss": 0.0382, "step": 3600 }, { "epoch": 45.69620253164557, "grad_norm": 1.147904396057129, "learning_rate": 9.865492554910633e-05, "loss": 0.0345, "step": 3610 }, { "epoch": 45.822784810126585, "grad_norm": 1.0307468175888062, "learning_rate": 9.864219792607262e-05, "loss": 0.0341, "step": 3620 }, { "epoch": 45.949367088607595, "grad_norm": 1.0161232948303223, "learning_rate": 9.862941119818177e-05, "loss": 0.034, "step": 3630 }, { "epoch": 46.075949367088604, "grad_norm": 0.95307856798172, "learning_rate": 9.861656538097086e-05, "loss": 0.0345, "step": 3640 }, { "epoch": 46.20253164556962, "grad_norm": 1.0072270631790161, "learning_rate": 9.860366049004877e-05, "loss": 0.0364, "step": 3650 }, { "epoch": 46.32911392405063, "grad_norm": 0.8868104219436646, "learning_rate": 9.859069654109615e-05, "loss": 0.035, "step": 3660 }, { "epoch": 46.45569620253165, "grad_norm": 1.1680771112442017, "learning_rate": 9.857767354986545e-05, "loss": 0.0351, "step": 3670 }, { "epoch": 46.58227848101266, "grad_norm": 1.096194863319397, "learning_rate": 9.856459153218078e-05, "loss": 0.033, "step": 3680 }, { "epoch": 46.70886075949367, "grad_norm": 1.1736159324645996, "learning_rate": 9.855145050393808e-05, "loss": 0.0348, "step": 3690 }, { "epoch": 46.835443037974684, "grad_norm": 1.0893644094467163, "learning_rate": 9.85382504811049e-05, "loss": 0.035, "step": 3700 }, { "epoch": 46.962025316455694, "grad_norm": 1.258373737335205, "learning_rate": 9.852499147972054e-05, "loss": 0.0359, "step": 3710 }, { "epoch": 47.08860759493671, "grad_norm": 1.0242760181427002, "learning_rate": 9.851167351589592e-05, "loss": 0.0351, "step": 3720 }, { "epoch": 47.21518987341772, "grad_norm": 1.1958013772964478, "learning_rate": 9.849829660581363e-05, "loss": 0.0342, "step": 3730 }, { "epoch": 47.34177215189873, "grad_norm": 1.050252914428711, "learning_rate": 9.848486076572787e-05, "loss": 0.0345, "step": 3740 }, { "epoch": 47.46835443037975, "grad_norm": 1.0325924158096313, "learning_rate": 9.847136601196446e-05, "loss": 0.0362, "step": 3750 }, { "epoch": 47.59493670886076, "grad_norm": 0.8337277770042419, "learning_rate": 9.845781236092078e-05, "loss": 0.0341, "step": 3760 }, { "epoch": 47.721518987341774, "grad_norm": 0.8350520730018616, "learning_rate": 9.844419982906583e-05, "loss": 0.0342, "step": 3770 }, { "epoch": 47.848101265822784, "grad_norm": 1.1395275592803955, "learning_rate": 9.843052843294008e-05, "loss": 0.0319, "step": 3780 }, { "epoch": 47.9746835443038, "grad_norm": 1.046377420425415, "learning_rate": 9.841679818915559e-05, "loss": 0.0364, "step": 3790 }, { "epoch": 48.10126582278481, "grad_norm": 1.0854600667953491, "learning_rate": 9.840300911439591e-05, "loss": 0.0359, "step": 3800 }, { "epoch": 48.22784810126582, "grad_norm": 1.0695265531539917, "learning_rate": 9.838916122541603e-05, "loss": 0.0312, "step": 3810 }, { "epoch": 48.35443037974684, "grad_norm": 1.0466622114181519, "learning_rate": 9.837525453904246e-05, "loss": 0.0344, "step": 3820 }, { "epoch": 48.48101265822785, "grad_norm": 1.0547373294830322, "learning_rate": 9.836128907217314e-05, "loss": 0.0349, "step": 3830 }, { "epoch": 48.607594936708864, "grad_norm": 0.8874145150184631, "learning_rate": 9.834726484177743e-05, "loss": 0.0331, "step": 3840 }, { "epoch": 48.734177215189874, "grad_norm": 0.8660915493965149, "learning_rate": 9.833318186489609e-05, "loss": 0.0328, "step": 3850 }, { "epoch": 48.860759493670884, "grad_norm": 1.0460058450698853, "learning_rate": 9.831904015864126e-05, "loss": 0.0332, "step": 3860 }, { "epoch": 48.9873417721519, "grad_norm": 1.1859245300292969, "learning_rate": 9.830483974019645e-05, "loss": 0.0366, "step": 3870 }, { "epoch": 49.11392405063291, "grad_norm": 1.153059720993042, "learning_rate": 9.82905806268165e-05, "loss": 0.0329, "step": 3880 }, { "epoch": 49.24050632911393, "grad_norm": 1.070103645324707, "learning_rate": 9.82762628358276e-05, "loss": 0.0315, "step": 3890 }, { "epoch": 49.36708860759494, "grad_norm": 1.0349938869476318, "learning_rate": 9.826188638462718e-05, "loss": 0.0321, "step": 3900 }, { "epoch": 49.49367088607595, "grad_norm": 1.0929532051086426, "learning_rate": 9.824745129068402e-05, "loss": 0.0359, "step": 3910 }, { "epoch": 49.620253164556964, "grad_norm": 0.9627532958984375, "learning_rate": 9.82329575715381e-05, "loss": 0.0389, "step": 3920 }, { "epoch": 49.74683544303797, "grad_norm": 0.9822688102722168, "learning_rate": 9.821840524480066e-05, "loss": 0.0293, "step": 3930 }, { "epoch": 49.87341772151899, "grad_norm": 1.000666856765747, "learning_rate": 9.820379432815414e-05, "loss": 0.0328, "step": 3940 }, { "epoch": 50.0, "grad_norm": 1.2742639780044556, "learning_rate": 9.81891248393522e-05, "loss": 0.032, "step": 3950 }, { "epoch": 50.12658227848101, "grad_norm": 0.9509206414222717, "learning_rate": 9.817439679621963e-05, "loss": 0.0351, "step": 3960 }, { "epoch": 50.25316455696203, "grad_norm": 1.0261927843093872, "learning_rate": 9.815961021665243e-05, "loss": 0.0336, "step": 3970 }, { "epoch": 50.379746835443036, "grad_norm": 1.0053622722625732, "learning_rate": 9.814476511861763e-05, "loss": 0.0401, "step": 3980 }, { "epoch": 50.50632911392405, "grad_norm": 1.0607666969299316, "learning_rate": 9.812986152015348e-05, "loss": 0.0373, "step": 3990 }, { "epoch": 50.63291139240506, "grad_norm": 1.0285210609436035, "learning_rate": 9.811489943936922e-05, "loss": 0.0319, "step": 4000 }, { "epoch": 50.75949367088607, "grad_norm": 1.1191110610961914, "learning_rate": 9.809987889444522e-05, "loss": 0.0355, "step": 4010 }, { "epoch": 50.88607594936709, "grad_norm": 1.038939118385315, "learning_rate": 9.808479990363282e-05, "loss": 0.0324, "step": 4020 }, { "epoch": 51.0126582278481, "grad_norm": 1.0432976484298706, "learning_rate": 9.806966248525445e-05, "loss": 0.0312, "step": 4030 }, { "epoch": 51.139240506329116, "grad_norm": 1.0056166648864746, "learning_rate": 9.805446665770348e-05, "loss": 0.0312, "step": 4040 }, { "epoch": 51.265822784810126, "grad_norm": 1.0049254894256592, "learning_rate": 9.803921243944429e-05, "loss": 0.0295, "step": 4050 }, { "epoch": 51.392405063291136, "grad_norm": 1.2176225185394287, "learning_rate": 9.802389984901218e-05, "loss": 0.0329, "step": 4060 }, { "epoch": 51.51898734177215, "grad_norm": 1.1809742450714111, "learning_rate": 9.80085289050134e-05, "loss": 0.0313, "step": 4070 }, { "epoch": 51.64556962025316, "grad_norm": 0.9751587510108948, "learning_rate": 9.799309962612508e-05, "loss": 0.0326, "step": 4080 }, { "epoch": 51.77215189873418, "grad_norm": 0.9921903610229492, "learning_rate": 9.797761203109527e-05, "loss": 0.0317, "step": 4090 }, { "epoch": 51.89873417721519, "grad_norm": 1.0834380388259888, "learning_rate": 9.796206613874283e-05, "loss": 0.0324, "step": 4100 }, { "epoch": 52.0253164556962, "grad_norm": 1.1114449501037598, "learning_rate": 9.794646196795754e-05, "loss": 0.0312, "step": 4110 }, { "epoch": 52.151898734177216, "grad_norm": 1.0670719146728516, "learning_rate": 9.793079953769987e-05, "loss": 0.0314, "step": 4120 }, { "epoch": 52.278481012658226, "grad_norm": 0.8970901370048523, "learning_rate": 9.79150788670012e-05, "loss": 0.0283, "step": 4130 }, { "epoch": 52.40506329113924, "grad_norm": 0.7845336198806763, "learning_rate": 9.78992999749636e-05, "loss": 0.0354, "step": 4140 }, { "epoch": 52.53164556962025, "grad_norm": 0.9588229656219482, "learning_rate": 9.788346288075994e-05, "loss": 0.0294, "step": 4150 }, { "epoch": 52.65822784810126, "grad_norm": 1.0077062845230103, "learning_rate": 9.786756760363373e-05, "loss": 0.0345, "step": 4160 }, { "epoch": 52.78481012658228, "grad_norm": 1.0113155841827393, "learning_rate": 9.78516141628993e-05, "loss": 0.0307, "step": 4170 }, { "epoch": 52.91139240506329, "grad_norm": 0.8903290033340454, "learning_rate": 9.783560257794154e-05, "loss": 0.0326, "step": 4180 }, { "epoch": 53.037974683544306, "grad_norm": 0.8688770532608032, "learning_rate": 9.781953286821603e-05, "loss": 0.0287, "step": 4190 }, { "epoch": 53.164556962025316, "grad_norm": 1.106698751449585, "learning_rate": 9.780340505324901e-05, "loss": 0.0298, "step": 4200 }, { "epoch": 53.29113924050633, "grad_norm": 0.9449043273925781, "learning_rate": 9.778721915263727e-05, "loss": 0.03, "step": 4210 }, { "epoch": 53.41772151898734, "grad_norm": 0.9414053559303284, "learning_rate": 9.777097518604824e-05, "loss": 0.0279, "step": 4220 }, { "epoch": 53.54430379746835, "grad_norm": 0.9773527979850769, "learning_rate": 9.775467317321984e-05, "loss": 0.0268, "step": 4230 }, { "epoch": 53.67088607594937, "grad_norm": 0.8971982002258301, "learning_rate": 9.773831313396055e-05, "loss": 0.0324, "step": 4240 }, { "epoch": 53.79746835443038, "grad_norm": 0.9644914865493774, "learning_rate": 9.77218950881494e-05, "loss": 0.0303, "step": 4250 }, { "epoch": 53.924050632911396, "grad_norm": 1.1617722511291504, "learning_rate": 9.770541905573583e-05, "loss": 0.0317, "step": 4260 }, { "epoch": 54.050632911392405, "grad_norm": 0.9576834440231323, "learning_rate": 9.768888505673976e-05, "loss": 0.0272, "step": 4270 }, { "epoch": 54.177215189873415, "grad_norm": 0.9766815900802612, "learning_rate": 9.767229311125162e-05, "loss": 0.0294, "step": 4280 }, { "epoch": 54.30379746835443, "grad_norm": 1.011382818222046, "learning_rate": 9.765564323943211e-05, "loss": 0.0309, "step": 4290 }, { "epoch": 54.43037974683544, "grad_norm": 1.0881396532058716, "learning_rate": 9.763893546151244e-05, "loss": 0.0308, "step": 4300 }, { "epoch": 54.55696202531646, "grad_norm": 0.8235928416252136, "learning_rate": 9.762216979779412e-05, "loss": 0.0323, "step": 4310 }, { "epoch": 54.68354430379747, "grad_norm": 0.9250511527061462, "learning_rate": 9.760534626864902e-05, "loss": 0.0282, "step": 4320 }, { "epoch": 54.81012658227848, "grad_norm": 0.8953907489776611, "learning_rate": 9.758846489451931e-05, "loss": 0.0321, "step": 4330 }, { "epoch": 54.936708860759495, "grad_norm": 0.9998877048492432, "learning_rate": 9.757152569591748e-05, "loss": 0.031, "step": 4340 }, { "epoch": 55.063291139240505, "grad_norm": 1.126808762550354, "learning_rate": 9.75545286934262e-05, "loss": 0.0301, "step": 4350 }, { "epoch": 55.18987341772152, "grad_norm": 1.1914352178573608, "learning_rate": 9.753747390769847e-05, "loss": 0.028, "step": 4360 }, { "epoch": 55.31645569620253, "grad_norm": 0.8055485486984253, "learning_rate": 9.752036135945744e-05, "loss": 0.0311, "step": 4370 }, { "epoch": 55.44303797468354, "grad_norm": 1.0468565225601196, "learning_rate": 9.750319106949649e-05, "loss": 0.0318, "step": 4380 }, { "epoch": 55.56962025316456, "grad_norm": 1.2089457511901855, "learning_rate": 9.748596305867913e-05, "loss": 0.0298, "step": 4390 }, { "epoch": 55.69620253164557, "grad_norm": 0.946148693561554, "learning_rate": 9.746867734793903e-05, "loss": 0.0318, "step": 4400 }, { "epoch": 55.822784810126585, "grad_norm": 0.7561853528022766, "learning_rate": 9.745133395827993e-05, "loss": 0.033, "step": 4410 }, { "epoch": 55.949367088607595, "grad_norm": 0.7411541938781738, "learning_rate": 9.743393291077572e-05, "loss": 0.0269, "step": 4420 }, { "epoch": 56.075949367088604, "grad_norm": 1.028990626335144, "learning_rate": 9.741647422657028e-05, "loss": 0.032, "step": 4430 }, { "epoch": 56.20253164556962, "grad_norm": 0.8544272184371948, "learning_rate": 9.739895792687758e-05, "loss": 0.0281, "step": 4440 }, { "epoch": 56.32911392405063, "grad_norm": 1.2221800088882446, "learning_rate": 9.738138403298157e-05, "loss": 0.0309, "step": 4450 }, { "epoch": 56.45569620253165, "grad_norm": 1.0337680578231812, "learning_rate": 9.736375256623619e-05, "loss": 0.0265, "step": 4460 }, { "epoch": 56.58227848101266, "grad_norm": 0.9238623976707458, "learning_rate": 9.734606354806533e-05, "loss": 0.0277, "step": 4470 }, { "epoch": 56.70886075949367, "grad_norm": 0.9508515000343323, "learning_rate": 9.73283169999628e-05, "loss": 0.0314, "step": 4480 }, { "epoch": 56.835443037974684, "grad_norm": 1.012923240661621, "learning_rate": 9.731051294349238e-05, "loss": 0.0283, "step": 4490 }, { "epoch": 56.962025316455694, "grad_norm": 0.9307712316513062, "learning_rate": 9.729265140028762e-05, "loss": 0.0327, "step": 4500 }, { "epoch": 57.08860759493671, "grad_norm": 0.8798244595527649, "learning_rate": 9.727473239205201e-05, "loss": 0.0258, "step": 4510 }, { "epoch": 57.21518987341772, "grad_norm": 0.9688798785209656, "learning_rate": 9.725675594055883e-05, "loss": 0.0262, "step": 4520 }, { "epoch": 57.34177215189873, "grad_norm": 0.87884122133255, "learning_rate": 9.723872206765116e-05, "loss": 0.0272, "step": 4530 }, { "epoch": 57.46835443037975, "grad_norm": 1.0193175077438354, "learning_rate": 9.722063079524185e-05, "loss": 0.0275, "step": 4540 }, { "epoch": 57.59493670886076, "grad_norm": 1.0105295181274414, "learning_rate": 9.720248214531351e-05, "loss": 0.0289, "step": 4550 }, { "epoch": 57.721518987341774, "grad_norm": 0.9404880404472351, "learning_rate": 9.718427613991848e-05, "loss": 0.0309, "step": 4560 }, { "epoch": 57.848101265822784, "grad_norm": 0.7972176671028137, "learning_rate": 9.716601280117873e-05, "loss": 0.0252, "step": 4570 }, { "epoch": 57.9746835443038, "grad_norm": 0.8033009767532349, "learning_rate": 9.714769215128596e-05, "loss": 0.031, "step": 4580 }, { "epoch": 58.10126582278481, "grad_norm": 0.8468573093414307, "learning_rate": 9.712931421250152e-05, "loss": 0.0319, "step": 4590 }, { "epoch": 58.22784810126582, "grad_norm": 0.9864641427993774, "learning_rate": 9.711087900715627e-05, "loss": 0.0285, "step": 4600 }, { "epoch": 58.35443037974684, "grad_norm": 0.9878087639808655, "learning_rate": 9.709238655765078e-05, "loss": 0.0311, "step": 4610 }, { "epoch": 58.48101265822785, "grad_norm": 0.7539659142494202, "learning_rate": 9.707383688645511e-05, "loss": 0.0277, "step": 4620 }, { "epoch": 58.607594936708864, "grad_norm": 1.0059280395507812, "learning_rate": 9.705523001610883e-05, "loss": 0.0288, "step": 4630 }, { "epoch": 58.734177215189874, "grad_norm": 0.8124339580535889, "learning_rate": 9.703656596922107e-05, "loss": 0.0269, "step": 4640 }, { "epoch": 58.860759493670884, "grad_norm": 0.8640549182891846, "learning_rate": 9.70178447684704e-05, "loss": 0.0283, "step": 4650 }, { "epoch": 58.9873417721519, "grad_norm": 0.7951909899711609, "learning_rate": 9.699906643660483e-05, "loss": 0.0292, "step": 4660 }, { "epoch": 59.11392405063291, "grad_norm": 0.7255719900131226, "learning_rate": 9.698023099644185e-05, "loss": 0.0293, "step": 4670 }, { "epoch": 59.24050632911393, "grad_norm": 0.8916126489639282, "learning_rate": 9.696133847086823e-05, "loss": 0.0267, "step": 4680 }, { "epoch": 59.36708860759494, "grad_norm": 0.8087840676307678, "learning_rate": 9.694238888284022e-05, "loss": 0.0332, "step": 4690 }, { "epoch": 59.49367088607595, "grad_norm": 0.7926869988441467, "learning_rate": 9.692338225538333e-05, "loss": 0.0272, "step": 4700 }, { "epoch": 59.620253164556964, "grad_norm": 0.7594752311706543, "learning_rate": 9.690431861159241e-05, "loss": 0.0279, "step": 4710 }, { "epoch": 59.74683544303797, "grad_norm": 0.9368493556976318, "learning_rate": 9.688519797463161e-05, "loss": 0.0291, "step": 4720 }, { "epoch": 59.87341772151899, "grad_norm": 0.9647102952003479, "learning_rate": 9.686602036773426e-05, "loss": 0.0283, "step": 4730 }, { "epoch": 60.0, "grad_norm": 0.8646212220191956, "learning_rate": 9.684678581420302e-05, "loss": 0.0267, "step": 4740 }, { "epoch": 60.12658227848101, "grad_norm": 0.9467335343360901, "learning_rate": 9.682749433740962e-05, "loss": 0.0301, "step": 4750 }, { "epoch": 60.25316455696203, "grad_norm": 0.7284456491470337, "learning_rate": 9.680814596079507e-05, "loss": 0.0288, "step": 4760 }, { "epoch": 60.379746835443036, "grad_norm": 0.9250410199165344, "learning_rate": 9.678874070786945e-05, "loss": 0.0306, "step": 4770 }, { "epoch": 60.50632911392405, "grad_norm": 0.942884087562561, "learning_rate": 9.676927860221199e-05, "loss": 0.0308, "step": 4780 }, { "epoch": 60.63291139240506, "grad_norm": 0.8738928437232971, "learning_rate": 9.674975966747097e-05, "loss": 0.03, "step": 4790 }, { "epoch": 60.75949367088607, "grad_norm": 0.75931715965271, "learning_rate": 9.673018392736374e-05, "loss": 0.0256, "step": 4800 }, { "epoch": 60.88607594936709, "grad_norm": 0.8450740575790405, "learning_rate": 9.671055140567667e-05, "loss": 0.033, "step": 4810 }, { "epoch": 61.0126582278481, "grad_norm": 0.7268543839454651, "learning_rate": 9.669086212626511e-05, "loss": 0.0278, "step": 4820 }, { "epoch": 61.139240506329116, "grad_norm": 0.9095327854156494, "learning_rate": 9.667111611305341e-05, "loss": 0.0288, "step": 4830 }, { "epoch": 61.265822784810126, "grad_norm": 0.8592759370803833, "learning_rate": 9.665131339003486e-05, "loss": 0.0292, "step": 4840 }, { "epoch": 61.392405063291136, "grad_norm": 0.8449200987815857, "learning_rate": 9.663145398127158e-05, "loss": 0.0257, "step": 4850 }, { "epoch": 61.51898734177215, "grad_norm": 0.7571454048156738, "learning_rate": 9.661153791089467e-05, "loss": 0.0298, "step": 4860 }, { "epoch": 61.64556962025316, "grad_norm": 0.5796782970428467, "learning_rate": 9.659156520310402e-05, "loss": 0.0293, "step": 4870 }, { "epoch": 61.77215189873418, "grad_norm": 0.8151237368583679, "learning_rate": 9.657153588216835e-05, "loss": 0.026, "step": 4880 }, { "epoch": 61.89873417721519, "grad_norm": 0.7195709347724915, "learning_rate": 9.655144997242516e-05, "loss": 0.0267, "step": 4890 }, { "epoch": 62.0253164556962, "grad_norm": 0.7935230135917664, "learning_rate": 9.653130749828075e-05, "loss": 0.027, "step": 4900 }, { "epoch": 62.151898734177216, "grad_norm": 0.7262299060821533, "learning_rate": 9.65111084842101e-05, "loss": 0.0265, "step": 4910 }, { "epoch": 62.278481012658226, "grad_norm": 0.9027156829833984, "learning_rate": 9.649085295475695e-05, "loss": 0.0313, "step": 4920 }, { "epoch": 62.40506329113924, "grad_norm": 1.0961835384368896, "learning_rate": 9.647054093453365e-05, "loss": 0.0309, "step": 4930 }, { "epoch": 62.53164556962025, "grad_norm": 0.7160388231277466, "learning_rate": 9.645017244822123e-05, "loss": 0.0276, "step": 4940 }, { "epoch": 62.65822784810126, "grad_norm": 0.8716493248939514, "learning_rate": 9.642974752056931e-05, "loss": 0.0318, "step": 4950 }, { "epoch": 62.78481012658228, "grad_norm": 0.8584858775138855, "learning_rate": 9.640926617639613e-05, "loss": 0.0301, "step": 4960 }, { "epoch": 62.91139240506329, "grad_norm": 0.7905063629150391, "learning_rate": 9.638872844058843e-05, "loss": 0.0279, "step": 4970 }, { "epoch": 63.037974683544306, "grad_norm": 0.9169068932533264, "learning_rate": 9.63681343381015e-05, "loss": 0.0288, "step": 4980 }, { "epoch": 63.164556962025316, "grad_norm": 0.8493922352790833, "learning_rate": 9.634748389395914e-05, "loss": 0.0266, "step": 4990 }, { "epoch": 63.29113924050633, "grad_norm": 0.8326159715652466, "learning_rate": 9.632677713325353e-05, "loss": 0.0291, "step": 5000 }, { "epoch": 63.41772151898734, "grad_norm": 0.7923072576522827, "learning_rate": 9.63060140811454e-05, "loss": 0.0272, "step": 5010 }, { "epoch": 63.54430379746835, "grad_norm": 0.8232184648513794, "learning_rate": 9.628519476286379e-05, "loss": 0.0302, "step": 5020 }, { "epoch": 63.67088607594937, "grad_norm": 0.814947247505188, "learning_rate": 9.626431920370612e-05, "loss": 0.0266, "step": 5030 }, { "epoch": 63.79746835443038, "grad_norm": 0.7652479410171509, "learning_rate": 9.624338742903819e-05, "loss": 0.0272, "step": 5040 }, { "epoch": 63.924050632911396, "grad_norm": 0.8352916240692139, "learning_rate": 9.622239946429406e-05, "loss": 0.0254, "step": 5050 }, { "epoch": 64.0506329113924, "grad_norm": 1.0361400842666626, "learning_rate": 9.620135533497609e-05, "loss": 0.0277, "step": 5060 }, { "epoch": 64.17721518987342, "grad_norm": 0.8133360743522644, "learning_rate": 9.61802550666549e-05, "loss": 0.0254, "step": 5070 }, { "epoch": 64.30379746835443, "grad_norm": 0.7156249284744263, "learning_rate": 9.615909868496928e-05, "loss": 0.0267, "step": 5080 }, { "epoch": 64.43037974683544, "grad_norm": 0.7597755789756775, "learning_rate": 9.613788621562622e-05, "loss": 0.0271, "step": 5090 }, { "epoch": 64.55696202531645, "grad_norm": 0.9343170523643494, "learning_rate": 9.611661768440093e-05, "loss": 0.0273, "step": 5100 }, { "epoch": 64.68354430379746, "grad_norm": 0.9620133638381958, "learning_rate": 9.609529311713661e-05, "loss": 0.0286, "step": 5110 }, { "epoch": 64.81012658227849, "grad_norm": 0.9584069848060608, "learning_rate": 9.607391253974466e-05, "loss": 0.0269, "step": 5120 }, { "epoch": 64.9367088607595, "grad_norm": 0.8843314051628113, "learning_rate": 9.605247597820448e-05, "loss": 0.0267, "step": 5130 }, { "epoch": 65.0632911392405, "grad_norm": 0.8985174298286438, "learning_rate": 9.603098345856354e-05, "loss": 0.0314, "step": 5140 }, { "epoch": 65.18987341772151, "grad_norm": 0.9333721995353699, "learning_rate": 9.600943500693725e-05, "loss": 0.0276, "step": 5150 }, { "epoch": 65.31645569620254, "grad_norm": 0.8594495058059692, "learning_rate": 9.598783064950902e-05, "loss": 0.0291, "step": 5160 }, { "epoch": 65.44303797468355, "grad_norm": 0.7762420177459717, "learning_rate": 9.596617041253018e-05, "loss": 0.0224, "step": 5170 }, { "epoch": 65.56962025316456, "grad_norm": 0.664027214050293, "learning_rate": 9.594445432231996e-05, "loss": 0.0226, "step": 5180 }, { "epoch": 65.69620253164557, "grad_norm": 0.7924083471298218, "learning_rate": 9.592268240526547e-05, "loss": 0.0265, "step": 5190 }, { "epoch": 65.82278481012658, "grad_norm": 0.777246356010437, "learning_rate": 9.590085468782162e-05, "loss": 0.0251, "step": 5200 }, { "epoch": 65.9493670886076, "grad_norm": 0.6887280344963074, "learning_rate": 9.587897119651116e-05, "loss": 0.0264, "step": 5210 }, { "epoch": 66.07594936708861, "grad_norm": 0.6895592212677002, "learning_rate": 9.585703195792459e-05, "loss": 0.0256, "step": 5220 }, { "epoch": 66.20253164556962, "grad_norm": 0.6821097731590271, "learning_rate": 9.583503699872016e-05, "loss": 0.0246, "step": 5230 }, { "epoch": 66.32911392405063, "grad_norm": 0.6516174077987671, "learning_rate": 9.581298634562381e-05, "loss": 0.0266, "step": 5240 }, { "epoch": 66.45569620253164, "grad_norm": 0.6771568655967712, "learning_rate": 9.579088002542917e-05, "loss": 0.0256, "step": 5250 }, { "epoch": 66.58227848101266, "grad_norm": 0.7670681476593018, "learning_rate": 9.57687180649975e-05, "loss": 0.0292, "step": 5260 }, { "epoch": 66.70886075949367, "grad_norm": 0.6776984930038452, "learning_rate": 9.574650049125768e-05, "loss": 0.0274, "step": 5270 }, { "epoch": 66.83544303797468, "grad_norm": 0.8211712837219238, "learning_rate": 9.572422733120614e-05, "loss": 0.0247, "step": 5280 }, { "epoch": 66.9620253164557, "grad_norm": 0.905326247215271, "learning_rate": 9.570189861190689e-05, "loss": 0.0247, "step": 5290 }, { "epoch": 67.0886075949367, "grad_norm": 0.7723165154457092, "learning_rate": 9.56795143604914e-05, "loss": 0.0259, "step": 5300 }, { "epoch": 67.21518987341773, "grad_norm": 0.6348652243614197, "learning_rate": 9.565707460415869e-05, "loss": 0.0239, "step": 5310 }, { "epoch": 67.34177215189874, "grad_norm": 0.7317135334014893, "learning_rate": 9.563457937017515e-05, "loss": 0.0259, "step": 5320 }, { "epoch": 67.46835443037975, "grad_norm": 0.9059149622917175, "learning_rate": 9.56120286858746e-05, "loss": 0.0242, "step": 5330 }, { "epoch": 67.59493670886076, "grad_norm": 0.8256801962852478, "learning_rate": 9.558942257865829e-05, "loss": 0.0276, "step": 5340 }, { "epoch": 67.72151898734177, "grad_norm": 0.7369053959846497, "learning_rate": 9.556676107599472e-05, "loss": 0.0233, "step": 5350 }, { "epoch": 67.84810126582279, "grad_norm": 0.632125198841095, "learning_rate": 9.554404420541978e-05, "loss": 0.0258, "step": 5360 }, { "epoch": 67.9746835443038, "grad_norm": 0.8224673271179199, "learning_rate": 9.55212719945366e-05, "loss": 0.025, "step": 5370 }, { "epoch": 68.10126582278481, "grad_norm": 0.9176068305969238, "learning_rate": 9.549844447101559e-05, "loss": 0.0242, "step": 5380 }, { "epoch": 68.22784810126582, "grad_norm": 0.7687156796455383, "learning_rate": 9.547556166259433e-05, "loss": 0.0242, "step": 5390 }, { "epoch": 68.35443037974683, "grad_norm": 0.9324635863304138, "learning_rate": 9.545262359707756e-05, "loss": 0.0261, "step": 5400 }, { "epoch": 68.48101265822785, "grad_norm": 0.7737944722175598, "learning_rate": 9.542963030233724e-05, "loss": 0.0261, "step": 5410 }, { "epoch": 68.60759493670886, "grad_norm": 0.6625587940216064, "learning_rate": 9.540658180631237e-05, "loss": 0.0265, "step": 5420 }, { "epoch": 68.73417721518987, "grad_norm": 0.6853529214859009, "learning_rate": 9.538347813700904e-05, "loss": 0.0247, "step": 5430 }, { "epoch": 68.86075949367088, "grad_norm": 0.7818745374679565, "learning_rate": 9.536031932250036e-05, "loss": 0.0238, "step": 5440 }, { "epoch": 68.9873417721519, "grad_norm": 0.81193608045578, "learning_rate": 9.533710539092653e-05, "loss": 0.0248, "step": 5450 }, { "epoch": 69.11392405063292, "grad_norm": 0.6590650677680969, "learning_rate": 9.531383637049464e-05, "loss": 0.022, "step": 5460 }, { "epoch": 69.24050632911393, "grad_norm": 0.6158607602119446, "learning_rate": 9.529051228947875e-05, "loss": 0.0251, "step": 5470 }, { "epoch": 69.36708860759494, "grad_norm": 0.6618325710296631, "learning_rate": 9.52671331762198e-05, "loss": 0.0244, "step": 5480 }, { "epoch": 69.49367088607595, "grad_norm": 0.8733943700790405, "learning_rate": 9.524369905912565e-05, "loss": 0.0238, "step": 5490 }, { "epoch": 69.62025316455696, "grad_norm": 0.8164533972740173, "learning_rate": 9.522020996667092e-05, "loss": 0.0299, "step": 5500 }, { "epoch": 69.74683544303798, "grad_norm": 0.7591565251350403, "learning_rate": 9.519666592739709e-05, "loss": 0.0224, "step": 5510 }, { "epoch": 69.87341772151899, "grad_norm": 0.8175694942474365, "learning_rate": 9.517306696991241e-05, "loss": 0.0245, "step": 5520 }, { "epoch": 70.0, "grad_norm": 0.7970036864280701, "learning_rate": 9.51494131228918e-05, "loss": 0.0228, "step": 5530 }, { "epoch": 70.12658227848101, "grad_norm": 0.6952484846115112, "learning_rate": 9.512570441507695e-05, "loss": 0.0269, "step": 5540 }, { "epoch": 70.25316455696202, "grad_norm": 0.845785915851593, "learning_rate": 9.510194087527615e-05, "loss": 0.0257, "step": 5550 }, { "epoch": 70.37974683544304, "grad_norm": 0.820184051990509, "learning_rate": 9.507812253236435e-05, "loss": 0.0285, "step": 5560 }, { "epoch": 70.50632911392405, "grad_norm": 0.6924116015434265, "learning_rate": 9.505424941528309e-05, "loss": 0.026, "step": 5570 }, { "epoch": 70.63291139240506, "grad_norm": 0.7388971447944641, "learning_rate": 9.503032155304046e-05, "loss": 0.031, "step": 5580 }, { "epoch": 70.75949367088607, "grad_norm": 0.7298076152801514, "learning_rate": 9.500633897471106e-05, "loss": 0.0251, "step": 5590 }, { "epoch": 70.88607594936708, "grad_norm": 0.697441577911377, "learning_rate": 9.498230170943596e-05, "loss": 0.0302, "step": 5600 }, { "epoch": 71.0126582278481, "grad_norm": 0.6868351101875305, "learning_rate": 9.495820978642275e-05, "loss": 0.0277, "step": 5610 }, { "epoch": 71.13924050632912, "grad_norm": 0.6569933295249939, "learning_rate": 9.493406323494535e-05, "loss": 0.0259, "step": 5620 }, { "epoch": 71.26582278481013, "grad_norm": 0.6692668199539185, "learning_rate": 9.490986208434413e-05, "loss": 0.0264, "step": 5630 }, { "epoch": 71.39240506329114, "grad_norm": 0.7959006428718567, "learning_rate": 9.488560636402577e-05, "loss": 0.0228, "step": 5640 }, { "epoch": 71.51898734177215, "grad_norm": 0.6626192927360535, "learning_rate": 9.486129610346321e-05, "loss": 0.0248, "step": 5650 }, { "epoch": 71.64556962025317, "grad_norm": 0.6167832612991333, "learning_rate": 9.483693133219576e-05, "loss": 0.0229, "step": 5660 }, { "epoch": 71.77215189873418, "grad_norm": 0.6407638788223267, "learning_rate": 9.481251207982888e-05, "loss": 0.0284, "step": 5670 }, { "epoch": 71.89873417721519, "grad_norm": 0.7757551670074463, "learning_rate": 9.47880383760343e-05, "loss": 0.023, "step": 5680 }, { "epoch": 72.0253164556962, "grad_norm": 0.7104263305664062, "learning_rate": 9.476351025054983e-05, "loss": 0.027, "step": 5690 }, { "epoch": 72.15189873417721, "grad_norm": 0.6697545051574707, "learning_rate": 9.473892773317952e-05, "loss": 0.0211, "step": 5700 }, { "epoch": 72.27848101265823, "grad_norm": 0.7519274950027466, "learning_rate": 9.471429085379338e-05, "loss": 0.0257, "step": 5710 }, { "epoch": 72.40506329113924, "grad_norm": 0.8248083591461182, "learning_rate": 9.468959964232757e-05, "loss": 0.0223, "step": 5720 }, { "epoch": 72.53164556962025, "grad_norm": 0.6892272233963013, "learning_rate": 9.466485412878425e-05, "loss": 0.0299, "step": 5730 }, { "epoch": 72.65822784810126, "grad_norm": 0.8287972211837769, "learning_rate": 9.464005434323154e-05, "loss": 0.0253, "step": 5740 }, { "epoch": 72.78481012658227, "grad_norm": 0.6319558024406433, "learning_rate": 9.461520031580352e-05, "loss": 0.0229, "step": 5750 }, { "epoch": 72.9113924050633, "grad_norm": 0.7285277843475342, "learning_rate": 9.459029207670019e-05, "loss": 0.0234, "step": 5760 }, { "epoch": 73.0379746835443, "grad_norm": 0.5910432934761047, "learning_rate": 9.456532965618737e-05, "loss": 0.0231, "step": 5770 }, { "epoch": 73.16455696202532, "grad_norm": 0.6281303763389587, "learning_rate": 9.454031308459681e-05, "loss": 0.0247, "step": 5780 }, { "epoch": 73.29113924050633, "grad_norm": 0.703631579875946, "learning_rate": 9.451524239232595e-05, "loss": 0.0214, "step": 5790 }, { "epoch": 73.41772151898734, "grad_norm": 0.7577108144760132, "learning_rate": 9.449011760983809e-05, "loss": 0.0247, "step": 5800 }, { "epoch": 73.54430379746836, "grad_norm": 0.6084796786308289, "learning_rate": 9.446493876766218e-05, "loss": 0.0255, "step": 5810 }, { "epoch": 73.67088607594937, "grad_norm": 0.6596516966819763, "learning_rate": 9.44397058963929e-05, "loss": 0.0235, "step": 5820 }, { "epoch": 73.79746835443038, "grad_norm": 0.8232873678207397, "learning_rate": 9.441441902669056e-05, "loss": 0.0213, "step": 5830 }, { "epoch": 73.92405063291139, "grad_norm": 0.7986706495285034, "learning_rate": 9.43890781892811e-05, "loss": 0.0239, "step": 5840 }, { "epoch": 74.0506329113924, "grad_norm": 0.6486222147941589, "learning_rate": 9.436368341495603e-05, "loss": 0.0245, "step": 5850 }, { "epoch": 74.17721518987342, "grad_norm": 0.686710774898529, "learning_rate": 9.43382347345724e-05, "loss": 0.0253, "step": 5860 }, { "epoch": 74.30379746835443, "grad_norm": 0.6755886077880859, "learning_rate": 9.431273217905273e-05, "loss": 0.0223, "step": 5870 }, { "epoch": 74.43037974683544, "grad_norm": 0.7151454091072083, "learning_rate": 9.428717577938504e-05, "loss": 0.0236, "step": 5880 }, { "epoch": 74.55696202531645, "grad_norm": 0.7202918529510498, "learning_rate": 9.426156556662276e-05, "loss": 0.0189, "step": 5890 }, { "epoch": 74.68354430379746, "grad_norm": 0.7045045495033264, "learning_rate": 9.423590157188474e-05, "loss": 0.0228, "step": 5900 }, { "epoch": 74.81012658227849, "grad_norm": 0.686638593673706, "learning_rate": 9.421018382635513e-05, "loss": 0.0246, "step": 5910 }, { "epoch": 74.9367088607595, "grad_norm": 0.7572327852249146, "learning_rate": 9.418441236128343e-05, "loss": 0.0223, "step": 5920 }, { "epoch": 75.0632911392405, "grad_norm": 0.7390560507774353, "learning_rate": 9.41585872079844e-05, "loss": 0.0257, "step": 5930 }, { "epoch": 75.18987341772151, "grad_norm": 0.6820474863052368, "learning_rate": 9.413270839783802e-05, "loss": 0.0248, "step": 5940 }, { "epoch": 75.31645569620254, "grad_norm": 0.7889032959938049, "learning_rate": 9.41067759622895e-05, "loss": 0.026, "step": 5950 }, { "epoch": 75.44303797468355, "grad_norm": 0.770007848739624, "learning_rate": 9.408078993284917e-05, "loss": 0.0225, "step": 5960 }, { "epoch": 75.56962025316456, "grad_norm": 0.7478076219558716, "learning_rate": 9.405475034109254e-05, "loss": 0.0226, "step": 5970 }, { "epoch": 75.69620253164557, "grad_norm": 0.5905894041061401, "learning_rate": 9.402865721866015e-05, "loss": 0.0225, "step": 5980 }, { "epoch": 75.82278481012658, "grad_norm": 0.5096355080604553, "learning_rate": 9.400251059725762e-05, "loss": 0.0235, "step": 5990 }, { "epoch": 75.9493670886076, "grad_norm": 0.6787786483764648, "learning_rate": 9.397631050865554e-05, "loss": 0.0208, "step": 6000 }, { "epoch": 76.07594936708861, "grad_norm": 0.7164995670318604, "learning_rate": 9.395005698468949e-05, "loss": 0.0261, "step": 6010 }, { "epoch": 76.20253164556962, "grad_norm": 0.7846658825874329, "learning_rate": 9.392375005726e-05, "loss": 0.0222, "step": 6020 }, { "epoch": 76.32911392405063, "grad_norm": 0.7093709707260132, "learning_rate": 9.389738975833242e-05, "loss": 0.0276, "step": 6030 }, { "epoch": 76.45569620253164, "grad_norm": 0.6661425232887268, "learning_rate": 9.387097611993707e-05, "loss": 0.0215, "step": 6040 }, { "epoch": 76.58227848101266, "grad_norm": 0.6477464437484741, "learning_rate": 9.384450917416894e-05, "loss": 0.0236, "step": 6050 }, { "epoch": 76.70886075949367, "grad_norm": 0.6013715267181396, "learning_rate": 9.381798895318792e-05, "loss": 0.023, "step": 6060 }, { "epoch": 76.83544303797468, "grad_norm": 0.5830410122871399, "learning_rate": 9.379141548921853e-05, "loss": 0.023, "step": 6070 }, { "epoch": 76.9620253164557, "grad_norm": 0.6925011277198792, "learning_rate": 9.376478881455009e-05, "loss": 0.0273, "step": 6080 }, { "epoch": 77.0886075949367, "grad_norm": 0.6104116439819336, "learning_rate": 9.373810896153647e-05, "loss": 0.0236, "step": 6090 }, { "epoch": 77.21518987341773, "grad_norm": 0.7319440841674805, "learning_rate": 9.371137596259623e-05, "loss": 0.0231, "step": 6100 }, { "epoch": 77.34177215189874, "grad_norm": 0.7523755431175232, "learning_rate": 9.368458985021248e-05, "loss": 0.0257, "step": 6110 }, { "epoch": 77.46835443037975, "grad_norm": 0.5654532313346863, "learning_rate": 9.365775065693287e-05, "loss": 0.0244, "step": 6120 }, { "epoch": 77.59493670886076, "grad_norm": 0.6036387085914612, "learning_rate": 9.363085841536957e-05, "loss": 0.0243, "step": 6130 }, { "epoch": 77.72151898734177, "grad_norm": 0.7049402594566345, "learning_rate": 9.360391315819917e-05, "loss": 0.0251, "step": 6140 }, { "epoch": 77.84810126582279, "grad_norm": 0.6080374717712402, "learning_rate": 9.357691491816269e-05, "loss": 0.0205, "step": 6150 }, { "epoch": 77.9746835443038, "grad_norm": 0.5685698390007019, "learning_rate": 9.354986372806557e-05, "loss": 0.0225, "step": 6160 }, { "epoch": 78.10126582278481, "grad_norm": 0.6853463649749756, "learning_rate": 9.352275962077752e-05, "loss": 0.02, "step": 6170 }, { "epoch": 78.22784810126582, "grad_norm": 0.6854147911071777, "learning_rate": 9.349560262923262e-05, "loss": 0.0213, "step": 6180 }, { "epoch": 78.35443037974683, "grad_norm": 0.713988721370697, "learning_rate": 9.346839278642913e-05, "loss": 0.0234, "step": 6190 }, { "epoch": 78.48101265822785, "grad_norm": 0.8338638544082642, "learning_rate": 9.344113012542963e-05, "loss": 0.024, "step": 6200 }, { "epoch": 78.60759493670886, "grad_norm": 0.7993277311325073, "learning_rate": 9.341381467936079e-05, "loss": 0.0255, "step": 6210 }, { "epoch": 78.73417721518987, "grad_norm": 0.7564109563827515, "learning_rate": 9.338644648141346e-05, "loss": 0.0228, "step": 6220 }, { "epoch": 78.86075949367088, "grad_norm": 0.6186654567718506, "learning_rate": 9.335902556484257e-05, "loss": 0.0226, "step": 6230 }, { "epoch": 78.9873417721519, "grad_norm": 0.6830978989601135, "learning_rate": 9.333155196296712e-05, "loss": 0.0225, "step": 6240 }, { "epoch": 79.11392405063292, "grad_norm": 0.700122594833374, "learning_rate": 9.330402570917016e-05, "loss": 0.0213, "step": 6250 }, { "epoch": 79.24050632911393, "grad_norm": 0.7704319953918457, "learning_rate": 9.327644683689865e-05, "loss": 0.0205, "step": 6260 }, { "epoch": 79.36708860759494, "grad_norm": 0.6628067493438721, "learning_rate": 9.324881537966354e-05, "loss": 0.0209, "step": 6270 }, { "epoch": 79.49367088607595, "grad_norm": 0.6484766006469727, "learning_rate": 9.322113137103964e-05, "loss": 0.0204, "step": 6280 }, { "epoch": 79.62025316455696, "grad_norm": 0.6751298904418945, "learning_rate": 9.319339484466564e-05, "loss": 0.0237, "step": 6290 }, { "epoch": 79.74683544303798, "grad_norm": 0.7381858825683594, "learning_rate": 9.316560583424403e-05, "loss": 0.0211, "step": 6300 }, { "epoch": 79.87341772151899, "grad_norm": 0.5551780462265015, "learning_rate": 9.313776437354108e-05, "loss": 0.0208, "step": 6310 }, { "epoch": 80.0, "grad_norm": 0.6717276573181152, "learning_rate": 9.31098704963868e-05, "loss": 0.0227, "step": 6320 }, { "epoch": 80.12658227848101, "grad_norm": 0.6242465376853943, "learning_rate": 9.308192423667485e-05, "loss": 0.0227, "step": 6330 }, { "epoch": 80.25316455696202, "grad_norm": 0.5524625778198242, "learning_rate": 9.305392562836262e-05, "loss": 0.0218, "step": 6340 }, { "epoch": 80.37974683544304, "grad_norm": 0.8116728067398071, "learning_rate": 9.3025874705471e-05, "loss": 0.0282, "step": 6350 }, { "epoch": 80.50632911392405, "grad_norm": 0.7211834192276001, "learning_rate": 9.299777150208457e-05, "loss": 0.021, "step": 6360 }, { "epoch": 80.63291139240506, "grad_norm": 0.6586140990257263, "learning_rate": 9.296961605235133e-05, "loss": 0.0224, "step": 6370 }, { "epoch": 80.75949367088607, "grad_norm": 0.7741222381591797, "learning_rate": 9.29414083904828e-05, "loss": 0.0227, "step": 6380 }, { "epoch": 80.88607594936708, "grad_norm": 0.6229574680328369, "learning_rate": 9.291314855075397e-05, "loss": 0.0257, "step": 6390 }, { "epoch": 81.0126582278481, "grad_norm": 0.7553236484527588, "learning_rate": 9.288483656750322e-05, "loss": 0.027, "step": 6400 }, { "epoch": 81.13924050632912, "grad_norm": 0.675486147403717, "learning_rate": 9.285647247513225e-05, "loss": 0.0256, "step": 6410 }, { "epoch": 81.26582278481013, "grad_norm": 0.6959460973739624, "learning_rate": 9.282805630810614e-05, "loss": 0.0239, "step": 6420 }, { "epoch": 81.39240506329114, "grad_norm": 0.7465101480484009, "learning_rate": 9.279958810095317e-05, "loss": 0.0207, "step": 6430 }, { "epoch": 81.51898734177215, "grad_norm": 0.6959761381149292, "learning_rate": 9.277106788826494e-05, "loss": 0.0235, "step": 6440 }, { "epoch": 81.64556962025317, "grad_norm": 0.5799683928489685, "learning_rate": 9.274249570469618e-05, "loss": 0.0207, "step": 6450 }, { "epoch": 81.77215189873418, "grad_norm": 0.6585679054260254, "learning_rate": 9.271387158496476e-05, "loss": 0.0231, "step": 6460 }, { "epoch": 81.89873417721519, "grad_norm": 0.6481782793998718, "learning_rate": 9.268519556385173e-05, "loss": 0.0195, "step": 6470 }, { "epoch": 82.0253164556962, "grad_norm": 0.4912411868572235, "learning_rate": 9.265646767620113e-05, "loss": 0.021, "step": 6480 }, { "epoch": 82.15189873417721, "grad_norm": 0.6294928789138794, "learning_rate": 9.262768795692006e-05, "loss": 0.0239, "step": 6490 }, { "epoch": 82.27848101265823, "grad_norm": 0.6218981742858887, "learning_rate": 9.259885644097861e-05, "loss": 0.0227, "step": 6500 }, { "epoch": 82.40506329113924, "grad_norm": 0.5318179726600647, "learning_rate": 9.256997316340976e-05, "loss": 0.0234, "step": 6510 }, { "epoch": 82.53164556962025, "grad_norm": 0.5888519883155823, "learning_rate": 9.254103815930943e-05, "loss": 0.0206, "step": 6520 }, { "epoch": 82.65822784810126, "grad_norm": 0.6435540318489075, "learning_rate": 9.251205146383636e-05, "loss": 0.0221, "step": 6530 }, { "epoch": 82.78481012658227, "grad_norm": 0.5584481358528137, "learning_rate": 9.248301311221216e-05, "loss": 0.018, "step": 6540 }, { "epoch": 82.9113924050633, "grad_norm": 0.5748832821846008, "learning_rate": 9.245392313972115e-05, "loss": 0.023, "step": 6550 }, { "epoch": 83.0379746835443, "grad_norm": 0.46790236234664917, "learning_rate": 9.242478158171037e-05, "loss": 0.021, "step": 6560 }, { "epoch": 83.16455696202532, "grad_norm": 0.6771383881568909, "learning_rate": 9.239558847358957e-05, "loss": 0.0208, "step": 6570 }, { "epoch": 83.29113924050633, "grad_norm": 0.5871838331222534, "learning_rate": 9.236634385083114e-05, "loss": 0.0181, "step": 6580 }, { "epoch": 83.41772151898734, "grad_norm": 0.5437034368515015, "learning_rate": 9.233704774897005e-05, "loss": 0.021, "step": 6590 }, { "epoch": 83.54430379746836, "grad_norm": 0.6254096627235413, "learning_rate": 9.230770020360382e-05, "loss": 0.023, "step": 6600 }, { "epoch": 83.67088607594937, "grad_norm": 0.6536136269569397, "learning_rate": 9.227830125039247e-05, "loss": 0.0219, "step": 6610 }, { "epoch": 83.79746835443038, "grad_norm": 0.6495211124420166, "learning_rate": 9.224885092505853e-05, "loss": 0.0252, "step": 6620 }, { "epoch": 83.92405063291139, "grad_norm": 0.551641047000885, "learning_rate": 9.221934926338691e-05, "loss": 0.0256, "step": 6630 }, { "epoch": 84.0506329113924, "grad_norm": 0.6160187125205994, "learning_rate": 9.21897963012249e-05, "loss": 0.0243, "step": 6640 }, { "epoch": 84.17721518987342, "grad_norm": 0.6102079749107361, "learning_rate": 9.216019207448217e-05, "loss": 0.0263, "step": 6650 }, { "epoch": 84.30379746835443, "grad_norm": 0.6204123497009277, "learning_rate": 9.21305366191306e-05, "loss": 0.0211, "step": 6660 }, { "epoch": 84.43037974683544, "grad_norm": 0.6499886512756348, "learning_rate": 9.21008299712044e-05, "loss": 0.0231, "step": 6670 }, { "epoch": 84.55696202531645, "grad_norm": 0.6145303249359131, "learning_rate": 9.207107216679993e-05, "loss": 0.0199, "step": 6680 }, { "epoch": 84.68354430379746, "grad_norm": 0.6036309003829956, "learning_rate": 9.204126324207574e-05, "loss": 0.0209, "step": 6690 }, { "epoch": 84.81012658227849, "grad_norm": 0.5562779903411865, "learning_rate": 9.201140323325248e-05, "loss": 0.0209, "step": 6700 }, { "epoch": 84.9367088607595, "grad_norm": 0.7573220729827881, "learning_rate": 9.198149217661287e-05, "loss": 0.0218, "step": 6710 }, { "epoch": 85.0632911392405, "grad_norm": 0.7077623009681702, "learning_rate": 9.195153010850167e-05, "loss": 0.0255, "step": 6720 }, { "epoch": 85.18987341772151, "grad_norm": 0.5214092135429382, "learning_rate": 9.192151706532562e-05, "loss": 0.0226, "step": 6730 }, { "epoch": 85.31645569620254, "grad_norm": 0.5969802141189575, "learning_rate": 9.189145308355337e-05, "loss": 0.0214, "step": 6740 }, { "epoch": 85.44303797468355, "grad_norm": 0.6398400664329529, "learning_rate": 9.186133819971555e-05, "loss": 0.029, "step": 6750 }, { "epoch": 85.56962025316456, "grad_norm": 0.6566268801689148, "learning_rate": 9.183117245040455e-05, "loss": 0.0229, "step": 6760 }, { "epoch": 85.69620253164557, "grad_norm": 0.5641177892684937, "learning_rate": 9.180095587227458e-05, "loss": 0.0237, "step": 6770 }, { "epoch": 85.82278481012658, "grad_norm": 0.6814992427825928, "learning_rate": 9.177068850204167e-05, "loss": 0.0229, "step": 6780 }, { "epoch": 85.9493670886076, "grad_norm": 0.7727667689323425, "learning_rate": 9.174037037648351e-05, "loss": 0.0208, "step": 6790 }, { "epoch": 86.07594936708861, "grad_norm": 0.617745578289032, "learning_rate": 9.171000153243947e-05, "loss": 0.0229, "step": 6800 }, { "epoch": 86.20253164556962, "grad_norm": 0.6856599450111389, "learning_rate": 9.167958200681057e-05, "loss": 0.0216, "step": 6810 }, { "epoch": 86.32911392405063, "grad_norm": 0.6238178610801697, "learning_rate": 9.164911183655943e-05, "loss": 0.024, "step": 6820 }, { "epoch": 86.45569620253164, "grad_norm": 0.6537659168243408, "learning_rate": 9.161859105871013e-05, "loss": 0.0219, "step": 6830 }, { "epoch": 86.58227848101266, "grad_norm": 0.62653648853302, "learning_rate": 9.158801971034832e-05, "loss": 0.0225, "step": 6840 }, { "epoch": 86.70886075949367, "grad_norm": 0.6961847543716431, "learning_rate": 9.155739782862107e-05, "loss": 0.0208, "step": 6850 }, { "epoch": 86.83544303797468, "grad_norm": 0.7345927357673645, "learning_rate": 9.152672545073687e-05, "loss": 0.0214, "step": 6860 }, { "epoch": 86.9620253164557, "grad_norm": 0.6405068039894104, "learning_rate": 9.149600261396551e-05, "loss": 0.0222, "step": 6870 }, { "epoch": 87.0886075949367, "grad_norm": 0.6547214388847351, "learning_rate": 9.146522935563817e-05, "loss": 0.0244, "step": 6880 }, { "epoch": 87.21518987341773, "grad_norm": 0.5308278799057007, "learning_rate": 9.143440571314723e-05, "loss": 0.0203, "step": 6890 }, { "epoch": 87.34177215189874, "grad_norm": 0.5365856289863586, "learning_rate": 9.140353172394636e-05, "loss": 0.0235, "step": 6900 }, { "epoch": 87.46835443037975, "grad_norm": 0.5892438292503357, "learning_rate": 9.137260742555032e-05, "loss": 0.0237, "step": 6910 }, { "epoch": 87.59493670886076, "grad_norm": 0.4782274663448334, "learning_rate": 9.13416328555351e-05, "loss": 0.0217, "step": 6920 }, { "epoch": 87.72151898734177, "grad_norm": 0.5714629292488098, "learning_rate": 9.131060805153769e-05, "loss": 0.0233, "step": 6930 }, { "epoch": 87.84810126582279, "grad_norm": 0.6861394047737122, "learning_rate": 9.127953305125618e-05, "loss": 0.019, "step": 6940 }, { "epoch": 87.9746835443038, "grad_norm": 0.6056827306747437, "learning_rate": 9.124840789244958e-05, "loss": 0.0231, "step": 6950 }, { "epoch": 88.10126582278481, "grad_norm": 0.6258984208106995, "learning_rate": 9.121723261293794e-05, "loss": 0.0212, "step": 6960 }, { "epoch": 88.22784810126582, "grad_norm": 0.569437563419342, "learning_rate": 9.118600725060214e-05, "loss": 0.0202, "step": 6970 }, { "epoch": 88.35443037974683, "grad_norm": 0.6166678071022034, "learning_rate": 9.115473184338392e-05, "loss": 0.023, "step": 6980 }, { "epoch": 88.48101265822785, "grad_norm": 0.5936673879623413, "learning_rate": 9.11234064292859e-05, "loss": 0.0189, "step": 6990 }, { "epoch": 88.60759493670886, "grad_norm": 0.6982902884483337, "learning_rate": 9.109203104637137e-05, "loss": 0.0197, "step": 7000 }, { "epoch": 88.73417721518987, "grad_norm": 0.6811900734901428, "learning_rate": 9.10606057327644e-05, "loss": 0.0231, "step": 7010 }, { "epoch": 88.86075949367088, "grad_norm": 0.6686020493507385, "learning_rate": 9.10291305266497e-05, "loss": 0.0205, "step": 7020 }, { "epoch": 88.9873417721519, "grad_norm": 0.6033003926277161, "learning_rate": 9.099760546627261e-05, "loss": 0.0196, "step": 7030 }, { "epoch": 89.11392405063292, "grad_norm": 0.5892189741134644, "learning_rate": 9.096603058993907e-05, "loss": 0.0202, "step": 7040 }, { "epoch": 89.24050632911393, "grad_norm": 0.5977049469947815, "learning_rate": 9.093440593601552e-05, "loss": 0.0209, "step": 7050 }, { "epoch": 89.36708860759494, "grad_norm": 0.5925251245498657, "learning_rate": 9.090273154292889e-05, "loss": 0.0184, "step": 7060 }, { "epoch": 89.49367088607595, "grad_norm": 0.5031836032867432, "learning_rate": 9.087100744916656e-05, "loss": 0.0244, "step": 7070 }, { "epoch": 89.62025316455696, "grad_norm": 0.48397743701934814, "learning_rate": 9.08392336932763e-05, "loss": 0.0204, "step": 7080 }, { "epoch": 89.74683544303798, "grad_norm": 0.49053770303726196, "learning_rate": 9.080741031386619e-05, "loss": 0.0218, "step": 7090 }, { "epoch": 89.87341772151899, "grad_norm": 0.47261273860931396, "learning_rate": 9.077553734960468e-05, "loss": 0.0198, "step": 7100 }, { "epoch": 90.0, "grad_norm": 0.4799606502056122, "learning_rate": 9.074361483922041e-05, "loss": 0.0225, "step": 7110 }, { "epoch": 90.12658227848101, "grad_norm": 0.5357427000999451, "learning_rate": 9.071164282150224e-05, "loss": 0.0195, "step": 7120 }, { "epoch": 90.25316455696202, "grad_norm": 0.5193285942077637, "learning_rate": 9.067962133529919e-05, "loss": 0.0226, "step": 7130 }, { "epoch": 90.37974683544304, "grad_norm": 0.608070969581604, "learning_rate": 9.064755041952036e-05, "loss": 0.019, "step": 7140 }, { "epoch": 90.50632911392405, "grad_norm": 0.4604995846748352, "learning_rate": 9.061543011313497e-05, "loss": 0.0176, "step": 7150 }, { "epoch": 90.63291139240506, "grad_norm": 0.6180056929588318, "learning_rate": 9.05832604551722e-05, "loss": 0.022, "step": 7160 }, { "epoch": 90.75949367088607, "grad_norm": 0.6685667634010315, "learning_rate": 9.055104148472124e-05, "loss": 0.0221, "step": 7170 }, { "epoch": 90.88607594936708, "grad_norm": 0.5730795860290527, "learning_rate": 9.051877324093114e-05, "loss": 0.0216, "step": 7180 }, { "epoch": 91.0126582278481, "grad_norm": 0.6245896220207214, "learning_rate": 9.04864557630109e-05, "loss": 0.0225, "step": 7190 }, { "epoch": 91.13924050632912, "grad_norm": 0.5161830186843872, "learning_rate": 9.045408909022928e-05, "loss": 0.02, "step": 7200 }, { "epoch": 91.26582278481013, "grad_norm": 0.5409154295921326, "learning_rate": 9.042167326191484e-05, "loss": 0.0219, "step": 7210 }, { "epoch": 91.39240506329114, "grad_norm": 0.6070626974105835, "learning_rate": 9.038920831745587e-05, "loss": 0.0179, "step": 7220 }, { "epoch": 91.51898734177215, "grad_norm": 0.5756113529205322, "learning_rate": 9.035669429630036e-05, "loss": 0.0234, "step": 7230 }, { "epoch": 91.64556962025317, "grad_norm": 0.6109285950660706, "learning_rate": 9.032413123795588e-05, "loss": 0.0241, "step": 7240 }, { "epoch": 91.77215189873418, "grad_norm": 0.5283324122428894, "learning_rate": 9.029151918198962e-05, "loss": 0.0184, "step": 7250 }, { "epoch": 91.89873417721519, "grad_norm": 0.6674916744232178, "learning_rate": 9.025885816802833e-05, "loss": 0.019, "step": 7260 }, { "epoch": 92.0253164556962, "grad_norm": 0.6662389039993286, "learning_rate": 9.02261482357582e-05, "loss": 0.0228, "step": 7270 }, { "epoch": 92.15189873417721, "grad_norm": 0.690048098564148, "learning_rate": 9.019338942492485e-05, "loss": 0.0215, "step": 7280 }, { "epoch": 92.27848101265823, "grad_norm": 0.6341032385826111, "learning_rate": 9.01605817753334e-05, "loss": 0.0216, "step": 7290 }, { "epoch": 92.40506329113924, "grad_norm": 0.5859357118606567, "learning_rate": 9.012772532684818e-05, "loss": 0.0199, "step": 7300 }, { "epoch": 92.53164556962025, "grad_norm": 0.5267744660377502, "learning_rate": 9.00948201193929e-05, "loss": 0.0203, "step": 7310 }, { "epoch": 92.65822784810126, "grad_norm": 0.5631044507026672, "learning_rate": 9.006186619295047e-05, "loss": 0.0237, "step": 7320 }, { "epoch": 92.78481012658227, "grad_norm": 0.730387270450592, "learning_rate": 9.002886358756305e-05, "loss": 0.0189, "step": 7330 }, { "epoch": 92.9113924050633, "grad_norm": 0.5004222989082336, "learning_rate": 8.999581234333188e-05, "loss": 0.0187, "step": 7340 }, { "epoch": 93.0379746835443, "grad_norm": 0.624485969543457, "learning_rate": 8.996271250041734e-05, "loss": 0.0191, "step": 7350 }, { "epoch": 93.16455696202532, "grad_norm": 0.5238460898399353, "learning_rate": 8.99295640990389e-05, "loss": 0.0281, "step": 7360 }, { "epoch": 93.29113924050633, "grad_norm": 0.5212045907974243, "learning_rate": 8.989636717947494e-05, "loss": 0.0251, "step": 7370 }, { "epoch": 93.41772151898734, "grad_norm": 0.5822661519050598, "learning_rate": 8.986312178206291e-05, "loss": 0.0224, "step": 7380 }, { "epoch": 93.54430379746836, "grad_norm": 0.6482012867927551, "learning_rate": 8.982982794719904e-05, "loss": 0.0228, "step": 7390 }, { "epoch": 93.67088607594937, "grad_norm": 0.5965848565101624, "learning_rate": 8.979648571533852e-05, "loss": 0.02, "step": 7400 }, { "epoch": 93.79746835443038, "grad_norm": 0.5986549854278564, "learning_rate": 8.976309512699528e-05, "loss": 0.0197, "step": 7410 }, { "epoch": 93.92405063291139, "grad_norm": 0.6057446002960205, "learning_rate": 8.972965622274205e-05, "loss": 0.0203, "step": 7420 }, { "epoch": 94.0506329113924, "grad_norm": 0.514458417892456, "learning_rate": 8.969616904321026e-05, "loss": 0.02, "step": 7430 }, { "epoch": 94.17721518987342, "grad_norm": 0.6588233709335327, "learning_rate": 8.966263362908998e-05, "loss": 0.0203, "step": 7440 }, { "epoch": 94.30379746835443, "grad_norm": 0.55894535779953, "learning_rate": 8.962905002112989e-05, "loss": 0.0196, "step": 7450 }, { "epoch": 94.43037974683544, "grad_norm": 0.5370065569877625, "learning_rate": 8.959541826013724e-05, "loss": 0.0231, "step": 7460 }, { "epoch": 94.55696202531645, "grad_norm": 0.6890499591827393, "learning_rate": 8.956173838697781e-05, "loss": 0.0197, "step": 7470 }, { "epoch": 94.68354430379746, "grad_norm": 0.5599539279937744, "learning_rate": 8.952801044257581e-05, "loss": 0.0188, "step": 7480 }, { "epoch": 94.81012658227849, "grad_norm": 0.5677613615989685, "learning_rate": 8.949423446791387e-05, "loss": 0.0178, "step": 7490 }, { "epoch": 94.9367088607595, "grad_norm": 0.5640727281570435, "learning_rate": 8.9460410504033e-05, "loss": 0.0221, "step": 7500 }, { "epoch": 95.0632911392405, "grad_norm": 0.6251025199890137, "learning_rate": 8.942653859203248e-05, "loss": 0.0183, "step": 7510 }, { "epoch": 95.18987341772151, "grad_norm": 0.4777739346027374, "learning_rate": 8.93926187730699e-05, "loss": 0.0213, "step": 7520 }, { "epoch": 95.31645569620254, "grad_norm": 0.44815030694007874, "learning_rate": 8.935865108836102e-05, "loss": 0.0201, "step": 7530 }, { "epoch": 95.44303797468355, "grad_norm": 0.5587257146835327, "learning_rate": 8.932463557917981e-05, "loss": 0.0183, "step": 7540 }, { "epoch": 95.56962025316456, "grad_norm": 0.6411269903182983, "learning_rate": 8.929057228685829e-05, "loss": 0.0195, "step": 7550 }, { "epoch": 95.69620253164557, "grad_norm": 0.5768268704414368, "learning_rate": 8.925646125278658e-05, "loss": 0.0249, "step": 7560 }, { "epoch": 95.82278481012658, "grad_norm": 0.5911860466003418, "learning_rate": 8.92223025184128e-05, "loss": 0.0192, "step": 7570 }, { "epoch": 95.9493670886076, "grad_norm": 0.539337694644928, "learning_rate": 8.918809612524303e-05, "loss": 0.0188, "step": 7580 }, { "epoch": 96.07594936708861, "grad_norm": 0.5694331526756287, "learning_rate": 8.91538421148413e-05, "loss": 0.0183, "step": 7590 }, { "epoch": 96.20253164556962, "grad_norm": 0.5195723176002502, "learning_rate": 8.91195405288294e-05, "loss": 0.0215, "step": 7600 }, { "epoch": 96.32911392405063, "grad_norm": 0.5895766615867615, "learning_rate": 8.908519140888705e-05, "loss": 0.0189, "step": 7610 }, { "epoch": 96.45569620253164, "grad_norm": 0.6082014441490173, "learning_rate": 8.905079479675164e-05, "loss": 0.0205, "step": 7620 }, { "epoch": 96.58227848101266, "grad_norm": 0.5652264952659607, "learning_rate": 8.901635073421831e-05, "loss": 0.0223, "step": 7630 }, { "epoch": 96.70886075949367, "grad_norm": 0.5413353443145752, "learning_rate": 8.898185926313983e-05, "loss": 0.0203, "step": 7640 }, { "epoch": 96.83544303797468, "grad_norm": 0.6881547570228577, "learning_rate": 8.894732042542659e-05, "loss": 0.0209, "step": 7650 }, { "epoch": 96.9620253164557, "grad_norm": 0.6985871195793152, "learning_rate": 8.891273426304654e-05, "loss": 0.0252, "step": 7660 }, { "epoch": 97.0886075949367, "grad_norm": 0.5874691605567932, "learning_rate": 8.887810081802515e-05, "loss": 0.0223, "step": 7670 }, { "epoch": 97.21518987341773, "grad_norm": 0.7350218296051025, "learning_rate": 8.884342013244529e-05, "loss": 0.0196, "step": 7680 }, { "epoch": 97.34177215189874, "grad_norm": 0.5900181531906128, "learning_rate": 8.880869224844725e-05, "loss": 0.0189, "step": 7690 }, { "epoch": 97.46835443037975, "grad_norm": 0.5944636464118958, "learning_rate": 8.877391720822875e-05, "loss": 0.025, "step": 7700 }, { "epoch": 97.59493670886076, "grad_norm": 0.6105281710624695, "learning_rate": 8.873909505404468e-05, "loss": 0.0222, "step": 7710 }, { "epoch": 97.72151898734177, "grad_norm": 0.5899769067764282, "learning_rate": 8.870422582820726e-05, "loss": 0.0177, "step": 7720 }, { "epoch": 97.84810126582279, "grad_norm": 0.523282527923584, "learning_rate": 8.866930957308588e-05, "loss": 0.02, "step": 7730 }, { "epoch": 97.9746835443038, "grad_norm": 0.5118725299835205, "learning_rate": 8.86343463311071e-05, "loss": 0.0182, "step": 7740 }, { "epoch": 98.10126582278481, "grad_norm": 0.5408750772476196, "learning_rate": 8.859933614475452e-05, "loss": 0.0197, "step": 7750 }, { "epoch": 98.22784810126582, "grad_norm": 0.43630915880203247, "learning_rate": 8.85642790565689e-05, "loss": 0.0179, "step": 7760 }, { "epoch": 98.35443037974683, "grad_norm": 0.5267261862754822, "learning_rate": 8.852917510914782e-05, "loss": 0.0249, "step": 7770 }, { "epoch": 98.48101265822785, "grad_norm": 0.5714665055274963, "learning_rate": 8.84940243451459e-05, "loss": 0.019, "step": 7780 }, { "epoch": 98.60759493670886, "grad_norm": 0.49981072545051575, "learning_rate": 8.845882680727469e-05, "loss": 0.0179, "step": 7790 }, { "epoch": 98.73417721518987, "grad_norm": 0.5933191180229187, "learning_rate": 8.842358253830246e-05, "loss": 0.0189, "step": 7800 }, { "epoch": 98.86075949367088, "grad_norm": 0.48989200592041016, "learning_rate": 8.838829158105434e-05, "loss": 0.0191, "step": 7810 }, { "epoch": 98.9873417721519, "grad_norm": 0.5612712502479553, "learning_rate": 8.835295397841217e-05, "loss": 0.0163, "step": 7820 }, { "epoch": 99.11392405063292, "grad_norm": 0.46296876668930054, "learning_rate": 8.831756977331447e-05, "loss": 0.0214, "step": 7830 }, { "epoch": 99.24050632911393, "grad_norm": 0.5805317163467407, "learning_rate": 8.828213900875638e-05, "loss": 0.018, "step": 7840 }, { "epoch": 99.36708860759494, "grad_norm": 0.5919309258460999, "learning_rate": 8.824666172778964e-05, "loss": 0.0189, "step": 7850 }, { "epoch": 99.49367088607595, "grad_norm": 0.6345791220664978, "learning_rate": 8.821113797352245e-05, "loss": 0.0207, "step": 7860 }, { "epoch": 99.62025316455696, "grad_norm": 0.5853433609008789, "learning_rate": 8.817556778911956e-05, "loss": 0.019, "step": 7870 }, { "epoch": 99.74683544303798, "grad_norm": 0.48441019654273987, "learning_rate": 8.81399512178021e-05, "loss": 0.0197, "step": 7880 }, { "epoch": 99.87341772151899, "grad_norm": 0.6547595262527466, "learning_rate": 8.810428830284752e-05, "loss": 0.0217, "step": 7890 }, { "epoch": 100.0, "grad_norm": 0.6075287461280823, "learning_rate": 8.806857908758967e-05, "loss": 0.0208, "step": 7900 }, { "epoch": 100.12658227848101, "grad_norm": 0.6695914268493652, "learning_rate": 8.80328236154186e-05, "loss": 0.0209, "step": 7910 }, { "epoch": 100.25316455696202, "grad_norm": 0.49227187037467957, "learning_rate": 8.799702192978056e-05, "loss": 0.0189, "step": 7920 }, { "epoch": 100.37974683544304, "grad_norm": 0.5238152742385864, "learning_rate": 8.7961174074178e-05, "loss": 0.0192, "step": 7930 }, { "epoch": 100.50632911392405, "grad_norm": 0.5055280923843384, "learning_rate": 8.792528009216942e-05, "loss": 0.0211, "step": 7940 }, { "epoch": 100.63291139240506, "grad_norm": 0.5583733320236206, "learning_rate": 8.788934002736943e-05, "loss": 0.0203, "step": 7950 }, { "epoch": 100.75949367088607, "grad_norm": 0.5992043018341064, "learning_rate": 8.785335392344857e-05, "loss": 0.0223, "step": 7960 }, { "epoch": 100.88607594936708, "grad_norm": 0.49284663796424866, "learning_rate": 8.781732182413336e-05, "loss": 0.0231, "step": 7970 }, { "epoch": 101.0126582278481, "grad_norm": 0.5273424983024597, "learning_rate": 8.778124377320618e-05, "loss": 0.0222, "step": 7980 }, { "epoch": 101.13924050632912, "grad_norm": 0.5754503011703491, "learning_rate": 8.774511981450529e-05, "loss": 0.0199, "step": 7990 }, { "epoch": 101.26582278481013, "grad_norm": 0.48712214827537537, "learning_rate": 8.770894999192467e-05, "loss": 0.0184, "step": 8000 }, { "epoch": 101.39240506329114, "grad_norm": 0.5029520392417908, "learning_rate": 8.767273434941412e-05, "loss": 0.0156, "step": 8010 }, { "epoch": 101.51898734177215, "grad_norm": 0.7971763610839844, "learning_rate": 8.7636472930979e-05, "loss": 0.0187, "step": 8020 }, { "epoch": 101.64556962025317, "grad_norm": 0.6544016599655151, "learning_rate": 8.76001657806804e-05, "loss": 0.0187, "step": 8030 }, { "epoch": 101.77215189873418, "grad_norm": 0.5481281876564026, "learning_rate": 8.756381294263491e-05, "loss": 0.0213, "step": 8040 }, { "epoch": 101.89873417721519, "grad_norm": 0.6812074780464172, "learning_rate": 8.752741446101464e-05, "loss": 0.0197, "step": 8050 }, { "epoch": 102.0253164556962, "grad_norm": 0.49700218439102173, "learning_rate": 8.749097038004722e-05, "loss": 0.0191, "step": 8060 }, { "epoch": 102.15189873417721, "grad_norm": 0.6452575325965881, "learning_rate": 8.745448074401561e-05, "loss": 0.0189, "step": 8070 }, { "epoch": 102.27848101265823, "grad_norm": 0.5273920297622681, "learning_rate": 8.741794559725818e-05, "loss": 0.0213, "step": 8080 }, { "epoch": 102.40506329113924, "grad_norm": 0.5765165090560913, "learning_rate": 8.738136498416856e-05, "loss": 0.0203, "step": 8090 }, { "epoch": 102.53164556962025, "grad_norm": 0.5375029444694519, "learning_rate": 8.734473894919564e-05, "loss": 0.0218, "step": 8100 }, { "epoch": 102.65822784810126, "grad_norm": 0.6136459708213806, "learning_rate": 8.730806753684353e-05, "loss": 0.0182, "step": 8110 }, { "epoch": 102.78481012658227, "grad_norm": 0.5399619936943054, "learning_rate": 8.727135079167144e-05, "loss": 0.0201, "step": 8120 }, { "epoch": 102.9113924050633, "grad_norm": 0.5928249955177307, "learning_rate": 8.723458875829368e-05, "loss": 0.0202, "step": 8130 }, { "epoch": 103.0379746835443, "grad_norm": 0.5636846423149109, "learning_rate": 8.719778148137958e-05, "loss": 0.0176, "step": 8140 }, { "epoch": 103.16455696202532, "grad_norm": 0.5451529026031494, "learning_rate": 8.716092900565346e-05, "loss": 0.0208, "step": 8150 }, { "epoch": 103.29113924050633, "grad_norm": 0.6147380471229553, "learning_rate": 8.712403137589456e-05, "loss": 0.0189, "step": 8160 }, { "epoch": 103.41772151898734, "grad_norm": 0.5506302714347839, "learning_rate": 8.708708863693697e-05, "loss": 0.0203, "step": 8170 }, { "epoch": 103.54430379746836, "grad_norm": 0.4755566418170929, "learning_rate": 8.70501008336696e-05, "loss": 0.0186, "step": 8180 }, { "epoch": 103.67088607594937, "grad_norm": 0.5311554670333862, "learning_rate": 8.701306801103611e-05, "loss": 0.0203, "step": 8190 }, { "epoch": 103.79746835443038, "grad_norm": 0.4316597878932953, "learning_rate": 8.697599021403489e-05, "loss": 0.019, "step": 8200 }, { "epoch": 103.92405063291139, "grad_norm": 0.5491881370544434, "learning_rate": 8.693886748771896e-05, "loss": 0.0188, "step": 8210 }, { "epoch": 104.0506329113924, "grad_norm": 0.5317553281784058, "learning_rate": 8.690169987719593e-05, "loss": 0.0161, "step": 8220 }, { "epoch": 104.17721518987342, "grad_norm": 0.45893168449401855, "learning_rate": 8.686448742762791e-05, "loss": 0.017, "step": 8230 }, { "epoch": 104.30379746835443, "grad_norm": 0.47043707966804504, "learning_rate": 8.682723018423161e-05, "loss": 0.0192, "step": 8240 }, { "epoch": 104.43037974683544, "grad_norm": 0.44072669744491577, "learning_rate": 8.678992819227804e-05, "loss": 0.02, "step": 8250 }, { "epoch": 104.55696202531645, "grad_norm": 0.6049776077270508, "learning_rate": 8.675258149709265e-05, "loss": 0.0169, "step": 8260 }, { "epoch": 104.68354430379746, "grad_norm": 0.40057843923568726, "learning_rate": 8.67151901440552e-05, "loss": 0.017, "step": 8270 }, { "epoch": 104.81012658227849, "grad_norm": 0.4883785843849182, "learning_rate": 8.667775417859971e-05, "loss": 0.0189, "step": 8280 }, { "epoch": 104.9367088607595, "grad_norm": 0.6331677436828613, "learning_rate": 8.664027364621441e-05, "loss": 0.0197, "step": 8290 }, { "epoch": 105.0632911392405, "grad_norm": 0.5700109601020813, "learning_rate": 8.660274859244168e-05, "loss": 0.0178, "step": 8300 }, { "epoch": 105.18987341772151, "grad_norm": 0.5666419863700867, "learning_rate": 8.656517906287799e-05, "loss": 0.0207, "step": 8310 }, { "epoch": 105.31645569620254, "grad_norm": 0.4792291522026062, "learning_rate": 8.652756510317387e-05, "loss": 0.0181, "step": 8320 }, { "epoch": 105.44303797468355, "grad_norm": 0.6192867159843445, "learning_rate": 8.648990675903382e-05, "loss": 0.019, "step": 8330 }, { "epoch": 105.56962025316456, "grad_norm": 0.5734803080558777, "learning_rate": 8.64522040762163e-05, "loss": 0.0177, "step": 8340 }, { "epoch": 105.69620253164557, "grad_norm": 0.48552823066711426, "learning_rate": 8.64144571005336e-05, "loss": 0.0208, "step": 8350 }, { "epoch": 105.82278481012658, "grad_norm": 0.5168567299842834, "learning_rate": 8.637666587785184e-05, "loss": 0.0186, "step": 8360 }, { "epoch": 105.9493670886076, "grad_norm": 0.3993741273880005, "learning_rate": 8.633883045409096e-05, "loss": 0.0161, "step": 8370 }, { "epoch": 106.07594936708861, "grad_norm": 0.4925747215747833, "learning_rate": 8.630095087522457e-05, "loss": 0.0174, "step": 8380 }, { "epoch": 106.20253164556962, "grad_norm": 0.5385444760322571, "learning_rate": 8.62630271872799e-05, "loss": 0.0184, "step": 8390 }, { "epoch": 106.32911392405063, "grad_norm": 0.5323207378387451, "learning_rate": 8.62250594363378e-05, "loss": 0.0188, "step": 8400 }, { "epoch": 106.45569620253164, "grad_norm": 0.48688721656799316, "learning_rate": 8.61870476685327e-05, "loss": 0.0212, "step": 8410 }, { "epoch": 106.58227848101266, "grad_norm": 0.5705913305282593, "learning_rate": 8.614899193005248e-05, "loss": 0.0177, "step": 8420 }, { "epoch": 106.70886075949367, "grad_norm": 0.6492370963096619, "learning_rate": 8.611089226713842e-05, "loss": 0.0172, "step": 8430 }, { "epoch": 106.83544303797468, "grad_norm": 0.4781129062175751, "learning_rate": 8.607274872608521e-05, "loss": 0.0193, "step": 8440 }, { "epoch": 106.9620253164557, "grad_norm": 0.5466167330741882, "learning_rate": 8.603456135324088e-05, "loss": 0.0256, "step": 8450 }, { "epoch": 107.0886075949367, "grad_norm": 0.5093086361885071, "learning_rate": 8.599633019500665e-05, "loss": 0.0172, "step": 8460 }, { "epoch": 107.21518987341773, "grad_norm": 0.5823283195495605, "learning_rate": 8.595805529783702e-05, "loss": 0.0178, "step": 8470 }, { "epoch": 107.34177215189874, "grad_norm": 0.5350452065467834, "learning_rate": 8.591973670823958e-05, "loss": 0.0192, "step": 8480 }, { "epoch": 107.46835443037975, "grad_norm": 0.5058180689811707, "learning_rate": 8.588137447277503e-05, "loss": 0.0189, "step": 8490 }, { "epoch": 107.59493670886076, "grad_norm": 0.602701723575592, "learning_rate": 8.584296863805709e-05, "loss": 0.022, "step": 8500 }, { "epoch": 107.72151898734177, "grad_norm": 0.5091608166694641, "learning_rate": 8.580451925075248e-05, "loss": 0.0184, "step": 8510 }, { "epoch": 107.84810126582279, "grad_norm": 0.4399580955505371, "learning_rate": 8.576602635758085e-05, "loss": 0.0177, "step": 8520 }, { "epoch": 107.9746835443038, "grad_norm": 0.49783119559288025, "learning_rate": 8.572749000531469e-05, "loss": 0.0176, "step": 8530 }, { "epoch": 108.10126582278481, "grad_norm": 0.5433919429779053, "learning_rate": 8.568891024077926e-05, "loss": 0.0209, "step": 8540 }, { "epoch": 108.22784810126582, "grad_norm": 0.4831826686859131, "learning_rate": 8.565028711085265e-05, "loss": 0.0168, "step": 8550 }, { "epoch": 108.35443037974683, "grad_norm": 0.5982249975204468, "learning_rate": 8.561162066246561e-05, "loss": 0.0199, "step": 8560 }, { "epoch": 108.48101265822785, "grad_norm": 0.5174546837806702, "learning_rate": 8.557291094260152e-05, "loss": 0.0241, "step": 8570 }, { "epoch": 108.60759493670886, "grad_norm": 0.6081712245941162, "learning_rate": 8.553415799829629e-05, "loss": 0.019, "step": 8580 }, { "epoch": 108.73417721518987, "grad_norm": 0.5239022970199585, "learning_rate": 8.549536187663847e-05, "loss": 0.0185, "step": 8590 }, { "epoch": 108.86075949367088, "grad_norm": 0.5084967613220215, "learning_rate": 8.545652262476897e-05, "loss": 0.0184, "step": 8600 }, { "epoch": 108.9873417721519, "grad_norm": 0.5009980797767639, "learning_rate": 8.541764028988115e-05, "loss": 0.0197, "step": 8610 }, { "epoch": 109.11392405063292, "grad_norm": 0.4464353919029236, "learning_rate": 8.537871491922071e-05, "loss": 0.0176, "step": 8620 }, { "epoch": 109.24050632911393, "grad_norm": 0.5392150282859802, "learning_rate": 8.533974656008566e-05, "loss": 0.0205, "step": 8630 }, { "epoch": 109.36708860759494, "grad_norm": 0.5773656368255615, "learning_rate": 8.530073525982621e-05, "loss": 0.0198, "step": 8640 }, { "epoch": 109.49367088607595, "grad_norm": 0.5628786087036133, "learning_rate": 8.526168106584476e-05, "loss": 0.0197, "step": 8650 }, { "epoch": 109.62025316455696, "grad_norm": 0.4881855249404907, "learning_rate": 8.522258402559586e-05, "loss": 0.0183, "step": 8660 }, { "epoch": 109.74683544303798, "grad_norm": 0.5412335991859436, "learning_rate": 8.518344418658613e-05, "loss": 0.0201, "step": 8670 }, { "epoch": 109.87341772151899, "grad_norm": 0.5480372309684753, "learning_rate": 8.51442615963741e-05, "loss": 0.0204, "step": 8680 }, { "epoch": 110.0, "grad_norm": 0.5014373064041138, "learning_rate": 8.510503630257034e-05, "loss": 0.0168, "step": 8690 }, { "epoch": 110.12658227848101, "grad_norm": 0.3906188905239105, "learning_rate": 8.50657683528373e-05, "loss": 0.0156, "step": 8700 }, { "epoch": 110.25316455696202, "grad_norm": 0.5298651456832886, "learning_rate": 8.502645779488922e-05, "loss": 0.0198, "step": 8710 }, { "epoch": 110.37974683544304, "grad_norm": 0.5590598583221436, "learning_rate": 8.498710467649213e-05, "loss": 0.0179, "step": 8720 }, { "epoch": 110.50632911392405, "grad_norm": 0.521360456943512, "learning_rate": 8.494770904546381e-05, "loss": 0.0174, "step": 8730 }, { "epoch": 110.63291139240506, "grad_norm": 0.4224398732185364, "learning_rate": 8.490827094967363e-05, "loss": 0.0189, "step": 8740 }, { "epoch": 110.75949367088607, "grad_norm": 0.47163379192352295, "learning_rate": 8.486879043704263e-05, "loss": 0.0155, "step": 8750 }, { "epoch": 110.88607594936708, "grad_norm": 0.45783278346061707, "learning_rate": 8.482926755554332e-05, "loss": 0.0167, "step": 8760 }, { "epoch": 111.0126582278481, "grad_norm": 0.5841414928436279, "learning_rate": 8.478970235319975e-05, "loss": 0.0207, "step": 8770 }, { "epoch": 111.13924050632912, "grad_norm": 0.5016734004020691, "learning_rate": 8.475009487808737e-05, "loss": 0.0203, "step": 8780 }, { "epoch": 111.26582278481013, "grad_norm": 0.5267652273178101, "learning_rate": 8.471044517833298e-05, "loss": 0.0195, "step": 8790 }, { "epoch": 111.39240506329114, "grad_norm": 0.5168585181236267, "learning_rate": 8.467075330211474e-05, "loss": 0.0148, "step": 8800 }, { "epoch": 111.51898734177215, "grad_norm": 0.4561016261577606, "learning_rate": 8.463101929766198e-05, "loss": 0.0155, "step": 8810 }, { "epoch": 111.64556962025317, "grad_norm": 0.5828018188476562, "learning_rate": 8.459124321325528e-05, "loss": 0.0185, "step": 8820 }, { "epoch": 111.77215189873418, "grad_norm": 0.5954235792160034, "learning_rate": 8.455142509722634e-05, "loss": 0.0171, "step": 8830 }, { "epoch": 111.89873417721519, "grad_norm": 0.45583489537239075, "learning_rate": 8.451156499795792e-05, "loss": 0.0222, "step": 8840 }, { "epoch": 112.0253164556962, "grad_norm": 0.5354249477386475, "learning_rate": 8.447166296388381e-05, "loss": 0.0216, "step": 8850 }, { "epoch": 112.15189873417721, "grad_norm": 0.4263935983181, "learning_rate": 8.443171904348872e-05, "loss": 0.0185, "step": 8860 }, { "epoch": 112.27848101265823, "grad_norm": 0.47236138582229614, "learning_rate": 8.43917332853083e-05, "loss": 0.0208, "step": 8870 }, { "epoch": 112.40506329113924, "grad_norm": 0.4560900926589966, "learning_rate": 8.435170573792901e-05, "loss": 0.018, "step": 8880 }, { "epoch": 112.53164556962025, "grad_norm": 0.45830559730529785, "learning_rate": 8.431163644998808e-05, "loss": 0.0218, "step": 8890 }, { "epoch": 112.65822784810126, "grad_norm": 0.5419922471046448, "learning_rate": 8.42715254701735e-05, "loss": 0.0202, "step": 8900 }, { "epoch": 112.78481012658227, "grad_norm": 0.49337872862815857, "learning_rate": 8.42313728472239e-05, "loss": 0.0175, "step": 8910 }, { "epoch": 112.9113924050633, "grad_norm": 0.43356600403785706, "learning_rate": 8.419117862992844e-05, "loss": 0.0199, "step": 8920 }, { "epoch": 113.0379746835443, "grad_norm": 0.5309543609619141, "learning_rate": 8.415094286712694e-05, "loss": 0.0233, "step": 8930 }, { "epoch": 113.16455696202532, "grad_norm": 0.5538792014122009, "learning_rate": 8.411066560770965e-05, "loss": 0.0179, "step": 8940 }, { "epoch": 113.29113924050633, "grad_norm": 0.5124020576477051, "learning_rate": 8.407034690061721e-05, "loss": 0.0177, "step": 8950 }, { "epoch": 113.41772151898734, "grad_norm": 0.6055654287338257, "learning_rate": 8.402998679484067e-05, "loss": 0.0187, "step": 8960 }, { "epoch": 113.54430379746836, "grad_norm": 0.6007237434387207, "learning_rate": 8.398958533942135e-05, "loss": 0.0205, "step": 8970 }, { "epoch": 113.67088607594937, "grad_norm": 0.534932017326355, "learning_rate": 8.394914258345083e-05, "loss": 0.0181, "step": 8980 }, { "epoch": 113.79746835443038, "grad_norm": 0.6152629256248474, "learning_rate": 8.390865857607089e-05, "loss": 0.0205, "step": 8990 }, { "epoch": 113.92405063291139, "grad_norm": 0.5029500722885132, "learning_rate": 8.386813336647339e-05, "loss": 0.0209, "step": 9000 }, { "epoch": 114.0506329113924, "grad_norm": 0.5263598561286926, "learning_rate": 8.38275670039003e-05, "loss": 0.0198, "step": 9010 }, { "epoch": 114.17721518987342, "grad_norm": 0.49263378977775574, "learning_rate": 8.378695953764356e-05, "loss": 0.0217, "step": 9020 }, { "epoch": 114.30379746835443, "grad_norm": 0.5535228252410889, "learning_rate": 8.374631101704509e-05, "loss": 0.0214, "step": 9030 }, { "epoch": 114.43037974683544, "grad_norm": 0.4578494131565094, "learning_rate": 8.370562149149665e-05, "loss": 0.017, "step": 9040 }, { "epoch": 114.55696202531645, "grad_norm": 0.5227771997451782, "learning_rate": 8.366489101043989e-05, "loss": 0.0167, "step": 9050 }, { "epoch": 114.68354430379746, "grad_norm": 0.49786606431007385, "learning_rate": 8.362411962336614e-05, "loss": 0.0177, "step": 9060 }, { "epoch": 114.81012658227849, "grad_norm": 0.5215317010879517, "learning_rate": 8.358330737981651e-05, "loss": 0.0183, "step": 9070 }, { "epoch": 114.9367088607595, "grad_norm": 0.512330174446106, "learning_rate": 8.35424543293817e-05, "loss": 0.018, "step": 9080 }, { "epoch": 115.0632911392405, "grad_norm": 0.5289077758789062, "learning_rate": 8.350156052170206e-05, "loss": 0.018, "step": 9090 }, { "epoch": 115.18987341772151, "grad_norm": 0.4465099573135376, "learning_rate": 8.346062600646738e-05, "loss": 0.0166, "step": 9100 }, { "epoch": 115.31645569620254, "grad_norm": 0.4556639790534973, "learning_rate": 8.341965083341697e-05, "loss": 0.0162, "step": 9110 }, { "epoch": 115.44303797468355, "grad_norm": 0.5509995222091675, "learning_rate": 8.337863505233953e-05, "loss": 0.018, "step": 9120 }, { "epoch": 115.56962025316456, "grad_norm": 0.5165451169013977, "learning_rate": 8.33375787130731e-05, "loss": 0.0182, "step": 9130 }, { "epoch": 115.69620253164557, "grad_norm": 0.5214295387268066, "learning_rate": 8.3296481865505e-05, "loss": 0.0162, "step": 9140 }, { "epoch": 115.82278481012658, "grad_norm": 0.4401211142539978, "learning_rate": 8.325534455957181e-05, "loss": 0.0206, "step": 9150 }, { "epoch": 115.9493670886076, "grad_norm": 0.4544669985771179, "learning_rate": 8.321416684525917e-05, "loss": 0.0158, "step": 9160 }, { "epoch": 116.07594936708861, "grad_norm": 0.3497403860092163, "learning_rate": 8.317294877260193e-05, "loss": 0.017, "step": 9170 }, { "epoch": 116.20253164556962, "grad_norm": 0.3067779839038849, "learning_rate": 8.313169039168394e-05, "loss": 0.016, "step": 9180 }, { "epoch": 116.32911392405063, "grad_norm": 0.507752537727356, "learning_rate": 8.3090391752638e-05, "loss": 0.0189, "step": 9190 }, { "epoch": 116.45569620253164, "grad_norm": 0.4631251096725464, "learning_rate": 8.304905290564586e-05, "loss": 0.0178, "step": 9200 }, { "epoch": 116.58227848101266, "grad_norm": 0.454550564289093, "learning_rate": 8.300767390093813e-05, "loss": 0.0189, "step": 9210 }, { "epoch": 116.70886075949367, "grad_norm": 0.47136279940605164, "learning_rate": 8.296625478879417e-05, "loss": 0.0175, "step": 9220 }, { "epoch": 116.83544303797468, "grad_norm": 0.5121127963066101, "learning_rate": 8.292479561954214e-05, "loss": 0.019, "step": 9230 }, { "epoch": 116.9620253164557, "grad_norm": 0.542431116104126, "learning_rate": 8.288329644355883e-05, "loss": 0.0157, "step": 9240 }, { "epoch": 117.0886075949367, "grad_norm": 0.4436202049255371, "learning_rate": 8.284175731126965e-05, "loss": 0.0182, "step": 9250 }, { "epoch": 117.21518987341773, "grad_norm": 0.4194943904876709, "learning_rate": 8.280017827314854e-05, "loss": 0.0202, "step": 9260 }, { "epoch": 117.34177215189874, "grad_norm": 0.5694897770881653, "learning_rate": 8.2758559379718e-05, "loss": 0.0229, "step": 9270 }, { "epoch": 117.46835443037975, "grad_norm": 0.4654391407966614, "learning_rate": 8.271690068154887e-05, "loss": 0.019, "step": 9280 }, { "epoch": 117.59493670886076, "grad_norm": 0.514228343963623, "learning_rate": 8.26752022292604e-05, "loss": 0.0175, "step": 9290 }, { "epoch": 117.72151898734177, "grad_norm": 0.5854518413543701, "learning_rate": 8.263346407352018e-05, "loss": 0.0196, "step": 9300 }, { "epoch": 117.84810126582279, "grad_norm": 0.4875151216983795, "learning_rate": 8.259168626504395e-05, "loss": 0.0179, "step": 9310 }, { "epoch": 117.9746835443038, "grad_norm": 0.4346241354942322, "learning_rate": 8.254986885459572e-05, "loss": 0.0164, "step": 9320 }, { "epoch": 118.10126582278481, "grad_norm": 0.6037802696228027, "learning_rate": 8.250801189298757e-05, "loss": 0.0178, "step": 9330 }, { "epoch": 118.22784810126582, "grad_norm": 0.4552825391292572, "learning_rate": 8.246611543107967e-05, "loss": 0.0232, "step": 9340 }, { "epoch": 118.35443037974683, "grad_norm": 0.461110383272171, "learning_rate": 8.242417951978018e-05, "loss": 0.0198, "step": 9350 }, { "epoch": 118.48101265822785, "grad_norm": 0.5787426829338074, "learning_rate": 8.238220421004517e-05, "loss": 0.0178, "step": 9360 }, { "epoch": 118.60759493670886, "grad_norm": 0.49664613604545593, "learning_rate": 8.234018955287859e-05, "loss": 0.0166, "step": 9370 }, { "epoch": 118.73417721518987, "grad_norm": 0.43062329292297363, "learning_rate": 8.229813559933224e-05, "loss": 0.0152, "step": 9380 }, { "epoch": 118.86075949367088, "grad_norm": 0.4897250831127167, "learning_rate": 8.22560424005056e-05, "loss": 0.0173, "step": 9390 }, { "epoch": 118.9873417721519, "grad_norm": 0.5963055491447449, "learning_rate": 8.22139100075459e-05, "loss": 0.0192, "step": 9400 }, { "epoch": 119.11392405063292, "grad_norm": 0.4497585892677307, "learning_rate": 8.217173847164799e-05, "loss": 0.0185, "step": 9410 }, { "epoch": 119.24050632911393, "grad_norm": 0.575033962726593, "learning_rate": 8.212952784405422e-05, "loss": 0.0154, "step": 9420 }, { "epoch": 119.36708860759494, "grad_norm": 0.5236964225769043, "learning_rate": 8.208727817605451e-05, "loss": 0.0167, "step": 9430 }, { "epoch": 119.49367088607595, "grad_norm": 0.43061500787734985, "learning_rate": 8.204498951898618e-05, "loss": 0.0171, "step": 9440 }, { "epoch": 119.62025316455696, "grad_norm": 0.4614903926849365, "learning_rate": 8.200266192423395e-05, "loss": 0.0163, "step": 9450 }, { "epoch": 119.74683544303798, "grad_norm": 0.4434010684490204, "learning_rate": 8.196029544322981e-05, "loss": 0.0187, "step": 9460 }, { "epoch": 119.87341772151899, "grad_norm": 0.44014132022857666, "learning_rate": 8.191789012745305e-05, "loss": 0.0166, "step": 9470 }, { "epoch": 120.0, "grad_norm": 0.49396848678588867, "learning_rate": 8.187544602843014e-05, "loss": 0.0169, "step": 9480 }, { "epoch": 120.12658227848101, "grad_norm": 0.41064581274986267, "learning_rate": 8.183296319773466e-05, "loss": 0.016, "step": 9490 }, { "epoch": 120.25316455696202, "grad_norm": 0.45776328444480896, "learning_rate": 8.179044168698721e-05, "loss": 0.014, "step": 9500 }, { "epoch": 120.37974683544304, "grad_norm": 0.45995309948921204, "learning_rate": 8.174788154785548e-05, "loss": 0.0177, "step": 9510 }, { "epoch": 120.50632911392405, "grad_norm": 0.4444408714771271, "learning_rate": 8.170528283205403e-05, "loss": 0.0176, "step": 9520 }, { "epoch": 120.63291139240506, "grad_norm": 0.5380370020866394, "learning_rate": 8.166264559134435e-05, "loss": 0.0146, "step": 9530 }, { "epoch": 120.75949367088607, "grad_norm": 0.5046033263206482, "learning_rate": 8.161996987753467e-05, "loss": 0.0177, "step": 9540 }, { "epoch": 120.88607594936708, "grad_norm": 0.4568360149860382, "learning_rate": 8.157725574248e-05, "loss": 0.0166, "step": 9550 }, { "epoch": 121.0126582278481, "grad_norm": 0.5330622792243958, "learning_rate": 8.153450323808209e-05, "loss": 0.015, "step": 9560 }, { "epoch": 121.13924050632912, "grad_norm": 0.4642433226108551, "learning_rate": 8.149171241628923e-05, "loss": 0.018, "step": 9570 }, { "epoch": 121.26582278481013, "grad_norm": 0.49931442737579346, "learning_rate": 8.144888332909631e-05, "loss": 0.0185, "step": 9580 }, { "epoch": 121.39240506329114, "grad_norm": 0.5414138436317444, "learning_rate": 8.14060160285447e-05, "loss": 0.0177, "step": 9590 }, { "epoch": 121.51898734177215, "grad_norm": 0.5153962969779968, "learning_rate": 8.136311056672223e-05, "loss": 0.0164, "step": 9600 }, { "epoch": 121.64556962025317, "grad_norm": 0.5080320835113525, "learning_rate": 8.132016699576308e-05, "loss": 0.0147, "step": 9610 }, { "epoch": 121.77215189873418, "grad_norm": 0.6011911034584045, "learning_rate": 8.127718536784771e-05, "loss": 0.0171, "step": 9620 }, { "epoch": 121.89873417721519, "grad_norm": 0.5309346914291382, "learning_rate": 8.12341657352029e-05, "loss": 0.0177, "step": 9630 }, { "epoch": 122.0253164556962, "grad_norm": 0.5072852373123169, "learning_rate": 8.119110815010151e-05, "loss": 0.017, "step": 9640 }, { "epoch": 122.15189873417721, "grad_norm": 0.593923807144165, "learning_rate": 8.114801266486261e-05, "loss": 0.0166, "step": 9650 }, { "epoch": 122.27848101265823, "grad_norm": 0.4367648959159851, "learning_rate": 8.110487933185122e-05, "loss": 0.0149, "step": 9660 }, { "epoch": 122.40506329113924, "grad_norm": 0.5385379791259766, "learning_rate": 8.106170820347847e-05, "loss": 0.018, "step": 9670 }, { "epoch": 122.53164556962025, "grad_norm": 0.4045659005641937, "learning_rate": 8.101849933220132e-05, "loss": 0.0169, "step": 9680 }, { "epoch": 122.65822784810126, "grad_norm": 0.49131640791893005, "learning_rate": 8.097525277052264e-05, "loss": 0.0167, "step": 9690 }, { "epoch": 122.78481012658227, "grad_norm": 0.5513209700584412, "learning_rate": 8.093196857099106e-05, "loss": 0.0166, "step": 9700 }, { "epoch": 122.9113924050633, "grad_norm": 0.453632652759552, "learning_rate": 8.088864678620096e-05, "loss": 0.0178, "step": 9710 }, { "epoch": 123.0379746835443, "grad_norm": 0.5156573057174683, "learning_rate": 8.084528746879243e-05, "loss": 0.017, "step": 9720 }, { "epoch": 123.16455696202532, "grad_norm": 0.5193892121315002, "learning_rate": 8.080189067145108e-05, "loss": 0.0152, "step": 9730 }, { "epoch": 123.29113924050633, "grad_norm": 0.46878692507743835, "learning_rate": 8.075845644690813e-05, "loss": 0.0211, "step": 9740 }, { "epoch": 123.41772151898734, "grad_norm": 0.4699380397796631, "learning_rate": 8.07149848479403e-05, "loss": 0.0196, "step": 9750 }, { "epoch": 123.54430379746836, "grad_norm": 0.3794056177139282, "learning_rate": 8.067147592736962e-05, "loss": 0.0166, "step": 9760 }, { "epoch": 123.67088607594937, "grad_norm": 0.4601663649082184, "learning_rate": 8.062792973806357e-05, "loss": 0.0173, "step": 9770 }, { "epoch": 123.79746835443038, "grad_norm": 0.504570722579956, "learning_rate": 8.058434633293485e-05, "loss": 0.0204, "step": 9780 }, { "epoch": 123.92405063291139, "grad_norm": 0.46002429723739624, "learning_rate": 8.054072576494143e-05, "loss": 0.019, "step": 9790 }, { "epoch": 124.0506329113924, "grad_norm": 0.5080515742301941, "learning_rate": 8.04970680870864e-05, "loss": 0.0148, "step": 9800 }, { "epoch": 124.17721518987342, "grad_norm": 0.40189141035079956, "learning_rate": 8.045337335241793e-05, "loss": 0.0144, "step": 9810 }, { "epoch": 124.30379746835443, "grad_norm": 0.47751927375793457, "learning_rate": 8.040964161402932e-05, "loss": 0.0145, "step": 9820 }, { "epoch": 124.43037974683544, "grad_norm": 0.39970335364341736, "learning_rate": 8.036587292505869e-05, "loss": 0.0155, "step": 9830 }, { "epoch": 124.55696202531645, "grad_norm": 0.4968959391117096, "learning_rate": 8.032206733868912e-05, "loss": 0.0167, "step": 9840 }, { "epoch": 124.68354430379746, "grad_norm": 0.3885106146335602, "learning_rate": 8.027822490814858e-05, "loss": 0.0188, "step": 9850 }, { "epoch": 124.81012658227849, "grad_norm": 0.53343665599823, "learning_rate": 8.023434568670972e-05, "loss": 0.0159, "step": 9860 }, { "epoch": 124.9367088607595, "grad_norm": 0.5037118792533875, "learning_rate": 8.019042972768992e-05, "loss": 0.0176, "step": 9870 }, { "epoch": 125.0632911392405, "grad_norm": 0.5777875781059265, "learning_rate": 8.014647708445124e-05, "loss": 0.0175, "step": 9880 }, { "epoch": 125.18987341772151, "grad_norm": 0.463262677192688, "learning_rate": 8.010248781040027e-05, "loss": 0.0175, "step": 9890 }, { "epoch": 125.31645569620254, "grad_norm": 0.492021381855011, "learning_rate": 8.005846195898814e-05, "loss": 0.0166, "step": 9900 }, { "epoch": 125.44303797468355, "grad_norm": 0.4883797764778137, "learning_rate": 8.00143995837104e-05, "loss": 0.0171, "step": 9910 }, { "epoch": 125.56962025316456, "grad_norm": 0.49586665630340576, "learning_rate": 7.997030073810698e-05, "loss": 0.0209, "step": 9920 }, { "epoch": 125.69620253164557, "grad_norm": 0.4464256167411804, "learning_rate": 7.992616547576217e-05, "loss": 0.0206, "step": 9930 }, { "epoch": 125.82278481012658, "grad_norm": 0.433688759803772, "learning_rate": 7.988199385030445e-05, "loss": 0.016, "step": 9940 }, { "epoch": 125.9493670886076, "grad_norm": 0.48456791043281555, "learning_rate": 7.98377859154065e-05, "loss": 0.0175, "step": 9950 }, { "epoch": 126.07594936708861, "grad_norm": 0.6700637340545654, "learning_rate": 7.979354172478516e-05, "loss": 0.0153, "step": 9960 }, { "epoch": 126.20253164556962, "grad_norm": 0.456901490688324, "learning_rate": 7.974926133220127e-05, "loss": 0.0145, "step": 9970 }, { "epoch": 126.32911392405063, "grad_norm": 0.463437020778656, "learning_rate": 7.970494479145967e-05, "loss": 0.0176, "step": 9980 }, { "epoch": 126.45569620253164, "grad_norm": 0.4274393618106842, "learning_rate": 7.966059215640919e-05, "loss": 0.0159, "step": 9990 }, { "epoch": 126.58227848101266, "grad_norm": 0.4323137700557709, "learning_rate": 7.961620348094241e-05, "loss": 0.0162, "step": 10000 }, { "epoch": 126.70886075949367, "grad_norm": 0.44687673449516296, "learning_rate": 7.957177881899577e-05, "loss": 0.015, "step": 10010 }, { "epoch": 126.83544303797468, "grad_norm": 0.4078390598297119, "learning_rate": 7.952731822454944e-05, "loss": 0.016, "step": 10020 }, { "epoch": 126.9620253164557, "grad_norm": 0.5254436731338501, "learning_rate": 7.948282175162722e-05, "loss": 0.0162, "step": 10030 }, { "epoch": 127.0886075949367, "grad_norm": 0.4744429290294647, "learning_rate": 7.943828945429652e-05, "loss": 0.0145, "step": 10040 }, { "epoch": 127.21518987341773, "grad_norm": 0.4113679826259613, "learning_rate": 7.939372138666827e-05, "loss": 0.0153, "step": 10050 }, { "epoch": 127.34177215189874, "grad_norm": 0.43026062846183777, "learning_rate": 7.934911760289692e-05, "loss": 0.0147, "step": 10060 }, { "epoch": 127.46835443037975, "grad_norm": 0.40424734354019165, "learning_rate": 7.930447815718022e-05, "loss": 0.0152, "step": 10070 }, { "epoch": 127.59493670886076, "grad_norm": 0.47085434198379517, "learning_rate": 7.925980310375934e-05, "loss": 0.0193, "step": 10080 }, { "epoch": 127.72151898734177, "grad_norm": 0.542140007019043, "learning_rate": 7.921509249691865e-05, "loss": 0.0173, "step": 10090 }, { "epoch": 127.84810126582279, "grad_norm": 0.6255098581314087, "learning_rate": 7.917034639098579e-05, "loss": 0.0183, "step": 10100 }, { "epoch": 127.9746835443038, "grad_norm": 0.515811026096344, "learning_rate": 7.912556484033146e-05, "loss": 0.017, "step": 10110 }, { "epoch": 128.1012658227848, "grad_norm": 0.537091851234436, "learning_rate": 7.908074789936952e-05, "loss": 0.016, "step": 10120 }, { "epoch": 128.22784810126583, "grad_norm": 0.4319365918636322, "learning_rate": 7.903589562255673e-05, "loss": 0.0148, "step": 10130 }, { "epoch": 128.35443037974684, "grad_norm": 0.41957905888557434, "learning_rate": 7.899100806439285e-05, "loss": 0.0153, "step": 10140 }, { "epoch": 128.48101265822785, "grad_norm": 0.4517355263233185, "learning_rate": 7.894608527942049e-05, "loss": 0.0136, "step": 10150 }, { "epoch": 128.60759493670886, "grad_norm": 0.45427197217941284, "learning_rate": 7.89011273222251e-05, "loss": 0.0146, "step": 10160 }, { "epoch": 128.73417721518987, "grad_norm": 0.4937154948711395, "learning_rate": 7.885613424743482e-05, "loss": 0.0208, "step": 10170 }, { "epoch": 128.86075949367088, "grad_norm": 0.4311061203479767, "learning_rate": 7.881110610972044e-05, "loss": 0.0154, "step": 10180 }, { "epoch": 128.9873417721519, "grad_norm": 0.4212472140789032, "learning_rate": 7.876604296379544e-05, "loss": 0.0141, "step": 10190 }, { "epoch": 129.1139240506329, "grad_norm": 0.4580206573009491, "learning_rate": 7.872094486441579e-05, "loss": 0.0154, "step": 10200 }, { "epoch": 129.2405063291139, "grad_norm": 0.4468308985233307, "learning_rate": 7.867581186637991e-05, "loss": 0.0162, "step": 10210 }, { "epoch": 129.36708860759492, "grad_norm": 0.5299113988876343, "learning_rate": 7.863064402452866e-05, "loss": 0.0182, "step": 10220 }, { "epoch": 129.49367088607596, "grad_norm": 0.4473611116409302, "learning_rate": 7.858544139374524e-05, "loss": 0.0151, "step": 10230 }, { "epoch": 129.62025316455697, "grad_norm": 0.5463567972183228, "learning_rate": 7.854020402895508e-05, "loss": 0.0153, "step": 10240 }, { "epoch": 129.74683544303798, "grad_norm": 0.36811283230781555, "learning_rate": 7.849493198512587e-05, "loss": 0.0136, "step": 10250 }, { "epoch": 129.873417721519, "grad_norm": 0.44092515110969543, "learning_rate": 7.84496253172674e-05, "loss": 0.0174, "step": 10260 }, { "epoch": 130.0, "grad_norm": 0.49921244382858276, "learning_rate": 7.840428408043155e-05, "loss": 0.0165, "step": 10270 }, { "epoch": 130.126582278481, "grad_norm": 0.5191583633422852, "learning_rate": 7.835890832971218e-05, "loss": 0.0175, "step": 10280 }, { "epoch": 130.25316455696202, "grad_norm": 0.5489046573638916, "learning_rate": 7.831349812024512e-05, "loss": 0.0181, "step": 10290 }, { "epoch": 130.37974683544303, "grad_norm": 0.4733566641807556, "learning_rate": 7.826805350720807e-05, "loss": 0.0172, "step": 10300 }, { "epoch": 130.50632911392404, "grad_norm": 0.49083974957466125, "learning_rate": 7.822257454582049e-05, "loss": 0.0162, "step": 10310 }, { "epoch": 130.63291139240508, "grad_norm": 0.37679848074913025, "learning_rate": 7.817706129134363e-05, "loss": 0.0131, "step": 10320 }, { "epoch": 130.7594936708861, "grad_norm": 0.42180535197257996, "learning_rate": 7.813151379908036e-05, "loss": 0.0187, "step": 10330 }, { "epoch": 130.8860759493671, "grad_norm": 0.40578222274780273, "learning_rate": 7.808593212437523e-05, "loss": 0.0157, "step": 10340 }, { "epoch": 131.0126582278481, "grad_norm": 0.5074722766876221, "learning_rate": 7.804031632261421e-05, "loss": 0.0161, "step": 10350 }, { "epoch": 131.13924050632912, "grad_norm": 0.4970320165157318, "learning_rate": 7.799466644922484e-05, "loss": 0.016, "step": 10360 }, { "epoch": 131.26582278481013, "grad_norm": 0.4276511073112488, "learning_rate": 7.794898255967602e-05, "loss": 0.0168, "step": 10370 }, { "epoch": 131.39240506329114, "grad_norm": 0.4498459994792938, "learning_rate": 7.790326470947795e-05, "loss": 0.0149, "step": 10380 }, { "epoch": 131.51898734177215, "grad_norm": 0.4638437330722809, "learning_rate": 7.785751295418217e-05, "loss": 0.0173, "step": 10390 }, { "epoch": 131.64556962025316, "grad_norm": 0.42757827043533325, "learning_rate": 7.781172734938136e-05, "loss": 0.0185, "step": 10400 }, { "epoch": 131.77215189873417, "grad_norm": 0.45879775285720825, "learning_rate": 7.776590795070933e-05, "loss": 0.0158, "step": 10410 }, { "epoch": 131.8987341772152, "grad_norm": 0.42917704582214355, "learning_rate": 7.772005481384099e-05, "loss": 0.0158, "step": 10420 }, { "epoch": 132.0253164556962, "grad_norm": 0.44322729110717773, "learning_rate": 7.767416799449223e-05, "loss": 0.0138, "step": 10430 }, { "epoch": 132.15189873417722, "grad_norm": 0.49916088581085205, "learning_rate": 7.762824754841984e-05, "loss": 0.0149, "step": 10440 }, { "epoch": 132.27848101265823, "grad_norm": 0.4484962522983551, "learning_rate": 7.758229353142152e-05, "loss": 0.0147, "step": 10450 }, { "epoch": 132.40506329113924, "grad_norm": 0.37972769141197205, "learning_rate": 7.753630599933572e-05, "loss": 0.0173, "step": 10460 }, { "epoch": 132.53164556962025, "grad_norm": 0.3645249009132385, "learning_rate": 7.749028500804161e-05, "loss": 0.0148, "step": 10470 }, { "epoch": 132.65822784810126, "grad_norm": 0.4645178020000458, "learning_rate": 7.744423061345906e-05, "loss": 0.0157, "step": 10480 }, { "epoch": 132.78481012658227, "grad_norm": 0.41793596744537354, "learning_rate": 7.73981428715485e-05, "loss": 0.0171, "step": 10490 }, { "epoch": 132.91139240506328, "grad_norm": 0.43585583567619324, "learning_rate": 7.735202183831085e-05, "loss": 0.0212, "step": 10500 }, { "epoch": 133.0379746835443, "grad_norm": 0.4226241111755371, "learning_rate": 7.730586756978757e-05, "loss": 0.0128, "step": 10510 }, { "epoch": 133.16455696202533, "grad_norm": 0.4316145181655884, "learning_rate": 7.72596801220604e-05, "loss": 0.0154, "step": 10520 }, { "epoch": 133.29113924050634, "grad_norm": 0.42912232875823975, "learning_rate": 7.721345955125147e-05, "loss": 0.017, "step": 10530 }, { "epoch": 133.41772151898735, "grad_norm": 0.4736458361148834, "learning_rate": 7.71672059135231e-05, "loss": 0.0184, "step": 10540 }, { "epoch": 133.54430379746836, "grad_norm": 0.42321062088012695, "learning_rate": 7.712091926507787e-05, "loss": 0.016, "step": 10550 }, { "epoch": 133.67088607594937, "grad_norm": 0.43422043323516846, "learning_rate": 7.70745996621584e-05, "loss": 0.0146, "step": 10560 }, { "epoch": 133.79746835443038, "grad_norm": 0.4272908866405487, "learning_rate": 7.702824716104735e-05, "loss": 0.016, "step": 10570 }, { "epoch": 133.9240506329114, "grad_norm": 0.40245044231414795, "learning_rate": 7.698186181806743e-05, "loss": 0.0151, "step": 10580 }, { "epoch": 134.0506329113924, "grad_norm": 0.3988349437713623, "learning_rate": 7.693544368958116e-05, "loss": 0.0161, "step": 10590 }, { "epoch": 134.1772151898734, "grad_norm": 0.40497061610221863, "learning_rate": 7.688899283199096e-05, "loss": 0.015, "step": 10600 }, { "epoch": 134.30379746835442, "grad_norm": 0.38293156027793884, "learning_rate": 7.684250930173901e-05, "loss": 0.0148, "step": 10610 }, { "epoch": 134.43037974683546, "grad_norm": 0.4745313823223114, "learning_rate": 7.679599315530718e-05, "loss": 0.0176, "step": 10620 }, { "epoch": 134.55696202531647, "grad_norm": 0.37231558561325073, "learning_rate": 7.674944444921695e-05, "loss": 0.0167, "step": 10630 }, { "epoch": 134.68354430379748, "grad_norm": 0.4632168412208557, "learning_rate": 7.670286324002944e-05, "loss": 0.0178, "step": 10640 }, { "epoch": 134.81012658227849, "grad_norm": 0.3522818982601166, "learning_rate": 7.665624958434514e-05, "loss": 0.0173, "step": 10650 }, { "epoch": 134.9367088607595, "grad_norm": 0.3578581213951111, "learning_rate": 7.66096035388041e-05, "loss": 0.0131, "step": 10660 }, { "epoch": 135.0632911392405, "grad_norm": 0.3611850440502167, "learning_rate": 7.656292516008563e-05, "loss": 0.0143, "step": 10670 }, { "epoch": 135.18987341772151, "grad_norm": 0.5362882018089294, "learning_rate": 7.651621450490837e-05, "loss": 0.0167, "step": 10680 }, { "epoch": 135.31645569620252, "grad_norm": 0.5681696534156799, "learning_rate": 7.646947163003017e-05, "loss": 0.0169, "step": 10690 }, { "epoch": 135.44303797468353, "grad_norm": 0.5313477516174316, "learning_rate": 7.642269659224803e-05, "loss": 0.0147, "step": 10700 }, { "epoch": 135.56962025316454, "grad_norm": 0.3750859200954437, "learning_rate": 7.637588944839803e-05, "loss": 0.0161, "step": 10710 }, { "epoch": 135.69620253164558, "grad_norm": 0.4500546455383301, "learning_rate": 7.632905025535528e-05, "loss": 0.017, "step": 10720 }, { "epoch": 135.8227848101266, "grad_norm": 0.3952364921569824, "learning_rate": 7.628217907003378e-05, "loss": 0.017, "step": 10730 }, { "epoch": 135.9493670886076, "grad_norm": 0.39828500151634216, "learning_rate": 7.623527594938649e-05, "loss": 0.0175, "step": 10740 }, { "epoch": 136.0759493670886, "grad_norm": 0.49094560742378235, "learning_rate": 7.618834095040509e-05, "loss": 0.0161, "step": 10750 }, { "epoch": 136.20253164556962, "grad_norm": 0.39237889647483826, "learning_rate": 7.614137413012e-05, "loss": 0.0169, "step": 10760 }, { "epoch": 136.32911392405063, "grad_norm": 0.4549959599971771, "learning_rate": 7.609437554560042e-05, "loss": 0.0145, "step": 10770 }, { "epoch": 136.45569620253164, "grad_norm": 0.423077791929245, "learning_rate": 7.604734525395398e-05, "loss": 0.013, "step": 10780 }, { "epoch": 136.58227848101265, "grad_norm": 0.4067600965499878, "learning_rate": 7.600028331232699e-05, "loss": 0.0159, "step": 10790 }, { "epoch": 136.70886075949366, "grad_norm": 0.34186238050460815, "learning_rate": 7.595318977790408e-05, "loss": 0.017, "step": 10800 }, { "epoch": 136.83544303797467, "grad_norm": 0.40938347578048706, "learning_rate": 7.590606470790836e-05, "loss": 0.0172, "step": 10810 }, { "epoch": 136.9620253164557, "grad_norm": 0.3891735076904297, "learning_rate": 7.585890815960125e-05, "loss": 0.0151, "step": 10820 }, { "epoch": 137.08860759493672, "grad_norm": 0.4977252185344696, "learning_rate": 7.581172019028238e-05, "loss": 0.0137, "step": 10830 }, { "epoch": 137.21518987341773, "grad_norm": 0.40480783581733704, "learning_rate": 7.576450085728959e-05, "loss": 0.0146, "step": 10840 }, { "epoch": 137.34177215189874, "grad_norm": 0.44416090846061707, "learning_rate": 7.571725021799885e-05, "loss": 0.017, "step": 10850 }, { "epoch": 137.46835443037975, "grad_norm": 0.4891066253185272, "learning_rate": 7.566996832982409e-05, "loss": 0.0148, "step": 10860 }, { "epoch": 137.59493670886076, "grad_norm": 0.42008283734321594, "learning_rate": 7.56226552502173e-05, "loss": 0.0157, "step": 10870 }, { "epoch": 137.72151898734177, "grad_norm": 0.4533681571483612, "learning_rate": 7.557531103666833e-05, "loss": 0.016, "step": 10880 }, { "epoch": 137.84810126582278, "grad_norm": 0.427062451839447, "learning_rate": 7.552793574670485e-05, "loss": 0.0152, "step": 10890 }, { "epoch": 137.9746835443038, "grad_norm": 0.44261685013771057, "learning_rate": 7.54805294378923e-05, "loss": 0.0154, "step": 10900 }, { "epoch": 138.1012658227848, "grad_norm": 0.41614580154418945, "learning_rate": 7.543309216783384e-05, "loss": 0.0158, "step": 10910 }, { "epoch": 138.22784810126583, "grad_norm": 0.42027121782302856, "learning_rate": 7.53856239941702e-05, "loss": 0.016, "step": 10920 }, { "epoch": 138.35443037974684, "grad_norm": 0.5299341678619385, "learning_rate": 7.533812497457972e-05, "loss": 0.0173, "step": 10930 }, { "epoch": 138.48101265822785, "grad_norm": 0.3959530293941498, "learning_rate": 7.529059516677814e-05, "loss": 0.0156, "step": 10940 }, { "epoch": 138.60759493670886, "grad_norm": 0.4369601607322693, "learning_rate": 7.524303462851872e-05, "loss": 0.015, "step": 10950 }, { "epoch": 138.73417721518987, "grad_norm": 0.3729081153869629, "learning_rate": 7.519544341759192e-05, "loss": 0.0141, "step": 10960 }, { "epoch": 138.86075949367088, "grad_norm": 0.4570898115634918, "learning_rate": 7.514782159182562e-05, "loss": 0.0188, "step": 10970 }, { "epoch": 138.9873417721519, "grad_norm": 0.4360170066356659, "learning_rate": 7.510016920908481e-05, "loss": 0.0162, "step": 10980 }, { "epoch": 139.1139240506329, "grad_norm": 0.517579972743988, "learning_rate": 7.505248632727158e-05, "loss": 0.0179, "step": 10990 }, { "epoch": 139.2405063291139, "grad_norm": 0.529423177242279, "learning_rate": 7.50047730043252e-05, "loss": 0.0156, "step": 11000 }, { "epoch": 139.36708860759492, "grad_norm": 0.40209805965423584, "learning_rate": 7.495702929822183e-05, "loss": 0.0156, "step": 11010 }, { "epoch": 139.49367088607596, "grad_norm": 0.4249514937400818, "learning_rate": 7.490925526697455e-05, "loss": 0.0157, "step": 11020 }, { "epoch": 139.62025316455697, "grad_norm": 0.4366181492805481, "learning_rate": 7.486145096863334e-05, "loss": 0.0143, "step": 11030 }, { "epoch": 139.74683544303798, "grad_norm": 0.4289974272251129, "learning_rate": 7.481361646128491e-05, "loss": 0.0148, "step": 11040 }, { "epoch": 139.873417721519, "grad_norm": 0.3991098999977112, "learning_rate": 7.476575180305271e-05, "loss": 0.0161, "step": 11050 }, { "epoch": 140.0, "grad_norm": 0.3731028437614441, "learning_rate": 7.471785705209682e-05, "loss": 0.0175, "step": 11060 }, { "epoch": 140.126582278481, "grad_norm": 0.39974701404571533, "learning_rate": 7.466993226661387e-05, "loss": 0.014, "step": 11070 }, { "epoch": 140.25316455696202, "grad_norm": 0.35126975178718567, "learning_rate": 7.462197750483699e-05, "loss": 0.016, "step": 11080 }, { "epoch": 140.37974683544303, "grad_norm": 0.42343541979789734, "learning_rate": 7.457399282503574e-05, "loss": 0.0146, "step": 11090 }, { "epoch": 140.50632911392404, "grad_norm": 0.4115615785121918, "learning_rate": 7.452597828551604e-05, "loss": 0.0146, "step": 11100 }, { "epoch": 140.63291139240508, "grad_norm": 0.4912974536418915, "learning_rate": 7.447793394462006e-05, "loss": 0.0147, "step": 11110 }, { "epoch": 140.7594936708861, "grad_norm": 0.44597724080085754, "learning_rate": 7.442985986072624e-05, "loss": 0.0162, "step": 11120 }, { "epoch": 140.8860759493671, "grad_norm": 0.42576494812965393, "learning_rate": 7.438175609224908e-05, "loss": 0.0157, "step": 11130 }, { "epoch": 141.0126582278481, "grad_norm": 0.46355491876602173, "learning_rate": 7.433362269763924e-05, "loss": 0.014, "step": 11140 }, { "epoch": 141.13924050632912, "grad_norm": 0.4213525950908661, "learning_rate": 7.428545973538329e-05, "loss": 0.0129, "step": 11150 }, { "epoch": 141.26582278481013, "grad_norm": 0.4413865804672241, "learning_rate": 7.42372672640038e-05, "loss": 0.0158, "step": 11160 }, { "epoch": 141.39240506329114, "grad_norm": 0.44507354497909546, "learning_rate": 7.418904534205917e-05, "loss": 0.0159, "step": 11170 }, { "epoch": 141.51898734177215, "grad_norm": 0.41539233922958374, "learning_rate": 7.414079402814356e-05, "loss": 0.0136, "step": 11180 }, { "epoch": 141.64556962025316, "grad_norm": 0.4126031696796417, "learning_rate": 7.409251338088687e-05, "loss": 0.0177, "step": 11190 }, { "epoch": 141.77215189873417, "grad_norm": 0.3674456775188446, "learning_rate": 7.404420345895467e-05, "loss": 0.0167, "step": 11200 }, { "epoch": 141.8987341772152, "grad_norm": 0.41905832290649414, "learning_rate": 7.399586432104804e-05, "loss": 0.0166, "step": 11210 }, { "epoch": 142.0253164556962, "grad_norm": 0.47269245982170105, "learning_rate": 7.394749602590358e-05, "loss": 0.0139, "step": 11220 }, { "epoch": 142.15189873417722, "grad_norm": 0.3746514916419983, "learning_rate": 7.389909863229336e-05, "loss": 0.0204, "step": 11230 }, { "epoch": 142.27848101265823, "grad_norm": 0.49761053919792175, "learning_rate": 7.385067219902477e-05, "loss": 0.0137, "step": 11240 }, { "epoch": 142.40506329113924, "grad_norm": 0.41309139132499695, "learning_rate": 7.380221678494049e-05, "loss": 0.0152, "step": 11250 }, { "epoch": 142.53164556962025, "grad_norm": 0.45250871777534485, "learning_rate": 7.37537324489184e-05, "loss": 0.014, "step": 11260 }, { "epoch": 142.65822784810126, "grad_norm": 0.4231874942779541, "learning_rate": 7.370521924987155e-05, "loss": 0.0144, "step": 11270 }, { "epoch": 142.78481012658227, "grad_norm": 0.3704090118408203, "learning_rate": 7.365667724674806e-05, "loss": 0.0124, "step": 11280 }, { "epoch": 142.91139240506328, "grad_norm": 0.38344863057136536, "learning_rate": 7.360810649853105e-05, "loss": 0.0137, "step": 11290 }, { "epoch": 143.0379746835443, "grad_norm": 0.3965582847595215, "learning_rate": 7.355950706423853e-05, "loss": 0.0137, "step": 11300 }, { "epoch": 143.16455696202533, "grad_norm": 0.45548728108406067, "learning_rate": 7.351087900292342e-05, "loss": 0.0144, "step": 11310 }, { "epoch": 143.29113924050634, "grad_norm": 0.41179877519607544, "learning_rate": 7.346222237367339e-05, "loss": 0.0137, "step": 11320 }, { "epoch": 143.41772151898735, "grad_norm": 0.29509609937667847, "learning_rate": 7.341353723561081e-05, "loss": 0.0144, "step": 11330 }, { "epoch": 143.54430379746836, "grad_norm": 0.2793385088443756, "learning_rate": 7.336482364789277e-05, "loss": 0.0155, "step": 11340 }, { "epoch": 143.67088607594937, "grad_norm": 0.3851372003555298, "learning_rate": 7.331608166971082e-05, "loss": 0.0169, "step": 11350 }, { "epoch": 143.79746835443038, "grad_norm": 0.4007650911808014, "learning_rate": 7.326731136029108e-05, "loss": 0.0156, "step": 11360 }, { "epoch": 143.9240506329114, "grad_norm": 0.41908618807792664, "learning_rate": 7.321851277889408e-05, "loss": 0.0169, "step": 11370 }, { "epoch": 144.0506329113924, "grad_norm": 0.4623008370399475, "learning_rate": 7.316968598481469e-05, "loss": 0.0158, "step": 11380 }, { "epoch": 144.1772151898734, "grad_norm": 0.50325608253479, "learning_rate": 7.312083103738206e-05, "loss": 0.0156, "step": 11390 }, { "epoch": 144.30379746835442, "grad_norm": 0.460562527179718, "learning_rate": 7.307194799595958e-05, "loss": 0.017, "step": 11400 }, { "epoch": 144.43037974683546, "grad_norm": 0.41903021931648254, "learning_rate": 7.302303691994473e-05, "loss": 0.0166, "step": 11410 }, { "epoch": 144.55696202531647, "grad_norm": 0.3798241913318634, "learning_rate": 7.29740978687691e-05, "loss": 0.0137, "step": 11420 }, { "epoch": 144.68354430379748, "grad_norm": 0.45348525047302246, "learning_rate": 7.292513090189825e-05, "loss": 0.0145, "step": 11430 }, { "epoch": 144.81012658227849, "grad_norm": 0.3600919246673584, "learning_rate": 7.287613607883163e-05, "loss": 0.0139, "step": 11440 }, { "epoch": 144.9367088607595, "grad_norm": 0.37354812026023865, "learning_rate": 7.282711345910263e-05, "loss": 0.0146, "step": 11450 }, { "epoch": 145.0632911392405, "grad_norm": 0.4909972846508026, "learning_rate": 7.27780631022783e-05, "loss": 0.0163, "step": 11460 }, { "epoch": 145.18987341772151, "grad_norm": 0.4186764061450958, "learning_rate": 7.272898506795948e-05, "loss": 0.0137, "step": 11470 }, { "epoch": 145.31645569620252, "grad_norm": 0.4236494302749634, "learning_rate": 7.267987941578058e-05, "loss": 0.0141, "step": 11480 }, { "epoch": 145.44303797468353, "grad_norm": 0.39548459649086, "learning_rate": 7.263074620540962e-05, "loss": 0.0137, "step": 11490 }, { "epoch": 145.56962025316454, "grad_norm": 0.4570174813270569, "learning_rate": 7.25815854965481e-05, "loss": 0.0147, "step": 11500 }, { "epoch": 145.69620253164558, "grad_norm": 0.43672898411750793, "learning_rate": 7.253239734893089e-05, "loss": 0.0143, "step": 11510 }, { "epoch": 145.8227848101266, "grad_norm": 0.3383423984050751, "learning_rate": 7.248318182232623e-05, "loss": 0.0157, "step": 11520 }, { "epoch": 145.9493670886076, "grad_norm": 0.3947126865386963, "learning_rate": 7.243393897653565e-05, "loss": 0.0172, "step": 11530 }, { "epoch": 146.0759493670886, "grad_norm": 0.43415772914886475, "learning_rate": 7.23846688713938e-05, "loss": 0.0172, "step": 11540 }, { "epoch": 146.20253164556962, "grad_norm": 0.4213506579399109, "learning_rate": 7.233537156676854e-05, "loss": 0.0159, "step": 11550 }, { "epoch": 146.32911392405063, "grad_norm": 0.4404980540275574, "learning_rate": 7.228604712256076e-05, "loss": 0.0162, "step": 11560 }, { "epoch": 146.45569620253164, "grad_norm": 0.44348180294036865, "learning_rate": 7.223669559870426e-05, "loss": 0.0138, "step": 11570 }, { "epoch": 146.58227848101265, "grad_norm": 0.3501180410385132, "learning_rate": 7.218731705516585e-05, "loss": 0.017, "step": 11580 }, { "epoch": 146.70886075949366, "grad_norm": 0.45616811513900757, "learning_rate": 7.21379115519451e-05, "loss": 0.0156, "step": 11590 }, { "epoch": 146.83544303797467, "grad_norm": 0.40642744302749634, "learning_rate": 7.20884791490743e-05, "loss": 0.0142, "step": 11600 }, { "epoch": 146.9620253164557, "grad_norm": 0.44371506571769714, "learning_rate": 7.203901990661857e-05, "loss": 0.0174, "step": 11610 }, { "epoch": 147.08860759493672, "grad_norm": 0.45646387338638306, "learning_rate": 7.198953388467549e-05, "loss": 0.0195, "step": 11620 }, { "epoch": 147.21518987341773, "grad_norm": 0.38989681005477905, "learning_rate": 7.194002114337526e-05, "loss": 0.019, "step": 11630 }, { "epoch": 147.34177215189874, "grad_norm": 0.38590875267982483, "learning_rate": 7.189048174288054e-05, "loss": 0.015, "step": 11640 }, { "epoch": 147.46835443037975, "grad_norm": 0.4439886212348938, "learning_rate": 7.184091574338636e-05, "loss": 0.0149, "step": 11650 }, { "epoch": 147.59493670886076, "grad_norm": 0.3688686490058899, "learning_rate": 7.179132320512009e-05, "loss": 0.0139, "step": 11660 }, { "epoch": 147.72151898734177, "grad_norm": 0.3829432427883148, "learning_rate": 7.174170418834134e-05, "loss": 0.0149, "step": 11670 }, { "epoch": 147.84810126582278, "grad_norm": 0.39028215408325195, "learning_rate": 7.169205875334189e-05, "loss": 0.0129, "step": 11680 }, { "epoch": 147.9746835443038, "grad_norm": 0.4368191063404083, "learning_rate": 7.164238696044562e-05, "loss": 0.0153, "step": 11690 }, { "epoch": 148.1012658227848, "grad_norm": 0.3947655260562897, "learning_rate": 7.159268887000846e-05, "loss": 0.0165, "step": 11700 }, { "epoch": 148.22784810126583, "grad_norm": 0.39236506819725037, "learning_rate": 7.154296454241827e-05, "loss": 0.0142, "step": 11710 }, { "epoch": 148.35443037974684, "grad_norm": 0.4236973226070404, "learning_rate": 7.149321403809479e-05, "loss": 0.0144, "step": 11720 }, { "epoch": 148.48101265822785, "grad_norm": 0.3683420717716217, "learning_rate": 7.144343741748956e-05, "loss": 0.0156, "step": 11730 }, { "epoch": 148.60759493670886, "grad_norm": 0.42502275109291077, "learning_rate": 7.139363474108589e-05, "loss": 0.0164, "step": 11740 }, { "epoch": 148.73417721518987, "grad_norm": 0.38866734504699707, "learning_rate": 7.134380606939874e-05, "loss": 0.015, "step": 11750 }, { "epoch": 148.86075949367088, "grad_norm": 0.4107532203197479, "learning_rate": 7.12939514629746e-05, "loss": 0.0147, "step": 11760 }, { "epoch": 148.9873417721519, "grad_norm": 0.31432482600212097, "learning_rate": 7.124407098239155e-05, "loss": 0.0137, "step": 11770 }, { "epoch": 149.1139240506329, "grad_norm": 0.37796247005462646, "learning_rate": 7.119416468825908e-05, "loss": 0.0127, "step": 11780 }, { "epoch": 149.2405063291139, "grad_norm": 0.3716771602630615, "learning_rate": 7.114423264121804e-05, "loss": 0.0154, "step": 11790 }, { "epoch": 149.36708860759492, "grad_norm": 0.47556039690971375, "learning_rate": 7.109427490194056e-05, "loss": 0.0161, "step": 11800 }, { "epoch": 149.49367088607596, "grad_norm": 0.41680851578712463, "learning_rate": 7.104429153113001e-05, "loss": 0.0159, "step": 11810 }, { "epoch": 149.62025316455697, "grad_norm": 0.38468143343925476, "learning_rate": 7.099428258952092e-05, "loss": 0.017, "step": 11820 }, { "epoch": 149.74683544303798, "grad_norm": 0.4678370952606201, "learning_rate": 7.094424813787883e-05, "loss": 0.0158, "step": 11830 }, { "epoch": 149.873417721519, "grad_norm": 0.42124563455581665, "learning_rate": 7.089418823700034e-05, "loss": 0.0154, "step": 11840 }, { "epoch": 150.0, "grad_norm": 0.3780612051486969, "learning_rate": 7.084410294771298e-05, "loss": 0.0146, "step": 11850 }, { "epoch": 150.126582278481, "grad_norm": 0.5106370449066162, "learning_rate": 7.079399233087504e-05, "loss": 0.0139, "step": 11860 }, { "epoch": 150.25316455696202, "grad_norm": 0.40100616216659546, "learning_rate": 7.074385644737568e-05, "loss": 0.0167, "step": 11870 }, { "epoch": 150.37974683544303, "grad_norm": 0.506669819355011, "learning_rate": 7.069369535813473e-05, "loss": 0.0169, "step": 11880 }, { "epoch": 150.50632911392404, "grad_norm": 0.3702039122581482, "learning_rate": 7.06435091241026e-05, "loss": 0.0149, "step": 11890 }, { "epoch": 150.63291139240508, "grad_norm": 0.39649462699890137, "learning_rate": 7.059329780626034e-05, "loss": 0.0156, "step": 11900 }, { "epoch": 150.7594936708861, "grad_norm": 0.35080549120903015, "learning_rate": 7.054306146561944e-05, "loss": 0.0153, "step": 11910 }, { "epoch": 150.8860759493671, "grad_norm": 0.34473860263824463, "learning_rate": 7.049280016322178e-05, "loss": 0.0163, "step": 11920 }, { "epoch": 151.0126582278481, "grad_norm": 0.3812321722507477, "learning_rate": 7.044251396013957e-05, "loss": 0.018, "step": 11930 }, { "epoch": 151.13924050632912, "grad_norm": 0.4634578824043274, "learning_rate": 7.039220291747528e-05, "loss": 0.0194, "step": 11940 }, { "epoch": 151.26582278481013, "grad_norm": 0.41969308257102966, "learning_rate": 7.034186709636159e-05, "loss": 0.0138, "step": 11950 }, { "epoch": 151.39240506329114, "grad_norm": 0.46813827753067017, "learning_rate": 7.029150655796129e-05, "loss": 0.0188, "step": 11960 }, { "epoch": 151.51898734177215, "grad_norm": 0.4036957621574402, "learning_rate": 7.024112136346712e-05, "loss": 0.0157, "step": 11970 }, { "epoch": 151.64556962025316, "grad_norm": 0.4250236451625824, "learning_rate": 7.01907115741019e-05, "loss": 0.0156, "step": 11980 }, { "epoch": 151.77215189873417, "grad_norm": 0.35539209842681885, "learning_rate": 7.014027725111825e-05, "loss": 0.014, "step": 11990 }, { "epoch": 151.8987341772152, "grad_norm": 0.38810375332832336, "learning_rate": 7.008981845579864e-05, "loss": 0.0171, "step": 12000 }, { "epoch": 152.0253164556962, "grad_norm": 0.4147331714630127, "learning_rate": 7.003933524945528e-05, "loss": 0.0169, "step": 12010 }, { "epoch": 152.15189873417722, "grad_norm": 0.3445667624473572, "learning_rate": 6.998882769342998e-05, "loss": 0.0143, "step": 12020 }, { "epoch": 152.27848101265823, "grad_norm": 0.3840770721435547, "learning_rate": 6.993829584909423e-05, "loss": 0.014, "step": 12030 }, { "epoch": 152.40506329113924, "grad_norm": 0.3643435537815094, "learning_rate": 6.988773977784895e-05, "loss": 0.0144, "step": 12040 }, { "epoch": 152.53164556962025, "grad_norm": 0.3728500306606293, "learning_rate": 6.983715954112454e-05, "loss": 0.0145, "step": 12050 }, { "epoch": 152.65822784810126, "grad_norm": 0.38859879970550537, "learning_rate": 6.978655520038079e-05, "loss": 0.0147, "step": 12060 }, { "epoch": 152.78481012658227, "grad_norm": 0.40717580914497375, "learning_rate": 6.97359268171067e-05, "loss": 0.0121, "step": 12070 }, { "epoch": 152.91139240506328, "grad_norm": 0.3889860212802887, "learning_rate": 6.968527445282056e-05, "loss": 0.0137, "step": 12080 }, { "epoch": 153.0379746835443, "grad_norm": 0.4082929790019989, "learning_rate": 6.963459816906977e-05, "loss": 0.0128, "step": 12090 }, { "epoch": 153.16455696202533, "grad_norm": 0.3575390875339508, "learning_rate": 6.958389802743077e-05, "loss": 0.014, "step": 12100 }, { "epoch": 153.29113924050634, "grad_norm": 0.39602920413017273, "learning_rate": 6.953317408950903e-05, "loss": 0.0142, "step": 12110 }, { "epoch": 153.41772151898735, "grad_norm": 0.5537218451499939, "learning_rate": 6.94824264169389e-05, "loss": 0.0142, "step": 12120 }, { "epoch": 153.54430379746836, "grad_norm": 0.340417742729187, "learning_rate": 6.94316550713836e-05, "loss": 0.014, "step": 12130 }, { "epoch": 153.67088607594937, "grad_norm": 0.3663218319416046, "learning_rate": 6.938086011453513e-05, "loss": 0.0138, "step": 12140 }, { "epoch": 153.79746835443038, "grad_norm": 0.4715401828289032, "learning_rate": 6.93300416081141e-05, "loss": 0.0133, "step": 12150 }, { "epoch": 153.9240506329114, "grad_norm": 0.41070887446403503, "learning_rate": 6.927919961386984e-05, "loss": 0.0156, "step": 12160 }, { "epoch": 154.0506329113924, "grad_norm": 0.4268708825111389, "learning_rate": 6.922833419358013e-05, "loss": 0.0128, "step": 12170 }, { "epoch": 154.1772151898734, "grad_norm": 0.46537744998931885, "learning_rate": 6.917744540905124e-05, "loss": 0.015, "step": 12180 }, { "epoch": 154.30379746835442, "grad_norm": 0.40864941477775574, "learning_rate": 6.912653332211787e-05, "loss": 0.0127, "step": 12190 }, { "epoch": 154.43037974683546, "grad_norm": 0.4742419123649597, "learning_rate": 6.9075597994643e-05, "loss": 0.0155, "step": 12200 }, { "epoch": 154.55696202531647, "grad_norm": 0.4097127616405487, "learning_rate": 6.902463948851787e-05, "loss": 0.0175, "step": 12210 }, { "epoch": 154.68354430379748, "grad_norm": 0.3509465456008911, "learning_rate": 6.897365786566184e-05, "loss": 0.0118, "step": 12220 }, { "epoch": 154.81012658227849, "grad_norm": 0.3665238618850708, "learning_rate": 6.892265318802242e-05, "loss": 0.015, "step": 12230 }, { "epoch": 154.9367088607595, "grad_norm": 0.37709513306617737, "learning_rate": 6.887162551757507e-05, "loss": 0.0154, "step": 12240 }, { "epoch": 155.0632911392405, "grad_norm": 0.33980217576026917, "learning_rate": 6.882057491632326e-05, "loss": 0.0122, "step": 12250 }, { "epoch": 155.18987341772151, "grad_norm": 0.3403623104095459, "learning_rate": 6.876950144629824e-05, "loss": 0.0133, "step": 12260 }, { "epoch": 155.31645569620252, "grad_norm": 0.3321526348590851, "learning_rate": 6.87184051695591e-05, "loss": 0.0174, "step": 12270 }, { "epoch": 155.44303797468353, "grad_norm": 0.4002130925655365, "learning_rate": 6.866728614819268e-05, "loss": 0.0119, "step": 12280 }, { "epoch": 155.56962025316454, "grad_norm": 0.3953796625137329, "learning_rate": 6.861614444431337e-05, "loss": 0.0109, "step": 12290 }, { "epoch": 155.69620253164558, "grad_norm": 0.3590865731239319, "learning_rate": 6.856498012006318e-05, "loss": 0.0171, "step": 12300 }, { "epoch": 155.8227848101266, "grad_norm": 0.4126300513744354, "learning_rate": 6.851379323761157e-05, "loss": 0.0154, "step": 12310 }, { "epoch": 155.9493670886076, "grad_norm": 0.36633118987083435, "learning_rate": 6.846258385915545e-05, "loss": 0.0137, "step": 12320 }, { "epoch": 156.0759493670886, "grad_norm": 0.43772730231285095, "learning_rate": 6.841135204691902e-05, "loss": 0.0123, "step": 12330 }, { "epoch": 156.20253164556962, "grad_norm": 0.3847324252128601, "learning_rate": 6.836009786315377e-05, "loss": 0.0142, "step": 12340 }, { "epoch": 156.32911392405063, "grad_norm": 0.410042941570282, "learning_rate": 6.830882137013839e-05, "loss": 0.0163, "step": 12350 }, { "epoch": 156.45569620253164, "grad_norm": 0.40545862913131714, "learning_rate": 6.825752263017863e-05, "loss": 0.014, "step": 12360 }, { "epoch": 156.58227848101265, "grad_norm": 0.3853617310523987, "learning_rate": 6.820620170560731e-05, "loss": 0.0128, "step": 12370 }, { "epoch": 156.70886075949366, "grad_norm": 0.3036900758743286, "learning_rate": 6.815485865878418e-05, "loss": 0.0121, "step": 12380 }, { "epoch": 156.83544303797467, "grad_norm": 0.3176374137401581, "learning_rate": 6.810349355209588e-05, "loss": 0.0134, "step": 12390 }, { "epoch": 156.9620253164557, "grad_norm": 0.36718934774398804, "learning_rate": 6.805210644795588e-05, "loss": 0.0141, "step": 12400 }, { "epoch": 157.08860759493672, "grad_norm": 0.33649110794067383, "learning_rate": 6.800069740880433e-05, "loss": 0.0116, "step": 12410 }, { "epoch": 157.21518987341773, "grad_norm": 0.46684637665748596, "learning_rate": 6.794926649710807e-05, "loss": 0.0137, "step": 12420 }, { "epoch": 157.34177215189874, "grad_norm": 0.38128161430358887, "learning_rate": 6.789781377536052e-05, "loss": 0.0163, "step": 12430 }, { "epoch": 157.46835443037975, "grad_norm": 0.30973461270332336, "learning_rate": 6.784633930608158e-05, "loss": 0.0122, "step": 12440 }, { "epoch": 157.59493670886076, "grad_norm": 0.30061039328575134, "learning_rate": 6.779484315181759e-05, "loss": 0.0122, "step": 12450 }, { "epoch": 157.72151898734177, "grad_norm": 0.3189159035682678, "learning_rate": 6.774332537514122e-05, "loss": 0.0116, "step": 12460 }, { "epoch": 157.84810126582278, "grad_norm": 0.39929258823394775, "learning_rate": 6.769178603865143e-05, "loss": 0.0155, "step": 12470 }, { "epoch": 157.9746835443038, "grad_norm": 0.3803061246871948, "learning_rate": 6.764022520497337e-05, "loss": 0.0119, "step": 12480 }, { "epoch": 158.1012658227848, "grad_norm": 0.3572578728199005, "learning_rate": 6.758864293675833e-05, "loss": 0.016, "step": 12490 }, { "epoch": 158.22784810126583, "grad_norm": 0.39943185448646545, "learning_rate": 6.753703929668362e-05, "loss": 0.0127, "step": 12500 }, { "epoch": 158.35443037974684, "grad_norm": 0.4240237772464752, "learning_rate": 6.748541434745253e-05, "loss": 0.0158, "step": 12510 }, { "epoch": 158.48101265822785, "grad_norm": 0.4813203513622284, "learning_rate": 6.743376815179424e-05, "loss": 0.0112, "step": 12520 }, { "epoch": 158.60759493670886, "grad_norm": 0.3731670379638672, "learning_rate": 6.738210077246376e-05, "loss": 0.015, "step": 12530 }, { "epoch": 158.73417721518987, "grad_norm": 0.4159543216228485, "learning_rate": 6.733041227224181e-05, "loss": 0.0117, "step": 12540 }, { "epoch": 158.86075949367088, "grad_norm": 0.4781552255153656, "learning_rate": 6.72787027139348e-05, "loss": 0.0131, "step": 12550 }, { "epoch": 158.9873417721519, "grad_norm": 0.42907628417015076, "learning_rate": 6.72269721603747e-05, "loss": 0.0167, "step": 12560 }, { "epoch": 159.1139240506329, "grad_norm": 0.44764983654022217, "learning_rate": 6.717522067441904e-05, "loss": 0.0139, "step": 12570 }, { "epoch": 159.2405063291139, "grad_norm": 0.4374462068080902, "learning_rate": 6.712344831895074e-05, "loss": 0.0163, "step": 12580 }, { "epoch": 159.36708860759492, "grad_norm": 0.40409055352211, "learning_rate": 6.707165515687811e-05, "loss": 0.0129, "step": 12590 }, { "epoch": 159.49367088607596, "grad_norm": 0.3582768142223358, "learning_rate": 6.70198412511347e-05, "loss": 0.0135, "step": 12600 }, { "epoch": 159.62025316455697, "grad_norm": 0.357154905796051, "learning_rate": 6.69680066646793e-05, "loss": 0.0106, "step": 12610 }, { "epoch": 159.74683544303798, "grad_norm": 0.31793463230133057, "learning_rate": 6.691615146049583e-05, "loss": 0.0125, "step": 12620 }, { "epoch": 159.873417721519, "grad_norm": 0.3811216652393341, "learning_rate": 6.686427570159324e-05, "loss": 0.0137, "step": 12630 }, { "epoch": 160.0, "grad_norm": 0.4236549437046051, "learning_rate": 6.681237945100548e-05, "loss": 0.014, "step": 12640 }, { "epoch": 160.126582278481, "grad_norm": 0.41608408093452454, "learning_rate": 6.676046277179139e-05, "loss": 0.0125, "step": 12650 }, { "epoch": 160.25316455696202, "grad_norm": 0.4236839711666107, "learning_rate": 6.670852572703462e-05, "loss": 0.0144, "step": 12660 }, { "epoch": 160.37974683544303, "grad_norm": 0.34447458386421204, "learning_rate": 6.665656837984359e-05, "loss": 0.0134, "step": 12670 }, { "epoch": 160.50632911392404, "grad_norm": 0.35967856645584106, "learning_rate": 6.660459079335135e-05, "loss": 0.013, "step": 12680 }, { "epoch": 160.63291139240508, "grad_norm": 0.39056503772735596, "learning_rate": 6.655259303071558e-05, "loss": 0.0168, "step": 12690 }, { "epoch": 160.7594936708861, "grad_norm": 0.3346683084964752, "learning_rate": 6.650057515511848e-05, "loss": 0.0118, "step": 12700 }, { "epoch": 160.8860759493671, "grad_norm": 0.42401671409606934, "learning_rate": 6.644853722976667e-05, "loss": 0.0143, "step": 12710 }, { "epoch": 161.0126582278481, "grad_norm": 0.4140920639038086, "learning_rate": 6.639647931789113e-05, "loss": 0.0163, "step": 12720 }, { "epoch": 161.13924050632912, "grad_norm": 0.39404335618019104, "learning_rate": 6.634440148274713e-05, "loss": 0.0162, "step": 12730 }, { "epoch": 161.26582278481013, "grad_norm": 0.3194509744644165, "learning_rate": 6.629230378761415e-05, "loss": 0.0116, "step": 12740 }, { "epoch": 161.39240506329114, "grad_norm": 0.35159990191459656, "learning_rate": 6.624018629579581e-05, "loss": 0.0138, "step": 12750 }, { "epoch": 161.51898734177215, "grad_norm": 0.3766539990901947, "learning_rate": 6.618804907061976e-05, "loss": 0.0162, "step": 12760 }, { "epoch": 161.64556962025316, "grad_norm": 0.39496761560440063, "learning_rate": 6.613589217543766e-05, "loss": 0.0154, "step": 12770 }, { "epoch": 161.77215189873417, "grad_norm": 0.3636765778064728, "learning_rate": 6.608371567362504e-05, "loss": 0.014, "step": 12780 }, { "epoch": 161.8987341772152, "grad_norm": 0.40954193472862244, "learning_rate": 6.60315196285813e-05, "loss": 0.0147, "step": 12790 }, { "epoch": 162.0253164556962, "grad_norm": 0.3889981806278229, "learning_rate": 6.597930410372954e-05, "loss": 0.0148, "step": 12800 }, { "epoch": 162.15189873417722, "grad_norm": 0.3397229313850403, "learning_rate": 6.592706916251653e-05, "loss": 0.0126, "step": 12810 }, { "epoch": 162.27848101265823, "grad_norm": 0.3194037675857544, "learning_rate": 6.587481486841267e-05, "loss": 0.0141, "step": 12820 }, { "epoch": 162.40506329113924, "grad_norm": 0.40126311779022217, "learning_rate": 6.582254128491184e-05, "loss": 0.0128, "step": 12830 }, { "epoch": 162.53164556962025, "grad_norm": 0.36203238368034363, "learning_rate": 6.577024847553139e-05, "loss": 0.0127, "step": 12840 }, { "epoch": 162.65822784810126, "grad_norm": 0.31912171840667725, "learning_rate": 6.571793650381201e-05, "loss": 0.0154, "step": 12850 }, { "epoch": 162.78481012658227, "grad_norm": 0.3694348931312561, "learning_rate": 6.566560543331766e-05, "loss": 0.0139, "step": 12860 }, { "epoch": 162.91139240506328, "grad_norm": 0.35618680715560913, "learning_rate": 6.561325532763554e-05, "loss": 0.0148, "step": 12870 }, { "epoch": 163.0379746835443, "grad_norm": 0.4433775842189789, "learning_rate": 6.556088625037597e-05, "loss": 0.0124, "step": 12880 }, { "epoch": 163.16455696202533, "grad_norm": 0.4143799841403961, "learning_rate": 6.55084982651723e-05, "loss": 0.0149, "step": 12890 }, { "epoch": 163.29113924050634, "grad_norm": 0.4060841202735901, "learning_rate": 6.54560914356809e-05, "loss": 0.0117, "step": 12900 }, { "epoch": 163.41772151898735, "grad_norm": 0.456158846616745, "learning_rate": 6.540366582558097e-05, "loss": 0.0107, "step": 12910 }, { "epoch": 163.54430379746836, "grad_norm": 0.39946040511131287, "learning_rate": 6.53512214985746e-05, "loss": 0.0146, "step": 12920 }, { "epoch": 163.67088607594937, "grad_norm": 0.370193749666214, "learning_rate": 6.529875851838658e-05, "loss": 0.0152, "step": 12930 }, { "epoch": 163.79746835443038, "grad_norm": 0.4503318667411804, "learning_rate": 6.52462769487644e-05, "loss": 0.0128, "step": 12940 }, { "epoch": 163.9240506329114, "grad_norm": 0.3279171586036682, "learning_rate": 6.519377685347807e-05, "loss": 0.0149, "step": 12950 }, { "epoch": 164.0506329113924, "grad_norm": 0.43185168504714966, "learning_rate": 6.514125829632021e-05, "loss": 0.0121, "step": 12960 }, { "epoch": 164.1772151898734, "grad_norm": 0.4366365075111389, "learning_rate": 6.508872134110577e-05, "loss": 0.0115, "step": 12970 }, { "epoch": 164.30379746835442, "grad_norm": 0.34837061166763306, "learning_rate": 6.503616605167213e-05, "loss": 0.0131, "step": 12980 }, { "epoch": 164.43037974683546, "grad_norm": 0.41766640543937683, "learning_rate": 6.498359249187893e-05, "loss": 0.0133, "step": 12990 }, { "epoch": 164.55696202531647, "grad_norm": 0.48624709248542786, "learning_rate": 6.493100072560799e-05, "loss": 0.0158, "step": 13000 }, { "epoch": 164.68354430379748, "grad_norm": 0.3370601236820221, "learning_rate": 6.487839081676327e-05, "loss": 0.0133, "step": 13010 }, { "epoch": 164.81012658227849, "grad_norm": 0.3415626585483551, "learning_rate": 6.482576282927076e-05, "loss": 0.0149, "step": 13020 }, { "epoch": 164.9367088607595, "grad_norm": 0.38370591402053833, "learning_rate": 6.477311682707845e-05, "loss": 0.0135, "step": 13030 }, { "epoch": 165.0632911392405, "grad_norm": 0.2910154461860657, "learning_rate": 6.472045287415616e-05, "loss": 0.013, "step": 13040 }, { "epoch": 165.18987341772151, "grad_norm": 0.40091145038604736, "learning_rate": 6.466777103449558e-05, "loss": 0.0155, "step": 13050 }, { "epoch": 165.31645569620252, "grad_norm": 0.383078008890152, "learning_rate": 6.461507137211012e-05, "loss": 0.014, "step": 13060 }, { "epoch": 165.44303797468353, "grad_norm": 0.3642904460430145, "learning_rate": 6.456235395103483e-05, "loss": 0.0108, "step": 13070 }, { "epoch": 165.56962025316454, "grad_norm": 0.32672637701034546, "learning_rate": 6.450961883532634e-05, "loss": 0.0153, "step": 13080 }, { "epoch": 165.69620253164558, "grad_norm": 0.38134273886680603, "learning_rate": 6.445686608906283e-05, "loss": 0.0142, "step": 13090 }, { "epoch": 165.8227848101266, "grad_norm": 0.419069766998291, "learning_rate": 6.44040957763438e-05, "loss": 0.0132, "step": 13100 }, { "epoch": 165.9493670886076, "grad_norm": 0.38638249039649963, "learning_rate": 6.435130796129018e-05, "loss": 0.0142, "step": 13110 }, { "epoch": 166.0759493670886, "grad_norm": 0.37674733996391296, "learning_rate": 6.429850270804416e-05, "loss": 0.0133, "step": 13120 }, { "epoch": 166.20253164556962, "grad_norm": 0.39023557305336, "learning_rate": 6.424568008076909e-05, "loss": 0.0137, "step": 13130 }, { "epoch": 166.32911392405063, "grad_norm": 0.3757597506046295, "learning_rate": 6.419284014364943e-05, "loss": 0.012, "step": 13140 }, { "epoch": 166.45569620253164, "grad_norm": 0.3624584972858429, "learning_rate": 6.41399829608907e-05, "loss": 0.0131, "step": 13150 }, { "epoch": 166.58227848101265, "grad_norm": 0.3110060691833496, "learning_rate": 6.408710859671938e-05, "loss": 0.0115, "step": 13160 }, { "epoch": 166.70886075949366, "grad_norm": 0.3314243257045746, "learning_rate": 6.403421711538278e-05, "loss": 0.0133, "step": 13170 }, { "epoch": 166.83544303797467, "grad_norm": 0.384164035320282, "learning_rate": 6.398130858114903e-05, "loss": 0.0132, "step": 13180 }, { "epoch": 166.9620253164557, "grad_norm": 0.39721524715423584, "learning_rate": 6.392838305830701e-05, "loss": 0.0152, "step": 13190 }, { "epoch": 167.08860759493672, "grad_norm": 0.34434759616851807, "learning_rate": 6.387544061116621e-05, "loss": 0.0105, "step": 13200 }, { "epoch": 167.21518987341773, "grad_norm": 0.374072790145874, "learning_rate": 6.382248130405671e-05, "loss": 0.0126, "step": 13210 }, { "epoch": 167.34177215189874, "grad_norm": 0.3913326561450958, "learning_rate": 6.376950520132905e-05, "loss": 0.0131, "step": 13220 }, { "epoch": 167.46835443037975, "grad_norm": 0.3194420337677002, "learning_rate": 6.371651236735418e-05, "loss": 0.0146, "step": 13230 }, { "epoch": 167.59493670886076, "grad_norm": 0.3965502679347992, "learning_rate": 6.366350286652341e-05, "loss": 0.0106, "step": 13240 }, { "epoch": 167.72151898734177, "grad_norm": 0.37987011671066284, "learning_rate": 6.361047676324827e-05, "loss": 0.0126, "step": 13250 }, { "epoch": 167.84810126582278, "grad_norm": 0.48695579171180725, "learning_rate": 6.355743412196047e-05, "loss": 0.0147, "step": 13260 }, { "epoch": 167.9746835443038, "grad_norm": 0.3365166187286377, "learning_rate": 6.350437500711184e-05, "loss": 0.013, "step": 13270 }, { "epoch": 168.1012658227848, "grad_norm": 0.3700025677680969, "learning_rate": 6.345129948317418e-05, "loss": 0.0127, "step": 13280 }, { "epoch": 168.22784810126583, "grad_norm": 0.42485636472702026, "learning_rate": 6.33982076146393e-05, "loss": 0.0154, "step": 13290 }, { "epoch": 168.35443037974684, "grad_norm": 0.37790730595588684, "learning_rate": 6.334509946601879e-05, "loss": 0.0146, "step": 13300 }, { "epoch": 168.48101265822785, "grad_norm": 0.4171140789985657, "learning_rate": 6.329197510184406e-05, "loss": 0.0125, "step": 13310 }, { "epoch": 168.60759493670886, "grad_norm": 0.34678107500076294, "learning_rate": 6.323883458666624e-05, "loss": 0.0169, "step": 13320 }, { "epoch": 168.73417721518987, "grad_norm": 0.3193938434123993, "learning_rate": 6.318567798505605e-05, "loss": 0.014, "step": 13330 }, { "epoch": 168.86075949367088, "grad_norm": 0.337452232837677, "learning_rate": 6.313250536160378e-05, "loss": 0.0133, "step": 13340 }, { "epoch": 168.9873417721519, "grad_norm": 0.3786640465259552, "learning_rate": 6.307931678091917e-05, "loss": 0.0135, "step": 13350 }, { "epoch": 169.1139240506329, "grad_norm": 0.4667029082775116, "learning_rate": 6.302611230763138e-05, "loss": 0.0159, "step": 13360 }, { "epoch": 169.2405063291139, "grad_norm": 0.2964588701725006, "learning_rate": 6.297289200638887e-05, "loss": 0.0128, "step": 13370 }, { "epoch": 169.36708860759492, "grad_norm": 0.3504825830459595, "learning_rate": 6.29196559418593e-05, "loss": 0.0136, "step": 13380 }, { "epoch": 169.49367088607596, "grad_norm": 0.2880951464176178, "learning_rate": 6.286640417872951e-05, "loss": 0.0126, "step": 13390 }, { "epoch": 169.62025316455697, "grad_norm": 0.3404616117477417, "learning_rate": 6.281313678170542e-05, "loss": 0.0141, "step": 13400 }, { "epoch": 169.74683544303798, "grad_norm": 0.3610069155693054, "learning_rate": 6.275985381551194e-05, "loss": 0.0137, "step": 13410 }, { "epoch": 169.873417721519, "grad_norm": 0.380180686712265, "learning_rate": 6.270655534489292e-05, "loss": 0.0141, "step": 13420 }, { "epoch": 170.0, "grad_norm": 0.3365721106529236, "learning_rate": 6.265324143461098e-05, "loss": 0.0161, "step": 13430 }, { "epoch": 170.126582278481, "grad_norm": 0.3966360092163086, "learning_rate": 6.259991214944758e-05, "loss": 0.0143, "step": 13440 }, { "epoch": 170.25316455696202, "grad_norm": 0.47980836033821106, "learning_rate": 6.254656755420283e-05, "loss": 0.0128, "step": 13450 }, { "epoch": 170.37974683544303, "grad_norm": 0.43515539169311523, "learning_rate": 6.249320771369545e-05, "loss": 0.0133, "step": 13460 }, { "epoch": 170.50632911392404, "grad_norm": 0.33169612288475037, "learning_rate": 6.243983269276263e-05, "loss": 0.0153, "step": 13470 }, { "epoch": 170.63291139240508, "grad_norm": 0.4359494149684906, "learning_rate": 6.238644255626012e-05, "loss": 0.0136, "step": 13480 }, { "epoch": 170.7594936708861, "grad_norm": 0.3960384726524353, "learning_rate": 6.233303736906193e-05, "loss": 0.0121, "step": 13490 }, { "epoch": 170.8860759493671, "grad_norm": 0.3292790651321411, "learning_rate": 6.227961719606039e-05, "loss": 0.0137, "step": 13500 }, { "epoch": 171.0126582278481, "grad_norm": 0.42343249917030334, "learning_rate": 6.222618210216608e-05, "loss": 0.0141, "step": 13510 }, { "epoch": 171.13924050632912, "grad_norm": 0.40693315863609314, "learning_rate": 6.217273215230767e-05, "loss": 0.012, "step": 13520 }, { "epoch": 171.26582278481013, "grad_norm": 0.3476593494415283, "learning_rate": 6.211926741143188e-05, "loss": 0.012, "step": 13530 }, { "epoch": 171.39240506329114, "grad_norm": 0.35678964853286743, "learning_rate": 6.20657879445034e-05, "loss": 0.0129, "step": 13540 }, { "epoch": 171.51898734177215, "grad_norm": 0.4077879786491394, "learning_rate": 6.201229381650485e-05, "loss": 0.0154, "step": 13550 }, { "epoch": 171.64556962025316, "grad_norm": 0.35880687832832336, "learning_rate": 6.195878509243661e-05, "loss": 0.0132, "step": 13560 }, { "epoch": 171.77215189873417, "grad_norm": 0.31374427676200867, "learning_rate": 6.190526183731685e-05, "loss": 0.014, "step": 13570 }, { "epoch": 171.8987341772152, "grad_norm": 0.3213214874267578, "learning_rate": 6.185172411618138e-05, "loss": 0.0136, "step": 13580 }, { "epoch": 172.0253164556962, "grad_norm": 0.3606400489807129, "learning_rate": 6.179817199408355e-05, "loss": 0.016, "step": 13590 }, { "epoch": 172.15189873417722, "grad_norm": 0.39809978008270264, "learning_rate": 6.174460553609426e-05, "loss": 0.0139, "step": 13600 }, { "epoch": 172.27848101265823, "grad_norm": 0.4173870384693146, "learning_rate": 6.16910248073018e-05, "loss": 0.0109, "step": 13610 }, { "epoch": 172.40506329113924, "grad_norm": 0.29863470792770386, "learning_rate": 6.16374298728118e-05, "loss": 0.012, "step": 13620 }, { "epoch": 172.53164556962025, "grad_norm": 0.32272323966026306, "learning_rate": 6.158382079774716e-05, "loss": 0.0135, "step": 13630 }, { "epoch": 172.65822784810126, "grad_norm": 0.3596499562263489, "learning_rate": 6.153019764724799e-05, "loss": 0.0128, "step": 13640 }, { "epoch": 172.78481012658227, "grad_norm": 0.3655839264392853, "learning_rate": 6.147656048647144e-05, "loss": 0.0155, "step": 13650 }, { "epoch": 172.91139240506328, "grad_norm": 0.3037557303905487, "learning_rate": 6.142290938059173e-05, "loss": 0.0134, "step": 13660 }, { "epoch": 173.0379746835443, "grad_norm": 0.36650341749191284, "learning_rate": 6.136924439480001e-05, "loss": 0.0131, "step": 13670 }, { "epoch": 173.16455696202533, "grad_norm": 0.28017693758010864, "learning_rate": 6.13155655943043e-05, "loss": 0.0159, "step": 13680 }, { "epoch": 173.29113924050634, "grad_norm": 0.39653053879737854, "learning_rate": 6.126187304432941e-05, "loss": 0.0137, "step": 13690 }, { "epoch": 173.41772151898735, "grad_norm": 0.31839877367019653, "learning_rate": 6.120816681011684e-05, "loss": 0.014, "step": 13700 }, { "epoch": 173.54430379746836, "grad_norm": 0.3524509370326996, "learning_rate": 6.115444695692474e-05, "loss": 0.0141, "step": 13710 }, { "epoch": 173.67088607594937, "grad_norm": 0.3582846224308014, "learning_rate": 6.110071355002779e-05, "loss": 0.011, "step": 13720 }, { "epoch": 173.79746835443038, "grad_norm": 0.32486140727996826, "learning_rate": 6.104696665471714e-05, "loss": 0.0165, "step": 13730 }, { "epoch": 173.9240506329114, "grad_norm": 0.34472721815109253, "learning_rate": 6.099320633630036e-05, "loss": 0.012, "step": 13740 }, { "epoch": 174.0506329113924, "grad_norm": 0.37753021717071533, "learning_rate": 6.093943266010128e-05, "loss": 0.0155, "step": 13750 }, { "epoch": 174.1772151898734, "grad_norm": 0.35820838809013367, "learning_rate": 6.088564569146e-05, "loss": 0.0141, "step": 13760 }, { "epoch": 174.30379746835442, "grad_norm": 0.3532885015010834, "learning_rate": 6.083184549573274e-05, "loss": 0.0114, "step": 13770 }, { "epoch": 174.43037974683546, "grad_norm": 0.4351333677768707, "learning_rate": 6.077803213829184e-05, "loss": 0.0114, "step": 13780 }, { "epoch": 174.55696202531647, "grad_norm": 0.39659926295280457, "learning_rate": 6.0724205684525594e-05, "loss": 0.0123, "step": 13790 }, { "epoch": 174.68354430379748, "grad_norm": 0.3569769561290741, "learning_rate": 6.0670366199838215e-05, "loss": 0.0112, "step": 13800 }, { "epoch": 174.81012658227849, "grad_norm": 0.28087517619132996, "learning_rate": 6.061651374964974e-05, "loss": 0.0114, "step": 13810 }, { "epoch": 174.9367088607595, "grad_norm": 0.3427102565765381, "learning_rate": 6.0562648399396004e-05, "loss": 0.0129, "step": 13820 }, { "epoch": 175.0632911392405, "grad_norm": 0.3861674666404724, "learning_rate": 6.050877021452845e-05, "loss": 0.0152, "step": 13830 }, { "epoch": 175.18987341772151, "grad_norm": 0.4563275873661041, "learning_rate": 6.04548792605142e-05, "loss": 0.0137, "step": 13840 }, { "epoch": 175.31645569620252, "grad_norm": 0.3508598804473877, "learning_rate": 6.04009756028358e-05, "loss": 0.0119, "step": 13850 }, { "epoch": 175.44303797468353, "grad_norm": 0.33098945021629333, "learning_rate": 6.0347059306991294e-05, "loss": 0.0108, "step": 13860 }, { "epoch": 175.56962025316454, "grad_norm": 0.3373817205429077, "learning_rate": 6.029313043849407e-05, "loss": 0.0152, "step": 13870 }, { "epoch": 175.69620253164558, "grad_norm": 0.3286502957344055, "learning_rate": 6.023918906287279e-05, "loss": 0.0134, "step": 13880 }, { "epoch": 175.8227848101266, "grad_norm": 0.34582969546318054, "learning_rate": 6.01852352456713e-05, "loss": 0.0126, "step": 13890 }, { "epoch": 175.9493670886076, "grad_norm": 0.3651049733161926, "learning_rate": 6.013126905244858e-05, "loss": 0.0111, "step": 13900 }, { "epoch": 176.0759493670886, "grad_norm": 0.4145332872867584, "learning_rate": 6.007729054877864e-05, "loss": 0.0151, "step": 13910 }, { "epoch": 176.20253164556962, "grad_norm": 0.42858168482780457, "learning_rate": 6.0023299800250466e-05, "loss": 0.0126, "step": 13920 }, { "epoch": 176.32911392405063, "grad_norm": 0.3498455584049225, "learning_rate": 5.9969296872467894e-05, "loss": 0.0111, "step": 13930 }, { "epoch": 176.45569620253164, "grad_norm": 0.3195134699344635, "learning_rate": 5.9915281831049585e-05, "loss": 0.0115, "step": 13940 }, { "epoch": 176.58227848101265, "grad_norm": 0.2993030250072479, "learning_rate": 5.98612547416289e-05, "loss": 0.0126, "step": 13950 }, { "epoch": 176.70886075949366, "grad_norm": 0.3521571457386017, "learning_rate": 5.9807215669853855e-05, "loss": 0.0131, "step": 13960 }, { "epoch": 176.83544303797467, "grad_norm": 0.4300849437713623, "learning_rate": 5.9753164681387e-05, "loss": 0.0144, "step": 13970 }, { "epoch": 176.9620253164557, "grad_norm": 0.4066919982433319, "learning_rate": 5.969910184190539e-05, "loss": 0.0123, "step": 13980 }, { "epoch": 177.08860759493672, "grad_norm": 0.4128744602203369, "learning_rate": 5.9645027217100477e-05, "loss": 0.0123, "step": 13990 }, { "epoch": 177.21518987341773, "grad_norm": 0.3746502101421356, "learning_rate": 5.959094087267804e-05, "loss": 0.0107, "step": 14000 }, { "epoch": 177.34177215189874, "grad_norm": 0.4071597158908844, "learning_rate": 5.9536842874358065e-05, "loss": 0.0139, "step": 14010 }, { "epoch": 177.46835443037975, "grad_norm": 0.37640246748924255, "learning_rate": 5.9482733287874734e-05, "loss": 0.0134, "step": 14020 }, { "epoch": 177.59493670886076, "grad_norm": 0.41664618253707886, "learning_rate": 5.942861217897631e-05, "loss": 0.0112, "step": 14030 }, { "epoch": 177.72151898734177, "grad_norm": 0.3471001088619232, "learning_rate": 5.9374479613425e-05, "loss": 0.01, "step": 14040 }, { "epoch": 177.84810126582278, "grad_norm": 0.3368377387523651, "learning_rate": 5.932033565699704e-05, "loss": 0.0143, "step": 14050 }, { "epoch": 177.9746835443038, "grad_norm": 0.3933947682380676, "learning_rate": 5.926618037548237e-05, "loss": 0.0114, "step": 14060 }, { "epoch": 178.1012658227848, "grad_norm": 0.3108864426612854, "learning_rate": 5.9212013834684824e-05, "loss": 0.0093, "step": 14070 }, { "epoch": 178.22784810126583, "grad_norm": 0.42528730630874634, "learning_rate": 5.9157836100421824e-05, "loss": 0.011, "step": 14080 }, { "epoch": 178.35443037974684, "grad_norm": 0.41127246618270874, "learning_rate": 5.910364723852444e-05, "loss": 0.0115, "step": 14090 }, { "epoch": 178.48101265822785, "grad_norm": 0.41398394107818604, "learning_rate": 5.904944731483723e-05, "loss": 0.0112, "step": 14100 }, { "epoch": 178.60759493670886, "grad_norm": 0.3648868501186371, "learning_rate": 5.899523639521825e-05, "loss": 0.0127, "step": 14110 }, { "epoch": 178.73417721518987, "grad_norm": 0.34424102306365967, "learning_rate": 5.894101454553883e-05, "loss": 0.0132, "step": 14120 }, { "epoch": 178.86075949367088, "grad_norm": 0.35566696524620056, "learning_rate": 5.888678183168368e-05, "loss": 0.0122, "step": 14130 }, { "epoch": 178.9873417721519, "grad_norm": 0.36553823947906494, "learning_rate": 5.883253831955061e-05, "loss": 0.0116, "step": 14140 }, { "epoch": 179.1139240506329, "grad_norm": 0.37443962693214417, "learning_rate": 5.8778284075050625e-05, "loss": 0.0099, "step": 14150 }, { "epoch": 179.2405063291139, "grad_norm": 0.3669402003288269, "learning_rate": 5.872401916410777e-05, "loss": 0.0126, "step": 14160 }, { "epoch": 179.36708860759492, "grad_norm": 0.369594007730484, "learning_rate": 5.8669743652659014e-05, "loss": 0.0139, "step": 14170 }, { "epoch": 179.49367088607596, "grad_norm": 0.34303954243659973, "learning_rate": 5.86154576066542e-05, "loss": 0.0115, "step": 14180 }, { "epoch": 179.62025316455697, "grad_norm": 0.4101557731628418, "learning_rate": 5.856116109205602e-05, "loss": 0.0135, "step": 14190 }, { "epoch": 179.74683544303798, "grad_norm": 0.33371439576148987, "learning_rate": 5.850685417483983e-05, "loss": 0.0125, "step": 14200 }, { "epoch": 179.873417721519, "grad_norm": 0.4084186851978302, "learning_rate": 5.8452536920993684e-05, "loss": 0.0142, "step": 14210 }, { "epoch": 180.0, "grad_norm": 0.3596045970916748, "learning_rate": 5.839820939651817e-05, "loss": 0.0137, "step": 14220 }, { "epoch": 180.126582278481, "grad_norm": 0.3151581883430481, "learning_rate": 5.834387166742632e-05, "loss": 0.0133, "step": 14230 }, { "epoch": 180.25316455696202, "grad_norm": 0.31056809425354004, "learning_rate": 5.828952379974364e-05, "loss": 0.0119, "step": 14240 }, { "epoch": 180.37974683544303, "grad_norm": 0.3574662506580353, "learning_rate": 5.8235165859507864e-05, "loss": 0.0136, "step": 14250 }, { "epoch": 180.50632911392404, "grad_norm": 0.3277854323387146, "learning_rate": 5.818079791276907e-05, "loss": 0.0113, "step": 14260 }, { "epoch": 180.63291139240508, "grad_norm": 0.3412717282772064, "learning_rate": 5.8126420025589415e-05, "loss": 0.0125, "step": 14270 }, { "epoch": 180.7594936708861, "grad_norm": 0.3323266804218292, "learning_rate": 5.807203226404313e-05, "loss": 0.0106, "step": 14280 }, { "epoch": 180.8860759493671, "grad_norm": 0.3303949236869812, "learning_rate": 5.801763469421652e-05, "loss": 0.0113, "step": 14290 }, { "epoch": 181.0126582278481, "grad_norm": 0.35645177960395813, "learning_rate": 5.796322738220774e-05, "loss": 0.012, "step": 14300 }, { "epoch": 181.13924050632912, "grad_norm": 0.3225765824317932, "learning_rate": 5.79088103941268e-05, "loss": 0.0154, "step": 14310 }, { "epoch": 181.26582278481013, "grad_norm": 0.31245294213294983, "learning_rate": 5.785438379609549e-05, "loss": 0.0132, "step": 14320 }, { "epoch": 181.39240506329114, "grad_norm": 0.3351185917854309, "learning_rate": 5.779994765424724e-05, "loss": 0.0094, "step": 14330 }, { "epoch": 181.51898734177215, "grad_norm": 0.33016082644462585, "learning_rate": 5.7745502034727114e-05, "loss": 0.0134, "step": 14340 }, { "epoch": 181.64556962025316, "grad_norm": 0.28369250893592834, "learning_rate": 5.769104700369164e-05, "loss": 0.0099, "step": 14350 }, { "epoch": 181.77215189873417, "grad_norm": 0.29150980710983276, "learning_rate": 5.7636582627308854e-05, "loss": 0.0114, "step": 14360 }, { "epoch": 181.8987341772152, "grad_norm": 0.33044004440307617, "learning_rate": 5.7582108971758095e-05, "loss": 0.0102, "step": 14370 }, { "epoch": 182.0253164556962, "grad_norm": 0.336203396320343, "learning_rate": 5.752762610322999e-05, "loss": 0.012, "step": 14380 }, { "epoch": 182.15189873417722, "grad_norm": 0.35593849420547485, "learning_rate": 5.747313408792636e-05, "loss": 0.0135, "step": 14390 }, { "epoch": 182.27848101265823, "grad_norm": 0.35130488872528076, "learning_rate": 5.741863299206014e-05, "loss": 0.0111, "step": 14400 }, { "epoch": 182.40506329113924, "grad_norm": 0.32270610332489014, "learning_rate": 5.73641228818553e-05, "loss": 0.0098, "step": 14410 }, { "epoch": 182.53164556962025, "grad_norm": 0.30183175206184387, "learning_rate": 5.730960382354676e-05, "loss": 0.0103, "step": 14420 }, { "epoch": 182.65822784810126, "grad_norm": 0.2594432830810547, "learning_rate": 5.725507588338035e-05, "loss": 0.0121, "step": 14430 }, { "epoch": 182.78481012658227, "grad_norm": 0.3750482499599457, "learning_rate": 5.7200539127612604e-05, "loss": 0.0135, "step": 14440 }, { "epoch": 182.91139240506328, "grad_norm": 0.33564645051956177, "learning_rate": 5.714599362251087e-05, "loss": 0.0117, "step": 14450 }, { "epoch": 183.0379746835443, "grad_norm": 0.29983973503112793, "learning_rate": 5.709143943435307e-05, "loss": 0.0127, "step": 14460 }, { "epoch": 183.16455696202533, "grad_norm": 0.3235190212726593, "learning_rate": 5.7036876629427646e-05, "loss": 0.0118, "step": 14470 }, { "epoch": 183.29113924050634, "grad_norm": 0.4560326337814331, "learning_rate": 5.698230527403361e-05, "loss": 0.0101, "step": 14480 }, { "epoch": 183.41772151898735, "grad_norm": 0.39121973514556885, "learning_rate": 5.6927725434480274e-05, "loss": 0.0115, "step": 14490 }, { "epoch": 183.54430379746836, "grad_norm": 0.3469381332397461, "learning_rate": 5.687313717708728e-05, "loss": 0.0136, "step": 14500 }, { "epoch": 183.67088607594937, "grad_norm": 0.28741002082824707, "learning_rate": 5.681854056818453e-05, "loss": 0.0129, "step": 14510 }, { "epoch": 183.79746835443038, "grad_norm": 0.37760820984840393, "learning_rate": 5.676393567411205e-05, "loss": 0.0108, "step": 14520 }, { "epoch": 183.9240506329114, "grad_norm": 0.383450448513031, "learning_rate": 5.670932256121991e-05, "loss": 0.0108, "step": 14530 }, { "epoch": 184.0506329113924, "grad_norm": 0.3323901891708374, "learning_rate": 5.6654701295868216e-05, "loss": 0.0105, "step": 14540 }, { "epoch": 184.1772151898734, "grad_norm": 0.37310487031936646, "learning_rate": 5.660007194442697e-05, "loss": 0.014, "step": 14550 }, { "epoch": 184.30379746835442, "grad_norm": 0.40965771675109863, "learning_rate": 5.6545434573275946e-05, "loss": 0.014, "step": 14560 }, { "epoch": 184.43037974683546, "grad_norm": 0.34807318449020386, "learning_rate": 5.649078924880472e-05, "loss": 0.011, "step": 14570 }, { "epoch": 184.55696202531647, "grad_norm": 0.41699716448783875, "learning_rate": 5.643613603741252e-05, "loss": 0.0115, "step": 14580 }, { "epoch": 184.68354430379748, "grad_norm": 0.3541872799396515, "learning_rate": 5.6381475005508154e-05, "loss": 0.0124, "step": 14590 }, { "epoch": 184.81012658227849, "grad_norm": 0.34003326296806335, "learning_rate": 5.632680621950992e-05, "loss": 0.0121, "step": 14600 }, { "epoch": 184.9367088607595, "grad_norm": 0.37839415669441223, "learning_rate": 5.627212974584555e-05, "loss": 0.0136, "step": 14610 }, { "epoch": 185.0632911392405, "grad_norm": 0.35986849665641785, "learning_rate": 5.62174456509521e-05, "loss": 0.0108, "step": 14620 }, { "epoch": 185.18987341772151, "grad_norm": 0.3412478268146515, "learning_rate": 5.616275400127594e-05, "loss": 0.0113, "step": 14630 }, { "epoch": 185.31645569620252, "grad_norm": 0.3110451102256775, "learning_rate": 5.610805486327254e-05, "loss": 0.0117, "step": 14640 }, { "epoch": 185.44303797468353, "grad_norm": 0.3230575919151306, "learning_rate": 5.6053348303406536e-05, "loss": 0.0109, "step": 14650 }, { "epoch": 185.56962025316454, "grad_norm": 0.3384734094142914, "learning_rate": 5.599863438815156e-05, "loss": 0.0113, "step": 14660 }, { "epoch": 185.69620253164558, "grad_norm": 0.32640302181243896, "learning_rate": 5.594391318399017e-05, "loss": 0.0128, "step": 14670 }, { "epoch": 185.8227848101266, "grad_norm": 0.3399411737918854, "learning_rate": 5.588918475741377e-05, "loss": 0.0122, "step": 14680 }, { "epoch": 185.9493670886076, "grad_norm": 0.41295719146728516, "learning_rate": 5.5834449174922586e-05, "loss": 0.0115, "step": 14690 }, { "epoch": 186.0759493670886, "grad_norm": 0.30270108580589294, "learning_rate": 5.57797065030255e-05, "loss": 0.0132, "step": 14700 }, { "epoch": 186.20253164556962, "grad_norm": 0.33755701780319214, "learning_rate": 5.5724956808240016e-05, "loss": 0.0105, "step": 14710 }, { "epoch": 186.32911392405063, "grad_norm": 0.34481528401374817, "learning_rate": 5.567020015709219e-05, "loss": 0.0115, "step": 14720 }, { "epoch": 186.45569620253164, "grad_norm": 0.32174789905548096, "learning_rate": 5.561543661611649e-05, "loss": 0.0128, "step": 14730 }, { "epoch": 186.58227848101265, "grad_norm": 0.3352152109146118, "learning_rate": 5.556066625185583e-05, "loss": 0.011, "step": 14740 }, { "epoch": 186.70886075949366, "grad_norm": 0.32554909586906433, "learning_rate": 5.550588913086131e-05, "loss": 0.0103, "step": 14750 }, { "epoch": 186.83544303797467, "grad_norm": 0.3953382968902588, "learning_rate": 5.545110531969234e-05, "loss": 0.0155, "step": 14760 }, { "epoch": 186.9620253164557, "grad_norm": 0.37090399861335754, "learning_rate": 5.539631488491641e-05, "loss": 0.0103, "step": 14770 }, { "epoch": 187.08860759493672, "grad_norm": 0.38582733273506165, "learning_rate": 5.534151789310904e-05, "loss": 0.0116, "step": 14780 }, { "epoch": 187.21518987341773, "grad_norm": 0.3668529689311981, "learning_rate": 5.528671441085376e-05, "loss": 0.0112, "step": 14790 }, { "epoch": 187.34177215189874, "grad_norm": 0.29654744267463684, "learning_rate": 5.523190450474197e-05, "loss": 0.0137, "step": 14800 }, { "epoch": 187.46835443037975, "grad_norm": 0.28987279534339905, "learning_rate": 5.5177088241372865e-05, "loss": 0.0107, "step": 14810 }, { "epoch": 187.59493670886076, "grad_norm": 0.3176715672016144, "learning_rate": 5.512226568735338e-05, "loss": 0.0098, "step": 14820 }, { "epoch": 187.72151898734177, "grad_norm": 0.3440273106098175, "learning_rate": 5.506743690929809e-05, "loss": 0.012, "step": 14830 }, { "epoch": 187.84810126582278, "grad_norm": 0.32233893871307373, "learning_rate": 5.501260197382912e-05, "loss": 0.0119, "step": 14840 }, { "epoch": 187.9746835443038, "grad_norm": 0.31110358238220215, "learning_rate": 5.49577609475761e-05, "loss": 0.0092, "step": 14850 }, { "epoch": 188.1012658227848, "grad_norm": 0.34649720788002014, "learning_rate": 5.490291389717603e-05, "loss": 0.0107, "step": 14860 }, { "epoch": 188.22784810126583, "grad_norm": 0.3389301002025604, "learning_rate": 5.484806088927329e-05, "loss": 0.0115, "step": 14870 }, { "epoch": 188.35443037974684, "grad_norm": 0.3511534333229065, "learning_rate": 5.479320199051942e-05, "loss": 0.0124, "step": 14880 }, { "epoch": 188.48101265822785, "grad_norm": 0.3172323703765869, "learning_rate": 5.473833726757314e-05, "loss": 0.0119, "step": 14890 }, { "epoch": 188.60759493670886, "grad_norm": 0.2608090043067932, "learning_rate": 5.4683466787100304e-05, "loss": 0.0134, "step": 14900 }, { "epoch": 188.73417721518987, "grad_norm": 0.33256855607032776, "learning_rate": 5.462859061577369e-05, "loss": 0.0136, "step": 14910 }, { "epoch": 188.86075949367088, "grad_norm": 0.26505687832832336, "learning_rate": 5.4573708820273026e-05, "loss": 0.0122, "step": 14920 }, { "epoch": 188.9873417721519, "grad_norm": 0.35226061940193176, "learning_rate": 5.451882146728488e-05, "loss": 0.0137, "step": 14930 }, { "epoch": 189.1139240506329, "grad_norm": 0.3581768870353699, "learning_rate": 5.446392862350255e-05, "loss": 0.0126, "step": 14940 }, { "epoch": 189.2405063291139, "grad_norm": 0.39807388186454773, "learning_rate": 5.440903035562603e-05, "loss": 0.0121, "step": 14950 }, { "epoch": 189.36708860759492, "grad_norm": 0.5599792003631592, "learning_rate": 5.435412673036188e-05, "loss": 0.0111, "step": 14960 }, { "epoch": 189.49367088607596, "grad_norm": 0.45097866654396057, "learning_rate": 5.429921781442318e-05, "loss": 0.0127, "step": 14970 }, { "epoch": 189.62025316455697, "grad_norm": 0.4085337519645691, "learning_rate": 5.424430367452945e-05, "loss": 0.0124, "step": 14980 }, { "epoch": 189.74683544303798, "grad_norm": 0.3457949757575989, "learning_rate": 5.418938437740655e-05, "loss": 0.0116, "step": 14990 }, { "epoch": 189.873417721519, "grad_norm": 0.3348681926727295, "learning_rate": 5.413445998978658e-05, "loss": 0.0119, "step": 15000 }, { "epoch": 190.0, "grad_norm": 0.3181767463684082, "learning_rate": 5.407953057840789e-05, "loss": 0.0145, "step": 15010 }, { "epoch": 190.126582278481, "grad_norm": 0.35724741220474243, "learning_rate": 5.4024596210014853e-05, "loss": 0.0134, "step": 15020 }, { "epoch": 190.25316455696202, "grad_norm": 0.35541409254074097, "learning_rate": 5.3969656951357935e-05, "loss": 0.0101, "step": 15030 }, { "epoch": 190.37974683544303, "grad_norm": 0.3280642628669739, "learning_rate": 5.3914712869193504e-05, "loss": 0.01, "step": 15040 }, { "epoch": 190.50632911392404, "grad_norm": 0.3126855194568634, "learning_rate": 5.385976403028381e-05, "loss": 0.0122, "step": 15050 }, { "epoch": 190.63291139240508, "grad_norm": 0.3132084906101227, "learning_rate": 5.3804810501396875e-05, "loss": 0.0124, "step": 15060 }, { "epoch": 190.7594936708861, "grad_norm": 0.35686469078063965, "learning_rate": 5.374985234930639e-05, "loss": 0.0142, "step": 15070 }, { "epoch": 190.8860759493671, "grad_norm": 0.2869809567928314, "learning_rate": 5.3694889640791724e-05, "loss": 0.0152, "step": 15080 }, { "epoch": 191.0126582278481, "grad_norm": 0.3343616724014282, "learning_rate": 5.3639922442637736e-05, "loss": 0.0147, "step": 15090 }, { "epoch": 191.13924050632912, "grad_norm": 0.4027363955974579, "learning_rate": 5.358495082163476e-05, "loss": 0.0124, "step": 15100 }, { "epoch": 191.26582278481013, "grad_norm": 0.37675368785858154, "learning_rate": 5.35299748445785e-05, "loss": 0.0136, "step": 15110 }, { "epoch": 191.39240506329114, "grad_norm": 0.35056763887405396, "learning_rate": 5.3474994578269944e-05, "loss": 0.0119, "step": 15120 }, { "epoch": 191.51898734177215, "grad_norm": 0.2745369076728821, "learning_rate": 5.34200100895153e-05, "loss": 0.0098, "step": 15130 }, { "epoch": 191.64556962025316, "grad_norm": 0.3264434337615967, "learning_rate": 5.3365021445125916e-05, "loss": 0.0125, "step": 15140 }, { "epoch": 191.77215189873417, "grad_norm": 0.35697224736213684, "learning_rate": 5.331002871191817e-05, "loss": 0.0105, "step": 15150 }, { "epoch": 191.8987341772152, "grad_norm": 0.37485021352767944, "learning_rate": 5.3255031956713443e-05, "loss": 0.0112, "step": 15160 }, { "epoch": 192.0253164556962, "grad_norm": 0.36961013078689575, "learning_rate": 5.320003124633795e-05, "loss": 0.0127, "step": 15170 }, { "epoch": 192.15189873417722, "grad_norm": 0.36818888783454895, "learning_rate": 5.314502664762274e-05, "loss": 0.0105, "step": 15180 }, { "epoch": 192.27848101265823, "grad_norm": 0.3086089789867401, "learning_rate": 5.3090018227403604e-05, "loss": 0.014, "step": 15190 }, { "epoch": 192.40506329113924, "grad_norm": 0.35843852162361145, "learning_rate": 5.3035006052520955e-05, "loss": 0.012, "step": 15200 }, { "epoch": 192.53164556962025, "grad_norm": 0.2798827886581421, "learning_rate": 5.297999018981976e-05, "loss": 0.0102, "step": 15210 }, { "epoch": 192.65822784810126, "grad_norm": 0.32826557755470276, "learning_rate": 5.2924970706149505e-05, "loss": 0.0108, "step": 15220 }, { "epoch": 192.78481012658227, "grad_norm": 0.3390084505081177, "learning_rate": 5.2869947668364014e-05, "loss": 0.0128, "step": 15230 }, { "epoch": 192.91139240506328, "grad_norm": 0.2811776101589203, "learning_rate": 5.2814921143321504e-05, "loss": 0.0124, "step": 15240 }, { "epoch": 193.0379746835443, "grad_norm": 0.3170446753501892, "learning_rate": 5.275989119788436e-05, "loss": 0.0113, "step": 15250 }, { "epoch": 193.16455696202533, "grad_norm": 0.28160640597343445, "learning_rate": 5.270485789891919e-05, "loss": 0.0114, "step": 15260 }, { "epoch": 193.29113924050634, "grad_norm": 0.3155539035797119, "learning_rate": 5.264982131329661e-05, "loss": 0.0139, "step": 15270 }, { "epoch": 193.41772151898735, "grad_norm": 0.3537997901439667, "learning_rate": 5.259478150789128e-05, "loss": 0.0127, "step": 15280 }, { "epoch": 193.54430379746836, "grad_norm": 0.37038084864616394, "learning_rate": 5.253973854958173e-05, "loss": 0.0134, "step": 15290 }, { "epoch": 193.67088607594937, "grad_norm": 0.36375394463539124, "learning_rate": 5.2484692505250376e-05, "loss": 0.0138, "step": 15300 }, { "epoch": 193.79746835443038, "grad_norm": 0.303254097700119, "learning_rate": 5.242964344178333e-05, "loss": 0.0109, "step": 15310 }, { "epoch": 193.9240506329114, "grad_norm": 0.32858750224113464, "learning_rate": 5.237459142607041e-05, "loss": 0.0081, "step": 15320 }, { "epoch": 194.0506329113924, "grad_norm": 0.4252813160419464, "learning_rate": 5.231953652500498e-05, "loss": 0.0118, "step": 15330 }, { "epoch": 194.1772151898734, "grad_norm": 0.3529892563819885, "learning_rate": 5.2264478805483976e-05, "loss": 0.0108, "step": 15340 }, { "epoch": 194.30379746835442, "grad_norm": 0.3591999411582947, "learning_rate": 5.220941833440768e-05, "loss": 0.0114, "step": 15350 }, { "epoch": 194.43037974683546, "grad_norm": 0.3486523926258087, "learning_rate": 5.215435517867977e-05, "loss": 0.0121, "step": 15360 }, { "epoch": 194.55696202531647, "grad_norm": 0.3774712383747101, "learning_rate": 5.209928940520719e-05, "loss": 0.014, "step": 15370 }, { "epoch": 194.68354430379748, "grad_norm": 0.3591398596763611, "learning_rate": 5.204422108090004e-05, "loss": 0.0106, "step": 15380 }, { "epoch": 194.81012658227849, "grad_norm": 0.3348822295665741, "learning_rate": 5.19891502726715e-05, "loss": 0.0139, "step": 15390 }, { "epoch": 194.9367088607595, "grad_norm": 0.37462225556373596, "learning_rate": 5.193407704743782e-05, "loss": 0.0119, "step": 15400 }, { "epoch": 195.0632911392405, "grad_norm": 0.3106846511363983, "learning_rate": 5.187900147211815e-05, "loss": 0.009, "step": 15410 }, { "epoch": 195.18987341772151, "grad_norm": 0.33471113443374634, "learning_rate": 5.182392361363453e-05, "loss": 0.0141, "step": 15420 }, { "epoch": 195.31645569620252, "grad_norm": 0.37945854663848877, "learning_rate": 5.176884353891172e-05, "loss": 0.0112, "step": 15430 }, { "epoch": 195.44303797468353, "grad_norm": 0.3865800201892853, "learning_rate": 5.171376131487722e-05, "loss": 0.0147, "step": 15440 }, { "epoch": 195.56962025316454, "grad_norm": 0.3404093086719513, "learning_rate": 5.165867700846113e-05, "loss": 0.0103, "step": 15450 }, { "epoch": 195.69620253164558, "grad_norm": 0.330578088760376, "learning_rate": 5.160359068659606e-05, "loss": 0.0107, "step": 15460 }, { "epoch": 195.8227848101266, "grad_norm": 0.3559316098690033, "learning_rate": 5.154850241621712e-05, "loss": 0.0098, "step": 15470 }, { "epoch": 195.9493670886076, "grad_norm": 0.36890703439712524, "learning_rate": 5.149341226426172e-05, "loss": 0.014, "step": 15480 }, { "epoch": 196.0759493670886, "grad_norm": 0.3620651960372925, "learning_rate": 5.14383202976696e-05, "loss": 0.0135, "step": 15490 }, { "epoch": 196.20253164556962, "grad_norm": 0.30868229269981384, "learning_rate": 5.1383226583382684e-05, "loss": 0.0144, "step": 15500 }, { "epoch": 196.32911392405063, "grad_norm": 0.3072099983692169, "learning_rate": 5.132813118834504e-05, "loss": 0.0113, "step": 15510 }, { "epoch": 196.45569620253164, "grad_norm": 0.3784283697605133, "learning_rate": 5.1273034179502777e-05, "loss": 0.0122, "step": 15520 }, { "epoch": 196.58227848101265, "grad_norm": 0.3220159709453583, "learning_rate": 5.1217935623803945e-05, "loss": 0.0109, "step": 15530 }, { "epoch": 196.70886075949366, "grad_norm": 0.34529390931129456, "learning_rate": 5.1162835588198476e-05, "loss": 0.012, "step": 15540 }, { "epoch": 196.83544303797467, "grad_norm": 0.3205519914627075, "learning_rate": 5.110773413963813e-05, "loss": 0.0116, "step": 15550 }, { "epoch": 196.9620253164557, "grad_norm": 0.28351905941963196, "learning_rate": 5.105263134507636e-05, "loss": 0.0133, "step": 15560 }, { "epoch": 197.08860759493672, "grad_norm": 0.45379310846328735, "learning_rate": 5.099752727146824e-05, "loss": 0.013, "step": 15570 }, { "epoch": 197.21518987341773, "grad_norm": 0.40407687425613403, "learning_rate": 5.094242198577042e-05, "loss": 0.0138, "step": 15580 }, { "epoch": 197.34177215189874, "grad_norm": 0.42557698488235474, "learning_rate": 5.088731555494102e-05, "loss": 0.0098, "step": 15590 }, { "epoch": 197.46835443037975, "grad_norm": 0.3407900333404541, "learning_rate": 5.0832208045939556e-05, "loss": 0.0117, "step": 15600 }, { "epoch": 197.59493670886076, "grad_norm": 0.33662182092666626, "learning_rate": 5.0777099525726844e-05, "loss": 0.011, "step": 15610 }, { "epoch": 197.72151898734177, "grad_norm": 0.27561673521995544, "learning_rate": 5.072199006126494e-05, "loss": 0.0103, "step": 15620 }, { "epoch": 197.84810126582278, "grad_norm": 0.3503524959087372, "learning_rate": 5.0666879719517026e-05, "loss": 0.0095, "step": 15630 }, { "epoch": 197.9746835443038, "grad_norm": 0.32814136147499084, "learning_rate": 5.061176856744737e-05, "loss": 0.0108, "step": 15640 }, { "epoch": 198.1012658227848, "grad_norm": 0.3789534866809845, "learning_rate": 5.0556656672021205e-05, "loss": 0.0112, "step": 15650 }, { "epoch": 198.22784810126583, "grad_norm": 0.3447735607624054, "learning_rate": 5.050154410020472e-05, "loss": 0.0115, "step": 15660 }, { "epoch": 198.35443037974684, "grad_norm": 0.33082953095436096, "learning_rate": 5.044643091896485e-05, "loss": 0.0094, "step": 15670 }, { "epoch": 198.48101265822785, "grad_norm": 0.26072585582733154, "learning_rate": 5.039131719526932e-05, "loss": 0.01, "step": 15680 }, { "epoch": 198.60759493670886, "grad_norm": 0.2785803973674774, "learning_rate": 5.03362029960865e-05, "loss": 0.0102, "step": 15690 }, { "epoch": 198.73417721518987, "grad_norm": 0.3131945729255676, "learning_rate": 5.028108838838533e-05, "loss": 0.0098, "step": 15700 }, { "epoch": 198.86075949367088, "grad_norm": 0.32273435592651367, "learning_rate": 5.0225973439135276e-05, "loss": 0.0116, "step": 15710 }, { "epoch": 198.9873417721519, "grad_norm": 0.3431546688079834, "learning_rate": 5.017085821530617e-05, "loss": 0.01, "step": 15720 }, { "epoch": 199.1139240506329, "grad_norm": 0.34846124053001404, "learning_rate": 5.011574278386822e-05, "loss": 0.0104, "step": 15730 }, { "epoch": 199.2405063291139, "grad_norm": 0.3178167939186096, "learning_rate": 5.006062721179189e-05, "loss": 0.011, "step": 15740 }, { "epoch": 199.36708860759492, "grad_norm": 0.3332348167896271, "learning_rate": 5.000551156604777e-05, "loss": 0.013, "step": 15750 }, { "epoch": 199.49367088607596, "grad_norm": 0.3281000554561615, "learning_rate": 4.9950395913606594e-05, "loss": 0.0111, "step": 15760 }, { "epoch": 199.62025316455697, "grad_norm": 0.3149600923061371, "learning_rate": 4.989528032143903e-05, "loss": 0.0127, "step": 15770 }, { "epoch": 199.74683544303798, "grad_norm": 0.3752437233924866, "learning_rate": 4.984016485651578e-05, "loss": 0.0106, "step": 15780 }, { "epoch": 199.873417721519, "grad_norm": 0.3456387519836426, "learning_rate": 4.9785049585807274e-05, "loss": 0.0132, "step": 15790 }, { "epoch": 200.0, "grad_norm": 0.36380478739738464, "learning_rate": 4.972993457628381e-05, "loss": 0.0134, "step": 15800 }, { "epoch": 200.126582278481, "grad_norm": 0.29669585824012756, "learning_rate": 4.9674819894915306e-05, "loss": 0.0107, "step": 15810 }, { "epoch": 200.25316455696202, "grad_norm": 0.30450087785720825, "learning_rate": 4.9619705608671265e-05, "loss": 0.0116, "step": 15820 }, { "epoch": 200.37974683544303, "grad_norm": 0.2805423140525818, "learning_rate": 4.956459178452079e-05, "loss": 0.0121, "step": 15830 }, { "epoch": 200.50632911392404, "grad_norm": 0.2983441650867462, "learning_rate": 4.950947848943235e-05, "loss": 0.0105, "step": 15840 }, { "epoch": 200.63291139240508, "grad_norm": 0.32280269265174866, "learning_rate": 4.9454365790373805e-05, "loss": 0.013, "step": 15850 }, { "epoch": 200.7594936708861, "grad_norm": 0.38296574354171753, "learning_rate": 4.939925375431226e-05, "loss": 0.0145, "step": 15860 }, { "epoch": 200.8860759493671, "grad_norm": 0.30299270153045654, "learning_rate": 4.9344142448214046e-05, "loss": 0.0107, "step": 15870 }, { "epoch": 201.0126582278481, "grad_norm": 0.2862353026866913, "learning_rate": 4.928903193904461e-05, "loss": 0.0122, "step": 15880 }, { "epoch": 201.13924050632912, "grad_norm": 0.290428102016449, "learning_rate": 4.923392229376842e-05, "loss": 0.0128, "step": 15890 }, { "epoch": 201.26582278481013, "grad_norm": 0.36965811252593994, "learning_rate": 4.9178813579348865e-05, "loss": 0.0117, "step": 15900 }, { "epoch": 201.39240506329114, "grad_norm": 0.3793191909790039, "learning_rate": 4.9123705862748253e-05, "loss": 0.0104, "step": 15910 }, { "epoch": 201.51898734177215, "grad_norm": 0.36568301916122437, "learning_rate": 4.9068599210927627e-05, "loss": 0.0108, "step": 15920 }, { "epoch": 201.64556962025316, "grad_norm": 0.3245100677013397, "learning_rate": 4.901349369084681e-05, "loss": 0.0115, "step": 15930 }, { "epoch": 201.77215189873417, "grad_norm": 0.30619117617607117, "learning_rate": 4.8958389369464156e-05, "loss": 0.0099, "step": 15940 }, { "epoch": 201.8987341772152, "grad_norm": 0.32532617449760437, "learning_rate": 4.890328631373666e-05, "loss": 0.0122, "step": 15950 }, { "epoch": 202.0253164556962, "grad_norm": 0.36702960729599, "learning_rate": 4.88481845906197e-05, "loss": 0.0094, "step": 15960 }, { "epoch": 202.15189873417722, "grad_norm": 0.3424702286720276, "learning_rate": 4.879308426706707e-05, "loss": 0.0125, "step": 15970 }, { "epoch": 202.27848101265823, "grad_norm": 0.361587256193161, "learning_rate": 4.873798541003084e-05, "loss": 0.0118, "step": 15980 }, { "epoch": 202.40506329113924, "grad_norm": 0.3625267744064331, "learning_rate": 4.868288808646136e-05, "loss": 0.0118, "step": 15990 }, { "epoch": 202.53164556962025, "grad_norm": 0.3653274178504944, "learning_rate": 4.862779236330705e-05, "loss": 0.0104, "step": 16000 }, { "epoch": 202.65822784810126, "grad_norm": 0.30712074041366577, "learning_rate": 4.8572698307514395e-05, "loss": 0.0095, "step": 16010 }, { "epoch": 202.78481012658227, "grad_norm": 0.32385173439979553, "learning_rate": 4.85176059860279e-05, "loss": 0.0114, "step": 16020 }, { "epoch": 202.91139240506328, "grad_norm": 0.2743152976036072, "learning_rate": 4.846251546578989e-05, "loss": 0.0105, "step": 16030 }, { "epoch": 203.0379746835443, "grad_norm": 0.3258954882621765, "learning_rate": 4.840742681374058e-05, "loss": 0.0111, "step": 16040 }, { "epoch": 203.16455696202533, "grad_norm": 0.3791365623474121, "learning_rate": 4.835234009681787e-05, "loss": 0.0112, "step": 16050 }, { "epoch": 203.29113924050634, "grad_norm": 0.3457777202129364, "learning_rate": 4.8297255381957296e-05, "loss": 0.0138, "step": 16060 }, { "epoch": 203.41772151898735, "grad_norm": 0.31710803508758545, "learning_rate": 4.824217273609198e-05, "loss": 0.0097, "step": 16070 }, { "epoch": 203.54430379746836, "grad_norm": 0.322417676448822, "learning_rate": 4.8187092226152555e-05, "loss": 0.0115, "step": 16080 }, { "epoch": 203.67088607594937, "grad_norm": 0.33732545375823975, "learning_rate": 4.8132013919067016e-05, "loss": 0.0138, "step": 16090 }, { "epoch": 203.79746835443038, "grad_norm": 0.3273940086364746, "learning_rate": 4.807693788176071e-05, "loss": 0.008, "step": 16100 }, { "epoch": 203.9240506329114, "grad_norm": 0.2506242096424103, "learning_rate": 4.8021864181156214e-05, "loss": 0.0106, "step": 16110 }, { "epoch": 204.0506329113924, "grad_norm": 0.37746289372444153, "learning_rate": 4.796679288417326e-05, "loss": 0.0106, "step": 16120 }, { "epoch": 204.1772151898734, "grad_norm": 0.3089601397514343, "learning_rate": 4.791172405772866e-05, "loss": 0.0113, "step": 16130 }, { "epoch": 204.30379746835442, "grad_norm": 0.2901926338672638, "learning_rate": 4.785665776873625e-05, "loss": 0.0109, "step": 16140 }, { "epoch": 204.43037974683546, "grad_norm": 0.3045850098133087, "learning_rate": 4.7801594084106763e-05, "loss": 0.0107, "step": 16150 }, { "epoch": 204.55696202531647, "grad_norm": 0.33266663551330566, "learning_rate": 4.774653307074775e-05, "loss": 0.0105, "step": 16160 }, { "epoch": 204.68354430379748, "grad_norm": 0.3048335015773773, "learning_rate": 4.769147479556355e-05, "loss": 0.0104, "step": 16170 }, { "epoch": 204.81012658227849, "grad_norm": 0.3192283511161804, "learning_rate": 4.763641932545515e-05, "loss": 0.0105, "step": 16180 }, { "epoch": 204.9367088607595, "grad_norm": 0.3534252643585205, "learning_rate": 4.758136672732013e-05, "loss": 0.0115, "step": 16190 }, { "epoch": 205.0632911392405, "grad_norm": 0.3770962953567505, "learning_rate": 4.752631706805261e-05, "loss": 0.0136, "step": 16200 }, { "epoch": 205.18987341772151, "grad_norm": 0.3540835678577423, "learning_rate": 4.7471270414543114e-05, "loss": 0.011, "step": 16210 }, { "epoch": 205.31645569620252, "grad_norm": 0.25436416268348694, "learning_rate": 4.741622683367849e-05, "loss": 0.0122, "step": 16220 }, { "epoch": 205.44303797468353, "grad_norm": 0.28054818511009216, "learning_rate": 4.736118639234191e-05, "loss": 0.0106, "step": 16230 }, { "epoch": 205.56962025316454, "grad_norm": 0.23580531775951385, "learning_rate": 4.730614915741267e-05, "loss": 0.0086, "step": 16240 }, { "epoch": 205.69620253164558, "grad_norm": 0.3487122058868408, "learning_rate": 4.7251115195766234e-05, "loss": 0.0129, "step": 16250 }, { "epoch": 205.8227848101266, "grad_norm": 0.3286904990673065, "learning_rate": 4.719608457427404e-05, "loss": 0.0114, "step": 16260 }, { "epoch": 205.9493670886076, "grad_norm": 0.30263185501098633, "learning_rate": 4.714105735980347e-05, "loss": 0.0133, "step": 16270 }, { "epoch": 206.0759493670886, "grad_norm": 0.27518653869628906, "learning_rate": 4.70860336192178e-05, "loss": 0.0111, "step": 16280 }, { "epoch": 206.20253164556962, "grad_norm": 0.359462171792984, "learning_rate": 4.7031013419376034e-05, "loss": 0.0163, "step": 16290 }, { "epoch": 206.32911392405063, "grad_norm": 0.30079495906829834, "learning_rate": 4.6975996827132926e-05, "loss": 0.0134, "step": 16300 }, { "epoch": 206.45569620253164, "grad_norm": 0.3655632436275482, "learning_rate": 4.692098390933883e-05, "loss": 0.0113, "step": 16310 }, { "epoch": 206.58227848101265, "grad_norm": 0.35251888632774353, "learning_rate": 4.6865974732839615e-05, "loss": 0.0113, "step": 16320 }, { "epoch": 206.70886075949366, "grad_norm": 0.2868507504463196, "learning_rate": 4.6810969364476615e-05, "loss": 0.0125, "step": 16330 }, { "epoch": 206.83544303797467, "grad_norm": 0.3432007133960724, "learning_rate": 4.675596787108653e-05, "loss": 0.0122, "step": 16340 }, { "epoch": 206.9620253164557, "grad_norm": 0.34621021151542664, "learning_rate": 4.670097031950139e-05, "loss": 0.0098, "step": 16350 }, { "epoch": 207.08860759493672, "grad_norm": 0.34690654277801514, "learning_rate": 4.664597677654838e-05, "loss": 0.0114, "step": 16360 }, { "epoch": 207.21518987341773, "grad_norm": 0.32281702756881714, "learning_rate": 4.659098730904986e-05, "loss": 0.0095, "step": 16370 }, { "epoch": 207.34177215189874, "grad_norm": 0.2852461636066437, "learning_rate": 4.6536001983823204e-05, "loss": 0.0099, "step": 16380 }, { "epoch": 207.46835443037975, "grad_norm": 0.29685139656066895, "learning_rate": 4.648102086768078e-05, "loss": 0.0135, "step": 16390 }, { "epoch": 207.59493670886076, "grad_norm": 0.30435270071029663, "learning_rate": 4.6426044027429786e-05, "loss": 0.0105, "step": 16400 }, { "epoch": 207.72151898734177, "grad_norm": 0.28674450516700745, "learning_rate": 4.637107152987234e-05, "loss": 0.0093, "step": 16410 }, { "epoch": 207.84810126582278, "grad_norm": 0.3024829924106598, "learning_rate": 4.631610344180515e-05, "loss": 0.0095, "step": 16420 }, { "epoch": 207.9746835443038, "grad_norm": 0.2879878282546997, "learning_rate": 4.626113983001965e-05, "loss": 0.0117, "step": 16430 }, { "epoch": 208.1012658227848, "grad_norm": 0.3264532685279846, "learning_rate": 4.620618076130182e-05, "loss": 0.0098, "step": 16440 }, { "epoch": 208.22784810126583, "grad_norm": 0.27269330620765686, "learning_rate": 4.6151226302432074e-05, "loss": 0.0117, "step": 16450 }, { "epoch": 208.35443037974684, "grad_norm": 0.2772877514362335, "learning_rate": 4.6096276520185304e-05, "loss": 0.0091, "step": 16460 }, { "epoch": 208.48101265822785, "grad_norm": 0.38392090797424316, "learning_rate": 4.604133148133066e-05, "loss": 0.0092, "step": 16470 }, { "epoch": 208.60759493670886, "grad_norm": 0.32810813188552856, "learning_rate": 4.598639125263155e-05, "loss": 0.0118, "step": 16480 }, { "epoch": 208.73417721518987, "grad_norm": 0.34088581800460815, "learning_rate": 4.593145590084553e-05, "loss": 0.0132, "step": 16490 }, { "epoch": 208.86075949367088, "grad_norm": 0.280286580324173, "learning_rate": 4.58765254927242e-05, "loss": 0.008, "step": 16500 }, { "epoch": 208.9873417721519, "grad_norm": 0.2948598265647888, "learning_rate": 4.582160009501323e-05, "loss": 0.0103, "step": 16510 }, { "epoch": 209.1139240506329, "grad_norm": 0.36244288086891174, "learning_rate": 4.5766679774452143e-05, "loss": 0.0098, "step": 16520 }, { "epoch": 209.2405063291139, "grad_norm": 0.3077971637248993, "learning_rate": 4.571176459777431e-05, "loss": 0.0103, "step": 16530 }, { "epoch": 209.36708860759492, "grad_norm": 0.33338987827301025, "learning_rate": 4.5656854631706846e-05, "loss": 0.0084, "step": 16540 }, { "epoch": 209.49367088607596, "grad_norm": 0.28029099106788635, "learning_rate": 4.560194994297054e-05, "loss": 0.0097, "step": 16550 }, { "epoch": 209.62025316455697, "grad_norm": 0.3045538365840912, "learning_rate": 4.554705059827974e-05, "loss": 0.0091, "step": 16560 }, { "epoch": 209.74683544303798, "grad_norm": 0.30498647689819336, "learning_rate": 4.5492156664342365e-05, "loss": 0.0094, "step": 16570 }, { "epoch": 209.873417721519, "grad_norm": 0.2446952909231186, "learning_rate": 4.5437268207859695e-05, "loss": 0.0109, "step": 16580 }, { "epoch": 210.0, "grad_norm": 0.29585596919059753, "learning_rate": 4.538238529552641e-05, "loss": 0.0106, "step": 16590 }, { "epoch": 210.126582278481, "grad_norm": 0.3206718862056732, "learning_rate": 4.5327507994030394e-05, "loss": 0.0141, "step": 16600 }, { "epoch": 210.25316455696202, "grad_norm": 0.24453701078891754, "learning_rate": 4.527263637005273e-05, "loss": 0.0102, "step": 16610 }, { "epoch": 210.37974683544303, "grad_norm": 0.31839683651924133, "learning_rate": 4.521777049026766e-05, "loss": 0.0093, "step": 16620 }, { "epoch": 210.50632911392404, "grad_norm": 0.2941882312297821, "learning_rate": 4.5162910421342386e-05, "loss": 0.0099, "step": 16630 }, { "epoch": 210.63291139240508, "grad_norm": 0.3530558943748474, "learning_rate": 4.510805622993706e-05, "loss": 0.0123, "step": 16640 }, { "epoch": 210.7594936708861, "grad_norm": 0.2879266142845154, "learning_rate": 4.505320798270467e-05, "loss": 0.0096, "step": 16650 }, { "epoch": 210.8860759493671, "grad_norm": 0.3052539825439453, "learning_rate": 4.499836574629105e-05, "loss": 0.0108, "step": 16660 }, { "epoch": 211.0126582278481, "grad_norm": 0.32487693428993225, "learning_rate": 4.494352958733465e-05, "loss": 0.012, "step": 16670 }, { "epoch": 211.13924050632912, "grad_norm": 0.3102487325668335, "learning_rate": 4.488869957246663e-05, "loss": 0.0113, "step": 16680 }, { "epoch": 211.26582278481013, "grad_norm": 0.3184799551963806, "learning_rate": 4.4833875768310575e-05, "loss": 0.0158, "step": 16690 }, { "epoch": 211.39240506329114, "grad_norm": 0.37000542879104614, "learning_rate": 4.47790582414826e-05, "loss": 0.0107, "step": 16700 }, { "epoch": 211.51898734177215, "grad_norm": 0.2917627990245819, "learning_rate": 4.472424705859114e-05, "loss": 0.0096, "step": 16710 }, { "epoch": 211.64556962025316, "grad_norm": 0.33236056566238403, "learning_rate": 4.466944228623701e-05, "loss": 0.0104, "step": 16720 }, { "epoch": 211.77215189873417, "grad_norm": 0.3376392722129822, "learning_rate": 4.461464399101312e-05, "loss": 0.0097, "step": 16730 }, { "epoch": 211.8987341772152, "grad_norm": 0.34000149369239807, "learning_rate": 4.45598522395046e-05, "loss": 0.0097, "step": 16740 }, { "epoch": 212.0253164556962, "grad_norm": 0.33971723914146423, "learning_rate": 4.4505067098288577e-05, "loss": 0.0132, "step": 16750 }, { "epoch": 212.15189873417722, "grad_norm": 0.31295913457870483, "learning_rate": 4.445028863393417e-05, "loss": 0.0096, "step": 16760 }, { "epoch": 212.27848101265823, "grad_norm": 0.2558313310146332, "learning_rate": 4.439551691300236e-05, "loss": 0.0108, "step": 16770 }, { "epoch": 212.40506329113924, "grad_norm": 0.30575650930404663, "learning_rate": 4.434075200204599e-05, "loss": 0.01, "step": 16780 }, { "epoch": 212.53164556962025, "grad_norm": 0.25703415274620056, "learning_rate": 4.428599396760956e-05, "loss": 0.0096, "step": 16790 }, { "epoch": 212.65822784810126, "grad_norm": 0.2696775496006012, "learning_rate": 4.423124287622926e-05, "loss": 0.0126, "step": 16800 }, { "epoch": 212.78481012658227, "grad_norm": 0.3072196841239929, "learning_rate": 4.417649879443282e-05, "loss": 0.0121, "step": 16810 }, { "epoch": 212.91139240506328, "grad_norm": 0.2774674594402313, "learning_rate": 4.4121761788739444e-05, "loss": 0.0119, "step": 16820 }, { "epoch": 213.0379746835443, "grad_norm": 0.3229420781135559, "learning_rate": 4.4067031925659805e-05, "loss": 0.012, "step": 16830 }, { "epoch": 213.16455696202533, "grad_norm": 0.30675584077835083, "learning_rate": 4.4012309271695816e-05, "loss": 0.0095, "step": 16840 }, { "epoch": 213.29113924050634, "grad_norm": 0.32656511664390564, "learning_rate": 4.395759389334067e-05, "loss": 0.0097, "step": 16850 }, { "epoch": 213.41772151898735, "grad_norm": 0.3297208547592163, "learning_rate": 4.3902885857078685e-05, "loss": 0.0092, "step": 16860 }, { "epoch": 213.54430379746836, "grad_norm": 0.2662786543369293, "learning_rate": 4.384818522938531e-05, "loss": 0.0105, "step": 16870 }, { "epoch": 213.67088607594937, "grad_norm": 0.27712446451187134, "learning_rate": 4.379349207672695e-05, "loss": 0.009, "step": 16880 }, { "epoch": 213.79746835443038, "grad_norm": 0.2762792110443115, "learning_rate": 4.373880646556098e-05, "loss": 0.0097, "step": 16890 }, { "epoch": 213.9240506329114, "grad_norm": 0.299276202917099, "learning_rate": 4.3684128462335533e-05, "loss": 0.0122, "step": 16900 }, { "epoch": 214.0506329113924, "grad_norm": 0.26527300477027893, "learning_rate": 4.362945813348955e-05, "loss": 0.0092, "step": 16910 }, { "epoch": 214.1772151898734, "grad_norm": 0.3162574768066406, "learning_rate": 4.357479554545262e-05, "loss": 0.0104, "step": 16920 }, { "epoch": 214.30379746835442, "grad_norm": 0.3184046447277069, "learning_rate": 4.352014076464499e-05, "loss": 0.0099, "step": 16930 }, { "epoch": 214.43037974683546, "grad_norm": 0.31981730461120605, "learning_rate": 4.346549385747734e-05, "loss": 0.0087, "step": 16940 }, { "epoch": 214.55696202531647, "grad_norm": 0.37613725662231445, "learning_rate": 4.34108548903508e-05, "loss": 0.0094, "step": 16950 }, { "epoch": 214.68354430379748, "grad_norm": 0.3335496783256531, "learning_rate": 4.335622392965689e-05, "loss": 0.0091, "step": 16960 }, { "epoch": 214.81012658227849, "grad_norm": 0.2712971568107605, "learning_rate": 4.330160104177738e-05, "loss": 0.008, "step": 16970 }, { "epoch": 214.9367088607595, "grad_norm": 0.28162387013435364, "learning_rate": 4.32469862930842e-05, "loss": 0.0106, "step": 16980 }, { "epoch": 215.0632911392405, "grad_norm": 0.31263211369514465, "learning_rate": 4.319237974993946e-05, "loss": 0.012, "step": 16990 }, { "epoch": 215.18987341772151, "grad_norm": 0.29545870423316956, "learning_rate": 4.3137781478695236e-05, "loss": 0.0115, "step": 17000 }, { "epoch": 215.31645569620252, "grad_norm": 0.3656616806983948, "learning_rate": 4.308319154569358e-05, "loss": 0.0113, "step": 17010 }, { "epoch": 215.44303797468353, "grad_norm": 0.3057821989059448, "learning_rate": 4.302861001726641e-05, "loss": 0.0112, "step": 17020 }, { "epoch": 215.56962025316454, "grad_norm": 0.3180609345436096, "learning_rate": 4.2974036959735415e-05, "loss": 0.0088, "step": 17030 }, { "epoch": 215.69620253164558, "grad_norm": 0.2612084746360779, "learning_rate": 4.2919472439412034e-05, "loss": 0.0105, "step": 17040 }, { "epoch": 215.8227848101266, "grad_norm": 0.2965889275074005, "learning_rate": 4.286491652259729e-05, "loss": 0.0091, "step": 17050 }, { "epoch": 215.9493670886076, "grad_norm": 0.29249411821365356, "learning_rate": 4.281036927558178e-05, "loss": 0.0117, "step": 17060 }, { "epoch": 216.0759493670886, "grad_norm": 0.29371654987335205, "learning_rate": 4.275583076464552e-05, "loss": 0.0089, "step": 17070 }, { "epoch": 216.20253164556962, "grad_norm": 0.32567262649536133, "learning_rate": 4.270130105605794e-05, "loss": 0.0107, "step": 17080 }, { "epoch": 216.32911392405063, "grad_norm": 0.3162460923194885, "learning_rate": 4.264678021607782e-05, "loss": 0.0105, "step": 17090 }, { "epoch": 216.45569620253164, "grad_norm": 0.3137592375278473, "learning_rate": 4.25922683109531e-05, "loss": 0.0104, "step": 17100 }, { "epoch": 216.58227848101265, "grad_norm": 0.35391223430633545, "learning_rate": 4.25377654069209e-05, "loss": 0.0118, "step": 17110 }, { "epoch": 216.70886075949366, "grad_norm": 0.3449254333972931, "learning_rate": 4.248327157020737e-05, "loss": 0.0085, "step": 17120 }, { "epoch": 216.83544303797467, "grad_norm": 0.3429417610168457, "learning_rate": 4.242878686702763e-05, "loss": 0.0102, "step": 17130 }, { "epoch": 216.9620253164557, "grad_norm": 0.2772759795188904, "learning_rate": 4.23743113635858e-05, "loss": 0.0115, "step": 17140 }, { "epoch": 217.08860759493672, "grad_norm": 0.3133218586444855, "learning_rate": 4.2319845126074705e-05, "loss": 0.0096, "step": 17150 }, { "epoch": 217.21518987341773, "grad_norm": 0.3321237564086914, "learning_rate": 4.226538822067598e-05, "loss": 0.0114, "step": 17160 }, { "epoch": 217.34177215189874, "grad_norm": 0.34603482484817505, "learning_rate": 4.2210940713559896e-05, "loss": 0.0097, "step": 17170 }, { "epoch": 217.46835443037975, "grad_norm": 0.3171890079975128, "learning_rate": 4.21565026708853e-05, "loss": 0.0114, "step": 17180 }, { "epoch": 217.59493670886076, "grad_norm": 0.28312256932258606, "learning_rate": 4.2102074158799535e-05, "loss": 0.0103, "step": 17190 }, { "epoch": 217.72151898734177, "grad_norm": 0.3206900954246521, "learning_rate": 4.204765524343841e-05, "loss": 0.0093, "step": 17200 }, { "epoch": 217.84810126582278, "grad_norm": 0.35816720128059387, "learning_rate": 4.1993245990926036e-05, "loss": 0.0102, "step": 17210 }, { "epoch": 217.9746835443038, "grad_norm": 0.4192253053188324, "learning_rate": 4.193884646737474e-05, "loss": 0.0125, "step": 17220 }, { "epoch": 218.1012658227848, "grad_norm": 0.3421322703361511, "learning_rate": 4.1884456738885124e-05, "loss": 0.01, "step": 17230 }, { "epoch": 218.22784810126583, "grad_norm": 0.31218990683555603, "learning_rate": 4.1830076871545795e-05, "loss": 0.0088, "step": 17240 }, { "epoch": 218.35443037974684, "grad_norm": 0.3308815062046051, "learning_rate": 4.177570693143346e-05, "loss": 0.0102, "step": 17250 }, { "epoch": 218.48101265822785, "grad_norm": 0.3286631107330322, "learning_rate": 4.172134698461271e-05, "loss": 0.0098, "step": 17260 }, { "epoch": 218.60759493670886, "grad_norm": 0.26476436853408813, "learning_rate": 4.166699709713599e-05, "loss": 0.0084, "step": 17270 }, { "epoch": 218.73417721518987, "grad_norm": 0.3410850167274475, "learning_rate": 4.1612657335043555e-05, "loss": 0.0102, "step": 17280 }, { "epoch": 218.86075949367088, "grad_norm": 0.2650614082813263, "learning_rate": 4.155832776436331e-05, "loss": 0.0084, "step": 17290 }, { "epoch": 218.9873417721519, "grad_norm": 0.2856336236000061, "learning_rate": 4.1504008451110845e-05, "loss": 0.0124, "step": 17300 }, { "epoch": 219.1139240506329, "grad_norm": 0.3257741928100586, "learning_rate": 4.1449699461289225e-05, "loss": 0.0117, "step": 17310 }, { "epoch": 219.2405063291139, "grad_norm": 0.3277989327907562, "learning_rate": 4.139540086088901e-05, "loss": 0.0098, "step": 17320 }, { "epoch": 219.36708860759492, "grad_norm": 0.3284159004688263, "learning_rate": 4.1341112715888106e-05, "loss": 0.0113, "step": 17330 }, { "epoch": 219.49367088607596, "grad_norm": 0.2810923457145691, "learning_rate": 4.128683509225172e-05, "loss": 0.0081, "step": 17340 }, { "epoch": 219.62025316455697, "grad_norm": 0.27373576164245605, "learning_rate": 4.123256805593231e-05, "loss": 0.0093, "step": 17350 }, { "epoch": 219.74683544303798, "grad_norm": 0.22727003693580627, "learning_rate": 4.117831167286943e-05, "loss": 0.0092, "step": 17360 }, { "epoch": 219.873417721519, "grad_norm": 0.29529863595962524, "learning_rate": 4.1124066008989685e-05, "loss": 0.0105, "step": 17370 }, { "epoch": 220.0, "grad_norm": 0.2924486994743347, "learning_rate": 4.106983113020669e-05, "loss": 0.0099, "step": 17380 }, { "epoch": 220.126582278481, "grad_norm": 0.2960222661495209, "learning_rate": 4.101560710242094e-05, "loss": 0.0099, "step": 17390 }, { "epoch": 220.25316455696202, "grad_norm": 0.30403977632522583, "learning_rate": 4.096139399151971e-05, "loss": 0.0147, "step": 17400 }, { "epoch": 220.37974683544303, "grad_norm": 0.35616791248321533, "learning_rate": 4.090719186337709e-05, "loss": 0.0093, "step": 17410 }, { "epoch": 220.50632911392404, "grad_norm": 0.37709954380989075, "learning_rate": 4.0853000783853746e-05, "loss": 0.009, "step": 17420 }, { "epoch": 220.63291139240508, "grad_norm": 0.30511274933815, "learning_rate": 4.0798820818796956e-05, "loss": 0.0099, "step": 17430 }, { "epoch": 220.7594936708861, "grad_norm": 0.2526765465736389, "learning_rate": 4.074465203404048e-05, "loss": 0.0079, "step": 17440 }, { "epoch": 220.8860759493671, "grad_norm": 0.2543767988681793, "learning_rate": 4.0690494495404495e-05, "loss": 0.0082, "step": 17450 }, { "epoch": 221.0126582278481, "grad_norm": 0.28053903579711914, "learning_rate": 4.0636348268695535e-05, "loss": 0.0115, "step": 17460 }, { "epoch": 221.13924050632912, "grad_norm": 0.25708815455436707, "learning_rate": 4.058221341970637e-05, "loss": 0.0102, "step": 17470 }, { "epoch": 221.26582278481013, "grad_norm": 0.3109745979309082, "learning_rate": 4.0528090014215945e-05, "loss": 0.0094, "step": 17480 }, { "epoch": 221.39240506329114, "grad_norm": 0.3241591453552246, "learning_rate": 4.047397811798929e-05, "loss": 0.0107, "step": 17490 }, { "epoch": 221.51898734177215, "grad_norm": 0.2530645728111267, "learning_rate": 4.041987779677745e-05, "loss": 0.0087, "step": 17500 }, { "epoch": 221.64556962025316, "grad_norm": 0.25837016105651855, "learning_rate": 4.036578911631745e-05, "loss": 0.0086, "step": 17510 }, { "epoch": 221.77215189873417, "grad_norm": 0.3073652684688568, "learning_rate": 4.031171214233211e-05, "loss": 0.0104, "step": 17520 }, { "epoch": 221.8987341772152, "grad_norm": 0.2609679698944092, "learning_rate": 4.025764694053008e-05, "loss": 0.0128, "step": 17530 }, { "epoch": 222.0253164556962, "grad_norm": 0.3045724332332611, "learning_rate": 4.020359357660566e-05, "loss": 0.0106, "step": 17540 }, { "epoch": 222.15189873417722, "grad_norm": 0.31132179498672485, "learning_rate": 4.014955211623875e-05, "loss": 0.0102, "step": 17550 }, { "epoch": 222.27848101265823, "grad_norm": 0.33697059750556946, "learning_rate": 4.0095522625094895e-05, "loss": 0.011, "step": 17560 }, { "epoch": 222.40506329113924, "grad_norm": 0.3218097984790802, "learning_rate": 4.004150516882497e-05, "loss": 0.0128, "step": 17570 }, { "epoch": 222.53164556962025, "grad_norm": 0.3043075203895569, "learning_rate": 3.99874998130653e-05, "loss": 0.0111, "step": 17580 }, { "epoch": 222.65822784810126, "grad_norm": 0.3013492822647095, "learning_rate": 3.9933506623437455e-05, "loss": 0.0111, "step": 17590 }, { "epoch": 222.78481012658227, "grad_norm": 0.29422232508659363, "learning_rate": 3.9879525665548276e-05, "loss": 0.011, "step": 17600 }, { "epoch": 222.91139240506328, "grad_norm": 0.2782425582408905, "learning_rate": 3.9825557004989704e-05, "loss": 0.0091, "step": 17610 }, { "epoch": 223.0379746835443, "grad_norm": 0.26068007946014404, "learning_rate": 3.977160070733877e-05, "loss": 0.0075, "step": 17620 }, { "epoch": 223.16455696202533, "grad_norm": 0.30522531270980835, "learning_rate": 3.971765683815746e-05, "loss": 0.0122, "step": 17630 }, { "epoch": 223.29113924050634, "grad_norm": 0.3440847396850586, "learning_rate": 3.966372546299265e-05, "loss": 0.0115, "step": 17640 }, { "epoch": 223.41772151898735, "grad_norm": 0.2915419638156891, "learning_rate": 3.960980664737604e-05, "loss": 0.0113, "step": 17650 }, { "epoch": 223.54430379746836, "grad_norm": 0.25979891419410706, "learning_rate": 3.955590045682408e-05, "loss": 0.011, "step": 17660 }, { "epoch": 223.67088607594937, "grad_norm": 0.27175500988960266, "learning_rate": 3.950200695683788e-05, "loss": 0.0081, "step": 17670 }, { "epoch": 223.79746835443038, "grad_norm": 0.33772432804107666, "learning_rate": 3.944812621290314e-05, "loss": 0.0093, "step": 17680 }, { "epoch": 223.9240506329114, "grad_norm": 0.2965388298034668, "learning_rate": 3.939425829049002e-05, "loss": 0.0133, "step": 17690 }, { "epoch": 224.0506329113924, "grad_norm": 0.29125261306762695, "learning_rate": 3.9340403255053135e-05, "loss": 0.0094, "step": 17700 }, { "epoch": 224.1772151898734, "grad_norm": 0.2612520754337311, "learning_rate": 3.9286561172031407e-05, "loss": 0.0085, "step": 17710 }, { "epoch": 224.30379746835442, "grad_norm": 0.2911588251590729, "learning_rate": 3.923273210684809e-05, "loss": 0.0087, "step": 17720 }, { "epoch": 224.43037974683546, "grad_norm": 0.30323147773742676, "learning_rate": 3.9178916124910555e-05, "loss": 0.0094, "step": 17730 }, { "epoch": 224.55696202531647, "grad_norm": 0.2743988633155823, "learning_rate": 3.9125113291610274e-05, "loss": 0.0086, "step": 17740 }, { "epoch": 224.68354430379748, "grad_norm": 0.2528955936431885, "learning_rate": 3.907132367232278e-05, "loss": 0.0087, "step": 17750 }, { "epoch": 224.81012658227849, "grad_norm": 0.28431466221809387, "learning_rate": 3.901754733240752e-05, "loss": 0.0086, "step": 17760 }, { "epoch": 224.9367088607595, "grad_norm": 0.263358473777771, "learning_rate": 3.896378433720785e-05, "loss": 0.0082, "step": 17770 }, { "epoch": 225.0632911392405, "grad_norm": 0.2795863449573517, "learning_rate": 3.891003475205086e-05, "loss": 0.0087, "step": 17780 }, { "epoch": 225.18987341772151, "grad_norm": 0.27657756209373474, "learning_rate": 3.885629864224736e-05, "loss": 0.0101, "step": 17790 }, { "epoch": 225.31645569620252, "grad_norm": 0.32800957560539246, "learning_rate": 3.880257607309178e-05, "loss": 0.009, "step": 17800 }, { "epoch": 225.44303797468353, "grad_norm": 0.2533341944217682, "learning_rate": 3.874886710986213e-05, "loss": 0.0107, "step": 17810 }, { "epoch": 225.56962025316454, "grad_norm": 0.3052198886871338, "learning_rate": 3.8695171817819834e-05, "loss": 0.0103, "step": 17820 }, { "epoch": 225.69620253164558, "grad_norm": 0.26483720541000366, "learning_rate": 3.8641490262209774e-05, "loss": 0.0082, "step": 17830 }, { "epoch": 225.8227848101266, "grad_norm": 0.32597649097442627, "learning_rate": 3.8587822508260084e-05, "loss": 0.0102, "step": 17840 }, { "epoch": 225.9493670886076, "grad_norm": 0.2988997995853424, "learning_rate": 3.853416862118214e-05, "loss": 0.011, "step": 17850 }, { "epoch": 226.0759493670886, "grad_norm": 0.28710466623306274, "learning_rate": 3.848052866617049e-05, "loss": 0.0088, "step": 17860 }, { "epoch": 226.20253164556962, "grad_norm": 0.28160789608955383, "learning_rate": 3.84269027084027e-05, "loss": 0.0097, "step": 17870 }, { "epoch": 226.32911392405063, "grad_norm": 0.25487589836120605, "learning_rate": 3.83732908130394e-05, "loss": 0.0094, "step": 17880 }, { "epoch": 226.45569620253164, "grad_norm": 0.28128525614738464, "learning_rate": 3.831969304522409e-05, "loss": 0.009, "step": 17890 }, { "epoch": 226.58227848101265, "grad_norm": 0.2586526870727539, "learning_rate": 3.826610947008312e-05, "loss": 0.009, "step": 17900 }, { "epoch": 226.70886075949366, "grad_norm": 0.24733519554138184, "learning_rate": 3.821254015272559e-05, "loss": 0.0092, "step": 17910 }, { "epoch": 226.83544303797467, "grad_norm": 0.3122856318950653, "learning_rate": 3.8158985158243214e-05, "loss": 0.0112, "step": 17920 }, { "epoch": 226.9620253164557, "grad_norm": 0.28570225834846497, "learning_rate": 3.810544455171044e-05, "loss": 0.0096, "step": 17930 }, { "epoch": 227.08860759493672, "grad_norm": 0.3233821392059326, "learning_rate": 3.8051918398184114e-05, "loss": 0.0093, "step": 17940 }, { "epoch": 227.21518987341773, "grad_norm": 0.23917058110237122, "learning_rate": 3.799840676270356e-05, "loss": 0.0089, "step": 17950 }, { "epoch": 227.34177215189874, "grad_norm": 0.2312076836824417, "learning_rate": 3.794490971029048e-05, "loss": 0.0089, "step": 17960 }, { "epoch": 227.46835443037975, "grad_norm": 0.25944533944129944, "learning_rate": 3.789142730594881e-05, "loss": 0.0111, "step": 17970 }, { "epoch": 227.59493670886076, "grad_norm": 0.24511098861694336, "learning_rate": 3.783795961466471e-05, "loss": 0.0072, "step": 17980 }, { "epoch": 227.72151898734177, "grad_norm": 0.30803102254867554, "learning_rate": 3.778450670140651e-05, "loss": 0.0085, "step": 17990 }, { "epoch": 227.84810126582278, "grad_norm": 0.26584866642951965, "learning_rate": 3.773106863112451e-05, "loss": 0.0096, "step": 18000 }, { "epoch": 227.9746835443038, "grad_norm": 0.2573884427547455, "learning_rate": 3.7677645468751e-05, "loss": 0.0104, "step": 18010 }, { "epoch": 228.1012658227848, "grad_norm": 0.23632918298244476, "learning_rate": 3.762423727920018e-05, "loss": 0.0089, "step": 18020 }, { "epoch": 228.22784810126583, "grad_norm": 0.26179760694503784, "learning_rate": 3.7570844127367995e-05, "loss": 0.0096, "step": 18030 }, { "epoch": 228.35443037974684, "grad_norm": 0.3590836226940155, "learning_rate": 3.751746607813221e-05, "loss": 0.01, "step": 18040 }, { "epoch": 228.48101265822785, "grad_norm": 0.361379474401474, "learning_rate": 3.746410319635217e-05, "loss": 0.0104, "step": 18050 }, { "epoch": 228.60759493670886, "grad_norm": 0.29509416222572327, "learning_rate": 3.7410755546868806e-05, "loss": 0.0076, "step": 18060 }, { "epoch": 228.73417721518987, "grad_norm": 0.3104882538318634, "learning_rate": 3.735742319450454e-05, "loss": 0.0099, "step": 18070 }, { "epoch": 228.86075949367088, "grad_norm": 0.330253541469574, "learning_rate": 3.7304106204063184e-05, "loss": 0.0092, "step": 18080 }, { "epoch": 228.9873417721519, "grad_norm": 0.31263238191604614, "learning_rate": 3.725080464032996e-05, "loss": 0.0104, "step": 18090 }, { "epoch": 229.1139240506329, "grad_norm": 0.29173219203948975, "learning_rate": 3.719751856807125e-05, "loss": 0.008, "step": 18100 }, { "epoch": 229.2405063291139, "grad_norm": 0.30526286363601685, "learning_rate": 3.71442480520347e-05, "loss": 0.0095, "step": 18110 }, { "epoch": 229.36708860759492, "grad_norm": 0.2821073532104492, "learning_rate": 3.709099315694897e-05, "loss": 0.0089, "step": 18120 }, { "epoch": 229.49367088607596, "grad_norm": 0.2831507623195648, "learning_rate": 3.703775394752378e-05, "loss": 0.0078, "step": 18130 }, { "epoch": 229.62025316455697, "grad_norm": 0.2567707896232605, "learning_rate": 3.698453048844983e-05, "loss": 0.0127, "step": 18140 }, { "epoch": 229.74683544303798, "grad_norm": 0.29834091663360596, "learning_rate": 3.693132284439861e-05, "loss": 0.0075, "step": 18150 }, { "epoch": 229.873417721519, "grad_norm": 0.3206382095813751, "learning_rate": 3.6878131080022414e-05, "loss": 0.0085, "step": 18160 }, { "epoch": 230.0, "grad_norm": 0.29707369208335876, "learning_rate": 3.682495525995429e-05, "loss": 0.0132, "step": 18170 }, { "epoch": 230.126582278481, "grad_norm": 0.27414989471435547, "learning_rate": 3.6771795448807846e-05, "loss": 0.0094, "step": 18180 }, { "epoch": 230.25316455696202, "grad_norm": 0.35453349351882935, "learning_rate": 3.671865171117724e-05, "loss": 0.0096, "step": 18190 }, { "epoch": 230.37974683544303, "grad_norm": 0.3058342933654785, "learning_rate": 3.666552411163718e-05, "loss": 0.0076, "step": 18200 }, { "epoch": 230.50632911392404, "grad_norm": 0.256648987531662, "learning_rate": 3.661241271474269e-05, "loss": 0.0095, "step": 18210 }, { "epoch": 230.63291139240508, "grad_norm": 0.3138805627822876, "learning_rate": 3.6559317585029116e-05, "loss": 0.0112, "step": 18220 }, { "epoch": 230.7594936708861, "grad_norm": 0.2989489734172821, "learning_rate": 3.650623878701204e-05, "loss": 0.0099, "step": 18230 }, { "epoch": 230.8860759493671, "grad_norm": 0.30132603645324707, "learning_rate": 3.645317638518721e-05, "loss": 0.0097, "step": 18240 }, { "epoch": 231.0126582278481, "grad_norm": 0.3159782886505127, "learning_rate": 3.640013044403046e-05, "loss": 0.0087, "step": 18250 }, { "epoch": 231.13924050632912, "grad_norm": 0.2615986168384552, "learning_rate": 3.634710102799761e-05, "loss": 0.0091, "step": 18260 }, { "epoch": 231.26582278481013, "grad_norm": 0.2860843241214752, "learning_rate": 3.6294088201524394e-05, "loss": 0.0102, "step": 18270 }, { "epoch": 231.39240506329114, "grad_norm": 0.29121387004852295, "learning_rate": 3.62410920290264e-05, "loss": 0.0098, "step": 18280 }, { "epoch": 231.51898734177215, "grad_norm": 0.30448639392852783, "learning_rate": 3.6188112574898954e-05, "loss": 0.0092, "step": 18290 }, { "epoch": 231.64556962025316, "grad_norm": 0.23892764747142792, "learning_rate": 3.6135149903517115e-05, "loss": 0.0098, "step": 18300 }, { "epoch": 231.77215189873417, "grad_norm": 0.2843843698501587, "learning_rate": 3.608220407923552e-05, "loss": 0.0107, "step": 18310 }, { "epoch": 231.8987341772152, "grad_norm": 0.27424585819244385, "learning_rate": 3.602927516638833e-05, "loss": 0.0088, "step": 18320 }, { "epoch": 232.0253164556962, "grad_norm": 0.31160926818847656, "learning_rate": 3.5976363229289165e-05, "loss": 0.0113, "step": 18330 }, { "epoch": 232.15189873417722, "grad_norm": 0.25160685181617737, "learning_rate": 3.5923468332231e-05, "loss": 0.0093, "step": 18340 }, { "epoch": 232.27848101265823, "grad_norm": 0.33534950017929077, "learning_rate": 3.587059053948616e-05, "loss": 0.008, "step": 18350 }, { "epoch": 232.40506329113924, "grad_norm": 0.29295480251312256, "learning_rate": 3.5817729915306134e-05, "loss": 0.009, "step": 18360 }, { "epoch": 232.53164556962025, "grad_norm": 0.291819304227829, "learning_rate": 3.5764886523921566e-05, "loss": 0.0091, "step": 18370 }, { "epoch": 232.65822784810126, "grad_norm": 0.2474658489227295, "learning_rate": 3.571206042954214e-05, "loss": 0.0112, "step": 18380 }, { "epoch": 232.78481012658227, "grad_norm": 0.2679266035556793, "learning_rate": 3.5659251696356566e-05, "loss": 0.0091, "step": 18390 }, { "epoch": 232.91139240506328, "grad_norm": 0.2768101394176483, "learning_rate": 3.56064603885324e-05, "loss": 0.0097, "step": 18400 }, { "epoch": 233.0379746835443, "grad_norm": 0.2789350152015686, "learning_rate": 3.5553686570216115e-05, "loss": 0.0121, "step": 18410 }, { "epoch": 233.16455696202533, "grad_norm": 0.24694028496742249, "learning_rate": 3.550093030553284e-05, "loss": 0.009, "step": 18420 }, { "epoch": 233.29113924050634, "grad_norm": 0.21386608481407166, "learning_rate": 3.544819165858642e-05, "loss": 0.0124, "step": 18430 }, { "epoch": 233.41772151898735, "grad_norm": 0.20741555094718933, "learning_rate": 3.539547069345926e-05, "loss": 0.0072, "step": 18440 }, { "epoch": 233.54430379746836, "grad_norm": 0.24006175994873047, "learning_rate": 3.5342767474212344e-05, "loss": 0.0097, "step": 18450 }, { "epoch": 233.67088607594937, "grad_norm": 0.24498556554317474, "learning_rate": 3.529008206488502e-05, "loss": 0.0094, "step": 18460 }, { "epoch": 233.79746835443038, "grad_norm": 0.2514749765396118, "learning_rate": 3.5237414529495055e-05, "loss": 0.0087, "step": 18470 }, { "epoch": 233.9240506329114, "grad_norm": 0.2481246143579483, "learning_rate": 3.5184764932038454e-05, "loss": 0.0087, "step": 18480 }, { "epoch": 234.0506329113924, "grad_norm": 0.27641263604164124, "learning_rate": 3.513213333648945e-05, "loss": 0.0088, "step": 18490 }, { "epoch": 234.1772151898734, "grad_norm": 0.2293470799922943, "learning_rate": 3.5079519806800374e-05, "loss": 0.0087, "step": 18500 }, { "epoch": 234.30379746835442, "grad_norm": 0.27615058422088623, "learning_rate": 3.502692440690165e-05, "loss": 0.0103, "step": 18510 }, { "epoch": 234.43037974683546, "grad_norm": 0.3012728989124298, "learning_rate": 3.497434720070165e-05, "loss": 0.0112, "step": 18520 }, { "epoch": 234.55696202531647, "grad_norm": 0.25920912623405457, "learning_rate": 3.4921788252086616e-05, "loss": 0.0081, "step": 18530 }, { "epoch": 234.68354430379748, "grad_norm": 0.279200941324234, "learning_rate": 3.486924762492065e-05, "loss": 0.0111, "step": 18540 }, { "epoch": 234.81012658227849, "grad_norm": 0.2922504246234894, "learning_rate": 3.4816725383045534e-05, "loss": 0.0087, "step": 18550 }, { "epoch": 234.9367088607595, "grad_norm": 0.2737603783607483, "learning_rate": 3.476422159028079e-05, "loss": 0.0091, "step": 18560 }, { "epoch": 235.0632911392405, "grad_norm": 0.2737991213798523, "learning_rate": 3.471173631042345e-05, "loss": 0.0116, "step": 18570 }, { "epoch": 235.18987341772151, "grad_norm": 0.22247722744941711, "learning_rate": 3.465926960724808e-05, "loss": 0.0074, "step": 18580 }, { "epoch": 235.31645569620252, "grad_norm": 0.27114611864089966, "learning_rate": 3.460682154450666e-05, "loss": 0.0103, "step": 18590 }, { "epoch": 235.44303797468353, "grad_norm": 0.31586602330207825, "learning_rate": 3.4554392185928564e-05, "loss": 0.0089, "step": 18600 }, { "epoch": 235.56962025316454, "grad_norm": 0.28127312660217285, "learning_rate": 3.450198159522037e-05, "loss": 0.0095, "step": 18610 }, { "epoch": 235.69620253164558, "grad_norm": 0.26625022292137146, "learning_rate": 3.444958983606592e-05, "loss": 0.008, "step": 18620 }, { "epoch": 235.8227848101266, "grad_norm": 0.28021228313446045, "learning_rate": 3.439721697212612e-05, "loss": 0.0089, "step": 18630 }, { "epoch": 235.9493670886076, "grad_norm": 0.2740342319011688, "learning_rate": 3.4344863067038954e-05, "loss": 0.0087, "step": 18640 }, { "epoch": 236.0759493670886, "grad_norm": 0.2790525257587433, "learning_rate": 3.429252818441935e-05, "loss": 0.0074, "step": 18650 }, { "epoch": 236.20253164556962, "grad_norm": 0.3122815489768982, "learning_rate": 3.4240212387859094e-05, "loss": 0.0078, "step": 18660 }, { "epoch": 236.32911392405063, "grad_norm": 0.221996009349823, "learning_rate": 3.418791574092686e-05, "loss": 0.0062, "step": 18670 }, { "epoch": 236.45569620253164, "grad_norm": 0.2536368668079376, "learning_rate": 3.413563830716796e-05, "loss": 0.0096, "step": 18680 }, { "epoch": 236.58227848101265, "grad_norm": 0.2628802955150604, "learning_rate": 3.408338015010445e-05, "loss": 0.01, "step": 18690 }, { "epoch": 236.70886075949366, "grad_norm": 0.286226361989975, "learning_rate": 3.40311413332349e-05, "loss": 0.0089, "step": 18700 }, { "epoch": 236.83544303797467, "grad_norm": 0.2878485918045044, "learning_rate": 3.3978921920034365e-05, "loss": 0.0079, "step": 18710 }, { "epoch": 236.9620253164557, "grad_norm": 0.29645559191703796, "learning_rate": 3.392672197395441e-05, "loss": 0.0093, "step": 18720 }, { "epoch": 237.08860759493672, "grad_norm": 0.3045327961444855, "learning_rate": 3.387454155842287e-05, "loss": 0.007, "step": 18730 }, { "epoch": 237.21518987341773, "grad_norm": 0.29658252000808716, "learning_rate": 3.382238073684386e-05, "loss": 0.0115, "step": 18740 }, { "epoch": 237.34177215189874, "grad_norm": 0.2417118102312088, "learning_rate": 3.377023957259771e-05, "loss": 0.01, "step": 18750 }, { "epoch": 237.46835443037975, "grad_norm": 0.2343144416809082, "learning_rate": 3.3718118129040835e-05, "loss": 0.0072, "step": 18760 }, { "epoch": 237.59493670886076, "grad_norm": 0.3019973039627075, "learning_rate": 3.3666016469505724e-05, "loss": 0.0089, "step": 18770 }, { "epoch": 237.72151898734177, "grad_norm": 0.26146426796913147, "learning_rate": 3.361393465730079e-05, "loss": 0.0091, "step": 18780 }, { "epoch": 237.84810126582278, "grad_norm": 0.2511689364910126, "learning_rate": 3.3561872755710366e-05, "loss": 0.0087, "step": 18790 }, { "epoch": 237.9746835443038, "grad_norm": 0.2769578993320465, "learning_rate": 3.350983082799456e-05, "loss": 0.0103, "step": 18800 }, { "epoch": 238.1012658227848, "grad_norm": 0.2558174729347229, "learning_rate": 3.34578089373892e-05, "loss": 0.0093, "step": 18810 }, { "epoch": 238.22784810126583, "grad_norm": 0.295673131942749, "learning_rate": 3.340580714710581e-05, "loss": 0.009, "step": 18820 }, { "epoch": 238.35443037974684, "grad_norm": 0.30333104729652405, "learning_rate": 3.3353825520331466e-05, "loss": 0.0085, "step": 18830 }, { "epoch": 238.48101265822785, "grad_norm": 0.2447376698255539, "learning_rate": 3.330186412022876e-05, "loss": 0.0094, "step": 18840 }, { "epoch": 238.60759493670886, "grad_norm": 0.31400397419929504, "learning_rate": 3.324992300993568e-05, "loss": 0.0101, "step": 18850 }, { "epoch": 238.73417721518987, "grad_norm": 0.2633711099624634, "learning_rate": 3.319800225256556e-05, "loss": 0.0087, "step": 18860 }, { "epoch": 238.86075949367088, "grad_norm": 0.2535097599029541, "learning_rate": 3.314610191120702e-05, "loss": 0.0077, "step": 18870 }, { "epoch": 238.9873417721519, "grad_norm": 0.2742028832435608, "learning_rate": 3.30942220489239e-05, "loss": 0.0078, "step": 18880 }, { "epoch": 239.1139240506329, "grad_norm": 0.2259722352027893, "learning_rate": 3.3042362728755086e-05, "loss": 0.0083, "step": 18890 }, { "epoch": 239.2405063291139, "grad_norm": 0.23254260420799255, "learning_rate": 3.299052401371456e-05, "loss": 0.0083, "step": 18900 }, { "epoch": 239.36708860759492, "grad_norm": 0.23429448902606964, "learning_rate": 3.293870596679125e-05, "loss": 0.0078, "step": 18910 }, { "epoch": 239.49367088607596, "grad_norm": 0.3073796033859253, "learning_rate": 3.288690865094895e-05, "loss": 0.0109, "step": 18920 }, { "epoch": 239.62025316455697, "grad_norm": 0.2427268922328949, "learning_rate": 3.283513212912632e-05, "loss": 0.0085, "step": 18930 }, { "epoch": 239.74683544303798, "grad_norm": 0.3033601939678192, "learning_rate": 3.2783376464236684e-05, "loss": 0.0067, "step": 18940 }, { "epoch": 239.873417721519, "grad_norm": 0.21725890040397644, "learning_rate": 3.273164171916806e-05, "loss": 0.0107, "step": 18950 }, { "epoch": 240.0, "grad_norm": 0.2677496671676636, "learning_rate": 3.267992795678306e-05, "loss": 0.0075, "step": 18960 }, { "epoch": 240.126582278481, "grad_norm": 0.26907581090927124, "learning_rate": 3.2628235239918744e-05, "loss": 0.011, "step": 18970 }, { "epoch": 240.25316455696202, "grad_norm": 0.28523656725883484, "learning_rate": 3.2576563631386695e-05, "loss": 0.0095, "step": 18980 }, { "epoch": 240.37974683544303, "grad_norm": 0.27582231163978577, "learning_rate": 3.252491319397275e-05, "loss": 0.01, "step": 18990 }, { "epoch": 240.50632911392404, "grad_norm": 0.2626091241836548, "learning_rate": 3.247328399043706e-05, "loss": 0.0081, "step": 19000 }, { "epoch": 240.63291139240508, "grad_norm": 0.2469189167022705, "learning_rate": 3.242167608351399e-05, "loss": 0.0071, "step": 19010 }, { "epoch": 240.7594936708861, "grad_norm": 0.3014425039291382, "learning_rate": 3.2370089535911986e-05, "loss": 0.0076, "step": 19020 }, { "epoch": 240.8860759493671, "grad_norm": 0.2632822096347809, "learning_rate": 3.23185244103136e-05, "loss": 0.0103, "step": 19030 }, { "epoch": 241.0126582278481, "grad_norm": 0.3048150837421417, "learning_rate": 3.22669807693753e-05, "loss": 0.0115, "step": 19040 }, { "epoch": 241.13924050632912, "grad_norm": 0.3198452591896057, "learning_rate": 3.2215458675727495e-05, "loss": 0.0095, "step": 19050 }, { "epoch": 241.26582278481013, "grad_norm": 0.23453065752983093, "learning_rate": 3.216395819197438e-05, "loss": 0.0118, "step": 19060 }, { "epoch": 241.39240506329114, "grad_norm": 0.27141454815864563, "learning_rate": 3.211247938069387e-05, "loss": 0.0096, "step": 19070 }, { "epoch": 241.51898734177215, "grad_norm": 0.31166329979896545, "learning_rate": 3.206102230443759e-05, "loss": 0.01, "step": 19080 }, { "epoch": 241.64556962025316, "grad_norm": 0.2914535105228424, "learning_rate": 3.2009587025730764e-05, "loss": 0.0086, "step": 19090 }, { "epoch": 241.77215189873417, "grad_norm": 0.28586384654045105, "learning_rate": 3.195817360707207e-05, "loss": 0.0109, "step": 19100 }, { "epoch": 241.8987341772152, "grad_norm": 0.2646012306213379, "learning_rate": 3.19067821109337e-05, "loss": 0.013, "step": 19110 }, { "epoch": 242.0253164556962, "grad_norm": 0.28943195939064026, "learning_rate": 3.185541259976114e-05, "loss": 0.0104, "step": 19120 }, { "epoch": 242.15189873417722, "grad_norm": 0.3506488800048828, "learning_rate": 3.180406513597316e-05, "loss": 0.0105, "step": 19130 }, { "epoch": 242.27848101265823, "grad_norm": 0.2698773145675659, "learning_rate": 3.1752739781961835e-05, "loss": 0.0107, "step": 19140 }, { "epoch": 242.40506329113924, "grad_norm": 0.27279406785964966, "learning_rate": 3.170143660009228e-05, "loss": 0.01, "step": 19150 }, { "epoch": 242.53164556962025, "grad_norm": 0.25900179147720337, "learning_rate": 3.16501556527027e-05, "loss": 0.0083, "step": 19160 }, { "epoch": 242.65822784810126, "grad_norm": 0.22975440323352814, "learning_rate": 3.1598897002104265e-05, "loss": 0.007, "step": 19170 }, { "epoch": 242.78481012658227, "grad_norm": 0.254923552274704, "learning_rate": 3.154766071058108e-05, "loss": 0.0086, "step": 19180 }, { "epoch": 242.91139240506328, "grad_norm": 0.28481870889663696, "learning_rate": 3.149644684039008e-05, "loss": 0.01, "step": 19190 }, { "epoch": 243.0379746835443, "grad_norm": 0.3279798626899719, "learning_rate": 3.144525545376095e-05, "loss": 0.0116, "step": 19200 }, { "epoch": 243.16455696202533, "grad_norm": 0.2906985282897949, "learning_rate": 3.139408661289603e-05, "loss": 0.0079, "step": 19210 }, { "epoch": 243.29113924050634, "grad_norm": 0.3142556846141815, "learning_rate": 3.134294037997032e-05, "loss": 0.0093, "step": 19220 }, { "epoch": 243.41772151898735, "grad_norm": 0.3119891583919525, "learning_rate": 3.129181681713127e-05, "loss": 0.0082, "step": 19230 }, { "epoch": 243.54430379746836, "grad_norm": 0.26203611493110657, "learning_rate": 3.1240715986498855e-05, "loss": 0.0089, "step": 19240 }, { "epoch": 243.67088607594937, "grad_norm": 0.29752159118652344, "learning_rate": 3.1189637950165394e-05, "loss": 0.0107, "step": 19250 }, { "epoch": 243.79746835443038, "grad_norm": 0.23636232316493988, "learning_rate": 3.1138582770195544e-05, "loss": 0.0101, "step": 19260 }, { "epoch": 243.9240506329114, "grad_norm": 0.3307333290576935, "learning_rate": 3.108755050862615e-05, "loss": 0.0072, "step": 19270 }, { "epoch": 244.0506329113924, "grad_norm": 0.36200520396232605, "learning_rate": 3.1036541227466204e-05, "loss": 0.0114, "step": 19280 }, { "epoch": 244.1772151898734, "grad_norm": 0.2722480297088623, "learning_rate": 3.0985554988696786e-05, "loss": 0.0086, "step": 19290 }, { "epoch": 244.30379746835442, "grad_norm": 0.2961176931858063, "learning_rate": 3.093459185427102e-05, "loss": 0.0103, "step": 19300 }, { "epoch": 244.43037974683546, "grad_norm": 0.262034147977829, "learning_rate": 3.0883651886113905e-05, "loss": 0.0088, "step": 19310 }, { "epoch": 244.55696202531647, "grad_norm": 0.3082742989063263, "learning_rate": 3.0832735146122294e-05, "loss": 0.0113, "step": 19320 }, { "epoch": 244.68354430379748, "grad_norm": 0.28890419006347656, "learning_rate": 3.078184169616485e-05, "loss": 0.0081, "step": 19330 }, { "epoch": 244.81012658227849, "grad_norm": 0.2981228828430176, "learning_rate": 3.073097159808187e-05, "loss": 0.0086, "step": 19340 }, { "epoch": 244.9367088607595, "grad_norm": 0.2784029245376587, "learning_rate": 3.068012491368537e-05, "loss": 0.0077, "step": 19350 }, { "epoch": 245.0632911392405, "grad_norm": 0.23518343269824982, "learning_rate": 3.062930170475885e-05, "loss": 0.0079, "step": 19360 }, { "epoch": 245.18987341772151, "grad_norm": 0.29056528210639954, "learning_rate": 3.057850203305729e-05, "loss": 0.0133, "step": 19370 }, { "epoch": 245.31645569620252, "grad_norm": 0.2936599850654602, "learning_rate": 3.052772596030708e-05, "loss": 0.0089, "step": 19380 }, { "epoch": 245.44303797468353, "grad_norm": 0.21361275017261505, "learning_rate": 3.0476973548205943e-05, "loss": 0.0108, "step": 19390 }, { "epoch": 245.56962025316454, "grad_norm": 0.2627056837081909, "learning_rate": 3.0426244858422847e-05, "loss": 0.0093, "step": 19400 }, { "epoch": 245.69620253164558, "grad_norm": 0.2566404938697815, "learning_rate": 3.0375539952597943e-05, "loss": 0.0099, "step": 19410 }, { "epoch": 245.8227848101266, "grad_norm": 0.28189438581466675, "learning_rate": 3.0324858892342468e-05, "loss": 0.009, "step": 19420 }, { "epoch": 245.9493670886076, "grad_norm": 0.24800638854503632, "learning_rate": 3.0274201739238672e-05, "loss": 0.008, "step": 19430 }, { "epoch": 246.0759493670886, "grad_norm": 0.262723833322525, "learning_rate": 3.0223568554839786e-05, "loss": 0.011, "step": 19440 }, { "epoch": 246.20253164556962, "grad_norm": 0.2238726168870926, "learning_rate": 3.0172959400669886e-05, "loss": 0.0063, "step": 19450 }, { "epoch": 246.32911392405063, "grad_norm": 0.2236589789390564, "learning_rate": 3.0122374338223902e-05, "loss": 0.008, "step": 19460 }, { "epoch": 246.45569620253164, "grad_norm": 0.18763937056064606, "learning_rate": 3.0071813428967427e-05, "loss": 0.0077, "step": 19470 }, { "epoch": 246.58227848101265, "grad_norm": 0.256940096616745, "learning_rate": 3.0021276734336746e-05, "loss": 0.0074, "step": 19480 }, { "epoch": 246.70886075949366, "grad_norm": 0.20036540925502777, "learning_rate": 2.997076431573871e-05, "loss": 0.0066, "step": 19490 }, { "epoch": 246.83544303797467, "grad_norm": 0.2259763777256012, "learning_rate": 2.9920276234550636e-05, "loss": 0.0064, "step": 19500 }, { "epoch": 246.9620253164557, "grad_norm": 0.25272318720817566, "learning_rate": 2.986981255212035e-05, "loss": 0.0106, "step": 19510 }, { "epoch": 247.08860759493672, "grad_norm": 0.2814439535140991, "learning_rate": 2.9819373329765977e-05, "loss": 0.0076, "step": 19520 }, { "epoch": 247.21518987341773, "grad_norm": 0.24862053990364075, "learning_rate": 2.9768958628775902e-05, "loss": 0.0091, "step": 19530 }, { "epoch": 247.34177215189874, "grad_norm": 0.26574379205703735, "learning_rate": 2.9718568510408762e-05, "loss": 0.0065, "step": 19540 }, { "epoch": 247.46835443037975, "grad_norm": 0.23034243285655975, "learning_rate": 2.9668203035893272e-05, "loss": 0.0083, "step": 19550 }, { "epoch": 247.59493670886076, "grad_norm": 0.24126489460468292, "learning_rate": 2.9617862266428287e-05, "loss": 0.0097, "step": 19560 }, { "epoch": 247.72151898734177, "grad_norm": 0.24485823512077332, "learning_rate": 2.9567546263182556e-05, "loss": 0.0099, "step": 19570 }, { "epoch": 247.84810126582278, "grad_norm": 0.22755372524261475, "learning_rate": 2.951725508729476e-05, "loss": 0.0075, "step": 19580 }, { "epoch": 247.9746835443038, "grad_norm": 0.23693515360355377, "learning_rate": 2.946698879987344e-05, "loss": 0.0068, "step": 19590 }, { "epoch": 248.1012658227848, "grad_norm": 0.2710379660129547, "learning_rate": 2.9416747461996853e-05, "loss": 0.0088, "step": 19600 }, { "epoch": 248.22784810126583, "grad_norm": 0.24100527167320251, "learning_rate": 2.9366531134712973e-05, "loss": 0.009, "step": 19610 }, { "epoch": 248.35443037974684, "grad_norm": 0.24519552290439606, "learning_rate": 2.9316339879039367e-05, "loss": 0.0071, "step": 19620 }, { "epoch": 248.48101265822785, "grad_norm": 0.24336639046669006, "learning_rate": 2.9266173755963167e-05, "loss": 0.0088, "step": 19630 }, { "epoch": 248.60759493670886, "grad_norm": 0.2306872457265854, "learning_rate": 2.9216032826440926e-05, "loss": 0.0063, "step": 19640 }, { "epoch": 248.73417721518987, "grad_norm": 0.22858956456184387, "learning_rate": 2.9165917151398592e-05, "loss": 0.0103, "step": 19650 }, { "epoch": 248.86075949367088, "grad_norm": 0.21148458123207092, "learning_rate": 2.9115826791731428e-05, "loss": 0.011, "step": 19660 }, { "epoch": 248.9873417721519, "grad_norm": 0.24481433629989624, "learning_rate": 2.906576180830398e-05, "loss": 0.0105, "step": 19670 }, { "epoch": 249.1139240506329, "grad_norm": 0.21403899788856506, "learning_rate": 2.9015722261949917e-05, "loss": 0.0071, "step": 19680 }, { "epoch": 249.2405063291139, "grad_norm": 0.3041815757751465, "learning_rate": 2.8965708213471986e-05, "loss": 0.0073, "step": 19690 }, { "epoch": 249.36708860759492, "grad_norm": 0.23664318025112152, "learning_rate": 2.8915719723641975e-05, "loss": 0.0073, "step": 19700 }, { "epoch": 249.49367088607596, "grad_norm": 0.21104472875595093, "learning_rate": 2.8865756853200604e-05, "loss": 0.007, "step": 19710 }, { "epoch": 249.62025316455697, "grad_norm": 0.2559050917625427, "learning_rate": 2.8815819662857503e-05, "loss": 0.0078, "step": 19720 }, { "epoch": 249.74683544303798, "grad_norm": 0.2936360836029053, "learning_rate": 2.876590821329105e-05, "loss": 0.0079, "step": 19730 }, { "epoch": 249.873417721519, "grad_norm": 0.2515782415866852, "learning_rate": 2.871602256514836e-05, "loss": 0.0083, "step": 19740 }, { "epoch": 250.0, "grad_norm": 0.212896466255188, "learning_rate": 2.8666162779045203e-05, "loss": 0.0083, "step": 19750 }, { "epoch": 250.126582278481, "grad_norm": 0.29056477546691895, "learning_rate": 2.8616328915565904e-05, "loss": 0.0112, "step": 19760 }, { "epoch": 250.25316455696202, "grad_norm": 0.2722928822040558, "learning_rate": 2.856652103526334e-05, "loss": 0.0086, "step": 19770 }, { "epoch": 250.37974683544303, "grad_norm": 0.26799166202545166, "learning_rate": 2.8516739198658755e-05, "loss": 0.0069, "step": 19780 }, { "epoch": 250.50632911392404, "grad_norm": 0.2632530629634857, "learning_rate": 2.846698346624177e-05, "loss": 0.0097, "step": 19790 }, { "epoch": 250.63291139240508, "grad_norm": 0.2965961992740631, "learning_rate": 2.841725389847032e-05, "loss": 0.0092, "step": 19800 }, { "epoch": 250.7594936708861, "grad_norm": 0.3060692250728607, "learning_rate": 2.8367550555770506e-05, "loss": 0.0091, "step": 19810 }, { "epoch": 250.8860759493671, "grad_norm": 0.26519718766212463, "learning_rate": 2.831787349853655e-05, "loss": 0.0075, "step": 19820 }, { "epoch": 251.0126582278481, "grad_norm": 0.2890629768371582, "learning_rate": 2.8268222787130806e-05, "loss": 0.0091, "step": 19830 }, { "epoch": 251.13924050632912, "grad_norm": 0.29644152522087097, "learning_rate": 2.821859848188355e-05, "loss": 0.0087, "step": 19840 }, { "epoch": 251.26582278481013, "grad_norm": 0.3089904487133026, "learning_rate": 2.816900064309299e-05, "loss": 0.008, "step": 19850 }, { "epoch": 251.39240506329114, "grad_norm": 0.24193193018436432, "learning_rate": 2.811942933102517e-05, "loss": 0.0089, "step": 19860 }, { "epoch": 251.51898734177215, "grad_norm": 0.24783702194690704, "learning_rate": 2.806988460591391e-05, "loss": 0.0089, "step": 19870 }, { "epoch": 251.64556962025316, "grad_norm": 0.28321996331214905, "learning_rate": 2.802036652796074e-05, "loss": 0.0082, "step": 19880 }, { "epoch": 251.77215189873417, "grad_norm": 0.26685547828674316, "learning_rate": 2.797087515733478e-05, "loss": 0.0079, "step": 19890 }, { "epoch": 251.8987341772152, "grad_norm": 0.2276250720024109, "learning_rate": 2.7921410554172723e-05, "loss": 0.0078, "step": 19900 }, { "epoch": 252.0253164556962, "grad_norm": 0.2563724219799042, "learning_rate": 2.787197277857871e-05, "loss": 0.0101, "step": 19910 }, { "epoch": 252.15189873417722, "grad_norm": 0.24989408254623413, "learning_rate": 2.782256189062429e-05, "loss": 0.0094, "step": 19920 }, { "epoch": 252.27848101265823, "grad_norm": 0.24762071669101715, "learning_rate": 2.777317795034839e-05, "loss": 0.0123, "step": 19930 }, { "epoch": 252.40506329113924, "grad_norm": 0.20848394930362701, "learning_rate": 2.7723821017757112e-05, "loss": 0.0077, "step": 19940 }, { "epoch": 252.53164556962025, "grad_norm": 0.2469085156917572, "learning_rate": 2.7674491152823822e-05, "loss": 0.0078, "step": 19950 }, { "epoch": 252.65822784810126, "grad_norm": 0.2688508629798889, "learning_rate": 2.7625188415488944e-05, "loss": 0.0117, "step": 19960 }, { "epoch": 252.78481012658227, "grad_norm": 0.2638944685459137, "learning_rate": 2.7575912865659924e-05, "loss": 0.0072, "step": 19970 }, { "epoch": 252.91139240506328, "grad_norm": 0.2711409032344818, "learning_rate": 2.7526664563211245e-05, "loss": 0.0083, "step": 19980 }, { "epoch": 253.0379746835443, "grad_norm": 0.230697900056839, "learning_rate": 2.7477443567984224e-05, "loss": 0.0077, "step": 19990 }, { "epoch": 253.16455696202533, "grad_norm": 0.28956612944602966, "learning_rate": 2.7428249939787e-05, "loss": 0.0092, "step": 20000 }, { "epoch": 253.29113924050634, "grad_norm": 0.2318604439496994, "learning_rate": 2.7379083738394484e-05, "loss": 0.0068, "step": 20010 }, { "epoch": 253.41772151898735, "grad_norm": 0.27510467171669006, "learning_rate": 2.732994502354823e-05, "loss": 0.0072, "step": 20020 }, { "epoch": 253.54430379746836, "grad_norm": 0.24221085011959076, "learning_rate": 2.72808338549564e-05, "loss": 0.0083, "step": 20030 }, { "epoch": 253.67088607594937, "grad_norm": 0.24647186696529388, "learning_rate": 2.723175029229374e-05, "loss": 0.0088, "step": 20040 }, { "epoch": 253.79746835443038, "grad_norm": 0.28152886033058167, "learning_rate": 2.718269439520138e-05, "loss": 0.0073, "step": 20050 }, { "epoch": 253.9240506329114, "grad_norm": 0.25091779232025146, "learning_rate": 2.7133666223286858e-05, "loss": 0.0069, "step": 20060 }, { "epoch": 254.0506329113924, "grad_norm": 0.25147396326065063, "learning_rate": 2.7084665836124006e-05, "loss": 0.0084, "step": 20070 }, { "epoch": 254.1772151898734, "grad_norm": 0.215467169880867, "learning_rate": 2.703569329325296e-05, "loss": 0.0079, "step": 20080 }, { "epoch": 254.30379746835442, "grad_norm": 0.26098984479904175, "learning_rate": 2.698674865417994e-05, "loss": 0.0076, "step": 20090 }, { "epoch": 254.43037974683546, "grad_norm": 0.22399897873401642, "learning_rate": 2.6937831978377325e-05, "loss": 0.0085, "step": 20100 }, { "epoch": 254.55696202531647, "grad_norm": 0.23172816634178162, "learning_rate": 2.6888943325283485e-05, "loss": 0.0092, "step": 20110 }, { "epoch": 254.68354430379748, "grad_norm": 0.3173345625400543, "learning_rate": 2.6840082754302733e-05, "loss": 0.0076, "step": 20120 }, { "epoch": 254.81012658227849, "grad_norm": 0.21272189915180206, "learning_rate": 2.679125032480525e-05, "loss": 0.0085, "step": 20130 }, { "epoch": 254.9367088607595, "grad_norm": 0.28530269861221313, "learning_rate": 2.674244609612708e-05, "loss": 0.0069, "step": 20140 }, { "epoch": 255.0632911392405, "grad_norm": 0.2824656367301941, "learning_rate": 2.6693670127569958e-05, "loss": 0.0081, "step": 20150 }, { "epoch": 255.18987341772151, "grad_norm": 0.2965713143348694, "learning_rate": 2.6644922478401268e-05, "loss": 0.0093, "step": 20160 }, { "epoch": 255.31645569620252, "grad_norm": 0.2776869535446167, "learning_rate": 2.6596203207854004e-05, "loss": 0.0099, "step": 20170 }, { "epoch": 255.44303797468353, "grad_norm": 0.26334139704704285, "learning_rate": 2.654751237512666e-05, "loss": 0.0082, "step": 20180 }, { "epoch": 255.56962025316454, "grad_norm": 0.29271239042282104, "learning_rate": 2.649885003938323e-05, "loss": 0.0085, "step": 20190 }, { "epoch": 255.69620253164558, "grad_norm": 0.24294693768024445, "learning_rate": 2.6450216259753e-05, "loss": 0.0075, "step": 20200 }, { "epoch": 255.8227848101266, "grad_norm": 0.3045782744884491, "learning_rate": 2.640161109533063e-05, "loss": 0.0106, "step": 20210 }, { "epoch": 255.9493670886076, "grad_norm": 0.3224192261695862, "learning_rate": 2.6353034605175936e-05, "loss": 0.0096, "step": 20220 }, { "epoch": 256.0759493670886, "grad_norm": 0.23440131545066833, "learning_rate": 2.630448684831398e-05, "loss": 0.0078, "step": 20230 }, { "epoch": 256.2025316455696, "grad_norm": 0.28184545040130615, "learning_rate": 2.625596788373482e-05, "loss": 0.0082, "step": 20240 }, { "epoch": 256.32911392405066, "grad_norm": 0.262582004070282, "learning_rate": 2.620747777039363e-05, "loss": 0.0073, "step": 20250 }, { "epoch": 256.45569620253167, "grad_norm": 0.255980908870697, "learning_rate": 2.6159016567210424e-05, "loss": 0.0081, "step": 20260 }, { "epoch": 256.5822784810127, "grad_norm": 0.21040263772010803, "learning_rate": 2.6110584333070153e-05, "loss": 0.0078, "step": 20270 }, { "epoch": 256.7088607594937, "grad_norm": 0.23297759890556335, "learning_rate": 2.6062181126822537e-05, "loss": 0.0074, "step": 20280 }, { "epoch": 256.8354430379747, "grad_norm": 0.27478930354118347, "learning_rate": 2.601380700728203e-05, "loss": 0.0073, "step": 20290 }, { "epoch": 256.9620253164557, "grad_norm": 0.255378782749176, "learning_rate": 2.596546203322777e-05, "loss": 0.0101, "step": 20300 }, { "epoch": 257.0886075949367, "grad_norm": 0.2768028974533081, "learning_rate": 2.5917146263403457e-05, "loss": 0.0084, "step": 20310 }, { "epoch": 257.2151898734177, "grad_norm": 0.21759451925754547, "learning_rate": 2.5868859756517296e-05, "loss": 0.006, "step": 20320 }, { "epoch": 257.34177215189874, "grad_norm": 0.22210387885570526, "learning_rate": 2.5820602571241947e-05, "loss": 0.0067, "step": 20330 }, { "epoch": 257.46835443037975, "grad_norm": 0.2044757455587387, "learning_rate": 2.577237476621442e-05, "loss": 0.0079, "step": 20340 }, { "epoch": 257.59493670886076, "grad_norm": 0.2677971124649048, "learning_rate": 2.572417640003609e-05, "loss": 0.0074, "step": 20350 }, { "epoch": 257.72151898734177, "grad_norm": 0.21751698851585388, "learning_rate": 2.5676007531272473e-05, "loss": 0.0094, "step": 20360 }, { "epoch": 257.8481012658228, "grad_norm": 0.2548682987689972, "learning_rate": 2.5627868218453332e-05, "loss": 0.0088, "step": 20370 }, { "epoch": 257.9746835443038, "grad_norm": 0.2800174653530121, "learning_rate": 2.5579758520072443e-05, "loss": 0.0111, "step": 20380 }, { "epoch": 258.1012658227848, "grad_norm": 0.2411888688802719, "learning_rate": 2.5531678494587615e-05, "loss": 0.0071, "step": 20390 }, { "epoch": 258.2278481012658, "grad_norm": 0.23628754913806915, "learning_rate": 2.5483628200420644e-05, "loss": 0.0087, "step": 20400 }, { "epoch": 258.3544303797468, "grad_norm": 0.21095001697540283, "learning_rate": 2.5435607695957154e-05, "loss": 0.0068, "step": 20410 }, { "epoch": 258.4810126582278, "grad_norm": 0.3182445764541626, "learning_rate": 2.5387617039546586e-05, "loss": 0.0073, "step": 20420 }, { "epoch": 258.60759493670884, "grad_norm": 0.24894607067108154, "learning_rate": 2.5339656289502106e-05, "loss": 0.007, "step": 20430 }, { "epoch": 258.73417721518985, "grad_norm": 0.2612587809562683, "learning_rate": 2.529172550410056e-05, "loss": 0.0081, "step": 20440 }, { "epoch": 258.8607594936709, "grad_norm": 0.1770034283399582, "learning_rate": 2.524382474158234e-05, "loss": 0.0083, "step": 20450 }, { "epoch": 258.9873417721519, "grad_norm": 0.29097896814346313, "learning_rate": 2.5195954060151433e-05, "loss": 0.0075, "step": 20460 }, { "epoch": 259.11392405063293, "grad_norm": 0.22325487434864044, "learning_rate": 2.5148113517975212e-05, "loss": 0.008, "step": 20470 }, { "epoch": 259.24050632911394, "grad_norm": 0.20969663560390472, "learning_rate": 2.5100303173184447e-05, "loss": 0.011, "step": 20480 }, { "epoch": 259.36708860759495, "grad_norm": 0.2429189682006836, "learning_rate": 2.5052523083873196e-05, "loss": 0.0077, "step": 20490 }, { "epoch": 259.49367088607596, "grad_norm": 0.16925111413002014, "learning_rate": 2.500477330809881e-05, "loss": 0.0061, "step": 20500 }, { "epoch": 259.62025316455697, "grad_norm": 0.26670610904693604, "learning_rate": 2.4957053903881733e-05, "loss": 0.0067, "step": 20510 }, { "epoch": 259.746835443038, "grad_norm": 0.19937902688980103, "learning_rate": 2.4909364929205576e-05, "loss": 0.0059, "step": 20520 }, { "epoch": 259.873417721519, "grad_norm": 0.21450598537921906, "learning_rate": 2.4861706442016923e-05, "loss": 0.0081, "step": 20530 }, { "epoch": 260.0, "grad_norm": 0.21705575287342072, "learning_rate": 2.481407850022533e-05, "loss": 0.0087, "step": 20540 }, { "epoch": 260.126582278481, "grad_norm": 0.2078782171010971, "learning_rate": 2.4766481161703213e-05, "loss": 0.0067, "step": 20550 }, { "epoch": 260.253164556962, "grad_norm": 0.24040336906909943, "learning_rate": 2.4718914484285876e-05, "loss": 0.0079, "step": 20560 }, { "epoch": 260.37974683544303, "grad_norm": 0.21381452679634094, "learning_rate": 2.467137852577129e-05, "loss": 0.0077, "step": 20570 }, { "epoch": 260.50632911392404, "grad_norm": 0.25708332657814026, "learning_rate": 2.4623873343920123e-05, "loss": 0.0087, "step": 20580 }, { "epoch": 260.63291139240505, "grad_norm": 0.1907951980829239, "learning_rate": 2.4576398996455656e-05, "loss": 0.0081, "step": 20590 }, { "epoch": 260.75949367088606, "grad_norm": 0.22575445473194122, "learning_rate": 2.452895554106368e-05, "loss": 0.0078, "step": 20600 }, { "epoch": 260.88607594936707, "grad_norm": 0.2163461595773697, "learning_rate": 2.4481543035392506e-05, "loss": 0.0091, "step": 20610 }, { "epoch": 261.0126582278481, "grad_norm": 0.2661060392856598, "learning_rate": 2.4434161537052774e-05, "loss": 0.006, "step": 20620 }, { "epoch": 261.1392405063291, "grad_norm": 0.20578700304031372, "learning_rate": 2.4386811103617474e-05, "loss": 0.0091, "step": 20630 }, { "epoch": 261.26582278481015, "grad_norm": 0.23480038344860077, "learning_rate": 2.4339491792621833e-05, "loss": 0.0073, "step": 20640 }, { "epoch": 261.39240506329116, "grad_norm": 0.2901012897491455, "learning_rate": 2.4292203661563312e-05, "loss": 0.0067, "step": 20650 }, { "epoch": 261.5189873417722, "grad_norm": 0.2532615065574646, "learning_rate": 2.424494676790141e-05, "loss": 0.0076, "step": 20660 }, { "epoch": 261.6455696202532, "grad_norm": 0.24196617305278778, "learning_rate": 2.4197721169057745e-05, "loss": 0.007, "step": 20670 }, { "epoch": 261.7721518987342, "grad_norm": 0.23630447685718536, "learning_rate": 2.4150526922415857e-05, "loss": 0.0082, "step": 20680 }, { "epoch": 261.8987341772152, "grad_norm": 0.2092188149690628, "learning_rate": 2.41033640853212e-05, "loss": 0.0087, "step": 20690 }, { "epoch": 262.0253164556962, "grad_norm": 0.21976418793201447, "learning_rate": 2.405623271508108e-05, "loss": 0.009, "step": 20700 }, { "epoch": 262.1518987341772, "grad_norm": 0.2344118356704712, "learning_rate": 2.4009132868964522e-05, "loss": 0.0091, "step": 20710 }, { "epoch": 262.27848101265823, "grad_norm": 0.22538962960243225, "learning_rate": 2.3962064604202327e-05, "loss": 0.0107, "step": 20720 }, { "epoch": 262.40506329113924, "grad_norm": 0.18538114428520203, "learning_rate": 2.3915027977986858e-05, "loss": 0.0079, "step": 20730 }, { "epoch": 262.53164556962025, "grad_norm": 0.22452445328235626, "learning_rate": 2.386802304747205e-05, "loss": 0.0068, "step": 20740 }, { "epoch": 262.65822784810126, "grad_norm": 0.20403075218200684, "learning_rate": 2.382104986977332e-05, "loss": 0.0076, "step": 20750 }, { "epoch": 262.7848101265823, "grad_norm": 0.2958781123161316, "learning_rate": 2.377410850196749e-05, "loss": 0.0059, "step": 20760 }, { "epoch": 262.9113924050633, "grad_norm": 0.2178170084953308, "learning_rate": 2.3727199001092797e-05, "loss": 0.0081, "step": 20770 }, { "epoch": 263.0379746835443, "grad_norm": 0.19557228684425354, "learning_rate": 2.3680321424148677e-05, "loss": 0.0083, "step": 20780 }, { "epoch": 263.1645569620253, "grad_norm": 0.21169111132621765, "learning_rate": 2.3633475828095793e-05, "loss": 0.0063, "step": 20790 }, { "epoch": 263.2911392405063, "grad_norm": 0.2359568029642105, "learning_rate": 2.358666226985599e-05, "loss": 0.0074, "step": 20800 }, { "epoch": 263.4177215189873, "grad_norm": 0.27106973528862, "learning_rate": 2.353988080631213e-05, "loss": 0.009, "step": 20810 }, { "epoch": 263.54430379746833, "grad_norm": 0.24633361399173737, "learning_rate": 2.3493131494308142e-05, "loss": 0.0074, "step": 20820 }, { "epoch": 263.67088607594934, "grad_norm": 0.26593634486198425, "learning_rate": 2.344641439064881e-05, "loss": 0.0092, "step": 20830 }, { "epoch": 263.7974683544304, "grad_norm": 0.2354051023721695, "learning_rate": 2.339972955209984e-05, "loss": 0.007, "step": 20840 }, { "epoch": 263.9240506329114, "grad_norm": 0.2296781986951828, "learning_rate": 2.335307703538771e-05, "loss": 0.0088, "step": 20850 }, { "epoch": 264.0506329113924, "grad_norm": 0.2300533950328827, "learning_rate": 2.330645689719962e-05, "loss": 0.0077, "step": 20860 }, { "epoch": 264.17721518987344, "grad_norm": 0.28882932662963867, "learning_rate": 2.3259869194183414e-05, "loss": 0.0084, "step": 20870 }, { "epoch": 264.30379746835445, "grad_norm": 0.24437253177165985, "learning_rate": 2.321331398294759e-05, "loss": 0.0065, "step": 20880 }, { "epoch": 264.43037974683546, "grad_norm": 0.2943979501724243, "learning_rate": 2.3166791320061094e-05, "loss": 0.0076, "step": 20890 }, { "epoch": 264.55696202531647, "grad_norm": 0.2746918797492981, "learning_rate": 2.312030126205335e-05, "loss": 0.0077, "step": 20900 }, { "epoch": 264.6835443037975, "grad_norm": 0.21956494450569153, "learning_rate": 2.3073843865414162e-05, "loss": 0.0082, "step": 20910 }, { "epoch": 264.8101265822785, "grad_norm": 0.23127739131450653, "learning_rate": 2.302741918659363e-05, "loss": 0.009, "step": 20920 }, { "epoch": 264.9367088607595, "grad_norm": 0.3210161030292511, "learning_rate": 2.2981027282002154e-05, "loss": 0.0073, "step": 20930 }, { "epoch": 265.0632911392405, "grad_norm": 0.2944880425930023, "learning_rate": 2.2934668208010236e-05, "loss": 0.0085, "step": 20940 }, { "epoch": 265.1898734177215, "grad_norm": 0.19440603256225586, "learning_rate": 2.2888342020948555e-05, "loss": 0.0088, "step": 20950 }, { "epoch": 265.3164556962025, "grad_norm": 0.21744424104690552, "learning_rate": 2.2842048777107783e-05, "loss": 0.0079, "step": 20960 }, { "epoch": 265.44303797468353, "grad_norm": 0.22037921845912933, "learning_rate": 2.2795788532738553e-05, "loss": 0.0081, "step": 20970 }, { "epoch": 265.56962025316454, "grad_norm": 0.2001199722290039, "learning_rate": 2.2749561344051468e-05, "loss": 0.0081, "step": 20980 }, { "epoch": 265.69620253164555, "grad_norm": 0.2544948160648346, "learning_rate": 2.2703367267216896e-05, "loss": 0.0086, "step": 20990 }, { "epoch": 265.82278481012656, "grad_norm": 0.22076863050460815, "learning_rate": 2.2657206358364997e-05, "loss": 0.0071, "step": 21000 }, { "epoch": 265.9493670886076, "grad_norm": 0.24846583604812622, "learning_rate": 2.261107867358563e-05, "loss": 0.0075, "step": 21010 }, { "epoch": 266.0759493670886, "grad_norm": 0.2774706184864044, "learning_rate": 2.2564984268928263e-05, "loss": 0.008, "step": 21020 }, { "epoch": 266.2025316455696, "grad_norm": 0.23554444313049316, "learning_rate": 2.251892320040198e-05, "loss": 0.0073, "step": 21030 }, { "epoch": 266.32911392405066, "grad_norm": 0.2322206050157547, "learning_rate": 2.2472895523975312e-05, "loss": 0.0098, "step": 21040 }, { "epoch": 266.45569620253167, "grad_norm": 0.23439063131809235, "learning_rate": 2.2426901295576212e-05, "loss": 0.0064, "step": 21050 }, { "epoch": 266.5822784810127, "grad_norm": 0.24742068350315094, "learning_rate": 2.2380940571092013e-05, "loss": 0.0073, "step": 21060 }, { "epoch": 266.7088607594937, "grad_norm": 0.2553226351737976, "learning_rate": 2.2335013406369302e-05, "loss": 0.0064, "step": 21070 }, { "epoch": 266.8354430379747, "grad_norm": 0.252151757478714, "learning_rate": 2.228911985721397e-05, "loss": 0.0088, "step": 21080 }, { "epoch": 266.9620253164557, "grad_norm": 0.2262161523103714, "learning_rate": 2.224325997939095e-05, "loss": 0.0066, "step": 21090 }, { "epoch": 267.0886075949367, "grad_norm": 0.23277647793293, "learning_rate": 2.219743382862437e-05, "loss": 0.0073, "step": 21100 }, { "epoch": 267.2151898734177, "grad_norm": 0.22278520464897156, "learning_rate": 2.2151641460597294e-05, "loss": 0.0064, "step": 21110 }, { "epoch": 267.34177215189874, "grad_norm": 0.21484656631946564, "learning_rate": 2.210588293095177e-05, "loss": 0.0074, "step": 21120 }, { "epoch": 267.46835443037975, "grad_norm": 0.21850945055484772, "learning_rate": 2.2060158295288714e-05, "loss": 0.0083, "step": 21130 }, { "epoch": 267.59493670886076, "grad_norm": 0.24112437665462494, "learning_rate": 2.2014467609167905e-05, "loss": 0.0069, "step": 21140 }, { "epoch": 267.72151898734177, "grad_norm": 0.19780471920967102, "learning_rate": 2.1968810928107806e-05, "loss": 0.0073, "step": 21150 }, { "epoch": 267.8481012658228, "grad_norm": 0.20252558588981628, "learning_rate": 2.1923188307585606e-05, "loss": 0.0075, "step": 21160 }, { "epoch": 267.9746835443038, "grad_norm": 0.20281201601028442, "learning_rate": 2.187759980303708e-05, "loss": 0.0071, "step": 21170 }, { "epoch": 268.1012658227848, "grad_norm": 0.2249174863100052, "learning_rate": 2.1832045469856545e-05, "loss": 0.0071, "step": 21180 }, { "epoch": 268.2278481012658, "grad_norm": 0.19959108531475067, "learning_rate": 2.1786525363396844e-05, "loss": 0.0071, "step": 21190 }, { "epoch": 268.3544303797468, "grad_norm": 0.21674089133739471, "learning_rate": 2.1741039538969183e-05, "loss": 0.0063, "step": 21200 }, { "epoch": 268.4810126582278, "grad_norm": 0.23497271537780762, "learning_rate": 2.169558805184313e-05, "loss": 0.0074, "step": 21210 }, { "epoch": 268.60759493670884, "grad_norm": 0.1847916841506958, "learning_rate": 2.165017095724651e-05, "loss": 0.0067, "step": 21220 }, { "epoch": 268.73417721518985, "grad_norm": 0.20221413671970367, "learning_rate": 2.1604788310365405e-05, "loss": 0.0081, "step": 21230 }, { "epoch": 268.8607594936709, "grad_norm": 0.21007698774337769, "learning_rate": 2.155944016634398e-05, "loss": 0.008, "step": 21240 }, { "epoch": 268.9873417721519, "grad_norm": 0.1611693948507309, "learning_rate": 2.1514126580284538e-05, "loss": 0.0063, "step": 21250 }, { "epoch": 269.11392405063293, "grad_norm": 0.19535894691944122, "learning_rate": 2.1468847607247344e-05, "loss": 0.0058, "step": 21260 }, { "epoch": 269.24050632911394, "grad_norm": 0.22502093017101288, "learning_rate": 2.1423603302250623e-05, "loss": 0.0061, "step": 21270 }, { "epoch": 269.36708860759495, "grad_norm": 0.23033283650875092, "learning_rate": 2.1378393720270468e-05, "loss": 0.0062, "step": 21280 }, { "epoch": 269.49367088607596, "grad_norm": 0.19351798295974731, "learning_rate": 2.133321891624076e-05, "loss": 0.0068, "step": 21290 }, { "epoch": 269.62025316455697, "grad_norm": 0.21589542925357819, "learning_rate": 2.1288078945053193e-05, "loss": 0.0068, "step": 21300 }, { "epoch": 269.746835443038, "grad_norm": 0.20192094147205353, "learning_rate": 2.1242973861557064e-05, "loss": 0.0106, "step": 21310 }, { "epoch": 269.873417721519, "grad_norm": 0.2195819765329361, "learning_rate": 2.1197903720559304e-05, "loss": 0.0067, "step": 21320 }, { "epoch": 270.0, "grad_norm": 0.3023253083229065, "learning_rate": 2.115286857682438e-05, "loss": 0.0077, "step": 21330 }, { "epoch": 270.126582278481, "grad_norm": 0.23457518219947815, "learning_rate": 2.1107868485074228e-05, "loss": 0.0071, "step": 21340 }, { "epoch": 270.253164556962, "grad_norm": 0.24176055192947388, "learning_rate": 2.1062903499988234e-05, "loss": 0.0097, "step": 21350 }, { "epoch": 270.37974683544303, "grad_norm": 0.22909735143184662, "learning_rate": 2.101797367620308e-05, "loss": 0.0081, "step": 21360 }, { "epoch": 270.50632911392404, "grad_norm": 0.20593304932117462, "learning_rate": 2.0973079068312713e-05, "loss": 0.0063, "step": 21370 }, { "epoch": 270.63291139240505, "grad_norm": 0.2113201767206192, "learning_rate": 2.0928219730868358e-05, "loss": 0.0068, "step": 21380 }, { "epoch": 270.75949367088606, "grad_norm": 0.23922507464885712, "learning_rate": 2.0883395718378302e-05, "loss": 0.0065, "step": 21390 }, { "epoch": 270.88607594936707, "grad_norm": 0.220464289188385, "learning_rate": 2.083860708530798e-05, "loss": 0.0083, "step": 21400 }, { "epoch": 271.0126582278481, "grad_norm": 0.17176516354084015, "learning_rate": 2.0793853886079796e-05, "loss": 0.0078, "step": 21410 }, { "epoch": 271.1392405063291, "grad_norm": 0.17638924717903137, "learning_rate": 2.074913617507309e-05, "loss": 0.0077, "step": 21420 }, { "epoch": 271.26582278481015, "grad_norm": 0.23490694165229797, "learning_rate": 2.070445400662411e-05, "loss": 0.008, "step": 21430 }, { "epoch": 271.39240506329116, "grad_norm": 0.20881083607673645, "learning_rate": 2.0659807435025907e-05, "loss": 0.007, "step": 21440 }, { "epoch": 271.5189873417722, "grad_norm": 0.21529953181743622, "learning_rate": 2.061519651452825e-05, "loss": 0.0063, "step": 21450 }, { "epoch": 271.6455696202532, "grad_norm": 0.22627398371696472, "learning_rate": 2.0570621299337656e-05, "loss": 0.0074, "step": 21460 }, { "epoch": 271.7721518987342, "grad_norm": 0.20526228845119476, "learning_rate": 2.0526081843617183e-05, "loss": 0.009, "step": 21470 }, { "epoch": 271.8987341772152, "grad_norm": 0.24524085223674774, "learning_rate": 2.0481578201486485e-05, "loss": 0.0059, "step": 21480 }, { "epoch": 272.0253164556962, "grad_norm": 0.22862288355827332, "learning_rate": 2.0437110427021677e-05, "loss": 0.0068, "step": 21490 }, { "epoch": 272.1518987341772, "grad_norm": 0.2416796088218689, "learning_rate": 2.0392678574255282e-05, "loss": 0.0063, "step": 21500 }, { "epoch": 272.27848101265823, "grad_norm": 0.2513695955276489, "learning_rate": 2.0348282697176217e-05, "loss": 0.0074, "step": 21510 }, { "epoch": 272.40506329113924, "grad_norm": 0.2672124207019806, "learning_rate": 2.030392284972964e-05, "loss": 0.0083, "step": 21520 }, { "epoch": 272.53164556962025, "grad_norm": 0.2505030632019043, "learning_rate": 2.0259599085816973e-05, "loss": 0.0066, "step": 21530 }, { "epoch": 272.65822784810126, "grad_norm": 0.2485566884279251, "learning_rate": 2.0215311459295755e-05, "loss": 0.0063, "step": 21540 }, { "epoch": 272.7848101265823, "grad_norm": 0.24291327595710754, "learning_rate": 2.0171060023979606e-05, "loss": 0.0071, "step": 21550 }, { "epoch": 272.9113924050633, "grad_norm": 0.18268580734729767, "learning_rate": 2.012684483363823e-05, "loss": 0.0071, "step": 21560 }, { "epoch": 273.0379746835443, "grad_norm": 0.20088249444961548, "learning_rate": 2.0082665941997237e-05, "loss": 0.0071, "step": 21570 }, { "epoch": 273.1645569620253, "grad_norm": 0.24291449785232544, "learning_rate": 2.0038523402738146e-05, "loss": 0.0074, "step": 21580 }, { "epoch": 273.2911392405063, "grad_norm": 0.24289298057556152, "learning_rate": 1.99944172694983e-05, "loss": 0.0076, "step": 21590 }, { "epoch": 273.4177215189873, "grad_norm": 0.2109215408563614, "learning_rate": 1.99503475958708e-05, "loss": 0.006, "step": 21600 }, { "epoch": 273.54430379746833, "grad_norm": 0.20237548649311066, "learning_rate": 1.9906314435404484e-05, "loss": 0.008, "step": 21610 }, { "epoch": 273.67088607594934, "grad_norm": 0.23042096197605133, "learning_rate": 1.986231784160378e-05, "loss": 0.0067, "step": 21620 }, { "epoch": 273.7974683544304, "grad_norm": 0.22426412999629974, "learning_rate": 1.9818357867928693e-05, "loss": 0.0072, "step": 21630 }, { "epoch": 273.9240506329114, "grad_norm": 0.21516898274421692, "learning_rate": 1.977443456779474e-05, "loss": 0.0065, "step": 21640 }, { "epoch": 274.0506329113924, "grad_norm": 0.22429385781288147, "learning_rate": 1.9730547994572858e-05, "loss": 0.007, "step": 21650 }, { "epoch": 274.17721518987344, "grad_norm": 0.22299319505691528, "learning_rate": 1.9686698201589393e-05, "loss": 0.0066, "step": 21660 }, { "epoch": 274.30379746835445, "grad_norm": 0.2536725103855133, "learning_rate": 1.9642885242125964e-05, "loss": 0.0066, "step": 21670 }, { "epoch": 274.43037974683546, "grad_norm": 0.22139832377433777, "learning_rate": 1.9599109169419466e-05, "loss": 0.0064, "step": 21680 }, { "epoch": 274.55696202531647, "grad_norm": 0.2205871194601059, "learning_rate": 1.9555370036661948e-05, "loss": 0.0057, "step": 21690 }, { "epoch": 274.6835443037975, "grad_norm": 0.23629650473594666, "learning_rate": 1.9511667897000574e-05, "loss": 0.0087, "step": 21700 }, { "epoch": 274.8101265822785, "grad_norm": 0.22450314462184906, "learning_rate": 1.946800280353755e-05, "loss": 0.0065, "step": 21710 }, { "epoch": 274.9367088607595, "grad_norm": 0.21622686088085175, "learning_rate": 1.9424374809330114e-05, "loss": 0.0083, "step": 21720 }, { "epoch": 275.0632911392405, "grad_norm": 0.24611051380634308, "learning_rate": 1.9380783967390382e-05, "loss": 0.0069, "step": 21730 }, { "epoch": 275.1898734177215, "grad_norm": 0.20551955699920654, "learning_rate": 1.933723033068533e-05, "loss": 0.0064, "step": 21740 }, { "epoch": 275.3164556962025, "grad_norm": 0.18701250851154327, "learning_rate": 1.929371395213674e-05, "loss": 0.0093, "step": 21750 }, { "epoch": 275.44303797468353, "grad_norm": 0.30709466338157654, "learning_rate": 1.9250234884621092e-05, "loss": 0.0079, "step": 21760 }, { "epoch": 275.56962025316454, "grad_norm": 0.25499433279037476, "learning_rate": 1.920679318096959e-05, "loss": 0.0064, "step": 21770 }, { "epoch": 275.69620253164555, "grad_norm": 0.24402444064617157, "learning_rate": 1.9163388893967982e-05, "loss": 0.0085, "step": 21780 }, { "epoch": 275.82278481012656, "grad_norm": 0.2140086591243744, "learning_rate": 1.9120022076356575e-05, "loss": 0.0053, "step": 21790 }, { "epoch": 275.9493670886076, "grad_norm": 0.23048371076583862, "learning_rate": 1.9076692780830114e-05, "loss": 0.007, "step": 21800 }, { "epoch": 276.0759493670886, "grad_norm": 0.30658403038978577, "learning_rate": 1.903340106003782e-05, "loss": 0.0078, "step": 21810 }, { "epoch": 276.2025316455696, "grad_norm": 0.2663652300834656, "learning_rate": 1.8990146966583183e-05, "loss": 0.0072, "step": 21820 }, { "epoch": 276.32911392405066, "grad_norm": 0.2359776794910431, "learning_rate": 1.8946930553024035e-05, "loss": 0.0076, "step": 21830 }, { "epoch": 276.45569620253167, "grad_norm": 0.23562267422676086, "learning_rate": 1.8903751871872378e-05, "loss": 0.0066, "step": 21840 }, { "epoch": 276.5822784810127, "grad_norm": 0.20510028302669525, "learning_rate": 1.8860610975594382e-05, "loss": 0.0094, "step": 21850 }, { "epoch": 276.7088607594937, "grad_norm": 0.20815660059452057, "learning_rate": 1.8817507916610307e-05, "loss": 0.0072, "step": 21860 }, { "epoch": 276.8354430379747, "grad_norm": 0.23317678272724152, "learning_rate": 1.8774442747294407e-05, "loss": 0.0079, "step": 21870 }, { "epoch": 276.9620253164557, "grad_norm": 0.21890705823898315, "learning_rate": 1.8731415519974965e-05, "loss": 0.0067, "step": 21880 }, { "epoch": 277.0886075949367, "grad_norm": 0.22660483419895172, "learning_rate": 1.86884262869341e-05, "loss": 0.0066, "step": 21890 }, { "epoch": 277.2151898734177, "grad_norm": 0.2095317244529724, "learning_rate": 1.8645475100407788e-05, "loss": 0.0059, "step": 21900 }, { "epoch": 277.34177215189874, "grad_norm": 0.21909138560295105, "learning_rate": 1.8602562012585768e-05, "loss": 0.0074, "step": 21910 }, { "epoch": 277.46835443037975, "grad_norm": 0.2610096037387848, "learning_rate": 1.8559687075611466e-05, "loss": 0.0063, "step": 21920 }, { "epoch": 277.59493670886076, "grad_norm": 0.2039376199245453, "learning_rate": 1.8516850341582017e-05, "loss": 0.0093, "step": 21930 }, { "epoch": 277.72151898734177, "grad_norm": 0.20586441457271576, "learning_rate": 1.847405186254807e-05, "loss": 0.0067, "step": 21940 }, { "epoch": 277.8481012658228, "grad_norm": 0.20985007286071777, "learning_rate": 1.843129169051379e-05, "loss": 0.0072, "step": 21950 }, { "epoch": 277.9746835443038, "grad_norm": 0.2363290935754776, "learning_rate": 1.8388569877436863e-05, "loss": 0.0058, "step": 21960 }, { "epoch": 278.1012658227848, "grad_norm": 0.29112669825553894, "learning_rate": 1.834588647522828e-05, "loss": 0.0079, "step": 21970 }, { "epoch": 278.2278481012658, "grad_norm": 0.2682948410511017, "learning_rate": 1.8303241535752435e-05, "loss": 0.0069, "step": 21980 }, { "epoch": 278.3544303797468, "grad_norm": 0.20352625846862793, "learning_rate": 1.8260635110826934e-05, "loss": 0.0071, "step": 21990 }, { "epoch": 278.4810126582278, "grad_norm": 0.21634633839130402, "learning_rate": 1.8218067252222597e-05, "loss": 0.0082, "step": 22000 }, { "epoch": 278.60759493670884, "grad_norm": 0.3100447952747345, "learning_rate": 1.817553801166339e-05, "loss": 0.0063, "step": 22010 }, { "epoch": 278.73417721518985, "grad_norm": 0.2333492934703827, "learning_rate": 1.8133047440826333e-05, "loss": 0.0073, "step": 22020 }, { "epoch": 278.8607594936709, "grad_norm": 0.25144821405410767, "learning_rate": 1.809059559134151e-05, "loss": 0.0072, "step": 22030 }, { "epoch": 278.9873417721519, "grad_norm": 0.23499730229377747, "learning_rate": 1.80481825147919e-05, "loss": 0.0061, "step": 22040 }, { "epoch": 279.11392405063293, "grad_norm": 0.24843591451644897, "learning_rate": 1.8005808262713396e-05, "loss": 0.0063, "step": 22050 }, { "epoch": 279.24050632911394, "grad_norm": 0.22221393883228302, "learning_rate": 1.7963472886594714e-05, "loss": 0.005, "step": 22060 }, { "epoch": 279.36708860759495, "grad_norm": 0.2559168040752411, "learning_rate": 1.7921176437877302e-05, "loss": 0.007, "step": 22070 }, { "epoch": 279.49367088607596, "grad_norm": 0.22362923622131348, "learning_rate": 1.7878918967955366e-05, "loss": 0.0074, "step": 22080 }, { "epoch": 279.62025316455697, "grad_norm": 0.2530349791049957, "learning_rate": 1.7836700528175693e-05, "loss": 0.0075, "step": 22090 }, { "epoch": 279.746835443038, "grad_norm": 0.20043236017227173, "learning_rate": 1.779452116983769e-05, "loss": 0.0061, "step": 22100 }, { "epoch": 279.873417721519, "grad_norm": 0.2244730293750763, "learning_rate": 1.775238094419325e-05, "loss": 0.007, "step": 22110 }, { "epoch": 280.0, "grad_norm": 0.18642902374267578, "learning_rate": 1.7710279902446708e-05, "loss": 0.0052, "step": 22120 }, { "epoch": 280.126582278481, "grad_norm": 0.17487481236457825, "learning_rate": 1.7668218095754795e-05, "loss": 0.0084, "step": 22130 }, { "epoch": 280.253164556962, "grad_norm": 0.2519301772117615, "learning_rate": 1.7626195575226594e-05, "loss": 0.0075, "step": 22140 }, { "epoch": 280.37974683544303, "grad_norm": 0.2038736641407013, "learning_rate": 1.758421239192343e-05, "loss": 0.0065, "step": 22150 }, { "epoch": 280.50632911392404, "grad_norm": 0.2194906324148178, "learning_rate": 1.7542268596858814e-05, "loss": 0.009, "step": 22160 }, { "epoch": 280.63291139240505, "grad_norm": 0.2593901753425598, "learning_rate": 1.750036424099841e-05, "loss": 0.0071, "step": 22170 }, { "epoch": 280.75949367088606, "grad_norm": 0.25842028856277466, "learning_rate": 1.7458499375259956e-05, "loss": 0.0069, "step": 22180 }, { "epoch": 280.88607594936707, "grad_norm": 0.20165292918682098, "learning_rate": 1.7416674050513245e-05, "loss": 0.0059, "step": 22190 }, { "epoch": 281.0126582278481, "grad_norm": 0.2721436023712158, "learning_rate": 1.737488831757997e-05, "loss": 0.0075, "step": 22200 }, { "epoch": 281.1392405063291, "grad_norm": 0.24349093437194824, "learning_rate": 1.733314222723373e-05, "loss": 0.0093, "step": 22210 }, { "epoch": 281.26582278481015, "grad_norm": 0.2329416275024414, "learning_rate": 1.7291435830199954e-05, "loss": 0.0064, "step": 22220 }, { "epoch": 281.39240506329116, "grad_norm": 0.17482773959636688, "learning_rate": 1.7249769177155878e-05, "loss": 0.0059, "step": 22230 }, { "epoch": 281.5189873417722, "grad_norm": 0.16695837676525116, "learning_rate": 1.720814231873038e-05, "loss": 0.0062, "step": 22240 }, { "epoch": 281.6455696202532, "grad_norm": 0.19561372697353363, "learning_rate": 1.716655530550405e-05, "loss": 0.006, "step": 22250 }, { "epoch": 281.7721518987342, "grad_norm": 0.2652120292186737, "learning_rate": 1.7125008188009017e-05, "loss": 0.0054, "step": 22260 }, { "epoch": 281.8987341772152, "grad_norm": 0.23556020855903625, "learning_rate": 1.708350101672894e-05, "loss": 0.0071, "step": 22270 }, { "epoch": 282.0253164556962, "grad_norm": 0.21659232676029205, "learning_rate": 1.7042033842098957e-05, "loss": 0.0078, "step": 22280 }, { "epoch": 282.1518987341772, "grad_norm": 0.19533947110176086, "learning_rate": 1.7000606714505567e-05, "loss": 0.0064, "step": 22290 }, { "epoch": 282.27848101265823, "grad_norm": 0.23111538589000702, "learning_rate": 1.695921968428668e-05, "loss": 0.0085, "step": 22300 }, { "epoch": 282.40506329113924, "grad_norm": 0.21231721341609955, "learning_rate": 1.6917872801731416e-05, "loss": 0.0078, "step": 22310 }, { "epoch": 282.53164556962025, "grad_norm": 0.22814083099365234, "learning_rate": 1.687656611708014e-05, "loss": 0.0071, "step": 22320 }, { "epoch": 282.65822784810126, "grad_norm": 0.25095003843307495, "learning_rate": 1.6835299680524368e-05, "loss": 0.0087, "step": 22330 }, { "epoch": 282.7848101265823, "grad_norm": 0.21991275250911713, "learning_rate": 1.679407354220669e-05, "loss": 0.0071, "step": 22340 }, { "epoch": 282.9113924050633, "grad_norm": 0.2681204676628113, "learning_rate": 1.6752887752220793e-05, "loss": 0.0093, "step": 22350 }, { "epoch": 283.0379746835443, "grad_norm": 0.22302553057670593, "learning_rate": 1.6711742360611277e-05, "loss": 0.0051, "step": 22360 }, { "epoch": 283.1645569620253, "grad_norm": 0.23259857296943665, "learning_rate": 1.6670637417373652e-05, "loss": 0.0076, "step": 22370 }, { "epoch": 283.2911392405063, "grad_norm": 0.241189107298851, "learning_rate": 1.6629572972454333e-05, "loss": 0.006, "step": 22380 }, { "epoch": 283.4177215189873, "grad_norm": 0.19611676037311554, "learning_rate": 1.6588549075750464e-05, "loss": 0.0088, "step": 22390 }, { "epoch": 283.54430379746833, "grad_norm": 0.2460181713104248, "learning_rate": 1.6547565777109976e-05, "loss": 0.0078, "step": 22400 }, { "epoch": 283.67088607594934, "grad_norm": 0.2416578233242035, "learning_rate": 1.650662312633143e-05, "loss": 0.0081, "step": 22410 }, { "epoch": 283.7974683544304, "grad_norm": 0.23584645986557007, "learning_rate": 1.6465721173164002e-05, "loss": 0.0081, "step": 22420 }, { "epoch": 283.9240506329114, "grad_norm": 0.2272181361913681, "learning_rate": 1.6424859967307428e-05, "loss": 0.0063, "step": 22430 }, { "epoch": 284.0506329113924, "grad_norm": 0.16933783888816833, "learning_rate": 1.6384039558411903e-05, "loss": 0.0085, "step": 22440 }, { "epoch": 284.17721518987344, "grad_norm": 0.1572342813014984, "learning_rate": 1.634325999607811e-05, "loss": 0.009, "step": 22450 }, { "epoch": 284.30379746835445, "grad_norm": 0.21011550724506378, "learning_rate": 1.6302521329857045e-05, "loss": 0.0089, "step": 22460 }, { "epoch": 284.43037974683546, "grad_norm": 0.24195048213005066, "learning_rate": 1.6261823609250027e-05, "loss": 0.0055, "step": 22470 }, { "epoch": 284.55696202531647, "grad_norm": 0.1969963014125824, "learning_rate": 1.6221166883708627e-05, "loss": 0.006, "step": 22480 }, { "epoch": 284.6835443037975, "grad_norm": 0.19438984990119934, "learning_rate": 1.6180551202634603e-05, "loss": 0.0047, "step": 22490 }, { "epoch": 284.8101265822785, "grad_norm": 0.20241424441337585, "learning_rate": 1.613997661537981e-05, "loss": 0.007, "step": 22500 }, { "epoch": 284.9367088607595, "grad_norm": 0.18976831436157227, "learning_rate": 1.609944317124624e-05, "loss": 0.0066, "step": 22510 }, { "epoch": 285.0632911392405, "grad_norm": 0.1911831498146057, "learning_rate": 1.605895091948582e-05, "loss": 0.0089, "step": 22520 }, { "epoch": 285.1898734177215, "grad_norm": 0.1589391976594925, "learning_rate": 1.6018499909300477e-05, "loss": 0.0052, "step": 22530 }, { "epoch": 285.3164556962025, "grad_norm": 0.1912136822938919, "learning_rate": 1.597809018984199e-05, "loss": 0.0077, "step": 22540 }, { "epoch": 285.44303797468353, "grad_norm": 0.1675831824541092, "learning_rate": 1.593772181021196e-05, "loss": 0.0074, "step": 22550 }, { "epoch": 285.56962025316454, "grad_norm": 0.22070971131324768, "learning_rate": 1.5897394819461813e-05, "loss": 0.0056, "step": 22560 }, { "epoch": 285.69620253164555, "grad_norm": 0.21519511938095093, "learning_rate": 1.5857109266592625e-05, "loss": 0.0088, "step": 22570 }, { "epoch": 285.82278481012656, "grad_norm": 0.1970653235912323, "learning_rate": 1.581686520055514e-05, "loss": 0.0074, "step": 22580 }, { "epoch": 285.9493670886076, "grad_norm": 0.2539706230163574, "learning_rate": 1.5776662670249704e-05, "loss": 0.0065, "step": 22590 }, { "epoch": 286.0759493670886, "grad_norm": 0.20779813826084137, "learning_rate": 1.573650172452615e-05, "loss": 0.0063, "step": 22600 }, { "epoch": 286.2025316455696, "grad_norm": 0.23930953443050385, "learning_rate": 1.5696382412183852e-05, "loss": 0.0099, "step": 22610 }, { "epoch": 286.32911392405066, "grad_norm": 0.21283602714538574, "learning_rate": 1.565630478197155e-05, "loss": 0.0078, "step": 22620 }, { "epoch": 286.45569620253167, "grad_norm": 0.19961011409759521, "learning_rate": 1.5616268882587332e-05, "loss": 0.0077, "step": 22630 }, { "epoch": 286.5822784810127, "grad_norm": 0.20723946392536163, "learning_rate": 1.5576274762678594e-05, "loss": 0.0065, "step": 22640 }, { "epoch": 286.7088607594937, "grad_norm": 0.18451733887195587, "learning_rate": 1.5536322470841952e-05, "loss": 0.0088, "step": 22650 }, { "epoch": 286.8354430379747, "grad_norm": 0.24036270380020142, "learning_rate": 1.549641205562324e-05, "loss": 0.0076, "step": 22660 }, { "epoch": 286.9620253164557, "grad_norm": 0.20089960098266602, "learning_rate": 1.5456543565517334e-05, "loss": 0.0089, "step": 22670 }, { "epoch": 287.0886075949367, "grad_norm": 0.23367980122566223, "learning_rate": 1.541671704896825e-05, "loss": 0.0061, "step": 22680 }, { "epoch": 287.2151898734177, "grad_norm": 0.20014259219169617, "learning_rate": 1.5376932554368938e-05, "loss": 0.0089, "step": 22690 }, { "epoch": 287.34177215189874, "grad_norm": 0.24201059341430664, "learning_rate": 1.5337190130061318e-05, "loss": 0.0079, "step": 22700 }, { "epoch": 287.46835443037975, "grad_norm": 0.23141804337501526, "learning_rate": 1.5297489824336165e-05, "loss": 0.0088, "step": 22710 }, { "epoch": 287.59493670886076, "grad_norm": 0.18728193640708923, "learning_rate": 1.5257831685433121e-05, "loss": 0.0075, "step": 22720 }, { "epoch": 287.72151898734177, "grad_norm": 0.1773281693458557, "learning_rate": 1.521821576154055e-05, "loss": 0.0059, "step": 22730 }, { "epoch": 287.8481012658228, "grad_norm": 0.15196380019187927, "learning_rate": 1.5178642100795542e-05, "loss": 0.008, "step": 22740 }, { "epoch": 287.9746835443038, "grad_norm": 0.15687055885791779, "learning_rate": 1.5139110751283819e-05, "loss": 0.0073, "step": 22750 }, { "epoch": 288.1012658227848, "grad_norm": 0.2123778611421585, "learning_rate": 1.5099621761039683e-05, "loss": 0.0072, "step": 22760 }, { "epoch": 288.2278481012658, "grad_norm": 0.17988114058971405, "learning_rate": 1.5060175178046016e-05, "loss": 0.0082, "step": 22770 }, { "epoch": 288.3544303797468, "grad_norm": 0.2024799883365631, "learning_rate": 1.5020771050234117e-05, "loss": 0.0071, "step": 22780 }, { "epoch": 288.4810126582278, "grad_norm": 0.22111448645591736, "learning_rate": 1.4981409425483716e-05, "loss": 0.0068, "step": 22790 }, { "epoch": 288.60759493670884, "grad_norm": 0.1860225945711136, "learning_rate": 1.4942090351622883e-05, "loss": 0.0058, "step": 22800 }, { "epoch": 288.73417721518985, "grad_norm": 0.21023403108119965, "learning_rate": 1.490281387642804e-05, "loss": 0.0064, "step": 22810 }, { "epoch": 288.8607594936709, "grad_norm": 0.17277805507183075, "learning_rate": 1.4863580047623765e-05, "loss": 0.0053, "step": 22820 }, { "epoch": 288.9873417721519, "grad_norm": 0.21212242543697357, "learning_rate": 1.4824388912882897e-05, "loss": 0.0048, "step": 22830 }, { "epoch": 289.11392405063293, "grad_norm": 0.22061187028884888, "learning_rate": 1.4785240519826344e-05, "loss": 0.0057, "step": 22840 }, { "epoch": 289.24050632911394, "grad_norm": 0.196391761302948, "learning_rate": 1.4746134916023097e-05, "loss": 0.0058, "step": 22850 }, { "epoch": 289.36708860759495, "grad_norm": 0.19789475202560425, "learning_rate": 1.4707072148990141e-05, "loss": 0.0058, "step": 22860 }, { "epoch": 289.49367088607596, "grad_norm": 0.19609320163726807, "learning_rate": 1.4668052266192423e-05, "loss": 0.0066, "step": 22870 }, { "epoch": 289.62025316455697, "grad_norm": 0.1996534913778305, "learning_rate": 1.4629075315042795e-05, "loss": 0.0071, "step": 22880 }, { "epoch": 289.746835443038, "grad_norm": 0.19824273884296417, "learning_rate": 1.4590141342901925e-05, "loss": 0.0065, "step": 22890 }, { "epoch": 289.873417721519, "grad_norm": 0.19837167859077454, "learning_rate": 1.4551250397078253e-05, "loss": 0.0073, "step": 22900 }, { "epoch": 290.0, "grad_norm": 0.20856693387031555, "learning_rate": 1.4512402524827945e-05, "loss": 0.0064, "step": 22910 }, { "epoch": 290.126582278481, "grad_norm": 0.22886618971824646, "learning_rate": 1.447359777335482e-05, "loss": 0.0078, "step": 22920 }, { "epoch": 290.253164556962, "grad_norm": 0.175782710313797, "learning_rate": 1.4434836189810335e-05, "loss": 0.0084, "step": 22930 }, { "epoch": 290.37974683544303, "grad_norm": 0.19943398237228394, "learning_rate": 1.4396117821293453e-05, "loss": 0.0072, "step": 22940 }, { "epoch": 290.50632911392404, "grad_norm": 0.18471243977546692, "learning_rate": 1.4357442714850633e-05, "loss": 0.0053, "step": 22950 }, { "epoch": 290.63291139240505, "grad_norm": 0.17647859454154968, "learning_rate": 1.43188109174758e-05, "loss": 0.0069, "step": 22960 }, { "epoch": 290.75949367088606, "grad_norm": 0.2126224935054779, "learning_rate": 1.4280222476110206e-05, "loss": 0.0069, "step": 22970 }, { "epoch": 290.88607594936707, "grad_norm": 0.195283904671669, "learning_rate": 1.4241677437642469e-05, "loss": 0.0075, "step": 22980 }, { "epoch": 291.0126582278481, "grad_norm": 0.18912570178508759, "learning_rate": 1.420317584890844e-05, "loss": 0.0071, "step": 22990 }, { "epoch": 291.1392405063291, "grad_norm": 0.19660158455371857, "learning_rate": 1.4164717756691176e-05, "loss": 0.0057, "step": 23000 }, { "epoch": 291.26582278481015, "grad_norm": 0.19949910044670105, "learning_rate": 1.4126303207720882e-05, "loss": 0.0065, "step": 23010 }, { "epoch": 291.39240506329116, "grad_norm": 0.198325514793396, "learning_rate": 1.408793224867484e-05, "loss": 0.0065, "step": 23020 }, { "epoch": 291.5189873417722, "grad_norm": 0.21163658797740936, "learning_rate": 1.4049604926177423e-05, "loss": 0.0067, "step": 23030 }, { "epoch": 291.6455696202532, "grad_norm": 0.22117452323436737, "learning_rate": 1.4011321286799916e-05, "loss": 0.0064, "step": 23040 }, { "epoch": 291.7721518987342, "grad_norm": 0.2166047990322113, "learning_rate": 1.3973081377060565e-05, "loss": 0.0071, "step": 23050 }, { "epoch": 291.8987341772152, "grad_norm": 0.2078629732131958, "learning_rate": 1.393488524342445e-05, "loss": 0.0064, "step": 23060 }, { "epoch": 292.0253164556962, "grad_norm": 0.20539872348308563, "learning_rate": 1.3896732932303486e-05, "loss": 0.0063, "step": 23070 }, { "epoch": 292.1518987341772, "grad_norm": 0.20712250471115112, "learning_rate": 1.3858624490056304e-05, "loss": 0.0074, "step": 23080 }, { "epoch": 292.27848101265823, "grad_norm": 0.17619094252586365, "learning_rate": 1.3820559962988299e-05, "loss": 0.007, "step": 23090 }, { "epoch": 292.40506329113924, "grad_norm": 0.2484229952096939, "learning_rate": 1.3782539397351418e-05, "loss": 0.0074, "step": 23100 }, { "epoch": 292.53164556962025, "grad_norm": 0.19956956803798676, "learning_rate": 1.3744562839344266e-05, "loss": 0.0071, "step": 23110 }, { "epoch": 292.65822784810126, "grad_norm": 0.19438035786151886, "learning_rate": 1.370663033511193e-05, "loss": 0.007, "step": 23120 }, { "epoch": 292.7848101265823, "grad_norm": 0.19934719800949097, "learning_rate": 1.3668741930745964e-05, "loss": 0.0066, "step": 23130 }, { "epoch": 292.9113924050633, "grad_norm": 0.1891486644744873, "learning_rate": 1.3630897672284382e-05, "loss": 0.0053, "step": 23140 }, { "epoch": 293.0379746835443, "grad_norm": 0.18786072731018066, "learning_rate": 1.3593097605711508e-05, "loss": 0.0103, "step": 23150 }, { "epoch": 293.1645569620253, "grad_norm": 0.255303293466568, "learning_rate": 1.3555341776957992e-05, "loss": 0.0074, "step": 23160 }, { "epoch": 293.2911392405063, "grad_norm": 0.21245059370994568, "learning_rate": 1.3517630231900724e-05, "loss": 0.0056, "step": 23170 }, { "epoch": 293.4177215189873, "grad_norm": 0.1970612108707428, "learning_rate": 1.3479963016362767e-05, "loss": 0.0063, "step": 23180 }, { "epoch": 293.54430379746833, "grad_norm": 0.21664385497570038, "learning_rate": 1.3442340176113377e-05, "loss": 0.0066, "step": 23190 }, { "epoch": 293.67088607594934, "grad_norm": 0.1987612396478653, "learning_rate": 1.340476175686784e-05, "loss": 0.0057, "step": 23200 }, { "epoch": 293.7974683544304, "grad_norm": 0.15714572370052338, "learning_rate": 1.3367227804287469e-05, "loss": 0.0051, "step": 23210 }, { "epoch": 293.9240506329114, "grad_norm": 0.1825714111328125, "learning_rate": 1.332973836397956e-05, "loss": 0.0069, "step": 23220 }, { "epoch": 294.0506329113924, "grad_norm": 0.18777988851070404, "learning_rate": 1.329229348149731e-05, "loss": 0.0054, "step": 23230 }, { "epoch": 294.17721518987344, "grad_norm": 0.24121315777301788, "learning_rate": 1.3254893202339796e-05, "loss": 0.0067, "step": 23240 }, { "epoch": 294.30379746835445, "grad_norm": 0.1954076588153839, "learning_rate": 1.3217537571951872e-05, "loss": 0.0065, "step": 23250 }, { "epoch": 294.43037974683546, "grad_norm": 0.1539987325668335, "learning_rate": 1.3180226635724169e-05, "loss": 0.0066, "step": 23260 }, { "epoch": 294.55696202531647, "grad_norm": 0.15902067720890045, "learning_rate": 1.314296043899298e-05, "loss": 0.0059, "step": 23270 }, { "epoch": 294.6835443037975, "grad_norm": 0.1913207769393921, "learning_rate": 1.3105739027040248e-05, "loss": 0.0086, "step": 23280 }, { "epoch": 294.8101265822785, "grad_norm": 0.1785464882850647, "learning_rate": 1.3068562445093491e-05, "loss": 0.0056, "step": 23290 }, { "epoch": 294.9367088607595, "grad_norm": 0.1987946629524231, "learning_rate": 1.3031430738325778e-05, "loss": 0.0071, "step": 23300 }, { "epoch": 295.0632911392405, "grad_norm": 0.16482235491275787, "learning_rate": 1.299434395185563e-05, "loss": 0.0073, "step": 23310 }, { "epoch": 295.1898734177215, "grad_norm": 0.16213025152683258, "learning_rate": 1.295730213074699e-05, "loss": 0.0056, "step": 23320 }, { "epoch": 295.3164556962025, "grad_norm": 0.1811683028936386, "learning_rate": 1.2920305320009152e-05, "loss": 0.0055, "step": 23330 }, { "epoch": 295.44303797468353, "grad_norm": 0.25266799330711365, "learning_rate": 1.2883353564596729e-05, "loss": 0.0079, "step": 23340 }, { "epoch": 295.56962025316454, "grad_norm": 0.17287743091583252, "learning_rate": 1.2846446909409604e-05, "loss": 0.0052, "step": 23350 }, { "epoch": 295.69620253164555, "grad_norm": 0.18710219860076904, "learning_rate": 1.280958539929284e-05, "loss": 0.0061, "step": 23360 }, { "epoch": 295.82278481012656, "grad_norm": 0.1910771131515503, "learning_rate": 1.277276907903664e-05, "loss": 0.0075, "step": 23370 }, { "epoch": 295.9493670886076, "grad_norm": 0.1810837686061859, "learning_rate": 1.2735997993376298e-05, "loss": 0.0058, "step": 23380 }, { "epoch": 296.0759493670886, "grad_norm": 0.1854669451713562, "learning_rate": 1.2699272186992166e-05, "loss": 0.0065, "step": 23390 }, { "epoch": 296.2025316455696, "grad_norm": 0.20895583927631378, "learning_rate": 1.2662591704509547e-05, "loss": 0.0056, "step": 23400 }, { "epoch": 296.32911392405066, "grad_norm": 0.1733980029821396, "learning_rate": 1.2625956590498711e-05, "loss": 0.0063, "step": 23410 }, { "epoch": 296.45569620253167, "grad_norm": 0.19691677391529083, "learning_rate": 1.2589366889474757e-05, "loss": 0.0085, "step": 23420 }, { "epoch": 296.5822784810127, "grad_norm": 0.21679015457630157, "learning_rate": 1.2552822645897621e-05, "loss": 0.0086, "step": 23430 }, { "epoch": 296.7088607594937, "grad_norm": 0.19976560771465302, "learning_rate": 1.2516323904172e-05, "loss": 0.0068, "step": 23440 }, { "epoch": 296.8354430379747, "grad_norm": 0.19947050511837006, "learning_rate": 1.2479870708647324e-05, "loss": 0.006, "step": 23450 }, { "epoch": 296.9620253164557, "grad_norm": 0.17064346373081207, "learning_rate": 1.2443463103617658e-05, "loss": 0.008, "step": 23460 }, { "epoch": 297.0886075949367, "grad_norm": 0.18824853003025055, "learning_rate": 1.240710113332167e-05, "loss": 0.0087, "step": 23470 }, { "epoch": 297.2151898734177, "grad_norm": 0.20929770171642303, "learning_rate": 1.2370784841942584e-05, "loss": 0.0065, "step": 23480 }, { "epoch": 297.34177215189874, "grad_norm": 0.1781093031167984, "learning_rate": 1.2334514273608117e-05, "loss": 0.0052, "step": 23490 }, { "epoch": 297.46835443037975, "grad_norm": 0.24165688455104828, "learning_rate": 1.2298289472390417e-05, "loss": 0.0072, "step": 23500 }, { "epoch": 297.59493670886076, "grad_norm": 0.1833139806985855, "learning_rate": 1.226211048230606e-05, "loss": 0.0071, "step": 23510 }, { "epoch": 297.72151898734177, "grad_norm": 0.25638410449028015, "learning_rate": 1.2225977347315921e-05, "loss": 0.0071, "step": 23520 }, { "epoch": 297.8481012658228, "grad_norm": 0.19157761335372925, "learning_rate": 1.2189890111325148e-05, "loss": 0.0063, "step": 23530 }, { "epoch": 297.9746835443038, "grad_norm": 0.20428746938705444, "learning_rate": 1.2153848818183161e-05, "loss": 0.0086, "step": 23540 }, { "epoch": 298.1012658227848, "grad_norm": 0.1802850216627121, "learning_rate": 1.2117853511683507e-05, "loss": 0.0066, "step": 23550 }, { "epoch": 298.2278481012658, "grad_norm": 0.17895947396755219, "learning_rate": 1.2081904235563906e-05, "loss": 0.007, "step": 23560 }, { "epoch": 298.3544303797468, "grad_norm": 0.20746862888336182, "learning_rate": 1.20460010335061e-05, "loss": 0.0089, "step": 23570 }, { "epoch": 298.4810126582278, "grad_norm": 0.20489901304244995, "learning_rate": 1.2010143949135865e-05, "loss": 0.0062, "step": 23580 }, { "epoch": 298.60759493670884, "grad_norm": 0.20002096891403198, "learning_rate": 1.1974333026022938e-05, "loss": 0.0059, "step": 23590 }, { "epoch": 298.73417721518985, "grad_norm": 0.16513076424598694, "learning_rate": 1.1938568307680964e-05, "loss": 0.0072, "step": 23600 }, { "epoch": 298.8607594936709, "grad_norm": 0.1615877002477646, "learning_rate": 1.1902849837567464e-05, "loss": 0.0065, "step": 23610 }, { "epoch": 298.9873417721519, "grad_norm": 0.18647892773151398, "learning_rate": 1.1867177659083739e-05, "loss": 0.0063, "step": 23620 }, { "epoch": 299.11392405063293, "grad_norm": 0.20896078646183014, "learning_rate": 1.1831551815574848e-05, "loss": 0.0061, "step": 23630 }, { "epoch": 299.24050632911394, "grad_norm": 0.14665962755680084, "learning_rate": 1.1795972350329554e-05, "loss": 0.0051, "step": 23640 }, { "epoch": 299.36708860759495, "grad_norm": 0.20831771194934845, "learning_rate": 1.176043930658024e-05, "loss": 0.0064, "step": 23650 }, { "epoch": 299.49367088607596, "grad_norm": 0.18268683552742004, "learning_rate": 1.1724952727502942e-05, "loss": 0.0092, "step": 23660 }, { "epoch": 299.62025316455697, "grad_norm": 0.19152849912643433, "learning_rate": 1.1689512656217178e-05, "loss": 0.0068, "step": 23670 }, { "epoch": 299.746835443038, "grad_norm": 0.15460996329784393, "learning_rate": 1.1654119135785962e-05, "loss": 0.0048, "step": 23680 }, { "epoch": 299.873417721519, "grad_norm": 0.21528024971485138, "learning_rate": 1.1618772209215794e-05, "loss": 0.0051, "step": 23690 }, { "epoch": 300.0, "grad_norm": 0.1804426610469818, "learning_rate": 1.1583471919456506e-05, "loss": 0.006, "step": 23700 }, { "epoch": 300.126582278481, "grad_norm": 0.15815366804599762, "learning_rate": 1.1548218309401266e-05, "loss": 0.0058, "step": 23710 }, { "epoch": 300.253164556962, "grad_norm": 0.1656663715839386, "learning_rate": 1.1513011421886555e-05, "loss": 0.0069, "step": 23720 }, { "epoch": 300.37974683544303, "grad_norm": 0.1637793332338333, "learning_rate": 1.1477851299692056e-05, "loss": 0.0072, "step": 23730 }, { "epoch": 300.50632911392404, "grad_norm": 0.17065970599651337, "learning_rate": 1.144273798554063e-05, "loss": 0.0062, "step": 23740 }, { "epoch": 300.63291139240505, "grad_norm": 0.18521709740161896, "learning_rate": 1.1407671522098262e-05, "loss": 0.0077, "step": 23750 }, { "epoch": 300.75949367088606, "grad_norm": 0.19371533393859863, "learning_rate": 1.1372651951974e-05, "loss": 0.008, "step": 23760 }, { "epoch": 300.88607594936707, "grad_norm": 0.17368175089359283, "learning_rate": 1.1337679317719952e-05, "loss": 0.0057, "step": 23770 }, { "epoch": 301.0126582278481, "grad_norm": 0.15246717631816864, "learning_rate": 1.1302753661831151e-05, "loss": 0.0064, "step": 23780 }, { "epoch": 301.1392405063291, "grad_norm": 0.2121478170156479, "learning_rate": 1.1267875026745561e-05, "loss": 0.005, "step": 23790 }, { "epoch": 301.26582278481015, "grad_norm": 0.20121456682682037, "learning_rate": 1.1233043454844016e-05, "loss": 0.006, "step": 23800 }, { "epoch": 301.39240506329116, "grad_norm": 0.22363021969795227, "learning_rate": 1.1198258988450144e-05, "loss": 0.0069, "step": 23810 }, { "epoch": 301.5189873417722, "grad_norm": 0.176567941904068, "learning_rate": 1.1163521669830368e-05, "loss": 0.0057, "step": 23820 }, { "epoch": 301.6455696202532, "grad_norm": 0.2271893471479416, "learning_rate": 1.1128831541193818e-05, "loss": 0.0062, "step": 23830 }, { "epoch": 301.7721518987342, "grad_norm": 0.1941230595111847, "learning_rate": 1.1094188644692255e-05, "loss": 0.0057, "step": 23840 }, { "epoch": 301.8987341772152, "grad_norm": 0.21078383922576904, "learning_rate": 1.1059593022420067e-05, "loss": 0.0066, "step": 23850 }, { "epoch": 302.0253164556962, "grad_norm": 0.19791778922080994, "learning_rate": 1.1025044716414184e-05, "loss": 0.0071, "step": 23860 }, { "epoch": 302.1518987341772, "grad_norm": 0.16863968968391418, "learning_rate": 1.0990543768654082e-05, "loss": 0.0058, "step": 23870 }, { "epoch": 302.27848101265823, "grad_norm": 0.17981772124767303, "learning_rate": 1.0956090221061655e-05, "loss": 0.0052, "step": 23880 }, { "epoch": 302.40506329113924, "grad_norm": 0.16035982966423035, "learning_rate": 1.0921684115501208e-05, "loss": 0.0053, "step": 23890 }, { "epoch": 302.53164556962025, "grad_norm": 0.15605801343917847, "learning_rate": 1.0887325493779405e-05, "loss": 0.0067, "step": 23900 }, { "epoch": 302.65822784810126, "grad_norm": 0.18390601873397827, "learning_rate": 1.0853014397645207e-05, "loss": 0.0054, "step": 23910 }, { "epoch": 302.7848101265823, "grad_norm": 0.1687825322151184, "learning_rate": 1.0818750868789829e-05, "loss": 0.0045, "step": 23920 }, { "epoch": 302.9113924050633, "grad_norm": 0.1831512749195099, "learning_rate": 1.0784534948846703e-05, "loss": 0.0062, "step": 23930 }, { "epoch": 303.0379746835443, "grad_norm": 0.2181122601032257, "learning_rate": 1.0750366679391393e-05, "loss": 0.0053, "step": 23940 }, { "epoch": 303.1645569620253, "grad_norm": 0.19644838571548462, "learning_rate": 1.0716246101941557e-05, "loss": 0.0069, "step": 23950 }, { "epoch": 303.2911392405063, "grad_norm": 0.1690131574869156, "learning_rate": 1.0682173257956934e-05, "loss": 0.0075, "step": 23960 }, { "epoch": 303.4177215189873, "grad_norm": 0.1685759723186493, "learning_rate": 1.064814818883922e-05, "loss": 0.0079, "step": 23970 }, { "epoch": 303.54430379746833, "grad_norm": 0.18856801092624664, "learning_rate": 1.06141709359321e-05, "loss": 0.006, "step": 23980 }, { "epoch": 303.67088607594934, "grad_norm": 0.221697136759758, "learning_rate": 1.058024154052114e-05, "loss": 0.0066, "step": 23990 }, { "epoch": 303.7974683544304, "grad_norm": 0.1934077888727188, "learning_rate": 1.0546360043833747e-05, "loss": 0.0058, "step": 24000 }, { "epoch": 303.9240506329114, "grad_norm": 0.19841203093528748, "learning_rate": 1.0512526487039138e-05, "loss": 0.0076, "step": 24010 }, { "epoch": 304.0506329113924, "grad_norm": 0.13954444229602814, "learning_rate": 1.0478740911248258e-05, "loss": 0.0054, "step": 24020 }, { "epoch": 304.17721518987344, "grad_norm": 0.15320032835006714, "learning_rate": 1.0445003357513793e-05, "loss": 0.0049, "step": 24030 }, { "epoch": 304.30379746835445, "grad_norm": 0.1589239239692688, "learning_rate": 1.0411313866830042e-05, "loss": 0.0047, "step": 24040 }, { "epoch": 304.43037974683546, "grad_norm": 0.17635506391525269, "learning_rate": 1.0377672480132915e-05, "loss": 0.006, "step": 24050 }, { "epoch": 304.55696202531647, "grad_norm": 0.19369831681251526, "learning_rate": 1.0344079238299864e-05, "loss": 0.0062, "step": 24060 }, { "epoch": 304.6835443037975, "grad_norm": 0.2033233791589737, "learning_rate": 1.0310534182149833e-05, "loss": 0.0061, "step": 24070 }, { "epoch": 304.8101265822785, "grad_norm": 0.1914629489183426, "learning_rate": 1.0277037352443258e-05, "loss": 0.0059, "step": 24080 }, { "epoch": 304.9367088607595, "grad_norm": 0.185069739818573, "learning_rate": 1.024358878988193e-05, "loss": 0.0059, "step": 24090 }, { "epoch": 305.0632911392405, "grad_norm": 0.19158966839313507, "learning_rate": 1.0210188535108995e-05, "loss": 0.0066, "step": 24100 }, { "epoch": 305.1898734177215, "grad_norm": 0.15044574439525604, "learning_rate": 1.0176836628708935e-05, "loss": 0.007, "step": 24110 }, { "epoch": 305.3164556962025, "grad_norm": 0.1742686629295349, "learning_rate": 1.0143533111207454e-05, "loss": 0.0049, "step": 24120 }, { "epoch": 305.44303797468353, "grad_norm": 0.16363699734210968, "learning_rate": 1.0110278023071446e-05, "loss": 0.0058, "step": 24130 }, { "epoch": 305.56962025316454, "grad_norm": 0.16186077892780304, "learning_rate": 1.007707140470901e-05, "loss": 0.0046, "step": 24140 }, { "epoch": 305.69620253164555, "grad_norm": 0.16183257102966309, "learning_rate": 1.0043913296469298e-05, "loss": 0.0046, "step": 24150 }, { "epoch": 305.82278481012656, "grad_norm": 0.15363585948944092, "learning_rate": 1.001080373864255e-05, "loss": 0.0092, "step": 24160 }, { "epoch": 305.9493670886076, "grad_norm": 0.22991040349006653, "learning_rate": 9.977742771459991e-06, "loss": 0.0064, "step": 24170 }, { "epoch": 306.0759493670886, "grad_norm": 0.17479632794857025, "learning_rate": 9.944730435093802e-06, "loss": 0.0061, "step": 24180 }, { "epoch": 306.2025316455696, "grad_norm": 0.21022330224514008, "learning_rate": 9.911766769657115e-06, "loss": 0.0048, "step": 24190 }, { "epoch": 306.32911392405066, "grad_norm": 0.15886478126049042, "learning_rate": 9.878851815203882e-06, "loss": 0.0085, "step": 24200 }, { "epoch": 306.45569620253167, "grad_norm": 0.1984962522983551, "learning_rate": 9.845985611728864e-06, "loss": 0.0071, "step": 24210 }, { "epoch": 306.5822784810127, "grad_norm": 0.1758348047733307, "learning_rate": 9.813168199167606e-06, "loss": 0.0068, "step": 24220 }, { "epoch": 306.7088607594937, "grad_norm": 0.18059398233890533, "learning_rate": 9.780399617396341e-06, "loss": 0.0056, "step": 24230 }, { "epoch": 306.8354430379747, "grad_norm": 0.15764451026916504, "learning_rate": 9.747679906232016e-06, "loss": 0.005, "step": 24240 }, { "epoch": 306.9620253164557, "grad_norm": 0.13394784927368164, "learning_rate": 9.715009105432138e-06, "loss": 0.0057, "step": 24250 }, { "epoch": 307.0886075949367, "grad_norm": 0.16080094873905182, "learning_rate": 9.682387254694835e-06, "loss": 0.0059, "step": 24260 }, { "epoch": 307.2151898734177, "grad_norm": 0.146935373544693, "learning_rate": 9.649814393658723e-06, "loss": 0.0065, "step": 24270 }, { "epoch": 307.34177215189874, "grad_norm": 0.14994172751903534, "learning_rate": 9.617290561902881e-06, "loss": 0.0057, "step": 24280 }, { "epoch": 307.46835443037975, "grad_norm": 0.13072243332862854, "learning_rate": 9.584815798946861e-06, "loss": 0.0064, "step": 24290 }, { "epoch": 307.59493670886076, "grad_norm": 0.17808814346790314, "learning_rate": 9.55239014425055e-06, "loss": 0.008, "step": 24300 }, { "epoch": 307.72151898734177, "grad_norm": 0.16428473591804504, "learning_rate": 9.520013637214176e-06, "loss": 0.0087, "step": 24310 }, { "epoch": 307.8481012658228, "grad_norm": 0.16332542896270752, "learning_rate": 9.48768631717824e-06, "loss": 0.0061, "step": 24320 }, { "epoch": 307.9746835443038, "grad_norm": 0.15124735236167908, "learning_rate": 9.455408223423496e-06, "loss": 0.0055, "step": 24330 }, { "epoch": 308.1012658227848, "grad_norm": 0.21117328107357025, "learning_rate": 9.423179395170845e-06, "loss": 0.0069, "step": 24340 }, { "epoch": 308.2278481012658, "grad_norm": 0.22945259511470795, "learning_rate": 9.390999871581391e-06, "loss": 0.0081, "step": 24350 }, { "epoch": 308.3544303797468, "grad_norm": 0.18880808353424072, "learning_rate": 9.358869691756273e-06, "loss": 0.0067, "step": 24360 }, { "epoch": 308.4810126582278, "grad_norm": 0.18102999031543732, "learning_rate": 9.326788894736688e-06, "loss": 0.0055, "step": 24370 }, { "epoch": 308.60759493670884, "grad_norm": 0.15931569039821625, "learning_rate": 9.294757519503811e-06, "loss": 0.005, "step": 24380 }, { "epoch": 308.73417721518985, "grad_norm": 0.20412302017211914, "learning_rate": 9.262775604978819e-06, "loss": 0.0062, "step": 24390 }, { "epoch": 308.8607594936709, "grad_norm": 0.1883237361907959, "learning_rate": 9.230843190022725e-06, "loss": 0.0071, "step": 24400 }, { "epoch": 308.9873417721519, "grad_norm": 0.18643110990524292, "learning_rate": 9.198960313436445e-06, "loss": 0.0077, "step": 24410 }, { "epoch": 309.11392405063293, "grad_norm": 0.16426318883895874, "learning_rate": 9.16712701396067e-06, "loss": 0.0052, "step": 24420 }, { "epoch": 309.24050632911394, "grad_norm": 0.1996685415506363, "learning_rate": 9.135343330275864e-06, "loss": 0.0065, "step": 24430 }, { "epoch": 309.36708860759495, "grad_norm": 0.20471204817295074, "learning_rate": 9.10360930100218e-06, "loss": 0.0076, "step": 24440 }, { "epoch": 309.49367088607596, "grad_norm": 0.19078756868839264, "learning_rate": 9.07192496469949e-06, "loss": 0.0094, "step": 24450 }, { "epoch": 309.62025316455697, "grad_norm": 0.18103377521038055, "learning_rate": 9.040290359867231e-06, "loss": 0.0042, "step": 24460 }, { "epoch": 309.746835443038, "grad_norm": 0.15824833512306213, "learning_rate": 9.008705524944438e-06, "loss": 0.0047, "step": 24470 }, { "epoch": 309.873417721519, "grad_norm": 0.1838928610086441, "learning_rate": 8.977170498309651e-06, "loss": 0.0066, "step": 24480 }, { "epoch": 310.0, "grad_norm": 0.18635115027427673, "learning_rate": 8.945685318280916e-06, "loss": 0.0045, "step": 24490 }, { "epoch": 310.126582278481, "grad_norm": 0.16623032093048096, "learning_rate": 8.914250023115672e-06, "loss": 0.0075, "step": 24500 }, { "epoch": 310.253164556962, "grad_norm": 0.1783362179994583, "learning_rate": 8.882864651010798e-06, "loss": 0.0073, "step": 24510 }, { "epoch": 310.37974683544303, "grad_norm": 0.19594724476337433, "learning_rate": 8.851529240102462e-06, "loss": 0.0058, "step": 24520 }, { "epoch": 310.50632911392404, "grad_norm": 0.17581480741500854, "learning_rate": 8.820243828466134e-06, "loss": 0.0054, "step": 24530 }, { "epoch": 310.63291139240505, "grad_norm": 0.17508432269096375, "learning_rate": 8.789008454116566e-06, "loss": 0.0048, "step": 24540 }, { "epoch": 310.75949367088606, "grad_norm": 0.12793472409248352, "learning_rate": 8.757823155007655e-06, "loss": 0.0066, "step": 24550 }, { "epoch": 310.88607594936707, "grad_norm": 0.16644707322120667, "learning_rate": 8.72668796903251e-06, "loss": 0.0063, "step": 24560 }, { "epoch": 311.0126582278481, "grad_norm": 0.14356161653995514, "learning_rate": 8.695602934023301e-06, "loss": 0.0051, "step": 24570 }, { "epoch": 311.1392405063291, "grad_norm": 0.1716286540031433, "learning_rate": 8.664568087751274e-06, "loss": 0.0076, "step": 24580 }, { "epoch": 311.26582278481015, "grad_norm": 0.1261943131685257, "learning_rate": 8.633583467926697e-06, "loss": 0.0046, "step": 24590 }, { "epoch": 311.39240506329116, "grad_norm": 0.1564132124185562, "learning_rate": 8.602649112198796e-06, "loss": 0.0055, "step": 24600 }, { "epoch": 311.5189873417722, "grad_norm": 0.13232894241809845, "learning_rate": 8.571765058155745e-06, "loss": 0.0052, "step": 24610 }, { "epoch": 311.6455696202532, "grad_norm": 0.16039985418319702, "learning_rate": 8.540931343324583e-06, "loss": 0.0064, "step": 24620 }, { "epoch": 311.7721518987342, "grad_norm": 0.15512284636497498, "learning_rate": 8.51014800517117e-06, "loss": 0.0054, "step": 24630 }, { "epoch": 311.8987341772152, "grad_norm": 0.20477168262004852, "learning_rate": 8.479415081100167e-06, "loss": 0.0058, "step": 24640 }, { "epoch": 312.0253164556962, "grad_norm": 0.17963004112243652, "learning_rate": 8.448732608454967e-06, "loss": 0.005, "step": 24650 }, { "epoch": 312.1518987341772, "grad_norm": 0.16528168320655823, "learning_rate": 8.418100624517688e-06, "loss": 0.0069, "step": 24660 }, { "epoch": 312.27848101265823, "grad_norm": 0.1637168675661087, "learning_rate": 8.387519166509062e-06, "loss": 0.0046, "step": 24670 }, { "epoch": 312.40506329113924, "grad_norm": 0.169535294175148, "learning_rate": 8.356988271588445e-06, "loss": 0.0045, "step": 24680 }, { "epoch": 312.53164556962025, "grad_norm": 0.17010201513767242, "learning_rate": 8.326507976853765e-06, "loss": 0.0054, "step": 24690 }, { "epoch": 312.65822784810126, "grad_norm": 0.18647658824920654, "learning_rate": 8.296078319341443e-06, "loss": 0.0052, "step": 24700 }, { "epoch": 312.7848101265823, "grad_norm": 0.1897420883178711, "learning_rate": 8.265699336026383e-06, "loss": 0.0085, "step": 24710 }, { "epoch": 312.9113924050633, "grad_norm": 0.1713140457868576, "learning_rate": 8.235371063821923e-06, "loss": 0.0057, "step": 24720 }, { "epoch": 313.0379746835443, "grad_norm": 0.18239274621009827, "learning_rate": 8.205093539579767e-06, "loss": 0.005, "step": 24730 }, { "epoch": 313.1645569620253, "grad_norm": 0.13971613347530365, "learning_rate": 8.174866800089963e-06, "loss": 0.0065, "step": 24740 }, { "epoch": 313.2911392405063, "grad_norm": 0.12720075249671936, "learning_rate": 8.144690882080853e-06, "loss": 0.0061, "step": 24750 }, { "epoch": 313.4177215189873, "grad_norm": 0.1430191695690155, "learning_rate": 8.114565822219006e-06, "loss": 0.0058, "step": 24760 }, { "epoch": 313.54430379746833, "grad_norm": 0.152175635099411, "learning_rate": 8.084491657109233e-06, "loss": 0.0076, "step": 24770 }, { "epoch": 313.67088607594934, "grad_norm": 0.1255338490009308, "learning_rate": 8.054468423294476e-06, "loss": 0.0061, "step": 24780 }, { "epoch": 313.7974683544304, "grad_norm": 0.1447785198688507, "learning_rate": 8.024496157255785e-06, "loss": 0.0066, "step": 24790 }, { "epoch": 313.9240506329114, "grad_norm": 0.15082353353500366, "learning_rate": 7.994574895412293e-06, "loss": 0.0055, "step": 24800 }, { "epoch": 314.0506329113924, "grad_norm": 0.15675869584083557, "learning_rate": 7.964704674121149e-06, "loss": 0.0063, "step": 24810 }, { "epoch": 314.17721518987344, "grad_norm": 0.1711164265871048, "learning_rate": 7.934885529677505e-06, "loss": 0.0076, "step": 24820 }, { "epoch": 314.30379746835445, "grad_norm": 0.15740670263767242, "learning_rate": 7.905117498314412e-06, "loss": 0.0066, "step": 24830 }, { "epoch": 314.43037974683546, "grad_norm": 0.15348297357559204, "learning_rate": 7.87540061620286e-06, "loss": 0.0051, "step": 24840 }, { "epoch": 314.55696202531647, "grad_norm": 0.17579086124897003, "learning_rate": 7.845734919451647e-06, "loss": 0.0053, "step": 24850 }, { "epoch": 314.6835443037975, "grad_norm": 0.20664216578006744, "learning_rate": 7.816120444107383e-06, "loss": 0.0093, "step": 24860 }, { "epoch": 314.8101265822785, "grad_norm": 0.1746351569890976, "learning_rate": 7.786557226154473e-06, "loss": 0.0061, "step": 24870 }, { "epoch": 314.9367088607595, "grad_norm": 0.19606824219226837, "learning_rate": 7.757045301514998e-06, "loss": 0.0072, "step": 24880 }, { "epoch": 315.0632911392405, "grad_norm": 0.16338488459587097, "learning_rate": 7.727584706048735e-06, "loss": 0.0057, "step": 24890 }, { "epoch": 315.1898734177215, "grad_norm": 0.1839679628610611, "learning_rate": 7.698175475553077e-06, "loss": 0.0051, "step": 24900 }, { "epoch": 315.3164556962025, "grad_norm": 0.16717249155044556, "learning_rate": 7.668817645763021e-06, "loss": 0.0081, "step": 24910 }, { "epoch": 315.44303797468353, "grad_norm": 0.14049240946769714, "learning_rate": 7.639511252351088e-06, "loss": 0.0064, "step": 24920 }, { "epoch": 315.56962025316454, "grad_norm": 0.16205894947052002, "learning_rate": 7.610256330927324e-06, "loss": 0.0058, "step": 24930 }, { "epoch": 315.69620253164555, "grad_norm": 0.2037530243396759, "learning_rate": 7.581052917039211e-06, "loss": 0.0079, "step": 24940 }, { "epoch": 315.82278481012656, "grad_norm": 0.14469651877880096, "learning_rate": 7.551901046171645e-06, "loss": 0.0045, "step": 24950 }, { "epoch": 315.9493670886076, "grad_norm": 0.16672328114509583, "learning_rate": 7.522800753746895e-06, "loss": 0.0081, "step": 24960 }, { "epoch": 316.0759493670886, "grad_norm": 0.15830017626285553, "learning_rate": 7.493752075124577e-06, "loss": 0.0058, "step": 24970 }, { "epoch": 316.2025316455696, "grad_norm": 0.14422041177749634, "learning_rate": 7.464755045601557e-06, "loss": 0.0056, "step": 24980 }, { "epoch": 316.32911392405066, "grad_norm": 0.13967949151992798, "learning_rate": 7.435809700411972e-06, "loss": 0.0048, "step": 24990 }, { "epoch": 316.45569620253167, "grad_norm": 0.1552359014749527, "learning_rate": 7.40691607472715e-06, "loss": 0.0084, "step": 25000 }, { "epoch": 316.5822784810127, "grad_norm": 0.17162996530532837, "learning_rate": 7.3780742036555606e-06, "loss": 0.0049, "step": 25010 }, { "epoch": 316.7088607594937, "grad_norm": 0.14805786311626434, "learning_rate": 7.349284122242783e-06, "loss": 0.0057, "step": 25020 }, { "epoch": 316.8354430379747, "grad_norm": 0.1102227121591568, "learning_rate": 7.320545865471512e-06, "loss": 0.006, "step": 25030 }, { "epoch": 316.9620253164557, "grad_norm": 0.16040530800819397, "learning_rate": 7.291859468261425e-06, "loss": 0.0073, "step": 25040 }, { "epoch": 317.0886075949367, "grad_norm": 0.1306684911251068, "learning_rate": 7.263224965469195e-06, "loss": 0.0054, "step": 25050 }, { "epoch": 317.2151898734177, "grad_norm": 0.12080492079257965, "learning_rate": 7.234642391888446e-06, "loss": 0.0063, "step": 25060 }, { "epoch": 317.34177215189874, "grad_norm": 0.18461105227470398, "learning_rate": 7.206111782249697e-06, "loss": 0.005, "step": 25070 }, { "epoch": 317.46835443037975, "grad_norm": 0.171108216047287, "learning_rate": 7.1776331712203385e-06, "loss": 0.0086, "step": 25080 }, { "epoch": 317.59493670886076, "grad_norm": 0.16349108517169952, "learning_rate": 7.149206593404561e-06, "loss": 0.0051, "step": 25090 }, { "epoch": 317.72151898734177, "grad_norm": 0.14132222533226013, "learning_rate": 7.120832083343337e-06, "loss": 0.006, "step": 25100 }, { "epoch": 317.8481012658228, "grad_norm": 0.14237745106220245, "learning_rate": 7.092509675514369e-06, "loss": 0.0073, "step": 25110 }, { "epoch": 317.9746835443038, "grad_norm": 0.14221704006195068, "learning_rate": 7.064239404332063e-06, "loss": 0.0046, "step": 25120 }, { "epoch": 318.1012658227848, "grad_norm": 0.13774845004081726, "learning_rate": 7.03602130414745e-06, "loss": 0.0061, "step": 25130 }, { "epoch": 318.2278481012658, "grad_norm": 0.16512569785118103, "learning_rate": 7.007855409248198e-06, "loss": 0.0062, "step": 25140 }, { "epoch": 318.3544303797468, "grad_norm": 0.10946137458086014, "learning_rate": 6.979741753858521e-06, "loss": 0.0071, "step": 25150 }, { "epoch": 318.4810126582278, "grad_norm": 0.1591956466436386, "learning_rate": 6.951680372139158e-06, "loss": 0.0063, "step": 25160 }, { "epoch": 318.60759493670884, "grad_norm": 0.13568179309368134, "learning_rate": 6.923671298187334e-06, "loss": 0.0057, "step": 25170 }, { "epoch": 318.73417721518985, "grad_norm": 0.15971961617469788, "learning_rate": 6.895714566036704e-06, "loss": 0.0057, "step": 25180 }, { "epoch": 318.8607594936709, "grad_norm": 0.16577520966529846, "learning_rate": 6.86781020965736e-06, "loss": 0.0057, "step": 25190 }, { "epoch": 318.9873417721519, "grad_norm": 0.130833700299263, "learning_rate": 6.839958262955709e-06, "loss": 0.0051, "step": 25200 }, { "epoch": 319.11392405063293, "grad_norm": 0.16621924936771393, "learning_rate": 6.8121587597744894e-06, "loss": 0.0062, "step": 25210 }, { "epoch": 319.24050632911394, "grad_norm": 0.17161312699317932, "learning_rate": 6.784411733892732e-06, "loss": 0.0069, "step": 25220 }, { "epoch": 319.36708860759495, "grad_norm": 0.17032331228256226, "learning_rate": 6.756717219025666e-06, "loss": 0.0049, "step": 25230 }, { "epoch": 319.49367088607596, "grad_norm": 0.14213185012340546, "learning_rate": 6.729075248824762e-06, "loss": 0.006, "step": 25240 }, { "epoch": 319.62025316455697, "grad_norm": 0.15185445547103882, "learning_rate": 6.701485856877615e-06, "loss": 0.0062, "step": 25250 }, { "epoch": 319.746835443038, "grad_norm": 0.15977594256401062, "learning_rate": 6.673949076707925e-06, "loss": 0.005, "step": 25260 }, { "epoch": 319.873417721519, "grad_norm": 0.14466224610805511, "learning_rate": 6.646464941775499e-06, "loss": 0.0069, "step": 25270 }, { "epoch": 320.0, "grad_norm": 0.1444135308265686, "learning_rate": 6.619033485476128e-06, "loss": 0.0048, "step": 25280 }, { "epoch": 320.126582278481, "grad_norm": 0.16629183292388916, "learning_rate": 6.59165474114164e-06, "loss": 0.006, "step": 25290 }, { "epoch": 320.253164556962, "grad_norm": 0.16540107131004333, "learning_rate": 6.564328742039783e-06, "loss": 0.0047, "step": 25300 }, { "epoch": 320.37974683544303, "grad_norm": 0.1754949986934662, "learning_rate": 6.537055521374219e-06, "loss": 0.006, "step": 25310 }, { "epoch": 320.50632911392404, "grad_norm": 0.15357054769992828, "learning_rate": 6.509835112284485e-06, "loss": 0.0049, "step": 25320 }, { "epoch": 320.63291139240505, "grad_norm": 0.18300503492355347, "learning_rate": 6.482667547845944e-06, "loss": 0.0056, "step": 25330 }, { "epoch": 320.75949367088606, "grad_norm": 0.15236763656139374, "learning_rate": 6.455552861069736e-06, "loss": 0.0049, "step": 25340 }, { "epoch": 320.88607594936707, "grad_norm": 0.129541277885437, "learning_rate": 6.428491084902788e-06, "loss": 0.0068, "step": 25350 }, { "epoch": 321.0126582278481, "grad_norm": 0.14029806852340698, "learning_rate": 6.401482252227698e-06, "loss": 0.0058, "step": 25360 }, { "epoch": 321.1392405063291, "grad_norm": 0.1662486493587494, "learning_rate": 6.37452639586274e-06, "loss": 0.0056, "step": 25370 }, { "epoch": 321.26582278481015, "grad_norm": 0.1574888676404953, "learning_rate": 6.347623548561826e-06, "loss": 0.0051, "step": 25380 }, { "epoch": 321.39240506329116, "grad_norm": 0.1293354481458664, "learning_rate": 6.320773743014441e-06, "loss": 0.0059, "step": 25390 }, { "epoch": 321.5189873417722, "grad_norm": 0.12155621498823166, "learning_rate": 6.293977011845648e-06, "loss": 0.0049, "step": 25400 }, { "epoch": 321.6455696202532, "grad_norm": 0.1537492573261261, "learning_rate": 6.2672333876159835e-06, "loss": 0.0055, "step": 25410 }, { "epoch": 321.7721518987342, "grad_norm": 0.12867368757724762, "learning_rate": 6.2405429028215e-06, "loss": 0.0064, "step": 25420 }, { "epoch": 321.8987341772152, "grad_norm": 0.13767535984516144, "learning_rate": 6.213905589893631e-06, "loss": 0.0059, "step": 25430 }, { "epoch": 322.0253164556962, "grad_norm": 0.12977494299411774, "learning_rate": 6.187321481199221e-06, "loss": 0.0047, "step": 25440 }, { "epoch": 322.1518987341772, "grad_norm": 0.13743150234222412, "learning_rate": 6.1607906090404895e-06, "loss": 0.0076, "step": 25450 }, { "epoch": 322.27848101265823, "grad_norm": 0.12735743820667267, "learning_rate": 6.134313005654929e-06, "loss": 0.0047, "step": 25460 }, { "epoch": 322.40506329113924, "grad_norm": 0.131357803940773, "learning_rate": 6.107888703215337e-06, "loss": 0.0065, "step": 25470 }, { "epoch": 322.53164556962025, "grad_norm": 0.16577552258968353, "learning_rate": 6.081517733829722e-06, "loss": 0.007, "step": 25480 }, { "epoch": 322.65822784810126, "grad_norm": 0.16870345175266266, "learning_rate": 6.055200129541294e-06, "loss": 0.0057, "step": 25490 }, { "epoch": 322.7848101265823, "grad_norm": 0.1601848304271698, "learning_rate": 6.0289359223284434e-06, "loss": 0.0075, "step": 25500 }, { "epoch": 322.9113924050633, "grad_norm": 0.17676764726638794, "learning_rate": 6.002725144104648e-06, "loss": 0.0043, "step": 25510 }, { "epoch": 323.0379746835443, "grad_norm": 0.18675799667835236, "learning_rate": 5.976567826718477e-06, "loss": 0.0057, "step": 25520 }, { "epoch": 323.1645569620253, "grad_norm": 0.12040503323078156, "learning_rate": 5.9504640019535315e-06, "loss": 0.0069, "step": 25530 }, { "epoch": 323.2911392405063, "grad_norm": 0.1441033035516739, "learning_rate": 5.924413701528414e-06, "loss": 0.0052, "step": 25540 }, { "epoch": 323.4177215189873, "grad_norm": 0.1397581845521927, "learning_rate": 5.898416957096703e-06, "loss": 0.0064, "step": 25550 }, { "epoch": 323.54430379746833, "grad_norm": 0.13791142404079437, "learning_rate": 5.872473800246913e-06, "loss": 0.0073, "step": 25560 }, { "epoch": 323.67088607594934, "grad_norm": 0.1450793296098709, "learning_rate": 5.846584262502402e-06, "loss": 0.0062, "step": 25570 }, { "epoch": 323.7974683544304, "grad_norm": 0.16298699378967285, "learning_rate": 5.820748375321411e-06, "loss": 0.005, "step": 25580 }, { "epoch": 323.9240506329114, "grad_norm": 0.1659737378358841, "learning_rate": 5.794966170096977e-06, "loss": 0.0084, "step": 25590 }, { "epoch": 324.0506329113924, "grad_norm": 0.14205290377140045, "learning_rate": 5.769237678156897e-06, "loss": 0.0064, "step": 25600 }, { "epoch": 324.17721518987344, "grad_norm": 0.17372924089431763, "learning_rate": 5.743562930763735e-06, "loss": 0.0062, "step": 25610 }, { "epoch": 324.30379746835445, "grad_norm": 0.1342843621969223, "learning_rate": 5.717941959114726e-06, "loss": 0.0047, "step": 25620 }, { "epoch": 324.43037974683546, "grad_norm": 0.1578548401594162, "learning_rate": 5.69237479434176e-06, "loss": 0.0062, "step": 25630 }, { "epoch": 324.55696202531647, "grad_norm": 0.12007880955934525, "learning_rate": 5.666861467511353e-06, "loss": 0.0048, "step": 25640 }, { "epoch": 324.6835443037975, "grad_norm": 0.1730392873287201, "learning_rate": 5.641402009624591e-06, "loss": 0.0045, "step": 25650 }, { "epoch": 324.8101265822785, "grad_norm": 0.1398620903491974, "learning_rate": 5.615996451617145e-06, "loss": 0.0048, "step": 25660 }, { "epoch": 324.9367088607595, "grad_norm": 0.131455659866333, "learning_rate": 5.590644824359148e-06, "loss": 0.0051, "step": 25670 }, { "epoch": 325.0632911392405, "grad_norm": 0.1569221317768097, "learning_rate": 5.56534715865521e-06, "loss": 0.0069, "step": 25680 }, { "epoch": 325.1898734177215, "grad_norm": 0.1423531472682953, "learning_rate": 5.5401034852443965e-06, "loss": 0.006, "step": 25690 }, { "epoch": 325.3164556962025, "grad_norm": 0.1217358261346817, "learning_rate": 5.514913834800134e-06, "loss": 0.0064, "step": 25700 }, { "epoch": 325.44303797468353, "grad_norm": 0.1392674595117569, "learning_rate": 5.489778237930238e-06, "loss": 0.0066, "step": 25710 }, { "epoch": 325.56962025316454, "grad_norm": 0.1286228448152542, "learning_rate": 5.464696725176827e-06, "loss": 0.0056, "step": 25720 }, { "epoch": 325.69620253164555, "grad_norm": 0.1480528563261032, "learning_rate": 5.439669327016294e-06, "loss": 0.0046, "step": 25730 }, { "epoch": 325.82278481012656, "grad_norm": 0.15678919851779938, "learning_rate": 5.4146960738592985e-06, "loss": 0.0067, "step": 25740 }, { "epoch": 325.9493670886076, "grad_norm": 0.1237073466181755, "learning_rate": 5.389776996050694e-06, "loss": 0.0052, "step": 25750 }, { "epoch": 326.0759493670886, "grad_norm": 0.15074756741523743, "learning_rate": 5.364912123869492e-06, "loss": 0.0055, "step": 25760 }, { "epoch": 326.2025316455696, "grad_norm": 0.12137246131896973, "learning_rate": 5.3401014875288864e-06, "loss": 0.0056, "step": 25770 }, { "epoch": 326.32911392405066, "grad_norm": 0.14563792943954468, "learning_rate": 5.315345117176129e-06, "loss": 0.0046, "step": 25780 }, { "epoch": 326.45569620253167, "grad_norm": 0.14906388521194458, "learning_rate": 5.2906430428925405e-06, "loss": 0.0083, "step": 25790 }, { "epoch": 326.5822784810127, "grad_norm": 0.1562327742576599, "learning_rate": 5.265995294693471e-06, "loss": 0.0052, "step": 25800 }, { "epoch": 326.7088607594937, "grad_norm": 0.14243538677692413, "learning_rate": 5.241401902528254e-06, "loss": 0.0045, "step": 25810 }, { "epoch": 326.8354430379747, "grad_norm": 0.12704849243164062, "learning_rate": 5.216862896280189e-06, "loss": 0.0054, "step": 25820 }, { "epoch": 326.9620253164557, "grad_norm": 0.12937863171100616, "learning_rate": 5.192378305766471e-06, "loss": 0.0045, "step": 25830 }, { "epoch": 327.0886075949367, "grad_norm": 0.1437894105911255, "learning_rate": 5.1679481607382065e-06, "loss": 0.0059, "step": 25840 }, { "epoch": 327.2151898734177, "grad_norm": 0.1342945694923401, "learning_rate": 5.1435724908803105e-06, "loss": 0.005, "step": 25850 }, { "epoch": 327.34177215189874, "grad_norm": 0.11765308678150177, "learning_rate": 5.119251325811514e-06, "loss": 0.0053, "step": 25860 }, { "epoch": 327.46835443037975, "grad_norm": 0.13937050104141235, "learning_rate": 5.094984695084348e-06, "loss": 0.0057, "step": 25870 }, { "epoch": 327.59493670886076, "grad_norm": 0.13857072591781616, "learning_rate": 5.070772628185039e-06, "loss": 0.0056, "step": 25880 }, { "epoch": 327.72151898734177, "grad_norm": 0.1303812563419342, "learning_rate": 5.046615154533535e-06, "loss": 0.0066, "step": 25890 }, { "epoch": 327.8481012658228, "grad_norm": 0.12289411574602127, "learning_rate": 5.022512303483451e-06, "loss": 0.0062, "step": 25900 }, { "epoch": 327.9746835443038, "grad_norm": 0.11964453756809235, "learning_rate": 4.998464104322015e-06, "loss": 0.0056, "step": 25910 }, { "epoch": 328.1012658227848, "grad_norm": 0.13190989196300507, "learning_rate": 4.974470586270047e-06, "loss": 0.0052, "step": 25920 }, { "epoch": 328.2278481012658, "grad_norm": 0.11752226948738098, "learning_rate": 4.950531778481965e-06, "loss": 0.0055, "step": 25930 }, { "epoch": 328.3544303797468, "grad_norm": 0.09534035623073578, "learning_rate": 4.926647710045651e-06, "loss": 0.0063, "step": 25940 }, { "epoch": 328.4810126582278, "grad_norm": 0.1278078854084015, "learning_rate": 4.9028184099825125e-06, "loss": 0.0046, "step": 25950 }, { "epoch": 328.60759493670884, "grad_norm": 0.1544378101825714, "learning_rate": 4.879043907247382e-06, "loss": 0.0059, "step": 25960 }, { "epoch": 328.73417721518985, "grad_norm": 0.15645229816436768, "learning_rate": 4.8553242307285414e-06, "loss": 0.0055, "step": 25970 }, { "epoch": 328.8607594936709, "grad_norm": 0.12890850007534027, "learning_rate": 4.8316594092476185e-06, "loss": 0.0081, "step": 25980 }, { "epoch": 328.9873417721519, "grad_norm": 0.11274803429841995, "learning_rate": 4.808049471559617e-06, "loss": 0.0082, "step": 25990 }, { "epoch": 329.11392405063293, "grad_norm": 0.17659111320972443, "learning_rate": 4.784494446352833e-06, "loss": 0.0049, "step": 26000 }, { "epoch": 329.24050632911394, "grad_norm": 0.16231757402420044, "learning_rate": 4.760994362248833e-06, "loss": 0.0057, "step": 26010 }, { "epoch": 329.36708860759495, "grad_norm": 0.13718581199645996, "learning_rate": 4.737549247802442e-06, "loss": 0.0053, "step": 26020 }, { "epoch": 329.49367088607596, "grad_norm": 0.11570290476083755, "learning_rate": 4.714159131501689e-06, "loss": 0.0056, "step": 26030 }, { "epoch": 329.62025316455697, "grad_norm": 0.13954885303974152, "learning_rate": 4.690824041767766e-06, "loss": 0.0062, "step": 26040 }, { "epoch": 329.746835443038, "grad_norm": 0.1176576167345047, "learning_rate": 4.6675440069550105e-06, "loss": 0.0048, "step": 26050 }, { "epoch": 329.873417721519, "grad_norm": 0.1385589987039566, "learning_rate": 4.64431905535086e-06, "loss": 0.0076, "step": 26060 }, { "epoch": 330.0, "grad_norm": 0.12526443600654602, "learning_rate": 4.621149215175808e-06, "loss": 0.0048, "step": 26070 }, { "epoch": 330.126582278481, "grad_norm": 0.15482352674007416, "learning_rate": 4.5980345145834154e-06, "loss": 0.0051, "step": 26080 }, { "epoch": 330.253164556962, "grad_norm": 0.14690259099006653, "learning_rate": 4.574974981660213e-06, "loss": 0.0044, "step": 26090 }, { "epoch": 330.37974683544303, "grad_norm": 0.11624438315629959, "learning_rate": 4.551970644425707e-06, "loss": 0.0047, "step": 26100 }, { "epoch": 330.50632911392404, "grad_norm": 0.1149243414402008, "learning_rate": 4.529021530832328e-06, "loss": 0.0055, "step": 26110 }, { "epoch": 330.63291139240505, "grad_norm": 0.1666371077299118, "learning_rate": 4.5061276687654306e-06, "loss": 0.0065, "step": 26120 }, { "epoch": 330.75949367088606, "grad_norm": 0.1636708378791809, "learning_rate": 4.4832890860431965e-06, "loss": 0.0046, "step": 26130 }, { "epoch": 330.88607594936707, "grad_norm": 0.13017812371253967, "learning_rate": 4.460505810416682e-06, "loss": 0.0048, "step": 26140 }, { "epoch": 331.0126582278481, "grad_norm": 0.1327931135892868, "learning_rate": 4.4377778695696984e-06, "loss": 0.0058, "step": 26150 }, { "epoch": 331.1392405063291, "grad_norm": 0.12333763390779495, "learning_rate": 4.415105291118843e-06, "loss": 0.0057, "step": 26160 }, { "epoch": 331.26582278481015, "grad_norm": 0.15918922424316406, "learning_rate": 4.392488102613435e-06, "loss": 0.0062, "step": 26170 }, { "epoch": 331.39240506329116, "grad_norm": 0.13602215051651, "learning_rate": 4.369926331535473e-06, "loss": 0.0045, "step": 26180 }, { "epoch": 331.5189873417722, "grad_norm": 0.15893089771270752, "learning_rate": 4.347420005299668e-06, "loss": 0.0049, "step": 26190 }, { "epoch": 331.6455696202532, "grad_norm": 0.13572467863559723, "learning_rate": 4.324969151253317e-06, "loss": 0.0064, "step": 26200 }, { "epoch": 331.7721518987342, "grad_norm": 0.1212175115942955, "learning_rate": 4.302573796676312e-06, "loss": 0.0048, "step": 26210 }, { "epoch": 331.8987341772152, "grad_norm": 0.13853046298027039, "learning_rate": 4.2802339687811385e-06, "loss": 0.0051, "step": 26220 }, { "epoch": 332.0253164556962, "grad_norm": 0.1305677592754364, "learning_rate": 4.257949694712771e-06, "loss": 0.0065, "step": 26230 }, { "epoch": 332.1518987341772, "grad_norm": 0.13974645733833313, "learning_rate": 4.235721001548726e-06, "loss": 0.0059, "step": 26240 }, { "epoch": 332.27848101265823, "grad_norm": 0.14298513531684875, "learning_rate": 4.21354791629896e-06, "loss": 0.005, "step": 26250 }, { "epoch": 332.40506329113924, "grad_norm": 0.13546736538410187, "learning_rate": 4.191430465905843e-06, "loss": 0.01, "step": 26260 }, { "epoch": 332.53164556962025, "grad_norm": 0.16632920503616333, "learning_rate": 4.169368677244184e-06, "loss": 0.0049, "step": 26270 }, { "epoch": 332.65822784810126, "grad_norm": 0.1434546560049057, "learning_rate": 4.147362577121122e-06, "loss": 0.0065, "step": 26280 }, { "epoch": 332.7848101265823, "grad_norm": 0.127992182970047, "learning_rate": 4.125412192276157e-06, "loss": 0.0047, "step": 26290 }, { "epoch": 332.9113924050633, "grad_norm": 0.09794028103351593, "learning_rate": 4.10351754938107e-06, "loss": 0.004, "step": 26300 }, { "epoch": 333.0379746835443, "grad_norm": 0.13576537370681763, "learning_rate": 4.081678675039913e-06, "loss": 0.01, "step": 26310 }, { "epoch": 333.1645569620253, "grad_norm": 0.11497242003679276, "learning_rate": 4.0598955957889876e-06, "loss": 0.0061, "step": 26320 }, { "epoch": 333.2911392405063, "grad_norm": 0.1308230608701706, "learning_rate": 4.038168338096776e-06, "loss": 0.0064, "step": 26330 }, { "epoch": 333.4177215189873, "grad_norm": 0.1566561609506607, "learning_rate": 4.016496928363944e-06, "loss": 0.008, "step": 26340 }, { "epoch": 333.54430379746833, "grad_norm": 0.11250434070825577, "learning_rate": 3.994881392923317e-06, "loss": 0.0055, "step": 26350 }, { "epoch": 333.67088607594934, "grad_norm": 0.16246090829372406, "learning_rate": 3.9733217580397945e-06, "loss": 0.0037, "step": 26360 }, { "epoch": 333.7974683544304, "grad_norm": 0.13760800659656525, "learning_rate": 3.951818049910366e-06, "loss": 0.0053, "step": 26370 }, { "epoch": 333.9240506329114, "grad_norm": 0.18463292717933655, "learning_rate": 3.930370294664071e-06, "loss": 0.0055, "step": 26380 }, { "epoch": 334.0506329113924, "grad_norm": 0.15437737107276917, "learning_rate": 3.908978518361938e-06, "loss": 0.0079, "step": 26390 }, { "epoch": 334.17721518987344, "grad_norm": 0.15155324339866638, "learning_rate": 3.887642746997017e-06, "loss": 0.008, "step": 26400 }, { "epoch": 334.30379746835445, "grad_norm": 0.1335071474313736, "learning_rate": 3.866363006494255e-06, "loss": 0.0054, "step": 26410 }, { "epoch": 334.43037974683546, "grad_norm": 0.13373427093029022, "learning_rate": 3.8451393227105725e-06, "loss": 0.0053, "step": 26420 }, { "epoch": 334.55696202531647, "grad_norm": 0.13028831779956818, "learning_rate": 3.823971721434727e-06, "loss": 0.0058, "step": 26430 }, { "epoch": 334.6835443037975, "grad_norm": 0.11144465208053589, "learning_rate": 3.8028602283873505e-06, "loss": 0.0063, "step": 26440 }, { "epoch": 334.8101265822785, "grad_norm": 0.10342356562614441, "learning_rate": 3.781804869220912e-06, "loss": 0.004, "step": 26450 }, { "epoch": 334.9367088607595, "grad_norm": 0.10112123191356659, "learning_rate": 3.760805669519646e-06, "loss": 0.0044, "step": 26460 }, { "epoch": 335.0632911392405, "grad_norm": 0.15911580622196198, "learning_rate": 3.7398626547995587e-06, "loss": 0.0061, "step": 26470 }, { "epoch": 335.1898734177215, "grad_norm": 0.11624888330698013, "learning_rate": 3.7189758505083995e-06, "loss": 0.0059, "step": 26480 }, { "epoch": 335.3164556962025, "grad_norm": 0.10547151416540146, "learning_rate": 3.6981452820255837e-06, "loss": 0.0063, "step": 26490 }, { "epoch": 335.44303797468353, "grad_norm": 0.10251038521528244, "learning_rate": 3.6773709746622363e-06, "loss": 0.0062, "step": 26500 }, { "epoch": 335.56962025316454, "grad_norm": 0.12080532312393188, "learning_rate": 3.6566529536610995e-06, "loss": 0.0075, "step": 26510 }, { "epoch": 335.69620253164555, "grad_norm": 0.1239936351776123, "learning_rate": 3.635991244196513e-06, "loss": 0.0048, "step": 26520 }, { "epoch": 335.82278481012656, "grad_norm": 0.12213125824928284, "learning_rate": 3.615385871374405e-06, "loss": 0.0082, "step": 26530 }, { "epoch": 335.9493670886076, "grad_norm": 0.11397616565227509, "learning_rate": 3.5948368602322367e-06, "loss": 0.0057, "step": 26540 }, { "epoch": 336.0759493670886, "grad_norm": 0.15911127626895905, "learning_rate": 3.5743442357390187e-06, "loss": 0.0056, "step": 26550 }, { "epoch": 336.2025316455696, "grad_norm": 0.15449991822242737, "learning_rate": 3.5539080227951937e-06, "loss": 0.0054, "step": 26560 }, { "epoch": 336.32911392405066, "grad_norm": 0.14498533308506012, "learning_rate": 3.533528246232709e-06, "loss": 0.0074, "step": 26570 }, { "epoch": 336.45569620253167, "grad_norm": 0.1599002331495285, "learning_rate": 3.5132049308149117e-06, "loss": 0.0075, "step": 26580 }, { "epoch": 336.5822784810127, "grad_norm": 0.12083196640014648, "learning_rate": 3.4929381012365425e-06, "loss": 0.0053, "step": 26590 }, { "epoch": 336.7088607594937, "grad_norm": 0.13910824060440063, "learning_rate": 3.472727782123697e-06, "loss": 0.0079, "step": 26600 }, { "epoch": 336.8354430379747, "grad_norm": 0.10904836654663086, "learning_rate": 3.4525739980338425e-06, "loss": 0.0063, "step": 26610 }, { "epoch": 336.9620253164557, "grad_norm": 0.12511228024959564, "learning_rate": 3.432476773455712e-06, "loss": 0.0057, "step": 26620 }, { "epoch": 337.0886075949367, "grad_norm": 0.13217656314373016, "learning_rate": 3.4124361328093378e-06, "loss": 0.0072, "step": 26630 }, { "epoch": 337.2151898734177, "grad_norm": 0.12094397842884064, "learning_rate": 3.392452100445975e-06, "loss": 0.0064, "step": 26640 }, { "epoch": 337.34177215189874, "grad_norm": 0.1573208123445511, "learning_rate": 3.3725247006481163e-06, "loss": 0.0061, "step": 26650 }, { "epoch": 337.46835443037975, "grad_norm": 0.12186241149902344, "learning_rate": 3.3526539576294313e-06, "loss": 0.0055, "step": 26660 }, { "epoch": 337.59493670886076, "grad_norm": 0.11359728127717972, "learning_rate": 3.332839895534745e-06, "loss": 0.0053, "step": 26670 }, { "epoch": 337.72151898734177, "grad_norm": 0.12208397686481476, "learning_rate": 3.3130825384400155e-06, "loss": 0.0053, "step": 26680 }, { "epoch": 337.8481012658228, "grad_norm": 0.12791426479816437, "learning_rate": 3.293381910352278e-06, "loss": 0.0055, "step": 26690 }, { "epoch": 337.9746835443038, "grad_norm": 0.12154512852430344, "learning_rate": 3.273738035209678e-06, "loss": 0.0055, "step": 26700 }, { "epoch": 338.1012658227848, "grad_norm": 0.1012389063835144, "learning_rate": 3.254150936881356e-06, "loss": 0.0067, "step": 26710 }, { "epoch": 338.2278481012658, "grad_norm": 0.12633997201919556, "learning_rate": 3.234620639167496e-06, "loss": 0.0059, "step": 26720 }, { "epoch": 338.3544303797468, "grad_norm": 0.17683708667755127, "learning_rate": 3.2151471657992482e-06, "loss": 0.0091, "step": 26730 }, { "epoch": 338.4810126582278, "grad_norm": 0.1588449627161026, "learning_rate": 3.195730540438718e-06, "loss": 0.0054, "step": 26740 }, { "epoch": 338.60759493670884, "grad_norm": 0.12716524302959442, "learning_rate": 3.1763707866789336e-06, "loss": 0.0049, "step": 26750 }, { "epoch": 338.73417721518985, "grad_norm": 0.12007599323987961, "learning_rate": 3.1570679280438165e-06, "loss": 0.0064, "step": 26760 }, { "epoch": 338.8607594936709, "grad_norm": 0.11041884869337082, "learning_rate": 3.1378219879881767e-06, "loss": 0.0066, "step": 26770 }, { "epoch": 338.9873417721519, "grad_norm": 0.11209219694137573, "learning_rate": 3.11863298989763e-06, "loss": 0.0052, "step": 26780 }, { "epoch": 339.11392405063293, "grad_norm": 0.13167117536067963, "learning_rate": 3.0995009570886303e-06, "loss": 0.0085, "step": 26790 }, { "epoch": 339.24050632911394, "grad_norm": 0.14794428646564484, "learning_rate": 3.0804259128083936e-06, "loss": 0.0066, "step": 26800 }, { "epoch": 339.36708860759495, "grad_norm": 0.11766718327999115, "learning_rate": 3.06140788023489e-06, "loss": 0.0047, "step": 26810 }, { "epoch": 339.49367088607596, "grad_norm": 0.10565811395645142, "learning_rate": 3.042446882476846e-06, "loss": 0.0051, "step": 26820 }, { "epoch": 339.62025316455697, "grad_norm": 0.11109215766191483, "learning_rate": 3.023542942573643e-06, "loss": 0.0061, "step": 26830 }, { "epoch": 339.746835443038, "grad_norm": 0.13532327115535736, "learning_rate": 3.004696083495351e-06, "loss": 0.0045, "step": 26840 }, { "epoch": 339.873417721519, "grad_norm": 0.16146673262119293, "learning_rate": 2.98590632814269e-06, "loss": 0.0062, "step": 26850 }, { "epoch": 340.0, "grad_norm": 0.093791663646698, "learning_rate": 2.9671736993469702e-06, "loss": 0.0067, "step": 26860 }, { "epoch": 340.126582278481, "grad_norm": 0.11665425449609756, "learning_rate": 2.9484982198701216e-06, "loss": 0.0048, "step": 26870 }, { "epoch": 340.253164556962, "grad_norm": 0.1436721682548523, "learning_rate": 2.929879912404604e-06, "loss": 0.0052, "step": 26880 }, { "epoch": 340.37974683544303, "grad_norm": 0.10135103762149811, "learning_rate": 2.9113187995734147e-06, "loss": 0.0048, "step": 26890 }, { "epoch": 340.50632911392404, "grad_norm": 0.09800107777118683, "learning_rate": 2.8928149039300524e-06, "loss": 0.0047, "step": 26900 }, { "epoch": 340.63291139240505, "grad_norm": 0.11774521321058273, "learning_rate": 2.8743682479584975e-06, "loss": 0.0054, "step": 26910 }, { "epoch": 340.75949367088606, "grad_norm": 0.10767583549022675, "learning_rate": 2.8559788540731824e-06, "loss": 0.0046, "step": 26920 }, { "epoch": 340.88607594936707, "grad_norm": 0.12925006449222565, "learning_rate": 2.8376467446189493e-06, "loss": 0.0062, "step": 26930 }, { "epoch": 341.0126582278481, "grad_norm": 0.09761989861726761, "learning_rate": 2.8193719418710406e-06, "loss": 0.0046, "step": 26940 }, { "epoch": 341.1392405063291, "grad_norm": 0.18212735652923584, "learning_rate": 2.8011544680350667e-06, "loss": 0.0047, "step": 26950 }, { "epoch": 341.26582278481015, "grad_norm": 0.10940950363874435, "learning_rate": 2.7829943452469753e-06, "loss": 0.0053, "step": 26960 }, { "epoch": 341.39240506329116, "grad_norm": 0.10064692795276642, "learning_rate": 2.7648915955730213e-06, "loss": 0.0041, "step": 26970 }, { "epoch": 341.5189873417722, "grad_norm": 0.11150985956192017, "learning_rate": 2.746846241009765e-06, "loss": 0.0069, "step": 26980 }, { "epoch": 341.6455696202532, "grad_norm": 0.1451924592256546, "learning_rate": 2.7288583034839945e-06, "loss": 0.0061, "step": 26990 }, { "epoch": 341.7721518987342, "grad_norm": 0.10254324227571487, "learning_rate": 2.7109278048527752e-06, "loss": 0.0059, "step": 27000 }, { "epoch": 341.8987341772152, "grad_norm": 0.10242682695388794, "learning_rate": 2.6930547669033413e-06, "loss": 0.0057, "step": 27010 }, { "epoch": 342.0253164556962, "grad_norm": 0.1194840595126152, "learning_rate": 2.675239211353109e-06, "loss": 0.0059, "step": 27020 }, { "epoch": 342.1518987341772, "grad_norm": 0.09401929378509521, "learning_rate": 2.6574811598496786e-06, "loss": 0.0072, "step": 27030 }, { "epoch": 342.27848101265823, "grad_norm": 0.11162663996219635, "learning_rate": 2.6397806339707456e-06, "loss": 0.0063, "step": 27040 }, { "epoch": 342.40506329113924, "grad_norm": 0.12268642336130142, "learning_rate": 2.6221376552241217e-06, "loss": 0.0058, "step": 27050 }, { "epoch": 342.53164556962025, "grad_norm": 0.10449032485485077, "learning_rate": 2.604552245047681e-06, "loss": 0.0057, "step": 27060 }, { "epoch": 342.65822784810126, "grad_norm": 0.10243873298168182, "learning_rate": 2.587024424809359e-06, "loss": 0.0071, "step": 27070 }, { "epoch": 342.7848101265823, "grad_norm": 0.11634361743927002, "learning_rate": 2.5695542158071186e-06, "loss": 0.0065, "step": 27080 }, { "epoch": 342.9113924050633, "grad_norm": 0.1032482162117958, "learning_rate": 2.5521416392689068e-06, "loss": 0.0052, "step": 27090 }, { "epoch": 343.0379746835443, "grad_norm": 0.10490545630455017, "learning_rate": 2.5347867163526384e-06, "loss": 0.0048, "step": 27100 }, { "epoch": 343.1645569620253, "grad_norm": 0.12026505172252655, "learning_rate": 2.517489468146189e-06, "loss": 0.0044, "step": 27110 }, { "epoch": 343.2911392405063, "grad_norm": 0.15310980379581451, "learning_rate": 2.500249915667341e-06, "loss": 0.005, "step": 27120 }, { "epoch": 343.4177215189873, "grad_norm": 0.11894989013671875, "learning_rate": 2.4830680798637817e-06, "loss": 0.0058, "step": 27130 }, { "epoch": 343.54430379746833, "grad_norm": 0.10515084117650986, "learning_rate": 2.465943981613056e-06, "loss": 0.0046, "step": 27140 }, { "epoch": 343.67088607594934, "grad_norm": 0.10613185167312622, "learning_rate": 2.448877641722569e-06, "loss": 0.0043, "step": 27150 }, { "epoch": 343.7974683544304, "grad_norm": 0.10683920234441757, "learning_rate": 2.431869080929522e-06, "loss": 0.0041, "step": 27160 }, { "epoch": 343.9240506329114, "grad_norm": 0.1204826757311821, "learning_rate": 2.4149183199009216e-06, "loss": 0.005, "step": 27170 }, { "epoch": 344.0506329113924, "grad_norm": 0.15476836264133453, "learning_rate": 2.3980253792335427e-06, "loss": 0.0074, "step": 27180 }, { "epoch": 344.17721518987344, "grad_norm": 0.10768409818410873, "learning_rate": 2.381190279453899e-06, "loss": 0.0047, "step": 27190 }, { "epoch": 344.30379746835445, "grad_norm": 0.10733425617218018, "learning_rate": 2.364413041018232e-06, "loss": 0.0058, "step": 27200 }, { "epoch": 344.43037974683546, "grad_norm": 0.12236811965703964, "learning_rate": 2.347693684312463e-06, "loss": 0.0062, "step": 27210 }, { "epoch": 344.55696202531647, "grad_norm": 0.10609418898820877, "learning_rate": 2.331032229652186e-06, "loss": 0.0052, "step": 27220 }, { "epoch": 344.6835443037975, "grad_norm": 0.1226658895611763, "learning_rate": 2.314428697282628e-06, "loss": 0.0048, "step": 27230 }, { "epoch": 344.8101265822785, "grad_norm": 0.09507697820663452, "learning_rate": 2.297883107378673e-06, "loss": 0.0048, "step": 27240 }, { "epoch": 344.9367088607595, "grad_norm": 0.10677468776702881, "learning_rate": 2.2813954800447513e-06, "loss": 0.0069, "step": 27250 }, { "epoch": 345.0632911392405, "grad_norm": 0.10341735184192657, "learning_rate": 2.2649658353148974e-06, "loss": 0.005, "step": 27260 }, { "epoch": 345.1898734177215, "grad_norm": 0.08672117441892624, "learning_rate": 2.2485941931526645e-06, "loss": 0.0048, "step": 27270 }, { "epoch": 345.3164556962025, "grad_norm": 0.0807858482003212, "learning_rate": 2.232280573451151e-06, "loss": 0.0063, "step": 27280 }, { "epoch": 345.44303797468353, "grad_norm": 0.10844763368368149, "learning_rate": 2.216024996032945e-06, "loss": 0.0052, "step": 27290 }, { "epoch": 345.56962025316454, "grad_norm": 0.0974959284067154, "learning_rate": 2.1998274806501074e-06, "loss": 0.007, "step": 27300 }, { "epoch": 345.69620253164555, "grad_norm": 0.12163127958774567, "learning_rate": 2.183688046984139e-06, "loss": 0.0045, "step": 27310 }, { "epoch": 345.82278481012656, "grad_norm": 0.128469318151474, "learning_rate": 2.167606714645981e-06, "loss": 0.0051, "step": 27320 }, { "epoch": 345.9493670886076, "grad_norm": 0.11873513460159302, "learning_rate": 2.151583503175958e-06, "loss": 0.0062, "step": 27330 }, { "epoch": 346.0759493670886, "grad_norm": 0.09558770060539246, "learning_rate": 2.1356184320437957e-06, "loss": 0.0057, "step": 27340 }, { "epoch": 346.2025316455696, "grad_norm": 0.09383939206600189, "learning_rate": 2.119711520648554e-06, "loss": 0.0045, "step": 27350 }, { "epoch": 346.32911392405066, "grad_norm": 0.10277149081230164, "learning_rate": 2.103862788318628e-06, "loss": 0.005, "step": 27360 }, { "epoch": 346.45569620253167, "grad_norm": 0.09752579033374786, "learning_rate": 2.088072254311729e-06, "loss": 0.0044, "step": 27370 }, { "epoch": 346.5822784810127, "grad_norm": 0.08444683998823166, "learning_rate": 2.0723399378148435e-06, "loss": 0.0045, "step": 27380 }, { "epoch": 346.7088607594937, "grad_norm": 0.07233065366744995, "learning_rate": 2.0566658579442065e-06, "loss": 0.0033, "step": 27390 }, { "epoch": 346.8354430379747, "grad_norm": 0.1142338365316391, "learning_rate": 2.0410500337453176e-06, "loss": 0.0053, "step": 27400 }, { "epoch": 346.9620253164557, "grad_norm": 0.11244018375873566, "learning_rate": 2.0254924841928645e-06, "loss": 0.0045, "step": 27410 }, { "epoch": 347.0886075949367, "grad_norm": 0.08523350208997726, "learning_rate": 2.009993228190754e-06, "loss": 0.0041, "step": 27420 }, { "epoch": 347.2151898734177, "grad_norm": 0.10464650392532349, "learning_rate": 1.9945522845720323e-06, "loss": 0.0059, "step": 27430 }, { "epoch": 347.34177215189874, "grad_norm": 0.10719295591115952, "learning_rate": 1.9791696720988963e-06, "loss": 0.0066, "step": 27440 }, { "epoch": 347.46835443037975, "grad_norm": 0.11305137723684311, "learning_rate": 1.9638454094626833e-06, "loss": 0.0071, "step": 27450 }, { "epoch": 347.59493670886076, "grad_norm": 0.12985502183437347, "learning_rate": 1.9485795152838104e-06, "loss": 0.0068, "step": 27460 }, { "epoch": 347.72151898734177, "grad_norm": 0.11050952225923538, "learning_rate": 1.933372008111778e-06, "loss": 0.0054, "step": 27470 }, { "epoch": 347.8481012658228, "grad_norm": 0.12003618478775024, "learning_rate": 1.918222906425143e-06, "loss": 0.0047, "step": 27480 }, { "epoch": 347.9746835443038, "grad_norm": 0.1027991995215416, "learning_rate": 1.903132228631488e-06, "loss": 0.0056, "step": 27490 }, { "epoch": 348.1012658227848, "grad_norm": 0.11893372982740402, "learning_rate": 1.8880999930674215e-06, "loss": 0.0042, "step": 27500 }, { "epoch": 348.2278481012658, "grad_norm": 0.07880426943302155, "learning_rate": 1.8731262179985166e-06, "loss": 0.0053, "step": 27510 }, { "epoch": 348.3544303797468, "grad_norm": 0.111236572265625, "learning_rate": 1.8582109216193244e-06, "loss": 0.0066, "step": 27520 }, { "epoch": 348.4810126582278, "grad_norm": 0.12316001951694489, "learning_rate": 1.8433541220533368e-06, "loss": 0.0045, "step": 27530 }, { "epoch": 348.60759493670884, "grad_norm": 0.1179247722029686, "learning_rate": 1.8285558373529577e-06, "loss": 0.0052, "step": 27540 }, { "epoch": 348.73417721518985, "grad_norm": 0.0857991948723793, "learning_rate": 1.8138160854995145e-06, "loss": 0.0042, "step": 27550 }, { "epoch": 348.8607594936709, "grad_norm": 0.14676876366138458, "learning_rate": 1.7991348844031864e-06, "loss": 0.0069, "step": 27560 }, { "epoch": 348.9873417721519, "grad_norm": 0.10391141474246979, "learning_rate": 1.78451225190302e-06, "loss": 0.0063, "step": 27570 }, { "epoch": 349.11392405063293, "grad_norm": 0.10452620685100555, "learning_rate": 1.7699482057668914e-06, "loss": 0.0038, "step": 27580 }, { "epoch": 349.24050632911394, "grad_norm": 0.07991216331720352, "learning_rate": 1.7554427636914895e-06, "loss": 0.0052, "step": 27590 }, { "epoch": 349.36708860759495, "grad_norm": 0.12368924915790558, "learning_rate": 1.7409959433022871e-06, "loss": 0.0061, "step": 27600 }, { "epoch": 349.49367088607596, "grad_norm": 0.10936478525400162, "learning_rate": 1.7266077621535481e-06, "loss": 0.0051, "step": 27610 }, { "epoch": 349.62025316455697, "grad_norm": 0.10885940492153168, "learning_rate": 1.7122782377282597e-06, "loss": 0.007, "step": 27620 }, { "epoch": 349.746835443038, "grad_norm": 0.12424590438604355, "learning_rate": 1.6980073874381497e-06, "loss": 0.0052, "step": 27630 }, { "epoch": 349.873417721519, "grad_norm": 0.1313592940568924, "learning_rate": 1.6837952286236413e-06, "loss": 0.0066, "step": 27640 }, { "epoch": 350.0, "grad_norm": 0.10458175837993622, "learning_rate": 1.6696417785538488e-06, "loss": 0.0058, "step": 27650 }, { "epoch": 350.126582278481, "grad_norm": 0.1011880487203598, "learning_rate": 1.6555470544265538e-06, "loss": 0.0055, "step": 27660 }, { "epoch": 350.253164556962, "grad_norm": 0.12016929686069489, "learning_rate": 1.6415110733681738e-06, "loss": 0.0044, "step": 27670 }, { "epoch": 350.37974683544303, "grad_norm": 0.0909065380692482, "learning_rate": 1.6275338524337436e-06, "loss": 0.0057, "step": 27680 }, { "epoch": 350.50632911392404, "grad_norm": 0.1050405353307724, "learning_rate": 1.6136154086069056e-06, "loss": 0.0069, "step": 27690 }, { "epoch": 350.63291139240505, "grad_norm": 0.0774710550904274, "learning_rate": 1.5997557587998868e-06, "loss": 0.0045, "step": 27700 }, { "epoch": 350.75949367088606, "grad_norm": 0.10304044187068939, "learning_rate": 1.5859549198534606e-06, "loss": 0.0078, "step": 27710 }, { "epoch": 350.88607594936707, "grad_norm": 0.12453177571296692, "learning_rate": 1.572212908536963e-06, "loss": 0.0097, "step": 27720 }, { "epoch": 351.0126582278481, "grad_norm": 0.08094476163387299, "learning_rate": 1.5585297415482202e-06, "loss": 0.0058, "step": 27730 }, { "epoch": 351.1392405063291, "grad_norm": 0.13122184574604034, "learning_rate": 1.5449054355135717e-06, "loss": 0.0049, "step": 27740 }, { "epoch": 351.26582278481015, "grad_norm": 0.12597815692424774, "learning_rate": 1.5313400069878415e-06, "loss": 0.006, "step": 27750 }, { "epoch": 351.39240506329116, "grad_norm": 0.09874460101127625, "learning_rate": 1.5178334724542887e-06, "loss": 0.0056, "step": 27760 }, { "epoch": 351.5189873417722, "grad_norm": 0.09363115578889847, "learning_rate": 1.504385848324641e-06, "loss": 0.0041, "step": 27770 }, { "epoch": 351.6455696202532, "grad_norm": 0.08699062466621399, "learning_rate": 1.490997150939033e-06, "loss": 0.0071, "step": 27780 }, { "epoch": 351.7721518987342, "grad_norm": 0.09194783866405487, "learning_rate": 1.4776673965659793e-06, "loss": 0.0052, "step": 27790 }, { "epoch": 351.8987341772152, "grad_norm": 0.11287857592105865, "learning_rate": 1.4643966014023957e-06, "loss": 0.0071, "step": 27800 }, { "epoch": 352.0253164556962, "grad_norm": 0.10496851801872253, "learning_rate": 1.4511847815735503e-06, "loss": 0.0044, "step": 27810 }, { "epoch": 352.1518987341772, "grad_norm": 0.0842103436589241, "learning_rate": 1.438031953133051e-06, "loss": 0.0065, "step": 27820 }, { "epoch": 352.27848101265823, "grad_norm": 0.0679529532790184, "learning_rate": 1.4249381320628197e-06, "loss": 0.0058, "step": 27830 }, { "epoch": 352.40506329113924, "grad_norm": 0.10208858549594879, "learning_rate": 1.4119033342730902e-06, "loss": 0.0035, "step": 27840 }, { "epoch": 352.53164556962025, "grad_norm": 0.08176101744174957, "learning_rate": 1.3989275756023656e-06, "loss": 0.004, "step": 27850 }, { "epoch": 352.65822784810126, "grad_norm": 0.0860704630613327, "learning_rate": 1.386010871817417e-06, "loss": 0.0047, "step": 27860 }, { "epoch": 352.7848101265823, "grad_norm": 0.10394398868083954, "learning_rate": 1.3731532386132616e-06, "loss": 0.0049, "step": 27870 }, { "epoch": 352.9113924050633, "grad_norm": 0.10694562643766403, "learning_rate": 1.360354691613136e-06, "loss": 0.0075, "step": 27880 }, { "epoch": 353.0379746835443, "grad_norm": 0.11968289315700531, "learning_rate": 1.3476152463684776e-06, "loss": 0.0064, "step": 27890 }, { "epoch": 353.1645569620253, "grad_norm": 0.12687140703201294, "learning_rate": 1.3349349183589155e-06, "loss": 0.007, "step": 27900 }, { "epoch": 353.2911392405063, "grad_norm": 0.11207157373428345, "learning_rate": 1.3223137229922356e-06, "loss": 0.007, "step": 27910 }, { "epoch": 353.4177215189873, "grad_norm": 0.10173141956329346, "learning_rate": 1.3097516756043981e-06, "loss": 0.0078, "step": 27920 }, { "epoch": 353.54430379746833, "grad_norm": 0.1118098795413971, "learning_rate": 1.29724879145946e-06, "loss": 0.0043, "step": 27930 }, { "epoch": 353.67088607594934, "grad_norm": 0.12065169215202332, "learning_rate": 1.284805085749613e-06, "loss": 0.0055, "step": 27940 }, { "epoch": 353.7974683544304, "grad_norm": 0.10536765307188034, "learning_rate": 1.2724205735951288e-06, "loss": 0.0044, "step": 27950 }, { "epoch": 353.9240506329114, "grad_norm": 0.11595168709754944, "learning_rate": 1.2600952700443591e-06, "loss": 0.0066, "step": 27960 }, { "epoch": 354.0506329113924, "grad_norm": 0.0998588502407074, "learning_rate": 1.247829190073707e-06, "loss": 0.0037, "step": 27970 }, { "epoch": 354.17721518987344, "grad_norm": 0.09125517308712006, "learning_rate": 1.2356223485876173e-06, "loss": 0.0053, "step": 27980 }, { "epoch": 354.30379746835445, "grad_norm": 0.0873604491353035, "learning_rate": 1.2234747604185526e-06, "loss": 0.008, "step": 27990 }, { "epoch": 354.43037974683546, "grad_norm": 0.10950903594493866, "learning_rate": 1.2113864403269836e-06, "loss": 0.0067, "step": 28000 }, { "epoch": 354.55696202531647, "grad_norm": 0.0881468653678894, "learning_rate": 1.1993574030013555e-06, "loss": 0.0064, "step": 28010 }, { "epoch": 354.6835443037975, "grad_norm": 0.09191849082708359, "learning_rate": 1.1873876630580816e-06, "loss": 0.0065, "step": 28020 }, { "epoch": 354.8101265822785, "grad_norm": 0.10559108853340149, "learning_rate": 1.1754772350415278e-06, "loss": 0.0041, "step": 28030 }, { "epoch": 354.9367088607595, "grad_norm": 0.09591516852378845, "learning_rate": 1.163626133423984e-06, "loss": 0.0072, "step": 28040 }, { "epoch": 355.0632911392405, "grad_norm": 0.08848045021295547, "learning_rate": 1.1518343726056591e-06, "loss": 0.0053, "step": 28050 }, { "epoch": 355.1898734177215, "grad_norm": 0.09290799498558044, "learning_rate": 1.1401019669146474e-06, "loss": 0.0068, "step": 28060 }, { "epoch": 355.3164556962025, "grad_norm": 0.07723705470561981, "learning_rate": 1.128428930606934e-06, "loss": 0.0047, "step": 28070 }, { "epoch": 355.44303797468353, "grad_norm": 0.07205020636320114, "learning_rate": 1.1168152778663621e-06, "loss": 0.0066, "step": 28080 }, { "epoch": 355.56962025316454, "grad_norm": 0.1556936800479889, "learning_rate": 1.10526102280461e-06, "loss": 0.0068, "step": 28090 }, { "epoch": 355.69620253164555, "grad_norm": 0.08063528686761856, "learning_rate": 1.0937661794611864e-06, "loss": 0.0043, "step": 28100 }, { "epoch": 355.82278481012656, "grad_norm": 0.09151024371385574, "learning_rate": 1.082330761803413e-06, "loss": 0.0074, "step": 28110 }, { "epoch": 355.9493670886076, "grad_norm": 0.0817791074514389, "learning_rate": 1.0709547837263966e-06, "loss": 0.0056, "step": 28120 }, { "epoch": 356.0759493670886, "grad_norm": 0.10495943576097488, "learning_rate": 1.059638259053025e-06, "loss": 0.0065, "step": 28130 }, { "epoch": 356.2025316455696, "grad_norm": 0.09613505750894547, "learning_rate": 1.0483812015339434e-06, "loss": 0.007, "step": 28140 }, { "epoch": 356.32911392405066, "grad_norm": 0.11180130392313004, "learning_rate": 1.037183624847543e-06, "loss": 0.0065, "step": 28150 }, { "epoch": 356.45569620253167, "grad_norm": 0.08608085662126541, "learning_rate": 1.0260455425999238e-06, "loss": 0.0051, "step": 28160 }, { "epoch": 356.5822784810127, "grad_norm": 0.07674264162778854, "learning_rate": 1.0149669683249208e-06, "loss": 0.0053, "step": 28170 }, { "epoch": 356.7088607594937, "grad_norm": 0.0920955166220665, "learning_rate": 1.003947915484027e-06, "loss": 0.0065, "step": 28180 }, { "epoch": 356.8354430379747, "grad_norm": 0.11676400899887085, "learning_rate": 9.92988397466449e-07, "loss": 0.0055, "step": 28190 }, { "epoch": 356.9620253164557, "grad_norm": 0.07894869893789291, "learning_rate": 9.820884275890286e-07, "loss": 0.0034, "step": 28200 }, { "epoch": 357.0886075949367, "grad_norm": 0.0711468905210495, "learning_rate": 9.712480190962546e-07, "loss": 0.0045, "step": 28210 }, { "epoch": 357.2151898734177, "grad_norm": 0.09721994400024414, "learning_rate": 9.604671851602464e-07, "loss": 0.0043, "step": 28220 }, { "epoch": 357.34177215189874, "grad_norm": 0.08209556341171265, "learning_rate": 9.497459388807306e-07, "loss": 0.0042, "step": 28230 }, { "epoch": 357.46835443037975, "grad_norm": 0.0681898295879364, "learning_rate": 9.390842932850364e-07, "loss": 0.006, "step": 28240 }, { "epoch": 357.59493670886076, "grad_norm": 0.0852547138929367, "learning_rate": 9.28482261328073e-07, "loss": 0.0079, "step": 28250 }, { "epoch": 357.72151898734177, "grad_norm": 0.1025208830833435, "learning_rate": 9.179398558923025e-07, "loss": 0.0078, "step": 28260 }, { "epoch": 357.8481012658228, "grad_norm": 0.0894632413983345, "learning_rate": 9.074570897877388e-07, "loss": 0.0062, "step": 28270 }, { "epoch": 357.9746835443038, "grad_norm": 0.0708136186003685, "learning_rate": 8.970339757519375e-07, "loss": 0.0042, "step": 28280 }, { "epoch": 358.1012658227848, "grad_norm": 0.08377724140882492, "learning_rate": 8.866705264499619e-07, "loss": 0.0052, "step": 28290 }, { "epoch": 358.2278481012658, "grad_norm": 0.058781322091817856, "learning_rate": 8.763667544743836e-07, "loss": 0.0041, "step": 28300 }, { "epoch": 358.3544303797468, "grad_norm": 0.10353133082389832, "learning_rate": 8.661226723452543e-07, "loss": 0.0063, "step": 28310 }, { "epoch": 358.4810126582278, "grad_norm": 0.08548717945814133, "learning_rate": 8.559382925101001e-07, "loss": 0.0046, "step": 28320 }, { "epoch": 358.60759493670884, "grad_norm": 0.07339710742235184, "learning_rate": 8.458136273438943e-07, "loss": 0.0041, "step": 28330 }, { "epoch": 358.73417721518985, "grad_norm": 0.07439198344945908, "learning_rate": 8.357486891490795e-07, "loss": 0.0063, "step": 28340 }, { "epoch": 358.8607594936709, "grad_norm": 0.08818480372428894, "learning_rate": 8.257434901554895e-07, "loss": 0.0045, "step": 28350 }, { "epoch": 358.9873417721519, "grad_norm": 0.11517839133739471, "learning_rate": 8.157980425203937e-07, "loss": 0.0057, "step": 28360 }, { "epoch": 359.11392405063293, "grad_norm": 0.07920888066291809, "learning_rate": 8.059123583284367e-07, "loss": 0.0043, "step": 28370 }, { "epoch": 359.24050632911394, "grad_norm": 0.11070065200328827, "learning_rate": 7.960864495916654e-07, "loss": 0.0055, "step": 28380 }, { "epoch": 359.36708860759495, "grad_norm": 0.07977204769849777, "learning_rate": 7.863203282494847e-07, "loss": 0.0066, "step": 28390 }, { "epoch": 359.49367088607596, "grad_norm": 0.099200040102005, "learning_rate": 7.766140061686522e-07, "loss": 0.0041, "step": 28400 }, { "epoch": 359.62025316455697, "grad_norm": 0.08919531852006912, "learning_rate": 7.669674951432615e-07, "loss": 0.0044, "step": 28410 }, { "epoch": 359.746835443038, "grad_norm": 0.19150136411190033, "learning_rate": 7.573808068947363e-07, "loss": 0.0042, "step": 28420 }, { "epoch": 359.873417721519, "grad_norm": 0.07611482590436935, "learning_rate": 7.478539530718087e-07, "loss": 0.0046, "step": 28430 }, { "epoch": 360.0, "grad_norm": 0.08659658581018448, "learning_rate": 7.383869452504965e-07, "loss": 0.0042, "step": 28440 }, { "epoch": 360.126582278481, "grad_norm": 0.0720798671245575, "learning_rate": 7.289797949341204e-07, "loss": 0.0047, "step": 28450 }, { "epoch": 360.253164556962, "grad_norm": 0.08066416531801224, "learning_rate": 7.196325135532423e-07, "loss": 0.0039, "step": 28460 }, { "epoch": 360.37974683544303, "grad_norm": 0.07771432399749756, "learning_rate": 7.10345112465699e-07, "loss": 0.0063, "step": 28470 }, { "epoch": 360.50632911392404, "grad_norm": 0.0962880402803421, "learning_rate": 7.011176029565525e-07, "loss": 0.0052, "step": 28480 }, { "epoch": 360.63291139240505, "grad_norm": 0.08181007206439972, "learning_rate": 6.919499962381004e-07, "loss": 0.0041, "step": 28490 }, { "epoch": 360.75949367088606, "grad_norm": 0.12477687746286392, "learning_rate": 6.828423034498488e-07, "loss": 0.0074, "step": 28500 }, { "epoch": 360.88607594936707, "grad_norm": 0.07802589982748032, "learning_rate": 6.737945356585007e-07, "loss": 0.0046, "step": 28510 }, { "epoch": 361.0126582278481, "grad_norm": 0.06652253121137619, "learning_rate": 6.648067038579508e-07, "loss": 0.0044, "step": 28520 }, { "epoch": 361.1392405063291, "grad_norm": 0.07158809155225754, "learning_rate": 6.558788189692578e-07, "loss": 0.0043, "step": 28530 }, { "epoch": 361.26582278481015, "grad_norm": 0.07322593033313751, "learning_rate": 6.470108918406492e-07, "loss": 0.0057, "step": 28540 }, { "epoch": 361.39240506329116, "grad_norm": 0.0717267245054245, "learning_rate": 6.382029332474893e-07, "loss": 0.004, "step": 28550 }, { "epoch": 361.5189873417722, "grad_norm": 0.09628582000732422, "learning_rate": 6.294549538922778e-07, "loss": 0.0054, "step": 28560 }, { "epoch": 361.6455696202532, "grad_norm": 0.07674375921487808, "learning_rate": 6.207669644046344e-07, "loss": 0.007, "step": 28570 }, { "epoch": 361.7721518987342, "grad_norm": 0.0808877944946289, "learning_rate": 6.121389753412865e-07, "loss": 0.0077, "step": 28580 }, { "epoch": 361.8987341772152, "grad_norm": 0.07622285187244415, "learning_rate": 6.035709971860592e-07, "loss": 0.0051, "step": 28590 }, { "epoch": 362.0253164556962, "grad_norm": 0.07772103697061539, "learning_rate": 5.950630403498469e-07, "loss": 0.0063, "step": 28600 }, { "epoch": 362.1518987341772, "grad_norm": 0.12592467665672302, "learning_rate": 5.866151151706189e-07, "loss": 0.0042, "step": 28610 }, { "epoch": 362.27848101265823, "grad_norm": 0.07711079716682434, "learning_rate": 5.782272319134086e-07, "loss": 0.0056, "step": 28620 }, { "epoch": 362.40506329113924, "grad_norm": 0.07687326520681381, "learning_rate": 5.698994007702796e-07, "loss": 0.0068, "step": 28630 }, { "epoch": 362.53164556962025, "grad_norm": 0.08277809619903564, "learning_rate": 5.616316318603321e-07, "loss": 0.0049, "step": 28640 }, { "epoch": 362.65822784810126, "grad_norm": 0.0809410959482193, "learning_rate": 5.534239352296799e-07, "loss": 0.0045, "step": 28650 }, { "epoch": 362.7848101265823, "grad_norm": 0.0688992440700531, "learning_rate": 5.452763208514621e-07, "loss": 0.0048, "step": 28660 }, { "epoch": 362.9113924050633, "grad_norm": 0.08862662315368652, "learning_rate": 5.371887986257873e-07, "loss": 0.007, "step": 28670 }, { "epoch": 363.0379746835443, "grad_norm": 0.0825532078742981, "learning_rate": 5.291613783797611e-07, "loss": 0.0042, "step": 28680 }, { "epoch": 363.1645569620253, "grad_norm": 0.07850083708763123, "learning_rate": 5.211940698674534e-07, "loss": 0.0062, "step": 28690 }, { "epoch": 363.2911392405063, "grad_norm": 0.06562197953462601, "learning_rate": 5.132868827698978e-07, "loss": 0.0046, "step": 28700 }, { "epoch": 363.4177215189873, "grad_norm": 0.07280188053846359, "learning_rate": 5.054398266950755e-07, "loss": 0.0048, "step": 28710 }, { "epoch": 363.54430379746833, "grad_norm": 0.07240143418312073, "learning_rate": 4.976529111778872e-07, "loss": 0.0046, "step": 28720 }, { "epoch": 363.67088607594934, "grad_norm": 0.05565370246767998, "learning_rate": 4.899261456801862e-07, "loss": 0.0042, "step": 28730 }, { "epoch": 363.7974683544304, "grad_norm": 0.06385821849107742, "learning_rate": 4.822595395907126e-07, "loss": 0.0062, "step": 28740 }, { "epoch": 363.9240506329114, "grad_norm": 0.13192683458328247, "learning_rate": 4.7465310222510886e-07, "loss": 0.0066, "step": 28750 }, { "epoch": 364.0506329113924, "grad_norm": 0.0976744145154953, "learning_rate": 4.6710684282593175e-07, "loss": 0.0072, "step": 28760 }, { "epoch": 364.17721518987344, "grad_norm": 0.07363884150981903, "learning_rate": 4.5962077056257993e-07, "loss": 0.0054, "step": 28770 }, { "epoch": 364.30379746835445, "grad_norm": 0.07456176728010178, "learning_rate": 4.521948945313492e-07, "loss": 0.006, "step": 28780 }, { "epoch": 364.43037974683546, "grad_norm": 0.07240016013383865, "learning_rate": 4.4482922375537195e-07, "loss": 0.0048, "step": 28790 }, { "epoch": 364.55696202531647, "grad_norm": 0.05740342289209366, "learning_rate": 4.375237671846333e-07, "loss": 0.0048, "step": 28800 }, { "epoch": 364.6835443037975, "grad_norm": 0.0678882896900177, "learning_rate": 4.302785336959547e-07, "loss": 0.0067, "step": 28810 }, { "epoch": 364.8101265822785, "grad_norm": 0.06608717143535614, "learning_rate": 4.2309353209297744e-07, "loss": 0.0065, "step": 28820 }, { "epoch": 364.9367088607595, "grad_norm": 0.07652057707309723, "learning_rate": 4.159687711061566e-07, "loss": 0.0048, "step": 28830 }, { "epoch": 365.0632911392405, "grad_norm": 0.10308805108070374, "learning_rate": 4.0890425939275055e-07, "loss": 0.0051, "step": 28840 }, { "epoch": 365.1898734177215, "grad_norm": 0.08024704456329346, "learning_rate": 4.0190000553679827e-07, "loss": 0.0059, "step": 28850 }, { "epoch": 365.3164556962025, "grad_norm": 0.074092797935009, "learning_rate": 3.9495601804913627e-07, "loss": 0.0062, "step": 28860 }, { "epoch": 365.44303797468353, "grad_norm": 0.06277543306350708, "learning_rate": 3.880723053673652e-07, "loss": 0.0075, "step": 28870 }, { "epoch": 365.56962025316454, "grad_norm": 0.11607486009597778, "learning_rate": 3.812488758558386e-07, "loss": 0.0068, "step": 28880 }, { "epoch": 365.69620253164555, "grad_norm": 0.06647367030382156, "learning_rate": 3.744857378056743e-07, "loss": 0.0039, "step": 28890 }, { "epoch": 365.82278481012656, "grad_norm": 0.08613641560077667, "learning_rate": 3.677828994347154e-07, "loss": 0.006, "step": 28900 }, { "epoch": 365.9493670886076, "grad_norm": 0.08467362076044083, "learning_rate": 3.61140368887547e-07, "loss": 0.0072, "step": 28910 }, { "epoch": 366.0759493670886, "grad_norm": 0.06083833426237106, "learning_rate": 3.545581542354681e-07, "loss": 0.0047, "step": 28920 }, { "epoch": 366.2025316455696, "grad_norm": 0.08816011995077133, "learning_rate": 3.480362634764922e-07, "loss": 0.0057, "step": 28930 }, { "epoch": 366.32911392405066, "grad_norm": 0.08151315152645111, "learning_rate": 3.4157470453533014e-07, "loss": 0.0052, "step": 28940 }, { "epoch": 366.45569620253167, "grad_norm": 0.07402793318033218, "learning_rate": 3.3517348526339034e-07, "loss": 0.0066, "step": 28950 }, { "epoch": 366.5822784810127, "grad_norm": 0.06824968010187149, "learning_rate": 3.288326134387454e-07, "loss": 0.007, "step": 28960 }, { "epoch": 366.7088607594937, "grad_norm": 0.09400804340839386, "learning_rate": 3.225520967661655e-07, "loss": 0.0055, "step": 28970 }, { "epoch": 366.8354430379747, "grad_norm": 0.08546137064695358, "learning_rate": 3.163319428770628e-07, "loss": 0.0046, "step": 28980 }, { "epoch": 366.9620253164557, "grad_norm": 0.09480928629636765, "learning_rate": 3.1017215932951374e-07, "loss": 0.0059, "step": 28990 }, { "epoch": 367.0886075949367, "grad_norm": 0.0519038662314415, "learning_rate": 3.040727536082366e-07, "loss": 0.0046, "step": 29000 }, { "epoch": 367.2151898734177, "grad_norm": 0.08063514530658722, "learning_rate": 2.980337331245864e-07, "loss": 0.0061, "step": 29010 }, { "epoch": 367.34177215189874, "grad_norm": 0.059054434299468994, "learning_rate": 2.9205510521653213e-07, "loss": 0.0054, "step": 29020 }, { "epoch": 367.46835443037975, "grad_norm": 0.0810805931687355, "learning_rate": 2.86136877148685e-07, "loss": 0.0047, "step": 29030 }, { "epoch": 367.59493670886076, "grad_norm": 0.08726964145898819, "learning_rate": 2.8027905611223704e-07, "loss": 0.0061, "step": 29040 }, { "epoch": 367.72151898734177, "grad_norm": 0.07720756530761719, "learning_rate": 2.7448164922500573e-07, "loss": 0.0068, "step": 29050 }, { "epoch": 367.8481012658228, "grad_norm": 0.08848274499177933, "learning_rate": 2.687446635313784e-07, "loss": 0.0064, "step": 29060 }, { "epoch": 367.9746835443038, "grad_norm": 0.10057349503040314, "learning_rate": 2.630681060023343e-07, "loss": 0.0057, "step": 29070 }, { "epoch": 368.1012658227848, "grad_norm": 0.07392734289169312, "learning_rate": 2.5745198353542833e-07, "loss": 0.006, "step": 29080 }, { "epoch": 368.2278481012658, "grad_norm": 0.07666134089231491, "learning_rate": 2.518963029547794e-07, "loss": 0.0045, "step": 29090 }, { "epoch": 368.3544303797468, "grad_norm": 0.08253086358308792, "learning_rate": 2.464010710110598e-07, "loss": 0.0055, "step": 29100 }, { "epoch": 368.4810126582278, "grad_norm": 0.08935759961605072, "learning_rate": 2.4096629438150054e-07, "loss": 0.0052, "step": 29110 }, { "epoch": 368.60759493670884, "grad_norm": 0.0982205867767334, "learning_rate": 2.3559197966985802e-07, "loss": 0.006, "step": 29120 }, { "epoch": 368.73417721518985, "grad_norm": 0.05435718968510628, "learning_rate": 2.3027813340644188e-07, "loss": 0.0059, "step": 29130 }, { "epoch": 368.8607594936709, "grad_norm": 0.0643111988902092, "learning_rate": 2.2502476204807055e-07, "loss": 0.0065, "step": 29140 }, { "epoch": 368.9873417721519, "grad_norm": 0.09041298925876617, "learning_rate": 2.1983187197808786e-07, "loss": 0.0068, "step": 29150 }, { "epoch": 369.11392405063293, "grad_norm": 0.09401947259902954, "learning_rate": 2.1469946950634644e-07, "loss": 0.0061, "step": 29160 }, { "epoch": 369.24050632911394, "grad_norm": 0.08318662643432617, "learning_rate": 2.096275608691911e-07, "loss": 0.0058, "step": 29170 }, { "epoch": 369.36708860759495, "grad_norm": 0.06575780361890793, "learning_rate": 2.046161522294754e-07, "loss": 0.0063, "step": 29180 }, { "epoch": 369.49367088607596, "grad_norm": 0.0840134397149086, "learning_rate": 1.9966524967653944e-07, "loss": 0.0046, "step": 29190 }, { "epoch": 369.62025316455697, "grad_norm": 0.07890450954437256, "learning_rate": 1.9477485922618222e-07, "loss": 0.0038, "step": 29200 }, { "epoch": 369.746835443038, "grad_norm": 0.07058586925268173, "learning_rate": 1.899449868207004e-07, "loss": 0.0074, "step": 29210 }, { "epoch": 369.873417721519, "grad_norm": 0.07745250314474106, "learning_rate": 1.851756383288439e-07, "loss": 0.0048, "step": 29220 }, { "epoch": 370.0, "grad_norm": 0.09552574157714844, "learning_rate": 1.8046681954581035e-07, "loss": 0.005, "step": 29230 }, { "epoch": 370.126582278481, "grad_norm": 0.05806674063205719, "learning_rate": 1.7581853619327294e-07, "loss": 0.0073, "step": 29240 }, { "epoch": 370.253164556962, "grad_norm": 0.07448924332857132, "learning_rate": 1.7123079391932472e-07, "loss": 0.0049, "step": 29250 }, { "epoch": 370.37974683544303, "grad_norm": 0.104945108294487, "learning_rate": 1.6670359829850657e-07, "loss": 0.0065, "step": 29260 }, { "epoch": 370.50632911392404, "grad_norm": 0.07071957737207413, "learning_rate": 1.6223695483179037e-07, "loss": 0.0059, "step": 29270 }, { "epoch": 370.63291139240505, "grad_norm": 0.09646380692720413, "learning_rate": 1.5783086894656795e-07, "loss": 0.0059, "step": 29280 }, { "epoch": 370.75949367088606, "grad_norm": 0.09460590779781342, "learning_rate": 1.5348534599665121e-07, "loss": 0.0063, "step": 29290 }, { "epoch": 370.88607594936707, "grad_norm": 0.07539906352758408, "learning_rate": 1.4920039126225527e-07, "loss": 0.0053, "step": 29300 }, { "epoch": 371.0126582278481, "grad_norm": 0.06359780579805374, "learning_rate": 1.449760099500097e-07, "loss": 0.0046, "step": 29310 }, { "epoch": 371.1392405063291, "grad_norm": 0.0941770002245903, "learning_rate": 1.4081220719293075e-07, "loss": 0.0071, "step": 29320 }, { "epoch": 371.26582278481015, "grad_norm": 0.0668116882443428, "learning_rate": 1.3670898805043797e-07, "loss": 0.0045, "step": 29330 }, { "epoch": 371.39240506329116, "grad_norm": 0.0869545117020607, "learning_rate": 1.326663575083209e-07, "loss": 0.0065, "step": 29340 }, { "epoch": 371.5189873417722, "grad_norm": 0.10116323828697205, "learning_rate": 1.2868432047876688e-07, "loss": 0.0066, "step": 29350 }, { "epoch": 371.6455696202532, "grad_norm": 0.06154251843690872, "learning_rate": 1.2476288180032213e-07, "loss": 0.0049, "step": 29360 }, { "epoch": 371.7721518987342, "grad_norm": 0.11325839906930923, "learning_rate": 1.209020462379029e-07, "loss": 0.0045, "step": 29370 }, { "epoch": 371.8987341772152, "grad_norm": 0.05853648483753204, "learning_rate": 1.1710181848278435e-07, "loss": 0.0043, "step": 29380 }, { "epoch": 372.0253164556962, "grad_norm": 0.09993967413902283, "learning_rate": 1.133622031526116e-07, "loss": 0.0053, "step": 29390 }, { "epoch": 372.1518987341772, "grad_norm": 0.07397576421499252, "learning_rate": 1.0968320479136095e-07, "loss": 0.0043, "step": 29400 }, { "epoch": 372.27848101265823, "grad_norm": 0.04662695899605751, "learning_rate": 1.0606482786936767e-07, "loss": 0.0056, "step": 29410 }, { "epoch": 372.40506329113924, "grad_norm": 0.08740434795618057, "learning_rate": 1.0250707678329808e-07, "loss": 0.0055, "step": 29420 }, { "epoch": 372.53164556962025, "grad_norm": 0.13144347071647644, "learning_rate": 9.900995585615525e-08, "loss": 0.0086, "step": 29430 }, { "epoch": 372.65822784810126, "grad_norm": 0.08211615681648254, "learning_rate": 9.55734693372734e-08, "loss": 0.0032, "step": 29440 }, { "epoch": 372.7848101265823, "grad_norm": 0.08405471593141556, "learning_rate": 9.219762140231236e-08, "loss": 0.0042, "step": 29450 }, { "epoch": 372.9113924050633, "grad_norm": 0.09529124200344086, "learning_rate": 8.888241615322978e-08, "loss": 0.0074, "step": 29460 }, { "epoch": 373.0379746835443, "grad_norm": 0.05982327088713646, "learning_rate": 8.562785761833114e-08, "loss": 0.0062, "step": 29470 }, { "epoch": 373.1645569620253, "grad_norm": 0.06740356236696243, "learning_rate": 8.243394975219753e-08, "loss": 0.0057, "step": 29480 }, { "epoch": 373.2911392405063, "grad_norm": 0.08174359053373337, "learning_rate": 7.930069643573568e-08, "loss": 0.0065, "step": 29490 }, { "epoch": 373.4177215189873, "grad_norm": 0.09358818829059601, "learning_rate": 7.622810147614456e-08, "loss": 0.0058, "step": 29500 }, { "epoch": 373.54430379746833, "grad_norm": 0.061015836894512177, "learning_rate": 7.321616860690995e-08, "loss": 0.0054, "step": 29510 }, { "epoch": 373.67088607594934, "grad_norm": 0.06957871466875076, "learning_rate": 7.026490148782095e-08, "loss": 0.0064, "step": 29520 }, { "epoch": 373.7974683544304, "grad_norm": 0.07339868694543839, "learning_rate": 6.737430370494236e-08, "loss": 0.0043, "step": 29530 }, { "epoch": 373.9240506329114, "grad_norm": 0.06945320218801498, "learning_rate": 6.454437877062569e-08, "loss": 0.0051, "step": 29540 }, { "epoch": 374.0506329113924, "grad_norm": 0.08459364622831345, "learning_rate": 6.177513012349256e-08, "loss": 0.0044, "step": 29550 }, { "epoch": 374.17721518987344, "grad_norm": 0.08550884574651718, "learning_rate": 5.9066561128445775e-08, "loss": 0.005, "step": 29560 }, { "epoch": 374.30379746835445, "grad_norm": 0.042560599744319916, "learning_rate": 5.6418675076641556e-08, "loss": 0.0046, "step": 29570 }, { "epoch": 374.43037974683546, "grad_norm": 0.06907189637422562, "learning_rate": 5.383147518552845e-08, "loss": 0.0083, "step": 29580 }, { "epoch": 374.55696202531647, "grad_norm": 0.05878376588225365, "learning_rate": 5.1304964598786196e-08, "loss": 0.0048, "step": 29590 }, { "epoch": 374.6835443037975, "grad_norm": 0.08457302302122116, "learning_rate": 4.883914638636467e-08, "loss": 0.0051, "step": 29600 }, { "epoch": 374.8101265822785, "grad_norm": 0.0804503932595253, "learning_rate": 4.643402354446713e-08, "loss": 0.0045, "step": 29610 }, { "epoch": 374.9367088607595, "grad_norm": 0.07814652472734451, "learning_rate": 4.4089598995544766e-08, "loss": 0.0053, "step": 29620 }, { "epoch": 375.0632911392405, "grad_norm": 0.06933733075857162, "learning_rate": 4.180587558829663e-08, "loss": 0.0073, "step": 29630 }, { "epoch": 375.1898734177215, "grad_norm": 0.09807969629764557, "learning_rate": 3.958285609765855e-08, "loss": 0.0063, "step": 29640 }, { "epoch": 375.3164556962025, "grad_norm": 0.08856277167797089, "learning_rate": 3.74205432248087e-08, "loss": 0.0045, "step": 29650 }, { "epoch": 375.44303797468353, "grad_norm": 0.0686030313372612, "learning_rate": 3.531893959716204e-08, "loss": 0.0042, "step": 29660 }, { "epoch": 375.56962025316454, "grad_norm": 0.063630610704422, "learning_rate": 3.327804776837029e-08, "loss": 0.004, "step": 29670 }, { "epoch": 375.69620253164555, "grad_norm": 0.06825961172580719, "learning_rate": 3.129787021829977e-08, "loss": 0.0046, "step": 29680 }, { "epoch": 375.82278481012656, "grad_norm": 0.06185897812247276, "learning_rate": 2.9378409353059133e-08, "loss": 0.0054, "step": 29690 }, { "epoch": 375.9493670886076, "grad_norm": 0.09014482796192169, "learning_rate": 2.7519667504971593e-08, "loss": 0.0061, "step": 29700 }, { "epoch": 376.0759493670886, "grad_norm": 0.08747068047523499, "learning_rate": 2.572164693258605e-08, "loss": 0.0057, "step": 29710 }, { "epoch": 376.2025316455696, "grad_norm": 0.05177748203277588, "learning_rate": 2.3984349820665996e-08, "loss": 0.0047, "step": 29720 }, { "epoch": 376.32911392405066, "grad_norm": 0.06003490090370178, "learning_rate": 2.2307778280189485e-08, "loss": 0.0044, "step": 29730 }, { "epoch": 376.45569620253167, "grad_norm": 0.08241970837116241, "learning_rate": 2.0691934348354704e-08, "loss": 0.0041, "step": 29740 }, { "epoch": 376.5822784810127, "grad_norm": 0.09010104089975357, "learning_rate": 1.9136819988557763e-08, "loss": 0.0058, "step": 29750 }, { "epoch": 376.7088607594937, "grad_norm": 0.08017411082983017, "learning_rate": 1.7642437090414908e-08, "loss": 0.0051, "step": 29760 }, { "epoch": 376.8354430379747, "grad_norm": 0.08756192773580551, "learning_rate": 1.6208787469734755e-08, "loss": 0.0053, "step": 29770 }, { "epoch": 376.9620253164557, "grad_norm": 0.08241076022386551, "learning_rate": 1.483587286854604e-08, "loss": 0.0067, "step": 29780 }, { "epoch": 377.0886075949367, "grad_norm": 0.07215216010808945, "learning_rate": 1.3523694955064336e-08, "loss": 0.0044, "step": 29790 }, { "epoch": 377.2151898734177, "grad_norm": 0.08288659155368805, "learning_rate": 1.2272255323708681e-08, "loss": 0.0055, "step": 29800 }, { "epoch": 377.34177215189874, "grad_norm": 0.10132910311222076, "learning_rate": 1.1081555495096042e-08, "loss": 0.007, "step": 29810 }, { "epoch": 377.46835443037975, "grad_norm": 0.11649405211210251, "learning_rate": 9.951596916041306e-09, "loss": 0.0044, "step": 29820 }, { "epoch": 377.59493670886076, "grad_norm": 0.06557465344667435, "learning_rate": 8.882380959551739e-09, "loss": 0.0046, "step": 29830 }, { "epoch": 377.72151898734177, "grad_norm": 0.04596668481826782, "learning_rate": 7.873908924821428e-09, "loss": 0.0058, "step": 29840 }, { "epoch": 377.8481012658228, "grad_norm": 0.06544525921344757, "learning_rate": 6.926182037242379e-09, "loss": 0.0062, "step": 29850 }, { "epoch": 377.9746835443038, "grad_norm": 0.09805713593959808, "learning_rate": 6.039201448387876e-09, "loss": 0.0064, "step": 29860 }, { "epoch": 378.1012658227848, "grad_norm": 0.04321904107928276, "learning_rate": 5.212968236029125e-09, "loss": 0.0054, "step": 29870 }, { "epoch": 378.2278481012658, "grad_norm": 0.10912930965423584, "learning_rate": 4.447483404118602e-09, "loss": 0.0061, "step": 29880 }, { "epoch": 378.3544303797468, "grad_norm": 0.06930802762508392, "learning_rate": 3.742747882784503e-09, "loss": 0.0073, "step": 29890 }, { "epoch": 378.4810126582278, "grad_norm": 0.0702086091041565, "learning_rate": 3.0987625283473987e-09, "loss": 0.0043, "step": 29900 }, { "epoch": 378.60759493670884, "grad_norm": 0.05954327434301376, "learning_rate": 2.515528123320232e-09, "loss": 0.0045, "step": 29910 }, { "epoch": 378.73417721518985, "grad_norm": 0.05428392440080643, "learning_rate": 1.9930453763750133e-09, "loss": 0.0068, "step": 29920 }, { "epoch": 378.8607594936709, "grad_norm": 0.08164291083812714, "learning_rate": 1.5313149223872281e-09, "loss": 0.0069, "step": 29930 }, { "epoch": 378.9873417721519, "grad_norm": 0.054868537932634354, "learning_rate": 1.1303373224025305e-09, "loss": 0.0056, "step": 29940 }, { "epoch": 379.11392405063293, "grad_norm": 0.08357033878564835, "learning_rate": 7.901130636367438e-10, "loss": 0.0069, "step": 29950 }, { "epoch": 379.24050632911394, "grad_norm": 0.08218730241060257, "learning_rate": 5.106425595036158e-10, "loss": 0.0038, "step": 29960 }, { "epoch": 379.36708860759495, "grad_norm": 0.07462327927350998, "learning_rate": 2.9192614958706286e-10, "loss": 0.0055, "step": 29970 }, { "epoch": 379.49367088607596, "grad_norm": 0.09697887301445007, "learning_rate": 1.3396409964117064e-10, "loss": 0.0048, "step": 29980 }, { "epoch": 379.62025316455697, "grad_norm": 0.057305701076984406, "learning_rate": 3.6756601606846574e-11, "loss": 0.004, "step": 29990 }, { "epoch": 379.746835443038, "grad_norm": 0.05732962489128113, "learning_rate": 3.0377360626943076e-13, "loss": 0.004, "step": 30000 }, { "epoch": 379.746835443038, "step": 30000, "total_flos": 0.0, "train_loss": 0.040401646179457504, "train_runtime": 27061.84, "train_samples_per_second": 141.897, "train_steps_per_second": 1.109 } ], "logging_steps": 10, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 380, "save_steps": 10000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 64, "trial_name": null, "trial_params": null }